scripts/kallsyms.c

   1 /* Generate assembler source containing symbol information
   2  *
   3  * Copyright 2002       by Kai Germaschewski
   4  *
   5  * This software may be used and distributed according to the terms
   6  * of the GNU General Public License, incorporated herein by reference.
   7  *
   8  * Usage: kallsyms [--all-symbols] [--absolute-percpu]
   9  *                         [--base-relative] in.map > out.S
  10  *
  11  *      Table compression uses all the unused char codes on the symbols and
  12  *  maps these to the most used substrings (tokens). For instance, it might
  13  *  map char code 0xF7 to represent "write_" and then in every symbol where
  14  *  "write_" appears it can be replaced by 0xF7, saving 5 bytes.
  15  *      The used codes themselves are also placed in the table so that the
  16  *  decompresion can work without "special cases".
  17  *      Applied to kernel symbols, this usually produces a compression ratio
  18  *  of about 50%.
  19  *
  20  */
  21
  22 #include <getopt.h>
  23 #include <stdbool.h>
  24 #include <stdio.h>
  25 #include <stdlib.h>
  26 #include <string.h>
  27 #include <ctype.h>
  28 #include <limits.h>
  29
  30 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0]))
  31
  32 #define _stringify_1(x) #x
  33 #define _stringify(x)   _stringify_1(x)
  34
  35 #define KSYM_NAME_LEN           512
  36
  37 /*
  38  * A substantially bigger size than the current maximum.
  39  *
  40  * It cannot be defined as an expression because it gets stringified
  41  * for the fscanf() format string. Therefore, a _Static_assert() is
  42  * used instead to maintain the relationship with KSYM_NAME_LEN.
  43  */
  44 #define KSYM_NAME_LEN_BUFFER    2048
  45 _Static_assert(
  46         KSYM_NAME_LEN_BUFFER == KSYM_NAME_LEN * 4,
  47         "Please keep KSYM_NAME_LEN_BUFFER in sync with KSYM_NAME_LEN"
  48 );
  49
  50 struct sym_entry {
  51         unsigned long long addr;
  52         unsigned int len;
  53         unsigned int seq;
  54         unsigned int start_pos;
  55         unsigned int percpu_absolute;
  56         unsigned char sym[];
  57 };
  58
  59 struct addr_range {
  60         const char *start_sym, *end_sym;
  61         unsigned long long start, end;
  62 };
  63
  64 static unsigned long long _text;
  65 static unsigned long long relative_base;
  66 static struct addr_range text_ranges[] = {
  67         { "_stext",     "_etext"     },
  68         { "_sinittext", "_einittext" },
  69 };
  70 #define text_range_text     (&text_ranges[0])
  71 #define text_range_inittext (&text_ranges[1])
  72
  73 static struct addr_range percpu_range = {
  74         "__per_cpu_start", "__per_cpu_end", -1ULL, 0
  75 };
  76
  77 static struct sym_entry **table;
  78 static unsigned int table_size, table_cnt;
  79 static int all_symbols;
  80 static int absolute_percpu;
  81 static int base_relative;
  82 static int lto_clang;
  83
  84 static int token_profit[0x10000];
  85
  86 /* the table that holds the result of the compression */
  87 static unsigned char best_table[256][2];
  88 static unsigned char best_table_len[256];
  89
  90
  91 static void usage(void)
  92 {
  93         fprintf(stderr, "Usage: kallsyms [--all-symbols] [--absolute-percpu] "
  94                         "[--base-relative] [--lto-clang] in.map > out.S\n");
  95         exit(1);
  96 }
  97
  98 static char *sym_name(const struct sym_entry *s)
  99 {
 100         return (char *)s->sym + 1;
 101 }
 102
 103 static bool is_ignored_symbol(const char *name, char type)
 104 {
 105         /* Symbol names that exactly match to the following are ignored.*/
 106         static const char * const ignored_symbols[] = {
 107                 /*
 108                  * Symbols which vary between passes. Passes 1 and 2 must have
 109                  * identical symbol lists. The kallsyms_* symbols below are
 110                  * only added after pass 1, they would be included in pass 2
 111                  * when --all-symbols is specified so exclude them to get a
 112                  * stable symbol list.
 113                  */
 114                 "kallsyms_addresses",
 115                 "kallsyms_offsets",
 116                 "kallsyms_relative_base",
 117                 "kallsyms_num_syms",
 118                 "kallsyms_names",
 119                 "kallsyms_markers",
 120                 "kallsyms_token_table",
 121                 "kallsyms_token_index",
 122                 "kallsyms_seqs_of_names",
 123                 /* Exclude linker generated symbols which vary between passes */
 124                 "_SDA_BASE_",           /* ppc */
 125                 "_SDA2_BASE_",          /* ppc */
 126                 NULL
 127         };
 128
 129         /* Symbol names that begin with the following are ignored.*/
 130         static const char * const ignored_prefixes[] = {
 131                 "__efistub_",           /* arm64 EFI stub namespace */
 132                 "__kvm_nvhe_$",         /* arm64 local symbols in non-VHE KVM namespace */
 133                 "__kvm_nvhe_.L",        /* arm64 local symbols in non-VHE KVM namespace */
 134                 "__AArch64ADRPThunk_",  /* arm64 lld */
 135                 "__ARMV5PILongThunk_",  /* arm lld */
 136                 "__ARMV7PILongThunk_",
 137                 "__ThumbV7PILongThunk_",
 138                 "__LA25Thunk_",         /* mips lld */
 139                 "__microLA25Thunk_",
 140                 "__kcfi_typeid_",       /* CFI type identifiers */
 141                 NULL
 142         };
 143
 144         /* Symbol names that end with the following are ignored.*/
 145         static const char * const ignored_suffixes[] = {
 146                 "_from_arm",            /* arm */
 147                 "_from_thumb",          /* arm */
 148                 "_veneer",              /* arm */
 149                 NULL
 150         };
 151
 152         /* Symbol names that contain the following are ignored.*/
 153         static const char * const ignored_matches[] = {
 154                 ".long_branch.",        /* ppc stub */
 155                 ".plt_branch.",         /* ppc stub */
 156                 NULL
 157         };
 158
 159         const char * const *p;
 160
 161         for (p = ignored_symbols; *p; p++)
 162                 if (!strcmp(name, *p))
 163                         return true;
 164
 165         for (p = ignored_prefixes; *p; p++)
 166                 if (!strncmp(name, *p, strlen(*p)))
 167                         return true;
 168
 169         for (p = ignored_suffixes; *p; p++) {
 170                 int l = strlen(name) - strlen(*p);
 171
 172                 if (l >= 0 && !strcmp(name + l, *p))
 173                         return true;
 174         }
 175
 176         for (p = ignored_matches; *p; p++) {
 177                 if (strstr(name, *p))
 178                         return true;
 179         }
 180
 181         if (type == 'U' || type == 'u')
 182                 return true;
 183         /* exclude debugging symbols */
 184         if (type == 'N' || type == 'n')
 185                 return true;
 186
 187         if (toupper(type) == 'A') {
 188                 /* Keep these useful absolute symbols */
 189                 if (strcmp(name, "__kernel_syscall_via_break") &&
 190                     strcmp(name, "__kernel_syscall_via_epc") &&
 191                     strcmp(name, "__kernel_sigtramp") &&
 192                     strcmp(name, "__gp"))
 193                         return true;
 194         }
 195
 196         return false;
 197 }
 198
 199 static void check_symbol_range(const char *sym, unsigned long long addr,
 200                                struct addr_range *ranges, int entries)
 201 {
 202         size_t i;
 203         struct addr_range *ar;
 204
 205         for (i = 0; i < entries; ++i) {
 206                 ar = &ranges[i];
 207
 208                 if (strcmp(sym, ar->start_sym) == 0) {
 209                         ar->start = addr;
 210                         return;
 211                 } else if (strcmp(sym, ar->end_sym) == 0) {
 212                         ar->end = addr;
 213                         return;
 214                 }
 215         }
 216 }
 217
 218 static struct sym_entry *read_symbol(FILE *in)
 219 {
 220         char name[KSYM_NAME_LEN_BUFFER+1], type;
 221         unsigned long long addr;
 222         unsigned int len;
 223         struct sym_entry *sym;
 224         int rc;
 225
 226         rc = fscanf(in, "%llx %c %" _stringify(KSYM_NAME_LEN_BUFFER) "s\n", &addr, &type, name);
 227         if (rc != 3) {
 228                 if (rc != EOF && fgets(name, ARRAY_SIZE(name), in) == NULL)
 229                         fprintf(stderr, "Read error or end of file.\n");
 230                 return NULL;
 231         }
 232         if (strlen(name) >= KSYM_NAME_LEN) {
 233                 fprintf(stderr, "Symbol %s too long for kallsyms (%zu >= %d).\n"
 234                                 "Please increase KSYM_NAME_LEN both in kernel and kallsyms.c\n",
 235                         name, strlen(name), KSYM_NAME_LEN);
 236                 return NULL;
 237         }
 238
 239         if (strcmp(name, "_text") == 0)
 240                 _text = addr;
 241
 242         /* Ignore most absolute/undefined (?) symbols. */
 243         if (is_ignored_symbol(name, type))
 244                 return NULL;
 245
 246         check_symbol_range(name, addr, text_ranges, ARRAY_SIZE(text_ranges));
 247         check_symbol_range(name, addr, &percpu_range, 1);
 248
 249         /* include the type field in the symbol name, so that it gets
 250          * compressed together */
 251
 252         len = strlen(name) + 1;
 253
 254         sym = malloc(sizeof(*sym) + len + 1);
 255         if (!sym) {
 256                 fprintf(stderr, "kallsyms failure: "
 257                         "unable to allocate required amount of memory\n");
 258                 exit(EXIT_FAILURE);
 259         }
 260         sym->addr = addr;
 261         sym->len = len;
 262         sym->sym[0] = type;
 263         strcpy(sym_name(sym), name);
 264         sym->percpu_absolute = 0;
 265
 266         return sym;
 267 }
 268
 269 static int symbol_in_range(const struct sym_entry *s,
 270                            const struct addr_range *ranges, int entries)
 271 {
 272         size_t i;
 273         const struct addr_range *ar;
 274
 275         for (i = 0; i < entries; ++i) {
 276                 ar = &ranges[i];
 277
 278                 if (s->addr >= ar->start && s->addr <= ar->end)
 279                         return 1;
 280         }
 281
 282         return 0;
 283 }
 284
 285 static int symbol_valid(const struct sym_entry *s)
 286 {
 287         const char *name = sym_name(s);
 288
 289         /* if --all-symbols is not specified, then symbols outside the text
 290          * and inittext sections are discarded */
 291         if (!all_symbols) {
 292                 if (symbol_in_range(s, text_ranges,
 293                                     ARRAY_SIZE(text_ranges)) == 0)
 294                         return 0;
 295                 /* Corner case.  Discard any symbols with the same value as
 296                  * _etext _einittext; they can move between pass 1 and 2 when
 297                  * the kallsyms data are added.  If these symbols move then
 298                  * they may get dropped in pass 2, which breaks the kallsyms
 299                  * rules.
 300                  */
 301                 if ((s->addr == text_range_text->end &&
 302                      strcmp(name, text_range_text->end_sym)) ||
 303                     (s->addr == text_range_inittext->end &&
 304                      strcmp(name, text_range_inittext->end_sym)))
 305                         return 0;
 306         }
 307
 308         return 1;
 309 }
 310
 311 /* remove all the invalid symbols from the table */
 312 static void shrink_table(void)
 313 {
 314         unsigned int i, pos;
 315
 316         pos = 0;
 317         for (i = 0; i < table_cnt; i++) {
 318                 if (symbol_valid(table[i])) {
 319                         if (pos != i)
 320                                 table[pos] = table[i];
 321                         pos++;
 322                 } else {
 323                         free(table[i]);
 324                 }
 325         }
 326         table_cnt = pos;
 327
 328         /* When valid symbol is not registered, exit to error */
 329         if (!table_cnt) {
 330                 fprintf(stderr, "No valid symbol.\n");
 331                 exit(1);
 332         }
 333 }
 334
 335 static void read_map(const char *in)
 336 {
 337         FILE *fp;
 338         struct sym_entry *sym;
 339
 340         fp = fopen(in, "r");
 341         if (!fp) {
 342                 perror(in);
 343                 exit(1);
 344         }
 345
 346         while (!feof(fp)) {
 347                 sym = read_symbol(fp);
 348                 if (!sym)
 349                         continue;
 350
 351                 sym->start_pos = table_cnt;
 352
 353                 if (table_cnt >= table_size) {
 354                         table_size += 10000;
 355                         table = realloc(table, sizeof(*table) * table_size);
 356                         if (!table) {
 357                                 fprintf(stderr, "out of memory\n");
 358                                 fclose(fp);
 359                                 exit (1);
 360                         }
 361                 }
 362
 363                 table[table_cnt++] = sym;
 364         }
 365
 366         fclose(fp);
 367 }
 368
 369 static void output_label(const char *label)
 370 {
 371         printf(".globl %s\n", label);
 372         printf("\tALGN\n");
 373         printf("%s:\n", label);
 374 }
 375
 376 /* Provide proper symbols relocatability by their '_text' relativeness. */
 377 static void output_address(unsigned long long addr)
 378 {
 379         if (_text <= addr)
 380                 printf("\tPTR\t_text + %#llx\n", addr - _text);
 381         else
 382                 printf("\tPTR\t_text - %#llx\n", _text - addr);
 383 }
 384
 385 /* uncompress a compressed symbol. When this function is called, the best table
 386  * might still be compressed itself, so the function needs to be recursive */
 387 static int expand_symbol(const unsigned char *data, int len, char *result)
 388 {
 389         int c, rlen, total=0;
 390
 391         while (len) {
 392                 c = *data;
 393                 /* if the table holds a single char that is the same as the one
 394                  * we are looking for, then end the search */
 395                 if (best_table[c][0]==c && best_table_len[c]==1) {
 396                         *result++ = c;
 397                         total++;
 398                 } else {
 399                         /* if not, recurse and expand */
 400                         rlen = expand_symbol(best_table[c], best_table_len[c], result);
 401                         total += rlen;
 402                         result += rlen;
 403                 }
 404                 data++;
 405                 len--;
 406         }
 407         *result=0;
 408
 409         return total;
 410 }
 411
 412 static int symbol_absolute(const struct sym_entry *s)
 413 {
 414         return s->percpu_absolute;
 415 }
 416
 417 static char * s_name(char *buf)
 418 {
 419         /* Skip the symbol type */
 420         return buf + 1;
 421 }
 422
 423 static void cleanup_symbol_name(char *s)
 424 {
 425         char *p;
 426
 427         if (!lto_clang)
 428                 return;
 429
 430         /*
 431          * ASCII[.]   = 2e
 432          * ASCII[0-9] = 30,39
 433          * ASCII[A-Z] = 41,5a
 434          * ASCII[_]   = 5f
 435          * ASCII[a-z] = 61,7a
 436          *
 437          * As above, replacing the first '.' in ".llvm." with '\0' does not
 438          * affect the main sorting, but it helps us with subsorting.
 439          */
 440         p = strstr(s, ".llvm.");
 441         if (p)
 442                 *p = '\0';
 443 }
 444
 445 static int compare_names(const void *a, const void *b)
 446 {
 447         int ret;
 448         char sa_namebuf[KSYM_NAME_LEN];
 449         char sb_namebuf[KSYM_NAME_LEN];
 450         const struct sym_entry *sa = *(const struct sym_entry **)a;
 451         const struct sym_entry *sb = *(const struct sym_entry **)b;
 452
 453         expand_symbol(sa->sym, sa->len, sa_namebuf);
 454         expand_symbol(sb->sym, sb->len, sb_namebuf);
 455         cleanup_symbol_name(s_name(sa_namebuf));
 456         cleanup_symbol_name(s_name(sb_namebuf));
 457         ret = strcmp(s_name(sa_namebuf), s_name(sb_namebuf));
 458         if (!ret) {
 459                 if (sa->addr > sb->addr)
 460                         return 1;
 461                 else if (sa->addr < sb->addr)
 462                         return -1;
 463
 464                 /* keep old order */
 465                 return (int)(sa->seq - sb->seq);
 466         }
 467
 468         return ret;
 469 }
 470
 471 static void sort_symbols_by_name(void)
 472 {
 473         qsort(table, table_cnt, sizeof(table[0]), compare_names);
 474 }
 475
 476 static void write_src(void)
 477 {
 478         unsigned int i, k, off;
 479         unsigned int best_idx[256];
 480         unsigned int *markers;
 481         char buf[KSYM_NAME_LEN];
 482
 483         printf("#include <asm/bitsperlong.h>\n");
 484         printf("#if BITS_PER_LONG == 64\n");
 485         printf("#define PTR .quad\n");
 486         printf("#define ALGN .balign 8\n");
 487         printf("#else\n");
 488         printf("#define PTR .long\n");
 489         printf("#define ALGN .balign 4\n");
 490         printf("#endif\n");
 491
 492         printf("\t.section .rodata, \"a\"\n");
 493
 494         if (!base_relative)
 495                 output_label("kallsyms_addresses");
 496         else
 497                 output_label("kallsyms_offsets");
 498
 499         for (i = 0; i < table_cnt; i++) {
 500                 if (base_relative) {
 501                         /*
 502                          * Use the offset relative to the lowest value
 503                          * encountered of all relative symbols, and emit
 504                          * non-relocatable fixed offsets that will be fixed
 505                          * up at runtime.
 506                          */
 507
 508                         long long offset;
 509                         int overflow;
 510
 511                         if (!absolute_percpu) {
 512                                 offset = table[i]->addr - relative_base;
 513                                 overflow = (offset < 0 || offset > UINT_MAX);
 514                         } else if (symbol_absolute(table[i])) {
 515                                 offset = table[i]->addr;
 516                                 overflow = (offset < 0 || offset > INT_MAX);
 517                         } else {
 518                                 offset = relative_base - table[i]->addr - 1;
 519                                 overflow = (offset < INT_MIN || offset >= 0);
 520                         }
 521                         if (overflow) {
 522                                 fprintf(stderr, "kallsyms failure: "
 523                                         "%s symbol value %#llx out of range in relative mode\n",
 524                                         symbol_absolute(table[i]) ? "absolute" : "relative",
 525                                         table[i]->addr);
 526                                 exit(EXIT_FAILURE);
 527                         }
 528                         printf("\t.long\t%#x\n", (int)offset);
 529                 } else if (!symbol_absolute(table[i])) {
 530                         output_address(table[i]->addr);
 531                 } else {
 532                         printf("\tPTR\t%#llx\n", table[i]->addr);
 533                 }
 534         }
 535         printf("\n");
 536
 537         if (base_relative) {
 538                 output_label("kallsyms_relative_base");
 539                 output_address(relative_base);
 540                 printf("\n");
 541         }
 542
 543         output_label("kallsyms_num_syms");
 544         printf("\t.long\t%u\n", table_cnt);
 545         printf("\n");
 546
 547         /* table of offset markers, that give the offset in the compressed stream
 548          * every 256 symbols */
 549         markers = malloc(sizeof(unsigned int) * ((table_cnt + 255) / 256));
 550         if (!markers) {
 551                 fprintf(stderr, "kallsyms failure: "
 552                         "unable to allocate required memory\n");
 553                 exit(EXIT_FAILURE);
 554         }
 555
 556         output_label("kallsyms_names");
 557         off = 0;
 558         for (i = 0; i < table_cnt; i++) {
 559                 if ((i & 0xFF) == 0)
 560                         markers[i >> 8] = off;
 561                 table[i]->seq = i;
 562
 563                 /* There cannot be any symbol of length zero. */
 564                 if (table[i]->len == 0) {
 565                         fprintf(stderr, "kallsyms failure: "
 566                                 "unexpected zero symbol length\n");
 567                         exit(EXIT_FAILURE);
 568                 }
 569
 570                 /* Only lengths that fit in up-to-two-byte ULEB128 are supported. */
 571                 if (table[i]->len > 0x3FFF) {
 572                         fprintf(stderr, "kallsyms failure: "
 573                                 "unexpected huge symbol length\n");
 574                         exit(EXIT_FAILURE);
 575                 }
 576
 577                 /* Encode length with ULEB128. */
 578                 if (table[i]->len <= 0x7F) {
 579                         /* Most symbols use a single byte for the length. */
 580                         printf("\t.byte 0x%02x", table[i]->len);
 581                         off += table[i]->len + 1;
 582                 } else {
 583                         /* "Big" symbols use two bytes. */
 584                         printf("\t.byte 0x%02x, 0x%02x",
 585                                 (table[i]->len & 0x7F) | 0x80,
 586                                 (table[i]->len >> 7) & 0x7F);
 587                         off += table[i]->len + 2;
 588                 }
 589                 for (k = 0; k < table[i]->len; k++)
 590                         printf(", 0x%02x", table[i]->sym[k]);
 591                 printf("\n");
 592         }
 593         printf("\n");
 594
 595         output_label("kallsyms_markers");
 596         for (i = 0; i < ((table_cnt + 255) >> 8); i++)
 597                 printf("\t.long\t%u\n", markers[i]);
 598         printf("\n");
 599
 600         free(markers);
 601
 602         sort_symbols_by_name();
 603         output_label("kallsyms_seqs_of_names");
 604         for (i = 0; i < table_cnt; i++)
 605                 printf("\t.long\t%u\n", table[i]->seq);
 606         printf("\n");
 607
 608         output_label("kallsyms_token_table");
 609         off = 0;
 610         for (i = 0; i < 256; i++) {
 611                 best_idx[i] = off;
 612                 expand_symbol(best_table[i], best_table_len[i], buf);
 613                 printf("\t.asciz\t\"%s\"\n", buf);
 614                 off += strlen(buf) + 1;
 615         }
 616         printf("\n");
 617
 618         output_label("kallsyms_token_index");
 619         for (i = 0; i < 256; i++)
 620                 printf("\t.short\t%d\n", best_idx[i]);
 621         printf("\n");
 622 }
 623
 624
 625 /* table lookup compression functions */
 626
 627 /* count all the possible tokens in a symbol */
 628 static void learn_symbol(const unsigned char *symbol, int len)
 629 {
 630         int i;
 631
 632         for (i = 0; i < len - 1; i++)
 633                 token_profit[ symbol[i] + (symbol[i + 1] << 8) ]++;
 634 }
 635
 636 /* decrease the count for all the possible tokens in a symbol */
 637 static void forget_symbol(const unsigned char *symbol, int len)
 638 {
 639         int i;
 640
 641         for (i = 0; i < len - 1; i++)
 642                 token_profit[ symbol[i] + (symbol[i + 1] << 8) ]--;
 643 }
 644
 645 /* do the initial token count */
 646 static void build_initial_tok_table(void)
 647 {
 648         unsigned int i;
 649
 650         for (i = 0; i < table_cnt; i++)
 651                 learn_symbol(table[i]->sym, table[i]->len);
 652 }
 653
 654 static unsigned char *find_token(unsigned char *str, int len,
 655                                  const unsigned char *token)
 656 {
 657         int i;
 658
 659         for (i = 0; i < len - 1; i++) {
 660                 if (str[i] == token[0] && str[i+1] == token[1])
 661                         return &str[i];
 662         }
 663         return NULL;
 664 }
 665
 666 /* replace a given token in all the valid symbols. Use the sampled symbols
 667  * to update the counts */
 668 static void compress_symbols(const unsigned char *str, int idx)
 669 {
 670         unsigned int i, len, size;
 671         unsigned char *p1, *p2;
 672
 673         for (i = 0; i < table_cnt; i++) {
 674
 675                 len = table[i]->len;
 676                 p1 = table[i]->sym;
 677
 678                 /* find the token on the symbol */
 679                 p2 = find_token(p1, len, str);
 680                 if (!p2) continue;
 681
 682                 /* decrease the counts for this symbol's tokens */
 683                 forget_symbol(table[i]->sym, len);
 684
 685                 size = len;
 686
 687                 do {
 688                         *p2 = idx;
 689                         p2++;
 690                         size -= (p2 - p1);
 691                         memmove(p2, p2 + 1, size);
 692                         p1 = p2;
 693                         len--;
 694
 695                         if (size < 2) break;
 696
 697                         /* find the token on the symbol */
 698                         p2 = find_token(p1, size, str);
 699
 700                 } while (p2);
 701
 702                 table[i]->len = len;
 703
 704                 /* increase the counts for this symbol's new tokens */
 705                 learn_symbol(table[i]->sym, len);
 706         }
 707 }
 708
 709 /* search the token with the maximum profit */
 710 static int find_best_token(void)
 711 {
 712         int i, best, bestprofit;
 713
 714         bestprofit=-10000;
 715         best = 0;
 716
 717         for (i = 0; i < 0x10000; i++) {
 718                 if (token_profit[i] > bestprofit) {
 719                         best = i;
 720                         bestprofit = token_profit[i];
 721                 }
 722         }
 723         return best;
 724 }
 725
 726 /* this is the core of the algorithm: calculate the "best" table */
 727 static void optimize_result(void)
 728 {
 729         int i, best;
 730
 731         /* using the '\0' symbol last allows compress_symbols to use standard
 732          * fast string functions */
 733         for (i = 255; i >= 0; i--) {
 734
 735                 /* if this table slot is empty (it is not used by an actual
 736                  * original char code */
 737                 if (!best_table_len[i]) {
 738
 739                         /* find the token with the best profit value */
 740                         best = find_best_token();
 741                         if (token_profit[best] == 0)
 742                                 break;
 743
 744                         /* place it in the "best" table */
 745                         best_table_len[i] = 2;
 746                         best_table[i][0] = best & 0xFF;
 747                         best_table[i][1] = (best >> 8) & 0xFF;
 748
 749                         /* replace this token in all the valid symbols */
 750                         compress_symbols(best_table[i], i);
 751                 }
 752         }
 753 }
 754
 755 /* start by placing the symbols that are actually used on the table */
 756 static void insert_real_symbols_in_table(void)
 757 {
 758         unsigned int i, j, c;
 759
 760         for (i = 0; i < table_cnt; i++) {
 761                 for (j = 0; j < table[i]->len; j++) {
 762                         c = table[i]->sym[j];
 763                         best_table[c][0]=c;
 764                         best_table_len[c]=1;
 765                 }
 766         }
 767 }
 768
 769 static void optimize_token_table(void)
 770 {
 771         build_initial_tok_table();
 772
 773         insert_real_symbols_in_table();
 774
 775         optimize_result();
 776 }
 777
 778 /* guess for "linker script provide" symbol */
 779 static int may_be_linker_script_provide_symbol(const struct sym_entry *se)
 780 {
 781         const char *symbol = sym_name(se);
 782         int len = se->len - 1;
 783
 784         if (len < 8)
 785                 return 0;
 786
 787         if (symbol[0] != '_' || symbol[1] != '_')
 788                 return 0;
 789
 790         /* __start_XXXXX */
 791         if (!memcmp(symbol + 2, "start_", 6))
 792                 return 1;
 793
 794         /* __stop_XXXXX */
 795         if (!memcmp(symbol + 2, "stop_", 5))
 796                 return 1;
 797
 798         /* __end_XXXXX */
 799         if (!memcmp(symbol + 2, "end_", 4))
 800                 return 1;
 801
 802         /* __XXXXX_start */
 803         if (!memcmp(symbol + len - 6, "_start", 6))
 804                 return 1;
 805
 806         /* __XXXXX_end */
 807         if (!memcmp(symbol + len - 4, "_end", 4))
 808                 return 1;
 809
 810         return 0;
 811 }
 812
 813 static int compare_symbols(const void *a, const void *b)
 814 {
 815         const struct sym_entry *sa = *(const struct sym_entry **)a;
 816         const struct sym_entry *sb = *(const struct sym_entry **)b;
 817         int wa, wb;
 818
 819         /* sort by address first */
 820         if (sa->addr > sb->addr)
 821                 return 1;
 822         if (sa->addr < sb->addr)
 823                 return -1;
 824
 825         /* sort by "weakness" type */
 826         wa = (sa->sym[0] == 'w') || (sa->sym[0] == 'W');
 827         wb = (sb->sym[0] == 'w') || (sb->sym[0] == 'W');
 828         if (wa != wb)
 829                 return wa - wb;
 830
 831         /* sort by "linker script provide" type */
 832         wa = may_be_linker_script_provide_symbol(sa);
 833         wb = may_be_linker_script_provide_symbol(sb);
 834         if (wa != wb)
 835                 return wa - wb;
 836
 837         /* sort by the number of prefix underscores */
 838         wa = strspn(sym_name(sa), "_");
 839         wb = strspn(sym_name(sb), "_");
 840         if (wa != wb)
 841                 return wa - wb;
 842
 843         /* sort by initial order, so that other symbols are left undisturbed */
 844         return sa->start_pos - sb->start_pos;
 845 }
 846
 847 static void sort_symbols(void)
 848 {
 849         qsort(table, table_cnt, sizeof(table[0]), compare_symbols);
 850 }
 851
 852 static void make_percpus_absolute(void)
 853 {
 854         unsigned int i;
 855
 856         for (i = 0; i < table_cnt; i++)
 857                 if (symbol_in_range(table[i], &percpu_range, 1)) {
 858                         /*
 859                          * Keep the 'A' override for percpu symbols to
 860                          * ensure consistent behavior compared to older
 861                          * versions of this tool.
 862                          */
 863                         table[i]->sym[0] = 'A';
 864                         table[i]->percpu_absolute = 1;
 865                 }
 866 }
 867
 868 /* find the minimum non-absolute symbol address */
 869 static void record_relative_base(void)
 870 {
 871         unsigned int i;
 872
 873         for (i = 0; i < table_cnt; i++)
 874                 if (!symbol_absolute(table[i])) {
 875                         /*
 876                          * The table is sorted by address.
 877                          * Take the first non-absolute symbol value.
 878                          */
 879                         relative_base = table[i]->addr;
 880                         return;
 881                 }
 882 }
 883
 884 int main(int argc, char **argv)
 885 {
 886         while (1) {
 887                 static struct option long_options[] = {
 888                         {"all-symbols",     no_argument, &all_symbols,     1},
 889                         {"absolute-percpu", no_argument, &absolute_percpu, 1},
 890                         {"base-relative",   no_argument, &base_relative,   1},
 891                         {"lto-clang",       no_argument, &lto_clang,       1},
 892                         {},
 893                 };
 894
 895                 int c = getopt_long(argc, argv, "", long_options, NULL);
 896
 897                 if (c == -1)
 898                         break;
 899                 if (c != 0)
 900                         usage();
 901         }
 902
 903         if (optind >= argc)
 904                 usage();
 905
 906         read_map(argv[optind]);
 907         shrink_table();
 908         if (absolute_percpu)
 909                 make_percpus_absolute();
 910         sort_symbols();
 911         if (base_relative)
 912                 record_relative_base();
 913         optimize_token_table();
 914         write_src();
 915
 916         return 0;
 917 }