| 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] in.map > out.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 <errno.h> |
| 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 | #include <xalloc.h> |
| 31 | |
| 32 | #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0])) |
| 33 | |
| 34 | #define KSYM_NAME_LEN 512 |
| 35 | |
| 36 | struct sym_entry { |
| 37 | unsigned long long addr; |
| 38 | unsigned int len; |
| 39 | unsigned int seq; |
| 40 | unsigned char sym[]; |
| 41 | }; |
| 42 | |
| 43 | struct addr_range { |
| 44 | const char *start_sym, *end_sym; |
| 45 | unsigned long long start, end; |
| 46 | }; |
| 47 | |
| 48 | static unsigned long long _text; |
| 49 | static unsigned long long relative_base; |
| 50 | static struct addr_range text_ranges[] = { |
| 51 | { "_stext" , "_etext" }, |
| 52 | { "_sinittext" , "_einittext" }, |
| 53 | }; |
| 54 | #define text_range_text (&text_ranges[0]) |
| 55 | #define text_range_inittext (&text_ranges[1]) |
| 56 | |
| 57 | static struct sym_entry **table; |
| 58 | static unsigned int table_size, table_cnt; |
| 59 | static int all_symbols; |
| 60 | |
| 61 | static int token_profit[0x10000]; |
| 62 | |
| 63 | /* the table that holds the result of the compression */ |
| 64 | static unsigned char best_table[256][2]; |
| 65 | static unsigned char best_table_len[256]; |
| 66 | |
| 67 | |
| 68 | static void usage(void) |
| 69 | { |
| 70 | fprintf(stderr, format: "Usage: kallsyms [--all-symbols] in.map > out.S\n" ); |
| 71 | exit(status: 1); |
| 72 | } |
| 73 | |
| 74 | static char *sym_name(const struct sym_entry *s) |
| 75 | { |
| 76 | return (char *)s->sym + 1; |
| 77 | } |
| 78 | |
| 79 | static bool is_ignored_symbol(const char *name, char type) |
| 80 | { |
| 81 | if (type == 'u' || type == 'n') |
| 82 | return true; |
| 83 | |
| 84 | if (toupper(type) == 'A') { |
| 85 | /* Keep these useful absolute symbols */ |
| 86 | if (strcmp(s1: name, s2: "__kernel_syscall_via_break" ) && |
| 87 | strcmp(s1: name, s2: "__kernel_syscall_via_epc" ) && |
| 88 | strcmp(s1: name, s2: "__kernel_sigtramp" ) && |
| 89 | strcmp(s1: name, s2: "__gp" )) |
| 90 | return true; |
| 91 | } |
| 92 | |
| 93 | return false; |
| 94 | } |
| 95 | |
| 96 | static void check_symbol_range(const char *sym, unsigned long long addr, |
| 97 | struct addr_range *ranges, int entries) |
| 98 | { |
| 99 | size_t i; |
| 100 | struct addr_range *ar; |
| 101 | |
| 102 | for (i = 0; i < entries; ++i) { |
| 103 | ar = &ranges[i]; |
| 104 | |
| 105 | if (strcmp(s1: sym, s2: ar->start_sym) == 0) { |
| 106 | ar->start = addr; |
| 107 | return; |
| 108 | } else if (strcmp(s1: sym, s2: ar->end_sym) == 0) { |
| 109 | ar->end = addr; |
| 110 | return; |
| 111 | } |
| 112 | } |
| 113 | } |
| 114 | |
| 115 | static struct sym_entry *read_symbol(FILE *in, char **buf, size_t *buf_len) |
| 116 | { |
| 117 | char *name, type, *p; |
| 118 | unsigned long long addr; |
| 119 | size_t len; |
| 120 | ssize_t readlen; |
| 121 | struct sym_entry *sym; |
| 122 | |
| 123 | errno = 0; |
| 124 | readlen = getline(lineptr: buf, n: buf_len, stream: in); |
| 125 | if (readlen < 0) { |
| 126 | if (errno) { |
| 127 | perror(s: "read_symbol" ); |
| 128 | exit(EXIT_FAILURE); |
| 129 | } |
| 130 | return NULL; |
| 131 | } |
| 132 | |
| 133 | if ((*buf)[readlen - 1] == '\n') |
| 134 | (*buf)[readlen - 1] = 0; |
| 135 | |
| 136 | addr = strtoull(nptr: *buf, endptr: &p, base: 16); |
| 137 | |
| 138 | if (*buf == p || *p++ != ' ' || !isascii((type = *p++)) || *p++ != ' ') { |
| 139 | fprintf(stderr, format: "line format error\n" ); |
| 140 | exit(EXIT_FAILURE); |
| 141 | } |
| 142 | |
| 143 | name = p; |
| 144 | len = strlen(s: name); |
| 145 | |
| 146 | if (len >= KSYM_NAME_LEN) { |
| 147 | fprintf(stderr, format: "Symbol %s too long for kallsyms (%zu >= %d).\n" |
| 148 | "Please increase KSYM_NAME_LEN both in kernel and kallsyms.c\n" , |
| 149 | name, len, KSYM_NAME_LEN); |
| 150 | return NULL; |
| 151 | } |
| 152 | |
| 153 | if (strcmp(s1: name, s2: "_text" ) == 0) |
| 154 | _text = addr; |
| 155 | |
| 156 | /* Ignore most absolute/undefined (?) symbols. */ |
| 157 | if (is_ignored_symbol(name, type)) |
| 158 | return NULL; |
| 159 | |
| 160 | check_symbol_range(sym: name, addr, ranges: text_ranges, ARRAY_SIZE(text_ranges)); |
| 161 | |
| 162 | /* include the type field in the symbol name, so that it gets |
| 163 | * compressed together */ |
| 164 | len++; |
| 165 | |
| 166 | sym = xmalloc(sizeof(*sym) + len + 1); |
| 167 | sym->addr = addr; |
| 168 | sym->len = len; |
| 169 | sym->sym[0] = type; |
| 170 | strcpy(dest: sym_name(s: sym), src: name); |
| 171 | |
| 172 | return sym; |
| 173 | } |
| 174 | |
| 175 | static int symbol_in_range(const struct sym_entry *s, |
| 176 | const struct addr_range *ranges, int entries) |
| 177 | { |
| 178 | size_t i; |
| 179 | const struct addr_range *ar; |
| 180 | |
| 181 | for (i = 0; i < entries; ++i) { |
| 182 | ar = &ranges[i]; |
| 183 | |
| 184 | if (s->addr >= ar->start && s->addr <= ar->end) |
| 185 | return 1; |
| 186 | } |
| 187 | |
| 188 | return 0; |
| 189 | } |
| 190 | |
| 191 | static bool string_starts_with(const char *s, const char *prefix) |
| 192 | { |
| 193 | return strncmp(s1: s, s2: prefix, n: strlen(s: prefix)) == 0; |
| 194 | } |
| 195 | |
| 196 | static int symbol_valid(const struct sym_entry *s) |
| 197 | { |
| 198 | const char *name = sym_name(s); |
| 199 | |
| 200 | /* if --all-symbols is not specified, then symbols outside the text |
| 201 | * and inittext sections are discarded */ |
| 202 | if (!all_symbols) { |
| 203 | /* |
| 204 | * Symbols starting with __start and __stop are used to denote |
| 205 | * section boundaries, and should always be included: |
| 206 | */ |
| 207 | if (string_starts_with(s: name, prefix: "__start_" ) || |
| 208 | string_starts_with(s: name, prefix: "__stop_" )) |
| 209 | return 1; |
| 210 | |
| 211 | if (symbol_in_range(s, ranges: text_ranges, |
| 212 | ARRAY_SIZE(text_ranges)) == 0) |
| 213 | return 0; |
| 214 | /* Corner case. Discard any symbols with the same value as |
| 215 | * _etext _einittext; they can move between pass 1 and 2 when |
| 216 | * the kallsyms data are added. If these symbols move then |
| 217 | * they may get dropped in pass 2, which breaks the kallsyms |
| 218 | * rules. |
| 219 | */ |
| 220 | if ((s->addr == text_range_text->end && |
| 221 | strcmp(s1: name, text_range_text->end_sym)) || |
| 222 | (s->addr == text_range_inittext->end && |
| 223 | strcmp(s1: name, text_range_inittext->end_sym))) |
| 224 | return 0; |
| 225 | } |
| 226 | |
| 227 | return 1; |
| 228 | } |
| 229 | |
| 230 | /* remove all the invalid symbols from the table */ |
| 231 | static void shrink_table(void) |
| 232 | { |
| 233 | unsigned int i, pos; |
| 234 | |
| 235 | pos = 0; |
| 236 | for (i = 0; i < table_cnt; i++) { |
| 237 | if (symbol_valid(s: table[i])) { |
| 238 | if (pos != i) |
| 239 | table[pos] = table[i]; |
| 240 | pos++; |
| 241 | } else { |
| 242 | free(ptr: table[i]); |
| 243 | } |
| 244 | } |
| 245 | table_cnt = pos; |
| 246 | } |
| 247 | |
| 248 | static void read_map(const char *in) |
| 249 | { |
| 250 | FILE *fp; |
| 251 | struct sym_entry *sym; |
| 252 | char *buf = NULL; |
| 253 | size_t buflen = 0; |
| 254 | |
| 255 | fp = fopen(filename: in, modes: "r" ); |
| 256 | if (!fp) { |
| 257 | perror(s: in); |
| 258 | exit(status: 1); |
| 259 | } |
| 260 | |
| 261 | while (!feof(stream: fp)) { |
| 262 | sym = read_symbol(in: fp, buf: &buf, buf_len: &buflen); |
| 263 | if (!sym) |
| 264 | continue; |
| 265 | |
| 266 | sym->seq = table_cnt; |
| 267 | |
| 268 | if (table_cnt >= table_size) { |
| 269 | table_size += 10000; |
| 270 | table = xrealloc(table, sizeof(*table) * table_size); |
| 271 | } |
| 272 | |
| 273 | table[table_cnt++] = sym; |
| 274 | } |
| 275 | |
| 276 | free(ptr: buf); |
| 277 | fclose(stream: fp); |
| 278 | } |
| 279 | |
| 280 | static void output_label(const char *label) |
| 281 | { |
| 282 | printf(format: ".globl %s\n" , label); |
| 283 | printf(format: "\tALGN\n" ); |
| 284 | printf(format: "%s:\n" , label); |
| 285 | } |
| 286 | |
| 287 | /* uncompress a compressed symbol. When this function is called, the best table |
| 288 | * might still be compressed itself, so the function needs to be recursive */ |
| 289 | static int expand_symbol(const unsigned char *data, int len, char *result) |
| 290 | { |
| 291 | int c, rlen, total=0; |
| 292 | |
| 293 | while (len) { |
| 294 | c = *data; |
| 295 | /* if the table holds a single char that is the same as the one |
| 296 | * we are looking for, then end the search */ |
| 297 | if (best_table[c][0]==c && best_table_len[c]==1) { |
| 298 | *result++ = c; |
| 299 | total++; |
| 300 | } else { |
| 301 | /* if not, recurse and expand */ |
| 302 | rlen = expand_symbol(data: best_table[c], len: best_table_len[c], result); |
| 303 | total += rlen; |
| 304 | result += rlen; |
| 305 | } |
| 306 | data++; |
| 307 | len--; |
| 308 | } |
| 309 | *result=0; |
| 310 | |
| 311 | return total; |
| 312 | } |
| 313 | |
| 314 | static int compare_names(const void *a, const void *b) |
| 315 | { |
| 316 | int ret; |
| 317 | const struct sym_entry *sa = *(const struct sym_entry **)a; |
| 318 | const struct sym_entry *sb = *(const struct sym_entry **)b; |
| 319 | |
| 320 | ret = strcmp(s1: sym_name(s: sa), s2: sym_name(s: sb)); |
| 321 | if (!ret) { |
| 322 | if (sa->addr > sb->addr) |
| 323 | return 1; |
| 324 | else if (sa->addr < sb->addr) |
| 325 | return -1; |
| 326 | |
| 327 | /* keep old order */ |
| 328 | return (int)(sa->seq - sb->seq); |
| 329 | } |
| 330 | |
| 331 | return ret; |
| 332 | } |
| 333 | |
| 334 | static void sort_symbols_by_name(void) |
| 335 | { |
| 336 | qsort(base: table, nmemb: table_cnt, size: sizeof(table[0]), compar: compare_names); |
| 337 | } |
| 338 | |
| 339 | static void write_src(void) |
| 340 | { |
| 341 | unsigned int i, k, off; |
| 342 | unsigned int best_idx[256]; |
| 343 | unsigned int *markers, markers_cnt; |
| 344 | char buf[KSYM_NAME_LEN]; |
| 345 | |
| 346 | printf(format: "#include <asm/bitsperlong.h>\n" ); |
| 347 | printf(format: "#if BITS_PER_LONG == 64\n" ); |
| 348 | printf(format: "#define PTR .quad\n" ); |
| 349 | printf(format: "#define ALGN .balign 8\n" ); |
| 350 | printf(format: "#else\n" ); |
| 351 | printf(format: "#define PTR .long\n" ); |
| 352 | printf(format: "#define ALGN .balign 4\n" ); |
| 353 | printf(format: "#endif\n" ); |
| 354 | |
| 355 | printf(format: "\t.section .rodata, \"a\"\n" ); |
| 356 | |
| 357 | output_label(label: "kallsyms_num_syms" ); |
| 358 | printf(format: "\t.long\t%u\n" , table_cnt); |
| 359 | printf(format: "\n" ); |
| 360 | |
| 361 | /* table of offset markers, that give the offset in the compressed stream |
| 362 | * every 256 symbols */ |
| 363 | markers_cnt = (table_cnt + 255) / 256; |
| 364 | markers = xmalloc(sizeof(*markers) * markers_cnt); |
| 365 | |
| 366 | output_label(label: "kallsyms_names" ); |
| 367 | off = 0; |
| 368 | for (i = 0; i < table_cnt; i++) { |
| 369 | if ((i & 0xFF) == 0) |
| 370 | markers[i >> 8] = off; |
| 371 | table[i]->seq = i; |
| 372 | |
| 373 | /* There cannot be any symbol of length zero. */ |
| 374 | if (table[i]->len == 0) { |
| 375 | fprintf(stderr, format: "kallsyms failure: " |
| 376 | "unexpected zero symbol length\n" ); |
| 377 | exit(EXIT_FAILURE); |
| 378 | } |
| 379 | |
| 380 | /* Only lengths that fit in up-to-two-byte ULEB128 are supported. */ |
| 381 | if (table[i]->len > 0x3FFF) { |
| 382 | fprintf(stderr, format: "kallsyms failure: " |
| 383 | "unexpected huge symbol length\n" ); |
| 384 | exit(EXIT_FAILURE); |
| 385 | } |
| 386 | |
| 387 | /* Encode length with ULEB128. */ |
| 388 | if (table[i]->len <= 0x7F) { |
| 389 | /* Most symbols use a single byte for the length. */ |
| 390 | printf(format: "\t.byte 0x%02x" , table[i]->len); |
| 391 | off += table[i]->len + 1; |
| 392 | } else { |
| 393 | /* "Big" symbols use two bytes. */ |
| 394 | printf(format: "\t.byte 0x%02x, 0x%02x" , |
| 395 | (table[i]->len & 0x7F) | 0x80, |
| 396 | (table[i]->len >> 7) & 0x7F); |
| 397 | off += table[i]->len + 2; |
| 398 | } |
| 399 | for (k = 0; k < table[i]->len; k++) |
| 400 | printf(format: ", 0x%02x" , table[i]->sym[k]); |
| 401 | |
| 402 | /* |
| 403 | * Now that we wrote out the compressed symbol name, restore the |
| 404 | * original name and print it in the comment. |
| 405 | */ |
| 406 | expand_symbol(data: table[i]->sym, len: table[i]->len, result: buf); |
| 407 | strcpy(dest: (char *)table[i]->sym, src: buf); |
| 408 | printf(format: "\t/* %s */\n" , table[i]->sym); |
| 409 | } |
| 410 | printf(format: "\n" ); |
| 411 | |
| 412 | output_label(label: "kallsyms_markers" ); |
| 413 | for (i = 0; i < markers_cnt; i++) |
| 414 | printf(format: "\t.long\t%u\n" , markers[i]); |
| 415 | printf(format: "\n" ); |
| 416 | |
| 417 | free(ptr: markers); |
| 418 | |
| 419 | output_label(label: "kallsyms_token_table" ); |
| 420 | off = 0; |
| 421 | for (i = 0; i < 256; i++) { |
| 422 | best_idx[i] = off; |
| 423 | expand_symbol(data: best_table[i], len: best_table_len[i], result: buf); |
| 424 | printf(format: "\t.asciz\t\"%s\"\n" , buf); |
| 425 | off += strlen(s: buf) + 1; |
| 426 | } |
| 427 | printf(format: "\n" ); |
| 428 | |
| 429 | output_label(label: "kallsyms_token_index" ); |
| 430 | for (i = 0; i < 256; i++) |
| 431 | printf(format: "\t.short\t%d\n" , best_idx[i]); |
| 432 | printf(format: "\n" ); |
| 433 | |
| 434 | output_label(label: "kallsyms_offsets" ); |
| 435 | |
| 436 | for (i = 0; i < table_cnt; i++) { |
| 437 | /* |
| 438 | * Use the offset relative to the lowest value |
| 439 | * encountered of all relative symbols, and emit |
| 440 | * non-relocatable fixed offsets that will be fixed |
| 441 | * up at runtime. |
| 442 | */ |
| 443 | |
| 444 | long long offset; |
| 445 | |
| 446 | offset = table[i]->addr - relative_base; |
| 447 | if (offset < 0 || offset > UINT_MAX) { |
| 448 | fprintf(stderr, format: "kallsyms failure: " |
| 449 | "relative symbol value %#llx out of range\n" , |
| 450 | table[i]->addr); |
| 451 | exit(EXIT_FAILURE); |
| 452 | } |
| 453 | printf(format: "\t.long\t%#x\t/* %s */\n" , (int)offset, table[i]->sym); |
| 454 | } |
| 455 | printf(format: "\n" ); |
| 456 | |
| 457 | output_label(label: "kallsyms_relative_base" ); |
| 458 | /* Provide proper symbols relocatability by their '_text' relativeness. */ |
| 459 | if (_text <= relative_base) |
| 460 | printf(format: "\tPTR\t_text + %#llx\n" , relative_base - _text); |
| 461 | else |
| 462 | printf(format: "\tPTR\t_text - %#llx\n" , _text - relative_base); |
| 463 | printf(format: "\n" ); |
| 464 | |
| 465 | sort_symbols_by_name(); |
| 466 | output_label(label: "kallsyms_seqs_of_names" ); |
| 467 | for (i = 0; i < table_cnt; i++) |
| 468 | printf(format: "\t.byte 0x%02x, 0x%02x, 0x%02x\t/* %s */\n" , |
| 469 | (unsigned char)(table[i]->seq >> 16), |
| 470 | (unsigned char)(table[i]->seq >> 8), |
| 471 | (unsigned char)(table[i]->seq >> 0), |
| 472 | table[i]->sym); |
| 473 | printf(format: "\n" ); |
| 474 | } |
| 475 | |
| 476 | |
| 477 | /* table lookup compression functions */ |
| 478 | |
| 479 | /* count all the possible tokens in a symbol */ |
| 480 | static void learn_symbol(const unsigned char *symbol, int len) |
| 481 | { |
| 482 | int i; |
| 483 | |
| 484 | for (i = 0; i < len - 1; i++) |
| 485 | token_profit[ symbol[i] + (symbol[i + 1] << 8) ]++; |
| 486 | } |
| 487 | |
| 488 | /* decrease the count for all the possible tokens in a symbol */ |
| 489 | static void forget_symbol(const unsigned char *symbol, int len) |
| 490 | { |
| 491 | int i; |
| 492 | |
| 493 | for (i = 0; i < len - 1; i++) |
| 494 | token_profit[ symbol[i] + (symbol[i + 1] << 8) ]--; |
| 495 | } |
| 496 | |
| 497 | /* do the initial token count */ |
| 498 | static void build_initial_token_table(void) |
| 499 | { |
| 500 | unsigned int i; |
| 501 | |
| 502 | for (i = 0; i < table_cnt; i++) |
| 503 | learn_symbol(symbol: table[i]->sym, len: table[i]->len); |
| 504 | } |
| 505 | |
| 506 | static unsigned char *find_token(unsigned char *str, int len, |
| 507 | const unsigned char *token) |
| 508 | { |
| 509 | int i; |
| 510 | |
| 511 | for (i = 0; i < len - 1; i++) { |
| 512 | if (str[i] == token[0] && str[i+1] == token[1]) |
| 513 | return &str[i]; |
| 514 | } |
| 515 | return NULL; |
| 516 | } |
| 517 | |
| 518 | /* replace a given token in all the valid symbols. Use the sampled symbols |
| 519 | * to update the counts */ |
| 520 | static void compress_symbols(const unsigned char *str, int idx) |
| 521 | { |
| 522 | unsigned int i, len, size; |
| 523 | unsigned char *p1, *p2; |
| 524 | |
| 525 | for (i = 0; i < table_cnt; i++) { |
| 526 | |
| 527 | len = table[i]->len; |
| 528 | p1 = table[i]->sym; |
| 529 | |
| 530 | /* find the token on the symbol */ |
| 531 | p2 = find_token(str: p1, len, token: str); |
| 532 | if (!p2) continue; |
| 533 | |
| 534 | /* decrease the counts for this symbol's tokens */ |
| 535 | forget_symbol(symbol: table[i]->sym, len); |
| 536 | |
| 537 | size = len; |
| 538 | |
| 539 | do { |
| 540 | *p2 = idx; |
| 541 | p2++; |
| 542 | size -= (p2 - p1); |
| 543 | memmove(dest: p2, src: p2 + 1, n: size); |
| 544 | p1 = p2; |
| 545 | len--; |
| 546 | |
| 547 | if (size < 2) break; |
| 548 | |
| 549 | /* find the token on the symbol */ |
| 550 | p2 = find_token(str: p1, len: size, token: str); |
| 551 | |
| 552 | } while (p2); |
| 553 | |
| 554 | table[i]->len = len; |
| 555 | |
| 556 | /* increase the counts for this symbol's new tokens */ |
| 557 | learn_symbol(symbol: table[i]->sym, len); |
| 558 | } |
| 559 | } |
| 560 | |
| 561 | /* search the token with the maximum profit */ |
| 562 | static int find_best_token(void) |
| 563 | { |
| 564 | int i, best, bestprofit; |
| 565 | |
| 566 | bestprofit=-10000; |
| 567 | best = 0; |
| 568 | |
| 569 | for (i = 0; i < 0x10000; i++) { |
| 570 | if (token_profit[i] > bestprofit) { |
| 571 | best = i; |
| 572 | bestprofit = token_profit[i]; |
| 573 | } |
| 574 | } |
| 575 | return best; |
| 576 | } |
| 577 | |
| 578 | /* this is the core of the algorithm: calculate the "best" table */ |
| 579 | static void optimize_result(void) |
| 580 | { |
| 581 | int i, best; |
| 582 | |
| 583 | /* using the '\0' symbol last allows compress_symbols to use standard |
| 584 | * fast string functions */ |
| 585 | for (i = 255; i >= 0; i--) { |
| 586 | |
| 587 | /* if this table slot is empty (it is not used by an actual |
| 588 | * original char code */ |
| 589 | if (!best_table_len[i]) { |
| 590 | |
| 591 | /* find the token with the best profit value */ |
| 592 | best = find_best_token(); |
| 593 | if (token_profit[best] == 0) |
| 594 | break; |
| 595 | |
| 596 | /* place it in the "best" table */ |
| 597 | best_table_len[i] = 2; |
| 598 | best_table[i][0] = best & 0xFF; |
| 599 | best_table[i][1] = (best >> 8) & 0xFF; |
| 600 | |
| 601 | /* replace this token in all the valid symbols */ |
| 602 | compress_symbols(str: best_table[i], idx: i); |
| 603 | } |
| 604 | } |
| 605 | } |
| 606 | |
| 607 | /* start by placing the symbols that are actually used on the table */ |
| 608 | static void insert_real_symbols_in_table(void) |
| 609 | { |
| 610 | unsigned int i, j, c; |
| 611 | |
| 612 | for (i = 0; i < table_cnt; i++) { |
| 613 | for (j = 0; j < table[i]->len; j++) { |
| 614 | c = table[i]->sym[j]; |
| 615 | best_table[c][0]=c; |
| 616 | best_table_len[c]=1; |
| 617 | } |
| 618 | } |
| 619 | } |
| 620 | |
| 621 | static void optimize_token_table(void) |
| 622 | { |
| 623 | build_initial_token_table(); |
| 624 | |
| 625 | insert_real_symbols_in_table(); |
| 626 | |
| 627 | optimize_result(); |
| 628 | } |
| 629 | |
| 630 | /* guess for "linker script provide" symbol */ |
| 631 | static int may_be_linker_script_provide_symbol(const struct sym_entry *se) |
| 632 | { |
| 633 | const char *symbol = sym_name(s: se); |
| 634 | int len = se->len - 1; |
| 635 | |
| 636 | if (len < 8) |
| 637 | return 0; |
| 638 | |
| 639 | if (symbol[0] != '_' || symbol[1] != '_') |
| 640 | return 0; |
| 641 | |
| 642 | /* __start_XXXXX */ |
| 643 | if (!memcmp(s1: symbol + 2, s2: "start_" , n: 6)) |
| 644 | return 1; |
| 645 | |
| 646 | /* __stop_XXXXX */ |
| 647 | if (!memcmp(s1: symbol + 2, s2: "stop_" , n: 5)) |
| 648 | return 1; |
| 649 | |
| 650 | /* __end_XXXXX */ |
| 651 | if (!memcmp(s1: symbol + 2, s2: "end_" , n: 4)) |
| 652 | return 1; |
| 653 | |
| 654 | /* __XXXXX_start */ |
| 655 | if (!memcmp(s1: symbol + len - 6, s2: "_start" , n: 6)) |
| 656 | return 1; |
| 657 | |
| 658 | /* __XXXXX_end */ |
| 659 | if (!memcmp(s1: symbol + len - 4, s2: "_end" , n: 4)) |
| 660 | return 1; |
| 661 | |
| 662 | return 0; |
| 663 | } |
| 664 | |
| 665 | static int compare_symbols(const void *a, const void *b) |
| 666 | { |
| 667 | const struct sym_entry *sa = *(const struct sym_entry **)a; |
| 668 | const struct sym_entry *sb = *(const struct sym_entry **)b; |
| 669 | int wa, wb; |
| 670 | |
| 671 | /* sort by address first */ |
| 672 | if (sa->addr > sb->addr) |
| 673 | return 1; |
| 674 | if (sa->addr < sb->addr) |
| 675 | return -1; |
| 676 | |
| 677 | /* sort by "weakness" type */ |
| 678 | wa = (sa->sym[0] == 'w') || (sa->sym[0] == 'W'); |
| 679 | wb = (sb->sym[0] == 'w') || (sb->sym[0] == 'W'); |
| 680 | if (wa != wb) |
| 681 | return wa - wb; |
| 682 | |
| 683 | /* sort by "linker script provide" type */ |
| 684 | wa = may_be_linker_script_provide_symbol(se: sa); |
| 685 | wb = may_be_linker_script_provide_symbol(se: sb); |
| 686 | if (wa != wb) |
| 687 | return wa - wb; |
| 688 | |
| 689 | /* sort by the number of prefix underscores */ |
| 690 | wa = strspn(s: sym_name(s: sa), accept: "_" ); |
| 691 | wb = strspn(s: sym_name(s: sb), accept: "_" ); |
| 692 | if (wa != wb) |
| 693 | return wa - wb; |
| 694 | |
| 695 | /* sort by initial order, so that other symbols are left undisturbed */ |
| 696 | return sa->seq - sb->seq; |
| 697 | } |
| 698 | |
| 699 | static void sort_symbols(void) |
| 700 | { |
| 701 | qsort(base: table, nmemb: table_cnt, size: sizeof(table[0]), compar: compare_symbols); |
| 702 | } |
| 703 | |
| 704 | /* find the minimum non-absolute symbol address */ |
| 705 | static void record_relative_base(void) |
| 706 | { |
| 707 | /* |
| 708 | * The table is sorted by address. |
| 709 | * Take the first symbol value. |
| 710 | */ |
| 711 | if (table_cnt) |
| 712 | relative_base = table[0]->addr; |
| 713 | } |
| 714 | |
| 715 | int main(int argc, char **argv) |
| 716 | { |
| 717 | while (1) { |
| 718 | static const struct option long_options[] = { |
| 719 | {.name: "all-symbols" , no_argument, .flag: &all_symbols, .val: 1}, |
| 720 | {}, |
| 721 | }; |
| 722 | |
| 723 | int c = getopt_long(argc: argc, argv: argv, shortopts: "" , longopts: long_options, NULL); |
| 724 | |
| 725 | if (c == -1) |
| 726 | break; |
| 727 | if (c != 0) |
| 728 | usage(); |
| 729 | } |
| 730 | |
| 731 | if (optind >= argc) |
| 732 | usage(); |
| 733 | |
| 734 | read_map(in: argv[optind]); |
| 735 | shrink_table(); |
| 736 | sort_symbols(); |
| 737 | record_relative_base(); |
| 738 | optimize_token_table(); |
| 739 | write_src(); |
| 740 | |
| 741 | return 0; |
| 742 | } |
| 743 | |