Currently, expand_symbol() is called many times to get the uncompressed symbol names for sorting, and also for adding comments. With the output order shuffled in the previous commit, the symbol data are now written in the following order: (1) kallsyms_num_syms (2) kallsyms_names <-- need compressed names (3) kallsyms_markers (4) kallsyms_token_table (5) kallsyms_token_index (6) kallsyms_addressed / kallsyms_offsets <-- need uncompressed names (for commenting) (7) kallsyms_relative_base (8) kallsyms_seq_of_names <-- need uncompressed names (for sorting) The compressed names are only needed by (2). Call expand_symbol() between (2) and (3) to restore the original symbol names. This requires just one expand_symbol() call for each symbol. Call cleanup_symbol_name() between (7) and (8) instead of during sorting. It is allowed to overwrite the ->sym field because (8) just outputs the index instead of the name of each symbol. Again, this requires just one cleanup_symbol_name() call for each symbol. This refactoring makes it ~30% faster. [Before] $ time scripts/kallsyms --all-symbols --absolute-percpu --base-relative \ .tmp_vmlinux.kallsyms2.syms >/dev/null real 0m1.027s user 0m1.010s sys 0m0.016s [After] $ time scripts/kallsyms --all-symbols --absolute-percpu --base-relative \ .tmp_vmlinux.kallsyms2.syms >/dev/null real 0m0.717s user 0m0.717s sys 0m0.000s
Showing the changes to scripts (1 of 1 file). Show all
Unified · Split
| @@ -335,19 +335,10 @@ static int symbol_absolute(const struct sym_entry *s) · 334 unmodified lines | |||
| 335 | return s->percpu_absolute; | 335 | return s->percpu_absolute; |
| 336 | } | 336 | } |
| 337 | 337 | ||
| 338 | static char * s_name(char *buf) | ||
| 339 | { | ||
| 340 | /* Skip the symbol type */ | ||
| 341 | return buf + 1; | ||
| 342 | } | ||
| 343 | |||
| 344 | static void cleanup_symbol_name(char *s) | 338 | static void cleanup_symbol_name(char *s) |
| 345 | { | 339 | { |
| 346 | char *p; | 340 | char *p; |
| 347 | 341 | ||
| 348 | if (!lto_clang) | ||
| 349 | return; | ||
| 350 | |||
| 351 | /* | 342 | /* |
| 352 | * ASCII[.] = 2e | 343 | * ASCII[.] = 2e |
| 353 | * ASCII[0-9] = 30,39 | 344 | * ASCII[0-9] = 30,39 |
| @@ -366,16 +357,10 @@ static void cleanup_symbol_name(char *s) · 12 unmodified lines | |||
| 366 | static int compare_names(const void *a, const void *b) | 357 | static int compare_names(const void *a, const void *b) |
| 367 | { | 358 | { |
| 368 | int ret; | 359 | int ret; |
| 369 | char sa_namebuf[KSYM_NAME_LEN]; | ||
| 370 | char sb_namebuf[KSYM_NAME_LEN]; | ||
| 371 | const struct sym_entry *sa = *(const struct sym_entry **)a; | 360 | const struct sym_entry *sa = *(const struct sym_entry **)a; |
| 372 | const struct sym_entry *sb = *(const struct sym_entry **)b; | 361 | const struct sym_entry *sb = *(const struct sym_entry **)b; |
| 373 | 362 | ||
| 374 | expand_symbol(sa->sym, sa->len, sa_namebuf); | 363 | ret = strcmp(sym_name(sa), sym_name(sb)); |
| 375 | expand_symbol(sb->sym, sb->len, sb_namebuf); | ||
| 376 | cleanup_symbol_name(s_name(sa_namebuf)); | ||
| 377 | cleanup_symbol_name(s_name(sb_namebuf)); | ||
| 378 | ret = strcmp(s_name(sa_namebuf), s_name(sb_namebuf)); | ||
| 379 | if (!ret) { | 364 | if (!ret) { |
| 380 | if (sa->addr > sb->addr) | 365 | if (sa->addr > sb->addr) |
| 381 | return 1; | 366 | return 1; |
| @@ -464,6 +449,15 @@ static void write_src(void) · 82 unmodified lines | |||
| 464 | } | 449 | } |
| 465 | printf("\n"); | 450 | printf("\n"); |
| 466 | 451 | ||
| 452 | /* | ||
| 453 | * Now that we wrote out the compressed symbol names, restore the | ||
| 454 | * original names, which are needed in some of the later steps. | ||
| 455 | */ | ||
| 456 | for (i = 0; i < table_cnt; i++) { | ||
| 457 | expand_symbol(table[i]->sym, table[i]->len, buf); | ||
| 458 | strcpy((char *)table[i]->sym, buf); | ||
| 459 | } | ||
| 460 | |||
| 467 | output_label("kallsyms_markers"); | 461 | output_label("kallsyms_markers"); |
| 468 | for (i = 0; i < ((table_cnt + 255) >> 8); i++) | 462 | for (i = 0; i < ((table_cnt + 255) >> 8); i++) |
| 469 | printf("\t.long\t%u\n", markers[i]); | 463 | printf("\t.long\t%u\n", markers[i]); |
| @@ -520,8 +514,7 @@ static void write_src(void) · 50 unmodified lines | |||
| 520 | table[i]->addr); | 514 | table[i]->addr); |
| 521 | exit(EXIT_FAILURE); | 515 | exit(EXIT_FAILURE); |
| 522 | } | 516 | } |
| 523 | expand_symbol(table[i]->sym, table[i]->len, buf); | 517 | printf("\t.long\t%#x /* %s */\n", (int)offset, table[i]->sym); |
| 524 | printf("\t.long\t%#x /* %s */\n", (int)offset, buf); | ||
| 525 | } else if (!symbol_absolute(table[i])) { | 518 | } else if (!symbol_absolute(table[i])) { |
| 526 | output_address(table[i]->addr); | 519 | output_address(table[i]->addr); |
| 527 | } else { | 520 | } else { |
| @@ -536,6 +529,10 @@ static void write_src(void) · 8 unmodified lines | |||
| 536 | printf("\n"); | 529 | printf("\n"); |
| 537 | } | 530 | } |
| 538 | 531 | ||
| 532 | if (lto_clang) | ||
| 533 | for (i = 0; i < table_cnt; i++) | ||
| 534 | cleanup_symbol_name((char *)table[i]->sym); | ||
| 535 | |||
| 539 | sort_symbols_by_name(); | 536 | sort_symbols_by_name(); |
| 540 | output_label("kallsyms_seqs_of_names"); | 537 | output_label("kallsyms_seqs_of_names"); |
| 541 | for (i = 0; i < table_cnt; i++) | 538 | for (i = 0; i < table_cnt; i++) |