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: nm -n vmlinux | scripts/kallsyms [--all-symbols] > symbols.S
   9  *
  10  *      Table compression uses all the unused char codes on the symbols and
  11  *  maps these to the most used substrings (tokens). For instance, it might
  12  *  map char code 0xF7 to represent "write_" and then in every symbol where
  13  *  "write_" appears it can be replaced by 0xF7, saving 5 bytes.
  14  *      The used codes themselves are also placed in the table so that the
  15  *  decompresion can work without "special cases".
  16  *      Applied to kernel symbols, this usually produces a compression ratio
  17  *  of about 50%.
  18  *
  19  */
  20
  21 #include <getopt.h>
  22 #include <stdbool.h>
  23 #include <stdio.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <ctype.h>
  27 #include <limits.h>
  28
  29 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0]))
  30
  31 #define _stringify_1(x) #x
  32 #define _stringify(x)   _stringify_1(x)
  33
  34 #define KSYM_NAME_LEN           512
  35
  36 /*
  37  * A substantially bigger size than the current maximum.
  38  *
  39  * It cannot be defined as an expression because it gets stringified
  40  * for the fscanf() format string. Therefore, a _Static_assert() is
  41  * used instead to maintain the relationship with KSYM_NAME_LEN.
  42  */
  43 #define KSYM_NAME_LEN_BUFFER    2048
  44 _Static_assert(
  45         KSYM_NAME_LEN_BUFFER == KSYM_NAME_LEN * 4,
  46         "Please keep KSYM_NAME_LEN_BUFFER in sync with KSYM_NAME_LEN"
  47 );
  48
  49 struct sym_entry {
  50         unsigned long long addr;
  51         unsigned int len;
  52         unsigned int seq;
  53         unsigned int start_pos;
  54         unsigned int percpu_absolute;
  55         unsigned char sym[];
  56 };
  57
  58 struct addr_range {
  59         const char *start_sym, *end_sym;
  60         unsigned long long start, end;
  61 };
  62
  63 static unsigned long long _text;
  64 static unsigned long long relative_base;
  65 static struct addr_range text_ranges[] = {
  66         { "_stext",     "_etext"     },
  67         { "_sinittext", "_einittext" },
  68 };
  69 #define text_range_text     (&text_ranges[0])
  70 #define text_range_inittext (&text_ranges[1])
  71
  72 static struct addr_range percpu_range = {
  73         "__per_cpu_start", "__per_cpu_end", -1ULL, 0
  74 };
  75
  76 static struct sym_entry **table;
  77 static unsigned int table_size, table_cnt;
  78 static int all_symbols;
  79 static int absolute_percpu;
  80 static int base_relative;
  81 static int lto_clang;
  82
  83 static int token_profit[0x10000];
  84
  85 /* the table that holds the result of the compression */
  86 static unsigned char best_table[256][2];
  87 static unsigned char best_table_len[256];
  88
  89
  90 static void usage(void)
  91 {
  92         fprintf(stderr, "Usage: kallsyms [--all-symbols] [--absolute-percpu] "
  93                         "[--base-relative] [--lto-clang] in.map > out.S\n");
  94         exit(1);
  95 }
  96
  97 static char *sym_name(const struct sym_entry *s)
  98 {
  99         return (char *)s->sym + 1;
 100 }
 101
 102 static bool is_ignored_symbol(const char *name, char type)
 103 {
 104         /* Symbol names that exactly match to the following are ignored.*/
 105         static const char * const ignored_symbols[] = {
 106                 /*
 107                  * Symbols which vary between passes. Passes 1 and 2 must have
 108                  * identical symbol lists. The kallsyms_* symbols below are
 109                  * only added after pass 1, they would be included in pass 2
 110                  * when --all-symbols is specified so exclude them to get a
 111                  * stable symbol list.
 112                  */
 113                 "kallsyms_addresses",
 114                 "kallsyms_offsets",
 115                 "kallsyms_relative_base",
 116                 "kallsyms_num_syms",
 117                 "kallsyms_names",
 118                 "kallsyms_markers",
 119                 "kallsyms_token_table",
 120                 "kallsyms_token_index",
 121                 /* Exclude linker generated symbols which vary between passes */
 122                 "_SDA_BASE_",           /* ppc */
 123                 "_SDA2_BASE_",          /* ppc */
 124                 NULL
 125         };
 126
 127         /* Symbol names that begin with the following are ignored.*/
 128         static const char * const ignored_prefixes[] = {
 129                 "__efistub_",           /* arm64 EFI stub namespace */
 130                 "__kvm_nvhe_$",         /* arm64 local symbols in non-VHE KVM namespace */
 131                 "__kvm_nvhe_.L",        /* arm64 local symbols in non-VHE KVM namespace */
 132                 "__AArch64ADRPThunk_",  /* arm64 lld */
 133                 "__ARMV5PILongThunk_",  /* arm lld */
 134                 "__ARMV7PILongThunk_",
 135                 "__ThumbV7PILongThunk_",
 136                 "__LA25Thunk_",         /* mips lld */
 137                 "__microLA25Thunk_",
 138                 "__kcfi_typeid_",       /* CFI type identifiers */
 139                 NULL
 140         };
 141
 142         /* Symbol names that end with the following are ignored.*/
 143         static const char * const ignored_suffixes[] = {
 144                 "_from_arm",            /* arm */
 145                 "_from_thumb",          /* arm */
 146                 "_veneer",              /* arm */
 147                 NULL
 148         };
 149
 150         /* Symbol names that contain the following are ignored.*/
 151         static const char * const ignored_matches[] = {
 152                 ".long_branch.",        /* ppc stub */
 153                 ".plt_branch.",         /* ppc stub */
 154                 NULL
 155         };
 156
 157         const char * const *p;
 158
 159         for (p = ignored_symbols; *p; p++)
 160                 if (!strcmp(name, *p))
 161                         return true;
 162
 163         for (p = ignored_prefixes; *p; p++)
 164                 if (!strncmp(name, *p, strlen(*p)))
 165                         return true;
 166
 167         for (p = ignored_suffixes; *p; p++) {
 168                 int l = strlen(name) - strlen(*p);
 169
 170                 if (l >= 0 && !strcmp(name + l, *p))
 171                         return true;
 172         }
 173
 174         for (p = ignored_matches; *p; p++) {
 175                 if (strstr(name, *p))
 176                         return true;
 177         }
 178
 179         if (type == 'U' || type == 'u')
 180                 return true;
 181         /* exclude debugging symbols */
 182         if (type == 'N' || type == 'n')
 183                 return true;
 184
 185         if (toupper(type) == 'A') {
 186                 /* Keep these useful absolute symbols */
 187                 if (strcmp(name, "__kernel_syscall_via_break") &&
 188                     strcmp(name, "__kernel_syscall_via_epc") &&
 189                     strcmp(name, "__kernel_sigtramp") &&
 190                     strcmp(name, "__gp"))
 191                         return true;
 192         }
 193
 194         return false;
 195 }
 196
 197 static void check_symbol_range(const char *sym, unsigned long long addr,
 198                                struct addr_range *ranges, int entries)
 199 {
 200         size_t i;
 201         struct addr_range *ar;
 202
 203         for (i = 0; i < entries; ++i) {
 204                 ar = &ranges[i];
 205
 206                 if (strcmp(sym, ar->start_sym) == 0) {
 207                         ar->start = addr;
 208                         return;
 209                 } else if (strcmp(sym, ar->end_sym) == 0) {
 210                         ar->end = addr;
 211                         return;
 212                 }
 213         }
 214 }
 215
 216 static struct sym_entry *read_symbol(FILE *in)
 217 {
 218         char name[KSYM_NAME_LEN_BUFFER+1], type;
 219         unsigned long long addr;
 220         unsigned int len;
 221         struct sym_entry *sym;
 222         int rc;
 223
 224         rc = fscanf(in, "%llx %c %" _stringify(KSYM_NAME_LEN_BUFFER) "s\n", &addr, &type, name);
 225         if (rc != 3) {
 226                 if (rc != EOF && fgets(name, ARRAY_SIZE(name), in) == NULL)
 227                         fprintf(stderr, "Read error or end of file.\n");
 228                 return NULL;
 229         }
 230         if (strlen(name) >= KSYM_NAME_LEN) {
 231                 fprintf(stderr, "Symbol %s too long for kallsyms (%zu >= %d).\n"
 232                                 "Please increase KSYM_NAME_LEN both in kernel and kallsyms.c\n",
 233                         name, strlen(name), KSYM_NAME_LEN);
 234                 return NULL;
 235         }
 236
 237         if (strcmp(name, "_text") == 0)
 238                 _text = addr;
 239
 240         /* Ignore most absolute/undefined (?) symbols. */
 241         if (is_ignored_symbol(name, type))
 242                 return NULL;
 243
 244         check_symbol_range(name, addr, text_ranges, ARRAY_SIZE(text_ranges));
 245         check_symbol_range(name, addr, &percpu_range, 1);
 246
 247         /* include the type field in the symbol name, so that it gets
 248          * compressed together */
 249
 250         len = strlen(name) + 1;
 251
 252         sym = malloc(sizeof(*sym) + len + 1);
 253         if (!sym) {
 254                 fprintf(stderr, "kallsyms failure: "
 255                         "unable to allocate required amount of memory\n");
 256                 exit(EXIT_FAILURE);
 257         }
 258         sym->addr = addr;
 259         sym->len = len;
 260         sym->sym[0] = type;
 261         strcpy(sym_name(sym), name);
 262         sym->percpu_absolute = 0;
 263
 264         return sym;
 265 }
 266
 267 static int symbol_in_range(const struct sym_entry *s,
 268                            const struct addr_range *ranges, int entries)
 269 {
 270         size_t i;
 271         const struct addr_range *ar;
 272
 273         for (i = 0; i < entries; ++i) {
 274                 ar = &ranges[i];
 275
 276                 if (s->addr >= ar->start && s->addr <= ar->end)
 277                         return 1;
 278         }
 279
 280         return 0;
 281 }
 282
 283 static int symbol_valid(const struct sym_entry *s)
 284 {
 285         const char *name = sym_name(s);
 286
 287         /* if --all-symbols is not specified, then symbols outside the text
 288          * and inittext sections are discarded */
 289         if (!all_symbols) {
 290                 if (symbol_in_range(s, text_ranges,
 291                                     ARRAY_SIZE(text_ranges)) == 0)
 292                         return 0;
 293                 /* Corner case.  Discard any symbols with the same value as
 294                  * _etext _einittext; they can move between pass 1 and 2 when
 295                  * the kallsyms data are added.  If these symbols move then
 296                  * they may get dropped in pass 2, which breaks the kallsyms
 297                  * rules.
 298                  */
 299                 if ((s->addr == text_range_text->end &&
 300                      strcmp(name, text_range_text->end_sym)) ||
 301                     (s->addr == text_range_inittext->end &&
 302                      strcmp(name, text_range_inittext->end_sym)))
 303                         return 0;
 304         }
 305
 306         return 1;
 307 }
 308
 309 /* remove all the invalid symbols from the table */
 310 static void shrink_table(void)
 311 {
 312         unsigned int i, pos;
 313
 314         pos = 0;
 315         for (i = 0; i < table_cnt; i++) {
 316                 if (symbol_valid(table[i])) {
 317                         if (pos != i)
 318                                 table[pos] = table[i];
 319                         pos++;
 320                 } else {
 321                         free(table[i]);
 322                 }
 323         }
 324         table_cnt = pos;
 325
 326         /* When valid symbol is not registered, exit to error */
 327         if (!table_cnt) {
 328                 fprintf(stderr, "No valid symbol.\n");
 329                 exit(1);
 330         }
 331 }
 332
 333 static void read_map(const char *in)
 334 {
 335         FILE *fp;
 336         struct sym_entry *sym;
 337
 338         fp = fopen(in, "r");
 339         if (!fp) {
 340                 perror(in);
 341                 exit(1);
 342         }
 343
 344         while (!feof(fp)) {
 345                 sym = read_symbol(fp);
 346                 if (!sym)
 347                         continue;
 348
 349                 sym->start_pos = table_cnt;
 350
 351                 if (table_cnt >= table_size) {
 352                         table_size += 10000;
 353                         table = realloc(table, sizeof(*table) * table_size);
 354                         if (!table) {
 355                                 fprintf(stderr, "out of memory\n");
 356                                 fclose(fp);
 357                                 exit (1);
 358                         }
 359                 }
 360
 361                 table[table_cnt++] = sym;
 362         }
 363
 364         fclose(fp);
 365 }
 366
 367 static void output_label(const char *label)
 368 {
 369         printf(".globl %s\n", label);
 370         printf("\tALGN\n");
 371         printf("%s:\n", label);
 372 }
 373
 374 /* Provide proper symbols relocatability by their '_text' relativeness. */
 375 static void output_address(unsigned long long addr)
 376 {
 377         if (_text <= addr)
 378                 printf("\tPTR\t_text + %#llx\n", addr - _text);
 379         else
 380                 printf("\tPTR\t_text - %#llx\n", _text - addr);
 381 }
 382
 383 /* uncompress a compressed symbol. When this function is called, the best table
 384  * might still be compressed itself, so the function needs to be recursive */
 385 static int expand_symbol(const unsigned char *data, int len, char *result)
 386 {
 387         int c, rlen, total=0;
 388
 389         while (len) {
 390                 c = *data;
 391                 /* if the table holds a single char that is the same as the one
 392                  * we are looking for, then end the search */
 393                 if (best_table[c][0]==c && best_table_len[c]==1) {
 394                         *result++ = c;
 395                         total++;
 396                 } else {
 397                         /* if not, recurse and expand */
 398                         rlen = expand_symbol(best_table[c], best_table_len[c], result);
 399                         total += rlen;
 400                         result += rlen;
 401                 }
 402                 data++;
 403                 len--;
 404         }
 405         *result=0;
 406
 407         return total;
 408 }
 409
 410 static int symbol_absolute(const struct sym_entry *s)
 411 {
 412         return s->percpu_absolute;
 413 }
 414
 415 static char * s_name(char *buf)
 416 {
 417         /* Skip the symbol type */
 418         return buf + 1;
 419 }
 420
 421 static void cleanup_symbol_name(char *s)
 422 {
 423         char *p;
 424
 425         if (!lto_clang)
 426                 return;
 427
 428         /*
 429          * ASCII[.]   = 2e
 430          * ASCII[0-9] = 30,39
 431          * ASCII[A-Z] = 41,5a
 432          * ASCII[_]   = 5f
 433          * ASCII[a-z] = 61,7a
 434          *
 435          * As above, replacing '.' with '\0' does not affect the main sorting,
 436          * but it helps us with subsorting.
 437          */
 438         p = strchr(s, '.');
 439         if (p)
 440                 *p = '\0';
 441 }
 442
 443 static int compare_names(const void *a, const void *b)
 444 {
 445         int ret;
 446         char sa_namebuf[KSYM_NAME_LEN];
 447         char sb_namebuf[KSYM_NAME_LEN];
 448         const struct sym_entry *sa = *(const struct sym_entry **)a;
 449         const struct sym_entry *sb = *(const struct sym_entry **)b;
 450
 451         expand_symbol(sa->sym, sa->len, sa_namebuf);
 452         expand_symbol(sb->sym, sb->len, sb_namebuf);
 453         cleanup_symbol_name(s_name(sa_namebuf));
 454         cleanup_symbol_name(s_name(sb_namebuf));
 455         ret = strcmp(s_name(sa_namebuf), s_name(sb_namebuf));
 456         if (!ret) {
 457                 if (sa->addr > sb->addr)
 458                         return 1;
 459                 else if (sa->addr < sb->addr)
 460                         return -1;
 461
 462                 /* keep old order */
 463                 return (int)(sa->seq - sb->seq);
 464         }
 465
 466         return ret;
 467 }
 468
 469 static void sort_symbols_by_name(void)
 470 {
 471         qsort(table, table_cnt, sizeof(table[0]), compare_names);
 472 }
 473
 474 static void write_src(void)
 475 {
 476         unsigned int i, k, off;
 477         unsigned int best_idx[256];
 478         unsigned int *markers;
 479         char buf[KSYM_NAME_LEN];
 480
 481         printf("#include <asm/bitsperlong.h>\n");
 482         printf("#if BITS_PER_LONG == 64\n");
 483         printf("#define PTR .quad\n");
 484         printf("#define ALGN .balign 8\n");
 485         printf("#else\n");
 486         printf("#define PTR .long\n");
 487         printf("#define ALGN .balign 4\n");
 488         printf("#endif\n");
 489
 490         printf("\t.section .rodata, \"a\"\n");
 491
 492         if (!base_relative)
 493                 output_label("kallsyms_addresses");
 494         else
 495                 output_label("kallsyms_offsets");
 496
 497         for (i = 0; i < table_cnt; i++) {
 498                 if (base_relative) {
 499                         /*
 500                          * Use the offset relative to the lowest value
 501                          * encountered of all relative symbols, and emit
 502                          * non-relocatable fixed offsets that will be fixed
 503                          * up at runtime.
 504                          */
 505
 506                         long long offset;
 507                         int overflow;
 508
 509                         if (!absolute_percpu) {
 510                                 offset = table[i]->addr - relative_base;
 511                                 overflow = (offset < 0 || offset > UINT_MAX);
 512                         } else if (symbol_absolute(table[i])) {
 513                                 offset = table[i]->addr;
 514                                 overflow = (offset < 0 || offset > INT_MAX);
 515                         } else {
 516                                 offset = relative_base - table[i]->addr - 1;
 517                                 overflow = (offset < INT_MIN || offset >= 0);
 518                         }
 519                         if (overflow) {
 520                                 fprintf(stderr, "kallsyms failure: "
 521                                         "%s symbol value %#llx out of range in relative mode\n",
 522                                         symbol_absolute(table[i]) ? "absolute" : "relative",
 523                                         table[i]->addr);
 524                                 exit(EXIT_FAILURE);
 525                         }
 526                         printf("\t.long\t%#x\n", (int)offset);
 527                 } else if (!symbol_absolute(table[i])) {
 528                         output_address(table[i]->addr);
 529                 } else {
 530                         printf("\tPTR\t%#llx\n", table[i]->addr);
 531                 }
 532         }
 533         printf("\n");
 534
 535         if (base_relative) {
 536                 output_label("kallsyms_relative_base");
 537                 output_address(relative_base);
 538                 printf("\n");
 539         }
 540
 541         output_label("kallsyms_num_syms");
 542         printf("\t.long\t%u\n", table_cnt);
 543         printf("\n");
 544
 545         /* table of offset markers, that give the offset in the compressed stream
 546          * every 256 symbols */
 547         markers = malloc(sizeof(unsigned int) * ((table_cnt + 255) / 256));
 548         if (!markers) {
 549                 fprintf(stderr, "kallsyms failure: "
 550                         "unable to allocate required memory\n");
 551                 exit(EXIT_FAILURE);
 552         }
 553
 554         output_label("kallsyms_names");
 555         off = 0;
 556         for (i = 0; i < table_cnt; i++) {
 557                 if ((i & 0xFF) == 0)
 558                         markers[i >> 8] = off;
 559                 table[i]->seq = i;
 560
 561                 /* There cannot be any symbol of length zero. */
 562                 if (table[i]->len == 0) {
 563                         fprintf(stderr, "kallsyms failure: "
 564                                 "unexpected zero symbol length\n");
 565                         exit(EXIT_FAILURE);
 566                 }
 567
 568                 /* Only lengths that fit in up-to-two-byte ULEB128 are supported. */
 569                 if (table[i]->len > 0x3FFF) {
 570                         fprintf(stderr, "kallsyms failure: "
 571                                 "unexpected huge symbol length\n");
 572                         exit(EXIT_FAILURE);
 573                 }
 574
 575                 /* Encode length with ULEB128. */
 576                 if (table[i]->len <= 0x7F) {
 577                         /* Most symbols use a single byte for the length. */
 578                         printf("\t.byte 0x%02x", table[i]->len);
 579                         off += table[i]->len + 1;
 580                 } else {
 581                         /* "Big" symbols use two bytes. */
 582                         printf("\t.byte 0x%02x, 0x%02x",
 583                                 (table[i]->len & 0x7F) | 0x80,
 584                                 (table[i]->len >> 7) & 0x7F);
 585                         off += table[i]->len + 2;
 586                 }
 587                 for (k = 0; k < table[i]->len; k++)
 588                         printf(", 0x%02x", table[i]->sym[k]);
 589                 printf("\n");
 590         }
 591         printf("\n");
 592
 593         output_label("kallsyms_markers");
 594         for (i = 0; i < ((table_cnt + 255) >> 8); i++)
 595                 printf("\t.long\t%u\n", markers[i]);
 596         printf("\n");
 597
 598         free(markers);
 599
 600         sort_symbols_by_name();
 601         output_label("kallsyms_seqs_of_names");
 602         for (i = 0; i < table_cnt; i++)
 603                 printf("\t.byte 0x%02x, 0x%02x, 0x%02x\n",
 604                         (unsigned char)(table[i]->seq >> 16),
 605                         (unsigned char)(table[i]->seq >> 8),
 606                         (unsigned char)(table[i]->seq >> 0));
 607         printf("\n");
 608
 609         output_label("kallsyms_token_table");
 610         off = 0;
 611         for (i = 0; i < 256; i++) {
 612                 best_idx[i] = off;
 613                 expand_symbol(best_table[i], best_table_len[i], buf);
 614                 printf("\t.asciz\t\"%s\"\n", buf);
 615                 off += strlen(buf) + 1;
 616         }
 617         printf("\n");
 618
 619         output_label("kallsyms_token_index");
 620         for (i = 0; i < 256; i++)
 621                 printf("\t.short\t%d\n", best_idx[i]);
 622         printf("\n");
 623 }
 624
 625
 626 /* table lookup compression functions */
 627
 628 /* count all the possible tokens in a symbol */
 629 static void learn_symbol(const unsigned char *symbol, int len)
 630 {
 631         int i;
 632
 633         for (i = 0; i < len - 1; i++)
 634                 token_profit[ symbol[i] + (symbol[i + 1] << 8) ]++;
 635 }
 636
 637 /* decrease the count for all the possible tokens in a symbol */
 638 static void forget_symbol(const unsigned char *symbol, int len)
 639 {
 640         int i;
 641
 642         for (i = 0; i < len - 1; i++)
 643                 token_profit[ symbol[i] + (symbol[i + 1] << 8) ]--;
 644 }
 645
 646 /* do the initial token count */
 647 static void build_initial_token_table(void)
 648 {
 649         unsigned int i;
 650
 651         for (i = 0; i < table_cnt; i++)
 652                 learn_symbol(table[i]->sym, table[i]->len);
 653 }
 654
 655 static unsigned char *find_token(unsigned char *str, int len,
 656                                  const unsigned char *token)
 657 {
 658         int i;
 659
 660         for (i = 0; i < len - 1; i++) {
 661                 if (str[i] == token[0] && str[i+1] == token[1])
 662                         return &str[i];
 663         }
 664         return NULL;
 665 }
 666
 667 /* replace a given token in all the valid symbols. Use the sampled symbols
 668  * to update the counts */
 669 static void compress_symbols(const unsigned char *str, int idx)
 670 {
 671         unsigned int i, len, size;
 672         unsigned char *p1, *p2;
 673
 674         for (i = 0; i < table_cnt; i++) {
 675
 676                 len = table[i]->len;
 677                 p1 = table[i]->sym;
 678
 679                 /* find the token on the symbol */
 680                 p2 = find_token(p1, len, str);
 681                 if (!p2) continue;
 682
 683                 /* decrease the counts for this symbol's tokens */
 684                 forget_symbol(table[i]->sym, len);
 685
 686                 size = len;
 687
 688                 do {
 689                         *p2 = idx;
 690                         p2++;
 691                         size -= (p2 - p1);
 692                         memmove(p2, p2 + 1, size);
 693                         p1 = p2;
 694                         len--;
 695
 696                         if (size < 2) break;
 697
 698                         /* find the token on the symbol */
 699                         p2 = find_token(p1, size, str);
 700
 701                 } while (p2);
 702
 703                 table[i]->len = len;
 704
 705                 /* increase the counts for this symbol's new tokens */
 706                 learn_symbol(table[i]->sym, len);
 707         }
 708 }
 709
 710 /* search the token with the maximum profit */
 711 static int find_best_token(void)
 712 {
 713         int i, best, bestprofit;
 714
 715         bestprofit=-10000;
 716         best = 0;
 717
 718         for (i = 0; i < 0x10000; i++) {
 719                 if (token_profit[i] > bestprofit) {
 720                         best = i;
 721                         bestprofit = token_profit[i];
 722                 }
 723         }
 724         return best;
 725 }
 726
 727 /* this is the core of the algorithm: calculate the "best" table */
 728 static void optimize_result(void)
 729 {
 730         int i, best;
 731
 732         /* using the '\0' symbol last allows compress_symbols to use standard
 733          * fast string functions */
 734         for (i = 255; i >= 0; i--) {
 735
 736                 /* if this table slot is empty (it is not used by an actual
 737                  * original char code */
 738                 if (!best_table_len[i]) {
 739
 740                         /* find the token with the best profit value */
 741                         best = find_best_token();
 742                         if (token_profit[best] == 0)
 743                                 break;
 744
 745                         /* place it in the "best" table */
 746                         best_table_len[i] = 2;
 747                         best_table[i][0] = best & 0xFF;
 748                         best_table[i][1] = (best >> 8) & 0xFF;
 749
 750                         /* replace this token in all the valid symbols */
 751                         compress_symbols(best_table[i], i);
 752                 }
 753         }
 754 }
 755
 756 /* start by placing the symbols that are actually used on the table */
 757 static void insert_real_symbols_in_table(void)
 758 {
 759         unsigned int i, j, c;
 760
 761         for (i = 0; i < table_cnt; i++) {
 762                 for (j = 0; j < table[i]->len; j++) {
 763                         c = table[i]->sym[j];
 764                         best_table[c][0]=c;
 765                         best_table_len[c]=1;
 766                 }
 767         }
 768 }
 769
 770 static void optimize_token_table(void)
 771 {
 772         build_initial_token_table();
 773
 774         insert_real_symbols_in_table();
 775
 776         optimize_result();
 777 }
 778
 779 /* guess for "linker script provide" symbol */
 780 static int may_be_linker_script_provide_symbol(const struct sym_entry *se)
 781 {
 782         const char *symbol = sym_name(se);
 783         int len = se->len - 1;
 784
 785         if (len < 8)
 786                 return 0;
 787
 788         if (symbol[0] != '_' || symbol[1] != '_')
 789                 return 0;
 790
 791         /* __start_XXXXX */
 792         if (!memcmp(symbol + 2, "start_", 6))
 793                 return 1;
 794
 795         /* __stop_XXXXX */
 796         if (!memcmp(symbol + 2, "stop_", 5))
 797                 return 1;
 798
 799         /* __end_XXXXX */
 800         if (!memcmp(symbol + 2, "end_", 4))
 801                 return 1;
 802
 803         /* __XXXXX_start */
 804         if (!memcmp(symbol + len - 6, "_start", 6))
 805                 return 1;
 806
 807         /* __XXXXX_end */
 808         if (!memcmp(symbol + len - 4, "_end", 4))
 809                 return 1;
 810
 811         return 0;
 812 }
 813
 814 static int compare_symbols(const void *a, const void *b)
 815 {
 816         const struct sym_entry *sa = *(const struct sym_entry **)a;
 817         const struct sym_entry *sb = *(const struct sym_entry **)b;
 818         int wa, wb;
 819
 820         /* sort by address first */
 821         if (sa->addr > sb->addr)
 822                 return 1;
 823         if (sa->addr < sb->addr)
 824                 return -1;
 825
 826         /* sort by "weakness" type */
 827         wa = (sa->sym[0] == 'w') || (sa->sym[0] == 'W');
 828         wb = (sb->sym[0] == 'w') || (sb->sym[0] == 'W');
 829         if (wa != wb)
 830                 return wa - wb;
 831
 832         /* sort by "linker script provide" type */
 833         wa = may_be_linker_script_provide_symbol(sa);
 834         wb = may_be_linker_script_provide_symbol(sb);
 835         if (wa != wb)
 836                 return wa - wb;
 837
 838         /* sort by the number of prefix underscores */
 839         wa = strspn(sym_name(sa), "_");
 840         wb = strspn(sym_name(sb), "_");
 841         if (wa != wb)
 842                 return wa - wb;
 843
 844         /* sort by initial order, so that other symbols are left undisturbed */
 845         return sa->start_pos - sb->start_pos;
 846 }
 847
 848 static void sort_symbols(void)
 849 {
 850         qsort(table, table_cnt, sizeof(table[0]), compare_symbols);
 851 }
 852
 853 static void make_percpus_absolute(void)
 854 {
 855         unsigned int i;
 856
 857         for (i = 0; i < table_cnt; i++)
 858                 if (symbol_in_range(table[i], &percpu_range, 1)) {
 859                         /*
 860                          * Keep the 'A' override for percpu symbols to
 861                          * ensure consistent behavior compared to older
 862                          * versions of this tool.
 863                          */
 864                         table[i]->sym[0] = 'A';
 865                         table[i]->percpu_absolute = 1;
 866                 }
 867 }
 868
 869 /* find the minimum non-absolute symbol address */
 870 static void record_relative_base(void)
 871 {
 872         unsigned int i;
 873
 874         for (i = 0; i < table_cnt; i++)
 875                 if (!symbol_absolute(table[i])) {
 876                         /*
 877                          * The table is sorted by address.
 878                          * Take the first non-absolute symbol value.
 879                          */
 880                         relative_base = table[i]->addr;
 881                         return;
 882                 }
 883 }
 884
 885 int main(int argc, char **argv)
 886 {
 887         while (1) {
 888                 static struct option long_options[] = {
 889                         {"all-symbols",     no_argument, &all_symbols,     1},
 890                         {"absolute-percpu", no_argument, &absolute_percpu, 1},
 891                         {"base-relative",   no_argument, &base_relative,   1},
 892                         {"lto-clang",       no_argument, &lto_clang,       1},
 893                         {},
 894                 };
 895
 896                 int c = getopt_long(argc, argv, "", long_options, NULL);
 897
 898                 if (c == -1)
 899                         break;
 900                 if (c != 0)
 901                         usage();
 902         }
 903
 904         if (optind >= argc)
 905                 usage();
 906
 907         read_map(argv[optind]);
 908         shrink_table();
 909         if (absolute_percpu)
 910                 make_percpus_absolute();
 911         sort_symbols();
 912         if (base_relative)
 913                 record_relative_base();
 914         optimize_token_table();
 915         write_src();
 916
 917         return 0;
 918 }