?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Improve texi2html -> makeinfo conversion
[tinycc.git] / tccelf.c
blob8af4bb63b3e5f69372d08095a8392d5b4933e80b
1 /*
2 * ELF file handling for TCC
3 *
4 * Copyright (c) 2001-2004 Fabrice Bellard
5 *
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
10 *
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 */
20
21 #include "tcc.h"
22
23 static int new_undef_sym = 0; /* Is there a new undefined sym since last new_undef_sym() */
24
25 ST_FUNC int put_elf_str(Section *s, const char *sym)
26 {
27 int offset, len;
28 char *ptr;
29
30 len = strlen(sym) + 1;
31 offset = s->data_offset;
32 ptr = section_ptr_add(s, len);
33 memcpy(ptr, sym, len);
34 return offset;
35 }
36
37 /* elf symbol hashing function */
38 static unsigned long elf_hash(const unsigned char *name)
39 {
40 unsigned long h = 0, g;
41
42 while (*name) {
43 h = (h << 4) + *name++;
44 g = h & 0xf0000000;
45 if (g)
46 h ^= g >> 24;
47 h &= ~g;
48 }
49 return h;
50 }
51
52 /* rebuild hash table of section s */
53 /* NOTE: we do factorize the hash table code to go faster */
54 static void rebuild_hash(Section *s, unsigned int nb_buckets)
55 {
56 ElfW(Sym) *sym;
57 int *ptr, *hash, nb_syms, sym_index, h;
58 char *strtab;
59
60 strtab = s->link->data;
61 nb_syms = s->data_offset / sizeof(ElfW(Sym));
62
63 s->hash->data_offset = 0;
64 ptr = section_ptr_add(s->hash, (2 + nb_buckets + nb_syms) * sizeof(int));
65 ptr[0] = nb_buckets;
66 ptr[1] = nb_syms;
67 ptr += 2;
68 hash = ptr;
69 memset(hash, 0, (nb_buckets + 1) * sizeof(int));
70 ptr += nb_buckets + 1;
71
72 sym = (ElfW(Sym) *)s->data + 1;
73 for(sym_index = 1; sym_index < nb_syms; sym_index++) {
74 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
75 h = elf_hash(strtab + sym->st_name) % nb_buckets;
76 *ptr = hash[h];
77 hash[h] = sym_index;
78 } else {
79 *ptr = 0;
80 }
81 ptr++;
82 sym++;
83 }
84 }
85
86 /* return the symbol number */
87 ST_FUNC int put_elf_sym(Section *s, addr_t value, unsigned long size,
88 int info, int other, int shndx, const char *name)
89 {
90 int name_offset, sym_index;
91 int nbuckets, h;
92 ElfW(Sym) *sym;
93 Section *hs;
94
95 sym = section_ptr_add(s, sizeof(ElfW(Sym)));
96 if (name)
97 name_offset = put_elf_str(s->link, name);
98 else
99 name_offset = 0;
100 /* XXX: endianness */
101 sym->st_name = name_offset;
102 sym->st_value = value;
103 sym->st_size = size;
104 sym->st_info = info;
105 sym->st_other = other;
106 sym->st_shndx = shndx;
107 sym_index = sym - (ElfW(Sym) *)s->data;
108 hs = s->hash;
109 if (hs) {
110 int *ptr, *base;
111 ptr = section_ptr_add(hs, sizeof(int));
112 base = (int *)hs->data;
113 /* only add global or weak symbols */
114 if (ELFW(ST_BIND)(info) != STB_LOCAL) {
115 /* add another hashing entry */
116 nbuckets = base[0];
117 h = elf_hash(name) % nbuckets;
118 *ptr = base[2 + h];
119 base[2 + h] = sym_index;
120 base[1]++;
121 /* we resize the hash table */
122 hs->nb_hashed_syms++;
123 if (hs->nb_hashed_syms > 2 * nbuckets) {
124 rebuild_hash(s, 2 * nbuckets);
125 }
126 } else {
127 *ptr = 0;
128 base[1]++;
129 }
130 }
131 return sym_index;
132 }
133
134 /* find global ELF symbol 'name' and return its index. Return 0 if not
135 found. */
136 ST_FUNC int find_elf_sym(Section *s, const char *name)
137 {
138 ElfW(Sym) *sym;
139 Section *hs;
140 int nbuckets, sym_index, h;
141 const char *name1;
142
143 hs = s->hash;
144 if (!hs)
145 return 0;
146 nbuckets = ((int *)hs->data)[0];
147 h = elf_hash(name) % nbuckets;
148 sym_index = ((int *)hs->data)[2 + h];
149 while (sym_index != 0) {
150 sym = &((ElfW(Sym) *)s->data)[sym_index];
151 name1 = s->link->data + sym->st_name;
152 if (!strcmp(name, name1))
153 return sym_index;
154 sym_index = ((int *)hs->data)[2 + nbuckets + sym_index];
155 }
156 return 0;
157 }
158
159 /* return elf symbol value, signal error if 'err' is nonzero */
160 ST_FUNC addr_t get_elf_sym_addr(TCCState *s, const char *name, int err)
161 {
162 int sym_index;
163 ElfW(Sym) *sym;
164
165 sym_index = find_elf_sym(s->symtab, name);
166 sym = &((ElfW(Sym) *)s->symtab->data)[sym_index];
167 if (!sym_index || sym->st_shndx == SHN_UNDEF) {
168 if (err)
169 tcc_error("%s not defined", name);
170 return 0;
171 }
172 return sym->st_value;
173 }
174
175 /* return elf symbol value */
176 LIBTCCAPI void *tcc_get_symbol(TCCState *s, const char *name)
177 {
178 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 0);
179 }
180
181 #ifdef TCC_IS_NATIVE
182 /* return elf symbol value or error */
183 ST_FUNC void* tcc_get_symbol_err(TCCState *s, const char *name)
184 {
185 return (void*)get_elf_sym_addr(s, name, 1);
186 }
187 #endif
188
189 /* add an elf symbol : check if it is already defined and patch
190 it. Return symbol index. NOTE that sh_num can be SHN_UNDEF. */
191 ST_FUNC int add_elf_sym(Section *s, addr_t value, unsigned long size,
192 int info, int other, int sh_num, const char *name)
193 {
194 ElfW(Sym) *esym;
195 int sym_bind, sym_index, sym_type, esym_bind;
196 unsigned char sym_vis, esym_vis, new_vis;
197
198 sym_bind = ELFW(ST_BIND)(info);
199 sym_type = ELFW(ST_TYPE)(info);
200 sym_vis = ELFW(ST_VISIBILITY)(other);
201
202 if (sym_bind != STB_LOCAL) {
203 /* we search global or weak symbols */
204 sym_index = find_elf_sym(s, name);
205 if (!sym_index)
206 goto do_def;
207 esym = &((ElfW(Sym) *)s->data)[sym_index];
208 if (esym->st_shndx != SHN_UNDEF) {
209 esym_bind = ELFW(ST_BIND)(esym->st_info);
210 /* propagate the most constraining visibility */
211 /* STV_DEFAULT(0)<STV_PROTECTED(3)<STV_HIDDEN(2)<STV_INTERNAL(1) */
212 esym_vis = ELFW(ST_VISIBILITY)(esym->st_other);
213 if (esym_vis == STV_DEFAULT) {
214 new_vis = sym_vis;
215 } else if (sym_vis == STV_DEFAULT) {
216 new_vis = esym_vis;
217 } else {
218 new_vis = (esym_vis < sym_vis) ? esym_vis : sym_vis;
219 }
220 esym->st_other = (esym->st_other & ~ELFW(ST_VISIBILITY)(-1))
221 | new_vis;
222 other = esym->st_other; /* in case we have to patch esym */
223 if (sh_num == SHN_UNDEF) {
224 /* ignore adding of undefined symbol if the
225 corresponding symbol is already defined */
226 } else if (sym_bind == STB_GLOBAL && esym_bind == STB_WEAK) {
227 /* global overrides weak, so patch */
228 goto do_patch;
229 } else if (sym_bind == STB_WEAK && esym_bind == STB_GLOBAL) {
230 /* weak is ignored if already global */
231 } else if (sym_bind == STB_WEAK && esym_bind == STB_WEAK) {
232 /* keep first-found weak definition, ignore subsequents */
233 } else if (sym_vis == STV_HIDDEN || sym_vis == STV_INTERNAL) {
234 /* ignore hidden symbols after */
235 } else if (esym->st_shndx == SHN_COMMON
236 && (sh_num < SHN_LORESERVE || sh_num == SHN_COMMON)) {
237 /* gr: Happens with 'tcc ... -static tcctest.c' on e.g. Ubuntu 6.01
238 No idea if this is the correct solution ... */
239 goto do_patch;
240 } else if (s == tcc_state->dynsymtab_section) {
241 /* we accept that two DLL define the same symbol */
242 } else {
243 #if 0
244 printf("new_bind=%x new_shndx=%x new_vis=%x old_bind=%x old_shndx=%x old_vis=%x\n",
245 sym_bind, sh_num, new_vis, esym_bind, esym->st_shndx, esym_vis);
246 #endif
247 tcc_error_noabort("'%s' defined twice", name);
248 }
249 } else {
250 do_patch:
251 esym->st_info = ELFW(ST_INFO)(sym_bind, sym_type);
252 esym->st_shndx = sh_num;
253 new_undef_sym = 1;
254 esym->st_value = value;
255 esym->st_size = size;
256 esym->st_other = other;
257 }
258 } else {
259 do_def:
260 sym_index = put_elf_sym(s, value, size,
261 ELFW(ST_INFO)(sym_bind, sym_type), other,
262 sh_num, name);
263 }
264 return sym_index;
265 }
266
267 /* put relocation */
268 ST_FUNC void put_elf_reloc(Section *symtab, Section *s, unsigned long offset,
269 int type, int symbol)
270 {
271 char buf[256];
272 Section *sr;
273 ElfW_Rel *rel;
274
275 sr = s->reloc;
276 if (!sr) {
277 /* if no relocation section, create it */
278 snprintf(buf, sizeof(buf), REL_SECTION_FMT, s->name);
279 /* if the symtab is allocated, then we consider the relocation
280 are also */
281 sr = new_section(tcc_state, buf, SHT_RELX, symtab->sh_flags);
282 sr->sh_entsize = sizeof(ElfW_Rel);
283 sr->link = symtab;
284 sr->sh_info = s->sh_num;
285 s->reloc = sr;
286 }
287 rel = section_ptr_add(sr, sizeof(ElfW_Rel));
288 rel->r_offset = offset;
289 rel->r_info = ELFW(R_INFO)(symbol, type);
290 #ifdef TCC_TARGET_X86_64
291 rel->r_addend = 0;
292 #endif
293 }
294
295 /* put stab debug information */
296
297 ST_FUNC void put_stabs(const char *str, int type, int other, int desc,
298 unsigned long value)
299 {
300 Stab_Sym *sym;
301
302 sym = section_ptr_add(stab_section, sizeof(Stab_Sym));
303 if (str) {
304 sym->n_strx = put_elf_str(stabstr_section, str);
305 } else {
306 sym->n_strx = 0;
307 }
308 sym->n_type = type;
309 sym->n_other = other;
310 sym->n_desc = desc;
311 sym->n_value = value;
312 }
313
314 ST_FUNC void put_stabs_r(const char *str, int type, int other, int desc,
315 unsigned long value, Section *sec, int sym_index)
316 {
317 put_stabs(str, type, other, desc, value);
318 put_elf_reloc(symtab_section, stab_section,
319 stab_section->data_offset - sizeof(unsigned int),
320 R_DATA_32, sym_index);
321 }
322
323 ST_FUNC void put_stabn(int type, int other, int desc, int value)
324 {
325 put_stabs(NULL, type, other, desc, value);
326 }
327
328 ST_FUNC void put_stabd(int type, int other, int desc)
329 {
330 put_stabs(NULL, type, other, desc, 0);
331 }
332
333 /* In an ELF file symbol table, the local symbols must appear below
334 the global and weak ones. Since TCC cannot sort it while generating
335 the code, we must do it after. All the relocation tables are also
336 modified to take into account the symbol table sorting */
337 static void sort_syms(TCCState *s1, Section *s)
338 {
339 int *old_to_new_syms;
340 ElfW(Sym) *new_syms;
341 int nb_syms, i;
342 ElfW(Sym) *p, *q;
343 ElfW_Rel *rel, *rel_end;
344 Section *sr;
345 int type, sym_index;
346
347 nb_syms = s->data_offset / sizeof(ElfW(Sym));
348 new_syms = tcc_malloc(nb_syms * sizeof(ElfW(Sym)));
349 old_to_new_syms = tcc_malloc(nb_syms * sizeof(int));
350
351 /* first pass for local symbols */
352 p = (ElfW(Sym) *)s->data;
353 q = new_syms;
354 for(i = 0; i < nb_syms; i++) {
355 if (ELFW(ST_BIND)(p->st_info) == STB_LOCAL) {
356 old_to_new_syms[i] = q - new_syms;
357 *q++ = *p;
358 }
359 p++;
360 }
361 /* save the number of local symbols in section header */
362 s->sh_info = q - new_syms;
363
364 /* then second pass for non local symbols */
365 p = (ElfW(Sym) *)s->data;
366 for(i = 0; i < nb_syms; i++) {
367 if (ELFW(ST_BIND)(p->st_info) != STB_LOCAL) {
368 old_to_new_syms[i] = q - new_syms;
369 *q++ = *p;
370 }
371 p++;
372 }
373
374 /* we copy the new symbols to the old */
375 memcpy(s->data, new_syms, nb_syms * sizeof(ElfW(Sym)));
376 tcc_free(new_syms);
377
378 /* now we modify all the relocations */
379 for(i = 1; i < s1->nb_sections; i++) {
380 sr = s1->sections[i];
381 if (sr->sh_type == SHT_RELX && sr->link == s) {
382 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
383 for(rel = (ElfW_Rel *)sr->data;
384 rel < rel_end;
385 rel++) {
386 sym_index = ELFW(R_SYM)(rel->r_info);
387 type = ELFW(R_TYPE)(rel->r_info);
388 sym_index = old_to_new_syms[sym_index];
389 rel->r_info = ELFW(R_INFO)(sym_index, type);
390 }
391 }
392 }
393
394 tcc_free(old_to_new_syms);
395 }
396
397 /* relocate common symbols in the .bss section */
398 ST_FUNC void relocate_common_syms(void)
399 {
400 ElfW(Sym) *sym, *sym_end;
401 unsigned long offset, align;
402
403 sym_end = (ElfW(Sym) *)(symtab_section->data + symtab_section->data_offset);
404 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
405 sym < sym_end;
406 sym++) {
407 if (sym->st_shndx == SHN_COMMON) {
408 /* align symbol */
409 align = sym->st_value;
410 offset = bss_section->data_offset;
411 offset = (offset + align - 1) & -align;
412 sym->st_value = offset;
413 sym->st_shndx = bss_section->sh_num;
414 offset += sym->st_size;
415 bss_section->data_offset = offset;
416 }
417 }
418 }
419
420 /* relocate symbol table, resolve undefined symbols if do_resolve is
421 true and output error if undefined symbol. */
422 ST_FUNC void relocate_syms(TCCState *s1, int do_resolve)
423 {
424 ElfW(Sym) *sym, *esym, *sym_end;
425 int sym_bind, sh_num, sym_index;
426 const char *name;
427
428 sym_end = (ElfW(Sym) *)(symtab_section->data + symtab_section->data_offset);
429 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
430 sym < sym_end;
431 sym++) {
432 sh_num = sym->st_shndx;
433 if (sh_num == SHN_UNDEF) {
434 name = strtab_section->data + sym->st_name;
435 if (do_resolve) {
436 #if defined TCC_IS_NATIVE && !defined _WIN32
437 void *addr;
438 name = symtab_section->link->data + sym->st_name;
439 addr = resolve_sym(s1, name);
440 if (addr) {
441 sym->st_value = (addr_t)addr;
442 goto found;
443 }
444 #endif
445 } else if (s1->dynsym) {
446 /* if dynamic symbol exist, then use it */
447 sym_index = find_elf_sym(s1->dynsym, name);
448 if (sym_index) {
449 esym = &((ElfW(Sym) *)s1->dynsym->data)[sym_index];
450 sym->st_value = esym->st_value;
451 goto found;
452 }
453 }
454 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
455 it */
456 if (!strcmp(name, "_fp_hw"))
457 goto found;
458 /* only weak symbols are accepted to be undefined. Their
459 value is zero */
460 sym_bind = ELFW(ST_BIND)(sym->st_info);
461 if (sym_bind == STB_WEAK) {
462 sym->st_value = 0;
463 } else {
464 tcc_error_noabort("undefined symbol '%s'", name);
465 }
466 } else if (sh_num < SHN_LORESERVE) {
467 /* add section base */
468 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
469 }
470 found: ;
471 }
472 }
473
474 #ifdef TCC_HAS_RUNTIME_PLTGOT
475 #ifdef TCC_TARGET_X86_64
476 #define JMP_TABLE_ENTRY_SIZE 14
477 static addr_t add_jmp_table(TCCState *s1, addr_t val)
478 {
479 char *p = s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset;
480 s1->runtime_plt_and_got_offset += JMP_TABLE_ENTRY_SIZE;
481 /* jmp *0x0(%rip) */
482 p[0] = 0xff;
483 p[1] = 0x25;
484 *(int *)(p + 2) = 0;
485 *(addr_t *)(p + 6) = val;
486 return (addr_t)p;
487 }
488
489 static addr_t add_got_table(TCCState *s1, addr_t val)
490 {
491 addr_t *p = (addr_t *)(s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset);
492 s1->runtime_plt_and_got_offset += sizeof(addr_t);
493 *p = val;
494 return (addr_t)p;
495 }
496 #elif defined TCC_TARGET_ARM
497 #define JMP_TABLE_ENTRY_SIZE 8
498 static addr_t add_jmp_table(TCCState *s1, int val)
499 {
500 uint32_t *p = (uint32_t *)(s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset);
501 s1->runtime_plt_and_got_offset += JMP_TABLE_ENTRY_SIZE;
502 /* ldr pc, [pc, #-4] */
503 p[0] = 0xE51FF004;
504 p[1] = val;
505 return (addr_t)p;
506 }
507 #endif
508 #endif /* def TCC_HAS_RUNTIME_PLTGOT */
509
510 /* relocate a given section (CPU dependent) */
511 ST_FUNC void relocate_section(TCCState *s1, Section *s)
512 {
513 Section *sr;
514 ElfW_Rel *rel, *rel_end, *qrel;
515 ElfW(Sym) *sym;
516 int type, sym_index;
517 unsigned char *ptr;
518 addr_t val, addr;
519 #if defined TCC_TARGET_I386 || defined TCC_TARGET_X86_64
520 int esym_index;
521 #endif
522
523 sr = s->reloc;
524 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
525 qrel = (ElfW_Rel *)sr->data;
526 for(rel = qrel;
527 rel < rel_end;
528 rel++) {
529 ptr = s->data + rel->r_offset;
530
531 sym_index = ELFW(R_SYM)(rel->r_info);
532 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
533 val = sym->st_value;
534 #ifdef TCC_TARGET_X86_64
535 val += rel->r_addend;
536 #endif
537 type = ELFW(R_TYPE)(rel->r_info);
538 addr = s->sh_addr + rel->r_offset;
539
540 /* CPU specific */
541 switch(type) {
542 #if defined(TCC_TARGET_I386)
543 case R_386_32:
544 if (s1->output_type == TCC_OUTPUT_DLL) {
545 esym_index = s1->symtab_to_dynsym[sym_index];
546 qrel->r_offset = rel->r_offset;
547 if (esym_index) {
548 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_32);
549 qrel++;
550 break;
551 } else {
552 qrel->r_info = ELFW(R_INFO)(0, R_386_RELATIVE);
553 qrel++;
554 }
555 }
556 *(int *)ptr += val;
557 break;
558 case R_386_PC32:
559 if (s1->output_type == TCC_OUTPUT_DLL) {
560 /* DLL relocation */
561 esym_index = s1->symtab_to_dynsym[sym_index];
562 if (esym_index) {
563 qrel->r_offset = rel->r_offset;
564 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_PC32);
565 qrel++;
566 break;
567 }
568 }
569 *(int *)ptr += val - addr;
570 break;
571 case R_386_PLT32:
572 *(int *)ptr += val - addr;
573 break;
574 case R_386_GLOB_DAT:
575 case R_386_JMP_SLOT:
576 *(int *)ptr = val;
577 break;
578 case R_386_GOTPC:
579 *(int *)ptr += s1->got->sh_addr - addr;
580 break;
581 case R_386_GOTOFF:
582 *(int *)ptr += val - s1->got->sh_addr;
583 break;
584 case R_386_GOT32:
585 /* we load the got offset */
586 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
587 break;
588 case R_386_16:
589 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY) {
590 output_file:
591 tcc_error("can only produce 16-bit binary files");
592 }
593 *(short *)ptr += val;
594 break;
595 case R_386_PC16:
596 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY)
597 goto output_file;
598 *(short *)ptr += val - addr;
599 break;
600 #elif defined(TCC_TARGET_ARM)
601 case R_ARM_PC24:
602 case R_ARM_CALL:
603 case R_ARM_JUMP24:
604 case R_ARM_PLT32:
605 {
606 int x, is_thumb, is_call, h, blx_avail;
607 x = (*(int *)ptr)&0xffffff;
608 (*(int *)ptr) &= 0xff000000;
609 if (x & 0x800000)
610 x -= 0x1000000;
611 x <<= 2;
612 blx_avail = (TCC_ARM_VERSION >= 5);
613 is_thumb = val & 1;
614 is_call = (type == R_ARM_CALL);
615 x += val - addr;
616 h = x & 2;
617 #ifdef TCC_HAS_RUNTIME_PLTGOT
618 if (s1->output_type == TCC_OUTPUT_MEMORY) {
619 if ((x & 3) || x >= 0x2000000 || x < -0x2000000)
620 if (!(x & 3) || !blx_avail || !is_call) {
621 x += add_jmp_table(s1, val) - val; /* add veneer */
622 is_thumb = 0; /* Veneer uses ARM instructions */
623 }
624 }
625 #endif
626 if ((x & 3) || x >= 0x2000000 || x < -0x2000000)
627 if (!(x & 3) || !blx_avail || !is_call)
628 tcc_error("can't relocate value at %x",addr);
629 x >>= 2;
630 x &= 0xffffff;
631 /* Only reached if blx is avail and it is a call */
632 if (is_thumb) {
633 x |= h << 24;
634 (*(int *)ptr) = 0xfa << 24; /* bl -> blx */
635 }
636 (*(int *)ptr) |= x;
637 }
638 break;
639 /* Since these relocations only concern Thumb-2 and blx instruction was
640 introduced before Thumb-2, we can assume blx is available and not
641 guard its use */
642 case R_ARM_THM_CALL:
643 case R_ARM_THM_JUMP24:
644 {
645 int x, hi, lo, s, j1, j2, i1, i2, imm10, imm11;
646 int to_thumb, is_call, to_plt, blx_bit = 1 << 12;
647 Section *plt;
648
649 /* weak reference */
650 if (sym->st_shndx == SHN_UNDEF &&
651 ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
652 break;
653
654 /* Get initial offset */
655 hi = (*(uint16_t *)ptr);
656 lo = (*(uint16_t *)(ptr+2));
657 s = (hi >> 10) & 1;
658 j1 = (lo >> 13) & 1;
659 j2 = (lo >> 11) & 1;
660 i1 = (j1 ^ s) ^ 1;
661 i2 = (j2 ^ s) ^ 1;
662 imm10 = hi & 0x3ff;
663 imm11 = lo & 0x7ff;
664 x = (s << 24) | (i1 << 23) | (i2 << 22) |
665 (imm10 << 12) | (imm11 << 1);
666 if (x & 0x01000000)
667 x -= 0x02000000;
668
669 /* Relocation infos */
670 to_thumb = val & 1;
671 plt = s1->plt;
672 to_plt = (val >= plt->sh_addr) &&
673 (val < plt->sh_addr + plt->data_offset);
674 is_call = (type == R_ARM_THM_CALL);
675
676 /* Compute final offset */
677 if (to_plt && !is_call) /* Point to 1st instr of Thumb stub */
678 x -= 4;
679 x += val - addr;
680 if (!to_thumb && is_call) {
681 blx_bit = 0; /* bl -> blx */
682 x = (x + 3) & -4; /* Compute offset from aligned PC */
683 }
684
685 /* Check that relocation is possible
686 * offset must not be out of range
687 * if target is to be entered in arm mode:
688 - bit 1 must not set
689 - instruction must be a call (bl) or a jump to PLT */
690 if (!to_thumb || x >= 0x1000000 || x < -0x1000000)
691 if (to_thumb || (val & 2) || (!is_call && !to_plt))
692 tcc_error("can't relocate value at %x",addr);
693
694 /* Compute and store final offset */
695 s = (x >> 24) & 1;
696 i1 = (x >> 23) & 1;
697 i2 = (x >> 22) & 1;
698 j1 = s ^ (i1 ^ 1);
699 j2 = s ^ (i2 ^ 1);
700 imm10 = (x >> 12) & 0x3ff;
701 imm11 = (x >> 1) & 0x7ff;
702 (*(uint16_t *)ptr) = (uint16_t) ((hi & 0xf800) |
703 (s << 10) | imm10);
704 (*(uint16_t *)(ptr+2)) = (uint16_t) ((lo & 0xc000) |
705 (j1 << 13) | blx_bit | (j2 << 11) |
706 imm11);
707 }
708 break;
709 case R_ARM_MOVT_ABS:
710 case R_ARM_MOVW_ABS_NC:
711 {
712 int x, imm4, imm12;
713 if (type == R_ARM_MOVT_ABS)
714 val >>= 16;
715 imm12 = val & 0xfff;
716 imm4 = (val >> 12) & 0xf;
717 x = (imm4 << 16) | imm12;
718 if (type == R_ARM_THM_MOVT_ABS)
719 *(int *)ptr |= x;
720 else
721 *(int *)ptr += x;
722 }
723 break;
724 case R_ARM_THM_MOVT_ABS:
725 case R_ARM_THM_MOVW_ABS_NC:
726 {
727 int x, i, imm4, imm3, imm8;
728 if (type == R_ARM_THM_MOVT_ABS)
729 val >>= 16;
730 imm8 = val & 0xff;
731 imm3 = (val >> 8) & 0x7;
732 i = (val >> 11) & 1;
733 imm4 = (val >> 12) & 0xf;
734 x = (imm3 << 28) | (imm8 << 16) | (i << 10) | imm4;
735 if (type == R_ARM_THM_MOVT_ABS)
736 *(int *)ptr |= x;
737 else
738 *(int *)ptr += x;
739 }
740 break;
741 case R_ARM_PREL31:
742 {
743 int x;
744 x = (*(int *)ptr) & 0x7fffffff;
745 (*(int *)ptr) &= 0x80000000;
746 x = (x * 2) / 2;
747 x += val - addr;
748 if((x^(x>>1))&0x40000000)
749 tcc_error("can't relocate value at %x",addr);
750 (*(int *)ptr) |= x & 0x7fffffff;
751 }
752 case R_ARM_ABS32:
753 *(int *)ptr += val;
754 break;
755 case R_ARM_REL32:
756 *(int *)ptr += val - addr;
757 break;
758 case R_ARM_BASE_PREL:
759 *(int *)ptr += s1->got->sh_addr - addr;
760 break;
761 case R_ARM_GOTOFF32:
762 *(int *)ptr += val - s1->got->sh_addr;
763 break;
764 case R_ARM_GOT_BREL:
765 /* we load the got offset */
766 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
767 break;
768 case R_ARM_COPY:
769 break;
770 case R_ARM_V4BX:
771 /* trade Thumb support for ARMv4 support */
772 if ((0x0ffffff0 & *(int*)ptr) == 0x012FFF10)
773 *(int*)ptr ^= 0xE12FFF10 ^ 0xE1A0F000; /* BX Rm -> MOV PC, Rm */
774 break;
775 default:
776 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
777 type, (unsigned)addr, ptr, (unsigned)val);
778 break;
779 #elif defined(TCC_TARGET_C67)
780 case R_C60_32:
781 *(int *)ptr += val;
782 break;
783 case R_C60LO16:
784 {
785 uint32_t orig;
786
787 /* put the low 16 bits of the absolute address */
788 // add to what is already there
789
790 orig = ((*(int *)(ptr )) >> 7) & 0xffff;
791 orig |= (((*(int *)(ptr+4)) >> 7) & 0xffff) << 16;
792
793 //patch both at once - assumes always in pairs Low - High
794
795 *(int *) ptr = (*(int *) ptr & (~(0xffff << 7)) ) | (((val+orig) & 0xffff) << 7);
796 *(int *)(ptr+4) = (*(int *)(ptr+4) & (~(0xffff << 7)) ) | ((((val+orig)>>16) & 0xffff) << 7);
797 }
798 break;
799 case R_C60HI16:
800 break;
801 default:
802 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
803 type, (unsigned)addr, ptr, (unsigned)val);
804 break;
805 #elif defined(TCC_TARGET_X86_64)
806 case R_X86_64_64:
807 if (s1->output_type == TCC_OUTPUT_DLL) {
808 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
809 qrel->r_addend = *(long long *)ptr + val;
810 qrel++;
811 }
812 *(long long *)ptr += val;
813 break;
814 case R_X86_64_32:
815 case R_X86_64_32S:
816 if (s1->output_type == TCC_OUTPUT_DLL) {
817 /* XXX: this logic may depend on TCC's codegen
818 now TCC uses R_X86_64_32 even for a 64bit pointer */
819 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
820 qrel->r_addend = *(int *)ptr + val;
821 qrel++;
822 }
823 *(int *)ptr += val;
824 break;
825
826 case R_X86_64_PC32:
827 if (s1->output_type == TCC_OUTPUT_DLL) {
828 /* DLL relocation */
829 esym_index = s1->symtab_to_dynsym[sym_index];
830 if (esym_index) {
831 qrel->r_offset = rel->r_offset;
832 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_PC32);
833 qrel->r_addend = *(int *)ptr;
834 qrel++;
835 break;
836 }
837 }
838 /* fall through */
839 case R_X86_64_PLT32: {
840 long long diff;
841 diff = (long long)val - addr;
842 if (diff <= -2147483647 || diff > 2147483647) {
843 #ifdef TCC_HAS_RUNTIME_PLTGOT
844 /* XXX: naive support for over 32bit jump */
845 if (s1->output_type == TCC_OUTPUT_MEMORY) {
846 val = (add_jmp_table(s1, val - rel->r_addend) +
847 rel->r_addend);
848 diff = val - addr;
849 }
850 #endif
851 if (diff <= -2147483647 || diff > 2147483647) {
852 tcc_error("internal error: relocation failed");
853 }
854 }
855 *(int *)ptr += diff;
856 }
857 break;
858 case R_X86_64_GLOB_DAT:
859 case R_X86_64_JUMP_SLOT:
860 /* They don't need addend */
861 *(int *)ptr = val - rel->r_addend;
862 break;
863 case R_X86_64_GOTPCREL:
864 #ifdef TCC_HAS_RUNTIME_PLTGOT
865 if (s1->output_type == TCC_OUTPUT_MEMORY) {
866 val = add_got_table(s1, val - rel->r_addend) + rel->r_addend;
867 *(int *)ptr += val - addr;
868 break;
869 }
870 #endif
871 *(int *)ptr += (s1->got->sh_addr - addr +
872 s1->sym_attrs[sym_index].got_offset - 4);
873 break;
874 case R_X86_64_GOTTPOFF:
875 *(int *)ptr += val - s1->got->sh_addr;
876 break;
877 case R_X86_64_GOT32:
878 /* we load the got offset */
879 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
880 break;
881 #else
882 #error unsupported processor
883 #endif
884 }
885 }
886 /* if the relocation is allocated, we change its symbol table */
887 if (sr->sh_flags & SHF_ALLOC)
888 sr->link = s1->dynsym;
889 }
890
891 /* relocate relocation table in 'sr' */
892 static void relocate_rel(TCCState *s1, Section *sr)
893 {
894 Section *s;
895 ElfW_Rel *rel, *rel_end;
896
897 s = s1->sections[sr->sh_info];
898 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
899 for(rel = (ElfW_Rel *)sr->data;
900 rel < rel_end;
901 rel++) {
902 rel->r_offset += s->sh_addr;
903 }
904 }
905
906 /* count the number of dynamic relocations so that we can reserve
907 their space */
908 static int prepare_dynamic_rel(TCCState *s1, Section *sr)
909 {
910 ElfW_Rel *rel, *rel_end;
911 int sym_index, esym_index, type, count;
912
913 count = 0;
914 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
915 for(rel = (ElfW_Rel *)sr->data; rel < rel_end; rel++) {
916 sym_index = ELFW(R_SYM)(rel->r_info);
917 type = ELFW(R_TYPE)(rel->r_info);
918 switch(type) {
919 #if defined(TCC_TARGET_I386)
920 case R_386_32:
921 #elif defined(TCC_TARGET_X86_64)
922 case R_X86_64_32:
923 case R_X86_64_32S:
924 case R_X86_64_64:
925 #endif
926 count++;
927 break;
928 #if defined(TCC_TARGET_I386)
929 case R_386_PC32:
930 #elif defined(TCC_TARGET_X86_64)
931 case R_X86_64_PC32:
932 #endif
933 esym_index = s1->symtab_to_dynsym[sym_index];
934 if (esym_index)
935 count++;
936 break;
937 default:
938 break;
939 }
940 }
941 if (count) {
942 /* allocate the section */
943 sr->sh_flags |= SHF_ALLOC;
944 sr->sh_size = count * sizeof(ElfW_Rel);
945 }
946 return count;
947 }
948
949 static struct sym_attr *alloc_sym_attr(TCCState *s1, int index)
950 {
951 int n;
952 struct sym_attr *tab;
953
954 if (index >= s1->nb_sym_attrs) {
955 /* find immediately bigger power of 2 and reallocate array */
956 n = 1;
957 while (index >= n)
958 n *= 2;
959 tab = tcc_realloc(s1->sym_attrs, n * sizeof(*s1->sym_attrs));
960 s1->sym_attrs = tab;
961 memset(s1->sym_attrs + s1->nb_sym_attrs, 0,
962 (n - s1->nb_sym_attrs) * sizeof(*s1->sym_attrs));
963 s1->nb_sym_attrs = n;
964 }
965 return &s1->sym_attrs[index];
966 }
967
968 /* XXX: suppress that */
969 static void put32(unsigned char *p, uint32_t val)
970 {
971 p[0] = val;
972 p[1] = val >> 8;
973 p[2] = val >> 16;
974 p[3] = val >> 24;
975 }
976
977 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_ARM) || \
978 defined(TCC_TARGET_X86_64)
979 static uint32_t get32(unsigned char *p)
980 {
981 return p[0] | (p[1] << 8) | (p[2] << 16) | (p[3] << 24);
982 }
983 #endif
984
985 static void build_got(TCCState *s1)
986 {
987 unsigned char *ptr;
988
989 /* if no got, then create it */
990 s1->got = new_section(s1, ".got", SHT_PROGBITS, SHF_ALLOC | SHF_WRITE);
991 s1->got->sh_entsize = 4;
992 add_elf_sym(symtab_section, 0, 4, ELFW(ST_INFO)(STB_GLOBAL, STT_OBJECT),
993 0, s1->got->sh_num, "_GLOBAL_OFFSET_TABLE_");
994 ptr = section_ptr_add(s1->got, 3 * PTR_SIZE);
995 #if PTR_SIZE == 4
996 /* keep space for _DYNAMIC pointer, if present */
997 put32(ptr, 0);
998 /* two dummy got entries */
999 put32(ptr + 4, 0);
1000 put32(ptr + 8, 0);
1001 #else
1002 /* keep space for _DYNAMIC pointer, if present */
1003 put32(ptr, 0);
1004 put32(ptr + 4, 0);
1005 /* two dummy got entries */
1006 put32(ptr + 8, 0);
1007 put32(ptr + 12, 0);
1008 put32(ptr + 16, 0);
1009 put32(ptr + 20, 0);
1010 #endif
1011 }
1012
1013 /* put a got entry corresponding to a symbol in symtab_section. 'size'
1014 and 'info' can be modifed if more precise info comes from the DLL */
1015 static void put_got_entry(TCCState *s1,
1016 int reloc_type, unsigned long size, int info,
1017 int sym_index)
1018 {
1019 int index;
1020 const char *name;
1021 ElfW(Sym) *sym;
1022 unsigned long offset;
1023 int *ptr;
1024
1025 if (!s1->got)
1026 build_got(s1);
1027
1028 /* if a got entry already exists for that symbol, no need to add one */
1029 if (sym_index < s1->nb_sym_attrs &&
1030 s1->sym_attrs[sym_index].got_offset)
1031 return;
1032
1033 alloc_sym_attr(s1, sym_index)->got_offset = s1->got->data_offset;
1034
1035 if (s1->dynsym) {
1036 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1037 name = symtab_section->link->data + sym->st_name;
1038 offset = sym->st_value;
1039 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_X86_64)
1040 if (reloc_type ==
1041 #ifdef TCC_TARGET_X86_64
1042 R_X86_64_JUMP_SLOT
1043 #else
1044 R_386_JMP_SLOT
1045 #endif
1046 ) {
1047 Section *plt;
1048 uint8_t *p;
1049 int modrm;
1050
1051 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1052 modrm = 0x25;
1053 #else
1054 /* if we build a DLL, we add a %ebx offset */
1055 if (s1->output_type == TCC_OUTPUT_DLL)
1056 modrm = 0xa3;
1057 else
1058 modrm = 0x25;
1059 #endif
1060
1061 /* add a PLT entry */
1062 plt = s1->plt;
1063 if (plt->data_offset == 0) {
1064 /* first plt entry */
1065 p = section_ptr_add(plt, 16);
1066 p[0] = 0xff; /* pushl got + PTR_SIZE */
1067 p[1] = modrm + 0x10;
1068 put32(p + 2, PTR_SIZE);
1069 p[6] = 0xff; /* jmp *(got + PTR_SIZE * 2) */
1070 p[7] = modrm;
1071 put32(p + 8, PTR_SIZE * 2);
1072 }
1073
1074 p = section_ptr_add(plt, 16);
1075 p[0] = 0xff; /* jmp *(got + x) */
1076 p[1] = modrm;
1077 put32(p + 2, s1->got->data_offset);
1078 p[6] = 0x68; /* push $xxx */
1079 put32(p + 7, (plt->data_offset - 32) >> 1);
1080 p[11] = 0xe9; /* jmp plt_start */
1081 put32(p + 12, -(plt->data_offset));
1082
1083 /* the symbol is modified so that it will be relocated to
1084 the PLT */
1085 #if !defined(TCC_OUTPUT_DLL_WITH_PLT)
1086 if (s1->output_type == TCC_OUTPUT_EXE)
1087 #endif
1088 offset = plt->data_offset - 16;
1089 }
1090 #elif defined(TCC_TARGET_ARM)
1091 if (reloc_type == R_ARM_JUMP_SLOT) {
1092 Section *plt;
1093 uint8_t *p;
1094
1095 /* if we build a DLL, we add a %ebx offset */
1096 if (s1->output_type == TCC_OUTPUT_DLL)
1097 tcc_error("DLLs unimplemented!");
1098
1099 /* add a PLT entry */
1100 plt = s1->plt;
1101 if (plt->data_offset == 0) {
1102 /* first plt entry */
1103 p = section_ptr_add(plt, 16);
1104 put32(p , 0xe52de004);
1105 put32(p + 4, 0xe59fe010);
1106 put32(p + 8, 0xe08fe00e);
1107 put32(p + 12, 0xe5bef008);
1108 }
1109
1110 if (s1->sym_attrs[sym_index].plt_thumb_stub) {
1111 p = section_ptr_add(plt, 20);
1112 put32(p , 0x4778); // bx pc
1113 put32(p+2, 0x46c0); // nop
1114 p += 4;
1115 } else
1116 p = section_ptr_add(plt, 16);
1117 put32(p , 0xe59fc004); // ldr ip, [pc, #4] // offset in GOT
1118 put32(p+4, 0xe08fc00c); // add ip, pc, ip // absolute address or offset
1119 put32(p+8, 0xe59cf000); // ldr pc, [ip] // load absolute address or load offset
1120 put32(p+12, s1->got->data_offset);
1121
1122 /* the symbol is modified so that it will be relocated to
1123 the PLT */
1124 if (s1->output_type == TCC_OUTPUT_EXE)
1125 offset = plt->data_offset - 16;
1126 }
1127 #elif defined(TCC_TARGET_C67)
1128 tcc_error("C67 got not implemented");
1129 #else
1130 #error unsupported CPU
1131 #endif
1132 index = put_elf_sym(s1->dynsym, offset,
1133 size, info, 0, sym->st_shndx, name);
1134 /* put a got entry */
1135 put_elf_reloc(s1->dynsym, s1->got,
1136 s1->got->data_offset,
1137 reloc_type, index);
1138 }
1139 ptr = section_ptr_add(s1->got, PTR_SIZE);
1140 *ptr = 0;
1141 }
1142
1143 /* build GOT and PLT entries */
1144 ST_FUNC void build_got_entries(TCCState *s1)
1145 {
1146 Section *s;
1147 ElfW_Rel *rel, *rel_end;
1148 ElfW(Sym) *sym;
1149 int i, type, reloc_type, sym_index;
1150
1151 for(i = 1; i < s1->nb_sections; i++) {
1152 s = s1->sections[i];
1153 if (s->sh_type != SHT_RELX)
1154 continue;
1155 /* no need to handle got relocations */
1156 if (s->link != symtab_section)
1157 continue;
1158 rel_end = (ElfW_Rel *)(s->data + s->data_offset);
1159 for(rel = (ElfW_Rel *)s->data;
1160 rel < rel_end;
1161 rel++) {
1162 type = ELFW(R_TYPE)(rel->r_info);
1163 switch(type) {
1164 #if defined(TCC_TARGET_I386)
1165 case R_386_GOT32:
1166 case R_386_GOTOFF:
1167 case R_386_GOTPC:
1168 case R_386_PLT32:
1169 if (!s1->got)
1170 build_got(s1);
1171 if (type == R_386_GOT32 || type == R_386_PLT32) {
1172 sym_index = ELFW(R_SYM)(rel->r_info);
1173 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1174 /* look at the symbol got offset. If none, then add one */
1175 if (type == R_386_GOT32)
1176 reloc_type = R_386_GLOB_DAT;
1177 else
1178 reloc_type = R_386_JMP_SLOT;
1179 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1180 sym_index);
1181 }
1182 break;
1183 #elif defined(TCC_TARGET_ARM)
1184 case R_ARM_GOT_BREL:
1185 case R_ARM_GOTOFF32:
1186 case R_ARM_BASE_PREL:
1187 case R_ARM_PLT32:
1188 if (!s1->got)
1189 build_got(s1);
1190 if (type == R_ARM_GOT_BREL || type == R_ARM_PLT32) {
1191 sym_index = ELFW(R_SYM)(rel->r_info);
1192 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1193 /* look at the symbol got offset. If none, then add one */
1194 if (type == R_ARM_GOT_BREL)
1195 reloc_type = R_ARM_GLOB_DAT;
1196 else
1197 reloc_type = R_ARM_JUMP_SLOT;
1198 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1199 sym_index);
1200 }
1201 break;
1202 #elif defined(TCC_TARGET_C67)
1203 case R_C60_GOT32:
1204 case R_C60_GOTOFF:
1205 case R_C60_GOTPC:
1206 case R_C60_PLT32:
1207 if (!s1->got)
1208 build_got(s1);
1209 if (type == R_C60_GOT32 || type == R_C60_PLT32) {
1210 sym_index = ELFW(R_SYM)(rel->r_info);
1211 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1212 /* look at the symbol got offset. If none, then add one */
1213 if (type == R_C60_GOT32)
1214 reloc_type = R_C60_GLOB_DAT;
1215 else
1216 reloc_type = R_C60_JMP_SLOT;
1217 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1218 sym_index);
1219 }
1220 break;
1221 #elif defined(TCC_TARGET_X86_64)
1222 case R_X86_64_GOT32:
1223 case R_X86_64_GOTTPOFF:
1224 case R_X86_64_GOTPCREL:
1225 case R_X86_64_PLT32:
1226 if (!s1->got)
1227 build_got(s1);
1228 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL ||
1229 type == R_X86_64_PLT32) {
1230 sym_index = ELFW(R_SYM)(rel->r_info);
1231 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1232 /* look at the symbol got offset. If none, then add one */
1233 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL)
1234 reloc_type = R_X86_64_GLOB_DAT;
1235 else
1236 reloc_type = R_X86_64_JUMP_SLOT;
1237 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1238 sym_index);
1239 }
1240 break;
1241 #else
1242 #error unsupported CPU
1243 #endif
1244 default:
1245 break;
1246 }
1247 }
1248 }
1249 }
1250
1251 ST_FUNC Section *new_symtab(TCCState *s1,
1252 const char *symtab_name, int sh_type, int sh_flags,
1253 const char *strtab_name,
1254 const char *hash_name, int hash_sh_flags)
1255 {
1256 Section *symtab, *strtab, *hash;
1257 int *ptr, nb_buckets;
1258
1259 symtab = new_section(s1, symtab_name, sh_type, sh_flags);
1260 symtab->sh_entsize = sizeof(ElfW(Sym));
1261 strtab = new_section(s1, strtab_name, SHT_STRTAB, sh_flags);
1262 put_elf_str(strtab, "");
1263 symtab->link = strtab;
1264 put_elf_sym(symtab, 0, 0, 0, 0, 0, NULL);
1265
1266 nb_buckets = 1;
1267
1268 hash = new_section(s1, hash_name, SHT_HASH, hash_sh_flags);
1269 hash->sh_entsize = sizeof(int);
1270 symtab->hash = hash;
1271 hash->link = symtab;
1272
1273 ptr = section_ptr_add(hash, (2 + nb_buckets + 1) * sizeof(int));
1274 ptr[0] = nb_buckets;
1275 ptr[1] = 1;
1276 memset(ptr + 2, 0, (nb_buckets + 1) * sizeof(int));
1277 return symtab;
1278 }
1279
1280 /* put dynamic tag */
1281 static void put_dt(Section *dynamic, int dt, addr_t val)
1282 {
1283 ElfW(Dyn) *dyn;
1284 dyn = section_ptr_add(dynamic, sizeof(ElfW(Dyn)));
1285 dyn->d_tag = dt;
1286 dyn->d_un.d_val = val;
1287 }
1288
1289 static void add_init_array_defines(TCCState *s1, const char *section_name)
1290 {
1291 Section *s;
1292 long end_offset;
1293 char sym_start[1024];
1294 char sym_end[1024];
1295
1296 snprintf(sym_start, sizeof(sym_start), "__%s_start", section_name + 1);
1297 snprintf(sym_end, sizeof(sym_end), "__%s_end", section_name + 1);
1298
1299 s = find_section(s1, section_name);
1300 if (!s) {
1301 end_offset = 0;
1302 s = data_section;
1303 } else {
1304 end_offset = s->data_offset;
1305 }
1306
1307 add_elf_sym(symtab_section,
1308 0, 0,
1309 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1310 s->sh_num, sym_start);
1311 add_elf_sym(symtab_section,
1312 end_offset, 0,
1313 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1314 s->sh_num, sym_end);
1315 }
1316
1317 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1318 {
1319 #ifdef CONFIG_TCC_BCHECK
1320 unsigned long *ptr;
1321 Section *init_section;
1322 unsigned char *pinit;
1323 int sym_index;
1324
1325 if (0 == s1->do_bounds_check)
1326 return;
1327
1328 /* XXX: add an object file to do that */
1329 ptr = section_ptr_add(bounds_section, sizeof(unsigned long));
1330 *ptr = 0;
1331 add_elf_sym(symtab_section, 0, 0,
1332 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1333 bounds_section->sh_num, "__bounds_start");
1334 #ifdef TCC_TARGET_I386
1335 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1336 /* add 'call __bound_init()' in .init section */
1337 init_section = find_section(s1, ".init");
1338 pinit = section_ptr_add(init_section, 5);
1339 pinit[0] = 0xe8;
1340 put32(pinit + 1, -4);
1341 sym_index = find_elf_sym(symtab_section, "__bound_init");
1342 put_elf_reloc(symtab_section, init_section,
1343 init_section->data_offset - 4, R_386_PC32, sym_index);
1344 }
1345 #endif
1346 #endif
1347 }
1348
1349 static inline int tcc_add_support(TCCState *s1, const char *filename)
1350 {
1351 char buf[1024];
1352 snprintf(buf, sizeof(buf), "%s/%s", s1->tcc_lib_path, filename);
1353 return tcc_add_file(s1, buf);
1354 }
1355
1356 /* add tcc runtime libraries */
1357 ST_FUNC void tcc_add_runtime(TCCState *s1)
1358 {
1359 tcc_add_bcheck(s1);
1360
1361 /* add libc */
1362 if (!s1->nostdlib) {
1363 tcc_add_library(s1, "c");
1364 #ifdef CONFIG_USE_LIBGCC
1365 tcc_add_file(s1, TCC_LIBGCC);
1366 #elif !defined WITHOUT_LIBTCC
1367 tcc_add_support(s1, "libtcc1.a");
1368 #endif
1369 /* add crt end if not memory output */
1370 if (s1->output_type != TCC_OUTPUT_MEMORY)
1371 tcc_add_crt(s1, "crtn.o");
1372 }
1373 }
1374
1375 /* add various standard linker symbols (must be done after the
1376 sections are filled (for example after allocating common
1377 symbols)) */
1378 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1379 {
1380 char buf[1024];
1381 int i;
1382 Section *s;
1383
1384 add_elf_sym(symtab_section,
1385 text_section->data_offset, 0,
1386 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1387 text_section->sh_num, "_etext");
1388 add_elf_sym(symtab_section,
1389 data_section->data_offset, 0,
1390 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1391 data_section->sh_num, "_edata");
1392 add_elf_sym(symtab_section,
1393 bss_section->data_offset, 0,
1394 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1395 bss_section->sh_num, "_end");
1396 /* horrible new standard ldscript defines */
1397 add_init_array_defines(s1, ".preinit_array");
1398 add_init_array_defines(s1, ".init_array");
1399 add_init_array_defines(s1, ".fini_array");
1400
1401 /* add start and stop symbols for sections whose name can be
1402 expressed in C */
1403 for(i = 1; i < s1->nb_sections; i++) {
1404 s = s1->sections[i];
1405 if (s->sh_type == SHT_PROGBITS &&
1406 (s->sh_flags & SHF_ALLOC)) {
1407 const char *p;
1408 int ch;
1409
1410 /* check if section name can be expressed in C */
1411 p = s->name;
1412 for(;;) {
1413 ch = *p;
1414 if (!ch)
1415 break;
1416 if (!isid(ch) && !isnum(ch))
1417 goto next_sec;
1418 p++;
1419 }
1420 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1421 add_elf_sym(symtab_section,
1422 0, 0,
1423 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1424 s->sh_num, buf);
1425 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1426 add_elf_sym(symtab_section,
1427 s->data_offset, 0,
1428 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1429 s->sh_num, buf);
1430 }
1431 next_sec: ;
1432 }
1433 }
1434
1435 static void tcc_output_binary(TCCState *s1, FILE *f,
1436 const int *section_order)
1437 {
1438 Section *s;
1439 int i, offset, size;
1440
1441 offset = 0;
1442 for(i=1;i<s1->nb_sections;i++) {
1443 s = s1->sections[section_order[i]];
1444 if (s->sh_type != SHT_NOBITS &&
1445 (s->sh_flags & SHF_ALLOC)) {
1446 while (offset < s->sh_offset) {
1447 fputc(0, f);
1448 offset++;
1449 }
1450 size = s->sh_size;
1451 fwrite(s->data, 1, size, f);
1452 offset += size;
1453 }
1454 }
1455 }
1456
1457 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1458 #define HAVE_PHDR 1
1459 #define EXTRA_RELITEMS 14
1460
1461 /* move the relocation value from .dynsym to .got */
1462 void patch_dynsym_undef(TCCState *s1, Section *s)
1463 {
1464 uint32_t *gotd = (void *)s1->got->data;
1465 ElfW(Sym) *sym, *sym_end;
1466
1467 gotd += 3; // dummy entries in .got
1468 /* relocate symbols in .dynsym */
1469 sym_end = (ElfW(Sym) *)(s->data + s->data_offset);
1470 for (sym = (ElfW(Sym) *)s->data + 1; sym < sym_end; sym++) {
1471 if (sym->st_shndx == SHN_UNDEF) {
1472 *gotd++ = sym->st_value + 6; // XXX 6 is magic ?
1473 sym->st_value = 0;
1474 }
1475 }
1476 }
1477 #else
1478 #define HAVE_PHDR 0
1479 #define EXTRA_RELITEMS 9
1480
1481 /* zero plt offsets of weak symbols in .dynsym */
1482 void patch_dynsym_undef(TCCState *s1, Section *s)
1483 {
1484 ElfW(Sym) *sym, *sym_end;
1485
1486 sym_end = (ElfW(Sym) *)(s->data + s->data_offset);
1487 for (sym = (ElfW(Sym) *)s->data + 1; sym < sym_end; sym++)
1488 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1489 sym->st_value = 0;
1490 }
1491 #endif
1492
1493 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1494 {
1495 int sym_index = ELFW(R_SYM) (rel->r_info);
1496 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1497 unsigned long offset;
1498
1499 if (sym_index >= s1->nb_sym_attrs)
1500 return;
1501 offset = s1->sym_attrs[sym_index].got_offset;
1502 section_reserve(s1->got, offset + PTR_SIZE);
1503 #ifdef TCC_TARGET_X86_64
1504 /* only works for x86-64 */
1505 put32(s1->got->data + offset + 4, sym->st_value >> 32);
1506 #endif
1507 put32(s1->got->data + offset, sym->st_value & 0xffffffff);
1508 }
1509
1510 ST_FUNC void fill_got(TCCState *s1)
1511 {
1512 Section *s;
1513 ElfW_Rel *rel, *rel_end;
1514 int i;
1515
1516 for(i = 1; i < s1->nb_sections; i++) {
1517 s = s1->sections[i];
1518 if (s->sh_type != SHT_RELX)
1519 continue;
1520 /* no need to handle got relocations */
1521 if (s->link != symtab_section)
1522 continue;
1523 rel_end = (ElfW_Rel *) (s->data + s->data_offset);
1524 for(rel = (ElfW_Rel *) s->data; rel < rel_end; rel++) {
1525 switch (ELFW(R_TYPE) (rel->r_info)) {
1526 case R_X86_64_GOT32:
1527 case R_X86_64_GOTPCREL:
1528 case R_X86_64_PLT32:
1529 fill_got_entry(s1, rel);
1530 break;
1531 }
1532 }
1533 }
1534 }
1535
1536
1537 /* output an ELF file */
1538 /* XXX: suppress unneeded sections */
1539 static int elf_output_file(TCCState *s1, const char *filename)
1540 {
1541 ElfW(Ehdr) ehdr;
1542 FILE *f;
1543 int fd, mode, ret;
1544 int *section_order;
1545 int shnum, i, phnum, file_offset, offset, size, j, sh_order_index, k;
1546 long long tmp;
1547 addr_t addr;
1548 Section *strsec, *s;
1549 ElfW(Shdr) shdr, *sh;
1550 ElfW(Phdr) *phdr, *ph;
1551 Section *interp, *dynamic, *dynstr;
1552 unsigned long saved_dynamic_data_offset;
1553 ElfW(Sym) *sym;
1554 int type, file_type;
1555 addr_t rel_addr, rel_size;
1556 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1557 addr_t bss_addr, bss_size;
1558 #endif
1559
1560 file_type = s1->output_type;
1561 s1->nb_errors = 0;
1562
1563 if (file_type != TCC_OUTPUT_OBJ) {
1564 tcc_add_runtime(s1);
1565 }
1566
1567 phdr = NULL;
1568 section_order = NULL;
1569 interp = NULL;
1570 dynamic = NULL;
1571 dynstr = NULL; /* avoid warning */
1572 saved_dynamic_data_offset = 0; /* avoid warning */
1573
1574 if (file_type != TCC_OUTPUT_OBJ) {
1575 relocate_common_syms();
1576
1577 tcc_add_linker_symbols(s1);
1578
1579 if (!s1->static_link) {
1580 const char *name;
1581 int sym_index, index;
1582 ElfW(Sym) *esym, *sym_end;
1583
1584 if (file_type == TCC_OUTPUT_EXE) {
1585 char *ptr;
1586 /* allow override the dynamic loader */
1587 const char *elfint = getenv("LD_SO");
1588 if (elfint == NULL)
1589 elfint = CONFIG_TCC_ELFINTERP;
1590 /* add interpreter section only if executable */
1591 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
1592 interp->sh_addralign = 1;
1593 ptr = section_ptr_add(interp, 1+strlen(elfint));
1594 strcpy(ptr, elfint);
1595 }
1596
1597 /* add dynamic symbol table */
1598 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
1599 ".dynstr",
1600 ".hash", SHF_ALLOC);
1601 dynstr = s1->dynsym->link;
1602
1603 /* add dynamic section */
1604 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
1605 SHF_ALLOC | SHF_WRITE);
1606 dynamic->link = dynstr;
1607 dynamic->sh_entsize = sizeof(ElfW(Dyn));
1608
1609 /* add PLT */
1610 s1->plt = new_section(s1, ".plt", SHT_PROGBITS,
1611 SHF_ALLOC | SHF_EXECINSTR);
1612 s1->plt->sh_entsize = 4;
1613
1614 build_got(s1);
1615
1616 /* scan for undefined symbols and see if they are in the
1617 dynamic symbols. If a symbol STT_FUNC or STT_GNU_IFUNC
1618 is found, then we add it in the PLT. If a symbol
1619 STT_OBJECT is found, we add it in the .bss section with
1620 a suitable relocation */
1621 sym_end = (ElfW(Sym) *)(symtab_section->data +
1622 symtab_section->data_offset);
1623 if (file_type == TCC_OUTPUT_EXE) {
1624 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
1625 sym < sym_end;
1626 sym++) {
1627 if (sym->st_shndx == SHN_UNDEF) {
1628 name = symtab_section->link->data + sym->st_name;
1629 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1630 if (sym_index) {
1631 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1632 type = ELFW(ST_TYPE)(esym->st_info);
1633 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1634 /* Indirect functions shall have STT_FUNC type
1635 * in executable dynsym section. Indeed, a dlsym
1636 * call following a lazy resolution would pick
1637 * the symbol value from the executable dynsym
1638 * entry which would contain the address of the
1639 * function wanted by the caller of dlsym
1640 * instead of the address of the function that
1641 * would return that address */
1642 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1643 ELFW(ST_INFO)(STB_GLOBAL,STT_FUNC),
1644 sym - (ElfW(Sym) *)symtab_section->data);
1645 } else if (type == STT_OBJECT) {
1646 unsigned long offset;
1647 ElfW(Sym) *dynsym, *dynsym_end;
1648 offset = bss_section->data_offset;
1649 /* XXX: which alignment ? */
1650 offset = (offset + 16 - 1) & -16;
1651 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1652 esym->st_info, 0,
1653 bss_section->sh_num, name);
1654 // Ensure R_COPY works for weak symbol aliases
1655 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1656 dynsym_end = (ElfW(Sym) *)
1657 (s1->dynsymtab_section->data +
1658 s1->dynsymtab_section->data_offset);
1659 for(dynsym = (ElfW(Sym) *)s1->dynsymtab_section->data + 1;
1660 dynsym < dynsym_end; dynsym++) {
1661 if ((dynsym->st_value == esym->st_value)
1662 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1663 char *dynname;
1664 dynname = s1->dynsymtab_section->link->data
1665 + dynsym->st_name;
1666 put_elf_sym(s1->dynsym, offset,
1667 dynsym->st_size,
1668 dynsym->st_info, 0,
1669 bss_section->sh_num,
1670 dynname);
1671 break;
1672 }
1673 }
1674 }
1675 put_elf_reloc(s1->dynsym, bss_section,
1676 offset, R_COPY, index);
1677 offset += esym->st_size;
1678 bss_section->data_offset = offset;
1679 }
1680 } else {
1681 /* STB_WEAK undefined symbols are accepted */
1682 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
1683 it */
1684 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1685 !strcmp(name, "_fp_hw")) {
1686 } else {
1687 tcc_error_noabort("undefined symbol '%s'", name);
1688 }
1689 }
1690 } else if (s1->rdynamic &&
1691 ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1692 /* if -rdynamic option, then export all non
1693 local symbols */
1694 name = symtab_section->link->data + sym->st_name;
1695 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1696 sym->st_info, 0,
1697 sym->st_shndx, name);
1698 }
1699 }
1700
1701 if (s1->nb_errors)
1702 goto fail;
1703
1704 /* now look at unresolved dynamic symbols and export
1705 corresponding symbol */
1706 sym_end = (ElfW(Sym) *)(s1->dynsymtab_section->data +
1707 s1->dynsymtab_section->data_offset);
1708 for(esym = (ElfW(Sym) *)s1->dynsymtab_section->data + 1;
1709 esym < sym_end;
1710 esym++) {
1711 if (esym->st_shndx == SHN_UNDEF) {
1712 name = s1->dynsymtab_section->link->data + esym->st_name;
1713 sym_index = find_elf_sym(symtab_section, name);
1714 if (sym_index) {
1715 /* XXX: avoid adding a symbol if already
1716 present because of -rdynamic ? */
1717 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1718 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1719 sym->st_info, 0,
1720 sym->st_shndx, name);
1721 } else {
1722 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1723 /* weak symbols can stay undefined */
1724 } else {
1725 tcc_warning("undefined dynamic symbol '%s'", name);
1726 }
1727 }
1728 }
1729 }
1730 } else {
1731 int nb_syms;
1732 /* shared library case : we simply export all the global symbols */
1733 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1734 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1735 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
1736 sym < sym_end;
1737 sym++) {
1738 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1739 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1740 if ((ELFW(ST_TYPE)(sym->st_info) == STT_FUNC ||
1741 ELFW(ST_TYPE)(sym->st_info) == STT_GNU_IFUNC)
1742 && sym->st_shndx == SHN_UNDEF) {
1743 int visibility = ELFW(ST_BIND)(sym->st_info);
1744 put_got_entry(s1, R_JMP_SLOT, sym->st_size,
1745 ELFW(ST_INFO)(visibility,STT_FUNC),
1746 sym - (ElfW(Sym) *)symtab_section->data);
1747 }
1748 else if (ELFW(ST_TYPE)(sym->st_info) == STT_OBJECT) {
1749 put_got_entry(s1, R_X86_64_GLOB_DAT, sym->st_size,
1750 sym->st_info,
1751 sym - (ElfW(Sym) *)symtab_section->data);
1752 }
1753 else
1754 #endif
1755 {
1756 name = symtab_section->link->data + sym->st_name;
1757 index = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1758 sym->st_info, 0,
1759 sym->st_shndx, name);
1760 s1->symtab_to_dynsym[sym -
1761 (ElfW(Sym) *)symtab_section->data] =
1762 index;
1763 }
1764 }
1765 }
1766 }
1767
1768 build_got_entries(s1);
1769
1770 /* add a list of needed dlls */
1771 for(i = 0; i < s1->nb_loaded_dlls; i++) {
1772 DLLReference *dllref = s1->loaded_dlls[i];
1773 if (dllref->level == 0)
1774 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
1775 }
1776
1777 if (s1->rpath)
1778 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
1779
1780 /* XXX: currently, since we do not handle PIC code, we
1781 must relocate the readonly segments */
1782 if (file_type == TCC_OUTPUT_DLL) {
1783 if (s1->soname)
1784 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
1785 put_dt(dynamic, DT_TEXTREL, 0);
1786 }
1787
1788 if (s1->symbolic)
1789 put_dt(dynamic, DT_SYMBOLIC, 0);
1790
1791 /* add necessary space for other entries */
1792 saved_dynamic_data_offset = dynamic->data_offset;
1793 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
1794 } else {
1795 /* still need to build got entries in case of static link */
1796 build_got_entries(s1);
1797 }
1798 }
1799
1800 memset(&ehdr, 0, sizeof(ehdr));
1801
1802 /* we add a section for symbols */
1803 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
1804 put_elf_str(strsec, "");
1805
1806 /* compute number of sections */
1807 shnum = s1->nb_sections;
1808
1809 /* this array is used to reorder sections in the output file */
1810 section_order = tcc_malloc(sizeof(int) * shnum);
1811 section_order[0] = 0;
1812 sh_order_index = 1;
1813
1814 /* compute number of program headers */
1815 switch(file_type) {
1816 default:
1817 case TCC_OUTPUT_OBJ:
1818 phnum = 0;
1819 break;
1820 case TCC_OUTPUT_EXE:
1821 if (!s1->static_link)
1822 phnum = 4 + HAVE_PHDR;
1823 else
1824 phnum = 2;
1825 break;
1826 case TCC_OUTPUT_DLL:
1827 phnum = 3;
1828 break;
1829 }
1830
1831 /* allocate strings for section names and decide if an unallocated
1832 section should be output */
1833 /* NOTE: the strsec section comes last, so its size is also
1834 correct ! */
1835 for(i = 1; i < s1->nb_sections; i++) {
1836 s = s1->sections[i];
1837 s->sh_name = put_elf_str(strsec, s->name);
1838 #if 0 /* gr */
1839 printf("section: f=%08x t=%08x i=%08x %s %s\n",
1840 s->sh_flags,
1841 s->sh_type,
1842 s->sh_info,
1843 s->name,
1844 s->reloc ? s->reloc->name : "n"
1845 );
1846 #endif
1847 /* when generating a DLL, we include relocations but we may
1848 patch them */
1849 if (file_type == TCC_OUTPUT_DLL &&
1850 s->sh_type == SHT_RELX &&
1851 !(s->sh_flags & SHF_ALLOC)) {
1852 /* //gr: avoid bogus relocs for empty (debug) sections */
1853 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
1854 prepare_dynamic_rel(s1, s);
1855 else if (s1->do_debug)
1856 s->sh_size = s->data_offset;
1857 } else if (s1->do_debug ||
1858 file_type == TCC_OUTPUT_OBJ ||
1859 (s->sh_flags & SHF_ALLOC) ||
1860 i == (s1->nb_sections - 1)) {
1861 /* we output all sections if debug or object file */
1862 s->sh_size = s->data_offset;
1863 }
1864 }
1865
1866 /* allocate program segment headers */
1867 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
1868
1869 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
1870 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
1871 } else {
1872 file_offset = 0;
1873 }
1874 if (phnum > 0) {
1875 /* compute section to program header mapping */
1876 if (s1->has_text_addr) {
1877 int a_offset, p_offset;
1878 addr = s1->text_addr;
1879 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
1880 ELF_PAGE_SIZE */
1881 a_offset = (int) (addr & (s1->section_align - 1));
1882 p_offset = file_offset & (s1->section_align - 1);
1883 if (a_offset < p_offset)
1884 a_offset += s1->section_align;
1885 file_offset += (a_offset - p_offset);
1886 } else {
1887 if (file_type == TCC_OUTPUT_DLL)
1888 addr = 0;
1889 else
1890 addr = ELF_START_ADDR;
1891 /* compute address after headers */
1892 addr += (file_offset & (s1->section_align - 1));
1893 }
1894
1895 /* dynamic relocation table information, for .dynamic section */
1896 rel_size = 0;
1897 rel_addr = 0;
1898
1899 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1900 bss_addr = bss_size = 0;
1901 #endif
1902 /* leave one program header for the program interpreter */
1903 ph = &phdr[0];
1904 if (interp)
1905 ph += 1 + HAVE_PHDR;
1906
1907 for(j = 0; j < 2; j++) {
1908 ph->p_type = PT_LOAD;
1909 if (j == 0)
1910 ph->p_flags = PF_R | PF_X;
1911 else
1912 ph->p_flags = PF_R | PF_W;
1913 ph->p_align = s1->section_align;
1914
1915 /* we do the following ordering: interp, symbol tables,
1916 relocations, progbits, nobits */
1917 /* XXX: do faster and simpler sorting */
1918 for(k = 0; k < 5; k++) {
1919 for(i = 1; i < s1->nb_sections; i++) {
1920 s = s1->sections[i];
1921 /* compute if section should be included */
1922 if (j == 0) {
1923 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1924 SHF_ALLOC)
1925 continue;
1926 } else {
1927 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1928 (SHF_ALLOC | SHF_WRITE))
1929 continue;
1930 }
1931 if (s == interp) {
1932 if (k != 0)
1933 continue;
1934 } else if (s->sh_type == SHT_DYNSYM ||
1935 s->sh_type == SHT_STRTAB ||
1936 s->sh_type == SHT_HASH) {
1937 if (k != 1)
1938 continue;
1939 } else if (s->sh_type == SHT_RELX) {
1940 if (k != 2)
1941 continue;
1942 } else if (s->sh_type == SHT_NOBITS) {
1943 if (k != 4)
1944 continue;
1945 } else {
1946 if (k != 3)
1947 continue;
1948 }
1949 section_order[sh_order_index++] = i;
1950
1951 /* section matches: we align it and add its size */
1952 tmp = addr;
1953 addr = (addr + s->sh_addralign - 1) &
1954 ~(s->sh_addralign - 1);
1955 file_offset += (int) ( addr - tmp );
1956 s->sh_offset = file_offset;
1957 s->sh_addr = addr;
1958
1959 /* update program header infos */
1960 if (ph->p_offset == 0) {
1961 ph->p_offset = file_offset;
1962 ph->p_vaddr = addr;
1963 ph->p_paddr = ph->p_vaddr;
1964 }
1965 /* update dynamic relocation infos */
1966 if (s->sh_type == SHT_RELX) {
1967 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1968 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) { // rel_size == 0) {
1969 rel_addr = addr;
1970 rel_size += s->sh_size; // XXX only first rel.
1971 }
1972 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) { // rel_size == 0) {
1973 bss_addr = addr;
1974 bss_size = s->sh_size; // XXX only first rel.
1975 }
1976 #else
1977 if (rel_size == 0)
1978 rel_addr = addr;
1979 rel_size += s->sh_size;
1980 #endif
1981 }
1982 addr += s->sh_size;
1983 if (s->sh_type != SHT_NOBITS)
1984 file_offset += s->sh_size;
1985 }
1986 }
1987 ph->p_filesz = file_offset - ph->p_offset;
1988 ph->p_memsz = addr - ph->p_vaddr;
1989 ph++;
1990 if (j == 0) {
1991 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
1992 /* if in the middle of a page, we duplicate the page in
1993 memory so that one copy is RX and the other is RW */
1994 if ((addr & (s1->section_align - 1)) != 0)
1995 addr += s1->section_align;
1996 } else {
1997 addr = (addr + s1->section_align - 1) & ~(s1->section_align - 1);
1998 file_offset = (file_offset + s1->section_align - 1) &
1999 ~(s1->section_align - 1);
2000 }
2001 }
2002 }
2003
2004 /* if interpreter, then add corresponing program header */
2005 if (interp) {
2006 ph = &phdr[0];
2007
2008 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2009 {
2010 int len = phnum * sizeof(ElfW(Phdr));
2011
2012 ph->p_type = PT_PHDR;
2013 ph->p_offset = sizeof(ElfW(Ehdr));
2014 ph->p_vaddr = interp->sh_addr - len;
2015 ph->p_paddr = ph->p_vaddr;
2016 ph->p_filesz = ph->p_memsz = len;
2017 ph->p_flags = PF_R | PF_X;
2018 ph->p_align = 4; // interp->sh_addralign;
2019 ph++;
2020 }
2021 #endif
2022
2023 ph->p_type = PT_INTERP;
2024 ph->p_offset = interp->sh_offset;
2025 ph->p_vaddr = interp->sh_addr;
2026 ph->p_paddr = ph->p_vaddr;
2027 ph->p_filesz = interp->sh_size;
2028 ph->p_memsz = interp->sh_size;
2029 ph->p_flags = PF_R;
2030 ph->p_align = interp->sh_addralign;
2031 }
2032
2033 /* if dynamic section, then add corresponing program header */
2034 if (dynamic) {
2035 ElfW(Sym) *sym_end;
2036
2037 ph = &phdr[phnum - 1];
2038
2039 ph->p_type = PT_DYNAMIC;
2040 ph->p_offset = dynamic->sh_offset;
2041 ph->p_vaddr = dynamic->sh_addr;
2042 ph->p_paddr = ph->p_vaddr;
2043 ph->p_filesz = dynamic->sh_size;
2044 ph->p_memsz = dynamic->sh_size;
2045 ph->p_flags = PF_R | PF_W;
2046 ph->p_align = dynamic->sh_addralign;
2047
2048 /* put GOT dynamic section address */
2049 put32(s1->got->data, dynamic->sh_addr);
2050
2051 /* relocate the PLT */
2052 if (file_type == TCC_OUTPUT_EXE
2053 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2054 || file_type == TCC_OUTPUT_DLL
2055 #endif
2056 ) {
2057 uint8_t *p, *p_end;
2058
2059 p = s1->plt->data;
2060 p_end = p + s1->plt->data_offset;
2061 if (p < p_end) {
2062 #if defined(TCC_TARGET_I386)
2063 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
2064 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
2065 p += 16;
2066 while (p < p_end) {
2067 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
2068 p += 16;
2069 }
2070 #elif defined(TCC_TARGET_X86_64)
2071 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
2072 put32(p + 2, get32(p + 2) + x);
2073 put32(p + 8, get32(p + 8) + x - 6);
2074 p += 16;
2075 while (p < p_end) {
2076 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
2077 p += 16;
2078 }
2079 #elif defined(TCC_TARGET_ARM)
2080 int x;
2081 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
2082 p += 16;
2083 while (p < p_end) {
2084 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
2085 p += 4;
2086 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
2087 p += 16;
2088 }
2089 #elif defined(TCC_TARGET_C67)
2090 /* XXX: TODO */
2091 #else
2092 #error unsupported CPU
2093 #endif
2094 }
2095 }
2096
2097 /* relocate symbols in .dynsym */
2098 sym_end = (ElfW(Sym) *)(s1->dynsym->data + s1->dynsym->data_offset);
2099 for(sym = (ElfW(Sym) *)s1->dynsym->data + 1;
2100 sym < sym_end;
2101 sym++) {
2102 if (sym->st_shndx == SHN_UNDEF) {
2103 /* relocate to the PLT if the symbol corresponds
2104 to a PLT entry */
2105 if (sym->st_value)
2106 sym->st_value += s1->plt->sh_addr;
2107 } else if (sym->st_shndx < SHN_LORESERVE) {
2108 /* do symbol relocation */
2109 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2110 }
2111 }
2112
2113 /* put dynamic section entries */
2114 dynamic->data_offset = saved_dynamic_data_offset;
2115 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2116 put_dt(dynamic, DT_STRTAB, dynstr->sh_addr);
2117 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2118 put_dt(dynamic, DT_STRSZ, dynstr->data_offset);
2119 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2120 #ifdef TCC_TARGET_X86_64
2121 put_dt(dynamic, DT_RELA, rel_addr);
2122 put_dt(dynamic, DT_RELASZ, rel_size);
2123 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2124 #else
2125 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2126 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2127 put_dt(dynamic, DT_PLTRELSZ, rel_size);
2128 put_dt(dynamic, DT_JMPREL, rel_addr);
2129 put_dt(dynamic, DT_PLTREL, DT_REL);
2130 put_dt(dynamic, DT_REL, bss_addr);
2131 put_dt(dynamic, DT_RELSZ, bss_size);
2132 #else
2133 put_dt(dynamic, DT_REL, rel_addr);
2134 put_dt(dynamic, DT_RELSZ, rel_size);
2135 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2136 #endif
2137 #endif
2138 if (s1->do_debug)
2139 put_dt(dynamic, DT_DEBUG, 0);
2140 put_dt(dynamic, DT_NULL, 0);
2141 }
2142
2143 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2144 ehdr.e_phnum = phnum;
2145 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2146 }
2147
2148 /* all other sections come after */
2149 for(i = 1; i < s1->nb_sections; i++) {
2150 s = s1->sections[i];
2151 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
2152 continue;
2153 section_order[sh_order_index++] = i;
2154
2155 file_offset = (file_offset + s->sh_addralign - 1) &
2156 ~(s->sh_addralign - 1);
2157 s->sh_offset = file_offset;
2158 if (s->sh_type != SHT_NOBITS)
2159 file_offset += s->sh_size;
2160 }
2161
2162 /* if building executable or DLL, then relocate each section
2163 except the GOT which is already relocated */
2164 if (file_type != TCC_OUTPUT_OBJ) {
2165 relocate_syms(s1, 0);
2166
2167 if (s1->nb_errors != 0) {
2168 fail:
2169 ret = -1;
2170 goto the_end;
2171 }
2172
2173 /* relocate sections */
2174 /* XXX: ignore sections with allocated relocations ? */
2175 for(i = 1; i < s1->nb_sections; i++) {
2176 s = s1->sections[i];
2177 if (s->reloc && s != s1->got)
2178 relocate_section(s1, s);
2179 }
2180
2181 /* relocate relocation entries if the relocation tables are
2182 allocated in the executable */
2183 for(i = 1; i < s1->nb_sections; i++) {
2184 s = s1->sections[i];
2185 if ((s->sh_flags & SHF_ALLOC) &&
2186 s->sh_type == SHT_RELX) {
2187 relocate_rel(s1, s);
2188 }
2189 }
2190
2191 /* get entry point address */
2192 if (file_type == TCC_OUTPUT_EXE)
2193 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2194 else
2195 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2196 }
2197 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2198 fill_got(s1);
2199
2200 /* write elf file */
2201 if (file_type == TCC_OUTPUT_OBJ)
2202 mode = 0666;
2203 else
2204 mode = 0777;
2205 unlink(filename);
2206 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2207 if (fd < 0) {
2208 tcc_error_noabort("could not write '%s'", filename);
2209 goto fail;
2210 }
2211 f = fdopen(fd, "wb");
2212 if (s1->verbose)
2213 printf("<- %s\n", filename);
2214
2215 #ifdef TCC_TARGET_COFF
2216 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF) {
2217 tcc_output_coff(s1, f);
2218 } else
2219 #endif
2220 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
2221 sort_syms(s1, symtab_section);
2222
2223 /* align to 4 */
2224 file_offset = (file_offset + 3) & -4;
2225
2226 /* fill header */
2227 ehdr.e_ident[0] = ELFMAG0;
2228 ehdr.e_ident[1] = ELFMAG1;
2229 ehdr.e_ident[2] = ELFMAG2;
2230 ehdr.e_ident[3] = ELFMAG3;
2231 ehdr.e_ident[4] = ELFCLASSW;
2232 ehdr.e_ident[5] = ELFDATA2LSB;
2233 ehdr.e_ident[6] = EV_CURRENT;
2234 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2235 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2236 #endif
2237 #ifdef TCC_TARGET_ARM
2238 #ifdef TCC_ARM_EABI
2239 ehdr.e_ident[EI_OSABI] = 0;
2240 ehdr.e_flags = 4 << 24;
2241 #else
2242 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2243 #endif
2244 #endif
2245 switch(file_type) {
2246 default:
2247 case TCC_OUTPUT_EXE:
2248 ehdr.e_type = ET_EXEC;
2249 break;
2250 case TCC_OUTPUT_DLL:
2251 ehdr.e_type = ET_DYN;
2252 break;
2253 case TCC_OUTPUT_OBJ:
2254 ehdr.e_type = ET_REL;
2255 break;
2256 }
2257 ehdr.e_machine = EM_TCC_TARGET;
2258 ehdr.e_version = EV_CURRENT;
2259 ehdr.e_shoff = file_offset;
2260 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2261 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2262 ehdr.e_shnum = shnum;
2263 ehdr.e_shstrndx = shnum - 1;
2264
2265 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2266 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2267 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2268
2269 for(i=1;i<s1->nb_sections;i++) {
2270 s = s1->sections[section_order[i]];
2271 if (s->sh_type != SHT_NOBITS) {
2272 if (s->sh_type == SHT_DYNSYM)
2273 patch_dynsym_undef(s1, s);
2274 while (offset < s->sh_offset) {
2275 fputc(0, f);
2276 offset++;
2277 }
2278 size = s->sh_size;
2279 fwrite(s->data, 1, size, f);
2280 offset += size;
2281 }
2282 }
2283
2284 /* output section headers */
2285 while (offset < ehdr.e_shoff) {
2286 fputc(0, f);
2287 offset++;
2288 }
2289
2290 for(i=0;i<s1->nb_sections;i++) {
2291 sh = &shdr;
2292 memset(sh, 0, sizeof(ElfW(Shdr)));
2293 s = s1->sections[i];
2294 if (s) {
2295 sh->sh_name = s->sh_name;
2296 sh->sh_type = s->sh_type;
2297 sh->sh_flags = s->sh_flags;
2298 sh->sh_entsize = s->sh_entsize;
2299 sh->sh_info = s->sh_info;
2300 if (s->link)
2301 sh->sh_link = s->link->sh_num;
2302 sh->sh_addralign = s->sh_addralign;
2303 sh->sh_addr = s->sh_addr;
2304 sh->sh_offset = s->sh_offset;
2305 sh->sh_size = s->sh_size;
2306 }
2307 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2308 }
2309 } else {
2310 tcc_output_binary(s1, f, section_order);
2311 }
2312 fclose(f);
2313
2314 ret = 0;
2315 the_end:
2316 tcc_free(s1->symtab_to_dynsym);
2317 tcc_free(section_order);
2318 tcc_free(phdr);
2319 tcc_free(s1->sym_attrs);
2320 return ret;
2321 }
2322
2323 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2324 {
2325 int ret;
2326 #ifdef TCC_TARGET_PE
2327 if (s->output_type != TCC_OUTPUT_OBJ) {
2328 ret = pe_output_file(s, filename);
2329 } else
2330 #endif
2331 {
2332 ret = elf_output_file(s, filename);
2333 }
2334 return ret;
2335 }
2336
2337 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2338 {
2339 void *data;
2340
2341 data = tcc_malloc(size);
2342 lseek(fd, file_offset, SEEK_SET);
2343 read(fd, data, size);
2344 return data;
2345 }
2346
2347 typedef struct SectionMergeInfo {
2348 Section *s; /* corresponding existing section */
2349 unsigned long offset; /* offset of the new section in the existing section */
2350 uint8_t new_section; /* true if section 's' was added */
2351 uint8_t link_once; /* true if link once section */
2352 } SectionMergeInfo;
2353
2354 /* load an object file and merge it with current files */
2355 /* XXX: handle correctly stab (debug) info */
2356 ST_FUNC int tcc_load_object_file(TCCState *s1,
2357 int fd, unsigned long file_offset)
2358 {
2359 ElfW(Ehdr) ehdr;
2360 ElfW(Shdr) *shdr, *sh;
2361 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2362 unsigned char *strsec, *strtab;
2363 int *old_to_new_syms;
2364 char *sh_name, *name;
2365 SectionMergeInfo *sm_table, *sm;
2366 ElfW(Sym) *sym, *symtab;
2367 ElfW_Rel *rel, *rel_end;
2368 Section *s;
2369
2370 int stab_index;
2371 int stabstr_index;
2372
2373 stab_index = stabstr_index = 0;
2374
2375 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2376 goto fail1;
2377 if (ehdr.e_ident[0] != ELFMAG0 ||
2378 ehdr.e_ident[1] != ELFMAG1 ||
2379 ehdr.e_ident[2] != ELFMAG2 ||
2380 ehdr.e_ident[3] != ELFMAG3)
2381 goto fail1;
2382 /* test if object file */
2383 if (ehdr.e_type != ET_REL)
2384 goto fail1;
2385 /* test CPU specific stuff */
2386 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2387 ehdr.e_machine != EM_TCC_TARGET) {
2388 fail1:
2389 tcc_error_noabort("invalid object file");
2390 return -1;
2391 }
2392 /* read sections */
2393 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2394 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2395 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2396
2397 /* load section names */
2398 sh = &shdr[ehdr.e_shstrndx];
2399 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2400
2401 /* load symtab and strtab */
2402 old_to_new_syms = NULL;
2403 symtab = NULL;
2404 strtab = NULL;
2405 nb_syms = 0;
2406 for(i = 1; i < ehdr.e_shnum; i++) {
2407 sh = &shdr[i];
2408 if (sh->sh_type == SHT_SYMTAB) {
2409 if (symtab) {
2410 tcc_error_noabort("object must contain only one symtab");
2411 fail:
2412 ret = -1;
2413 goto the_end;
2414 }
2415 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2416 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2417 sm_table[i].s = symtab_section;
2418
2419 /* now load strtab */
2420 sh = &shdr[sh->sh_link];
2421 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2422 }
2423 }
2424
2425 /* now examine each section and try to merge its content with the
2426 ones in memory */
2427 for(i = 1; i < ehdr.e_shnum; i++) {
2428 /* no need to examine section name strtab */
2429 if (i == ehdr.e_shstrndx)
2430 continue;
2431 sh = &shdr[i];
2432 sh_name = strsec + sh->sh_name;
2433 /* ignore sections types we do not handle */
2434 if (sh->sh_type != SHT_PROGBITS &&
2435 sh->sh_type != SHT_RELX &&
2436 #ifdef TCC_ARM_EABI
2437 sh->sh_type != SHT_ARM_EXIDX &&
2438 #endif
2439 sh->sh_type != SHT_NOBITS &&
2440 sh->sh_type != SHT_PREINIT_ARRAY &&
2441 sh->sh_type != SHT_INIT_ARRAY &&
2442 sh->sh_type != SHT_FINI_ARRAY &&
2443 strcmp(sh_name, ".stabstr")
2444 )
2445 continue;
2446 if (sh->sh_addralign < 1)
2447 sh->sh_addralign = 1;
2448 /* find corresponding section, if any */
2449 for(j = 1; j < s1->nb_sections;j++) {
2450 s = s1->sections[j];
2451 if (!strcmp(s->name, sh_name)) {
2452 if (!strncmp(sh_name, ".gnu.linkonce",
2453 sizeof(".gnu.linkonce") - 1)) {
2454 /* if a 'linkonce' section is already present, we
2455 do not add it again. It is a little tricky as
2456 symbols can still be defined in
2457 it. */
2458 sm_table[i].link_once = 1;
2459 goto next;
2460 } else {
2461 goto found;
2462 }
2463 }
2464 }
2465 /* not found: create new section */
2466 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2467 /* take as much info as possible from the section. sh_link and
2468 sh_info will be updated later */
2469 s->sh_addralign = sh->sh_addralign;
2470 s->sh_entsize = sh->sh_entsize;
2471 sm_table[i].new_section = 1;
2472 found:
2473 if (sh->sh_type != s->sh_type) {
2474 tcc_error_noabort("invalid section type");
2475 goto fail;
2476 }
2477
2478 /* align start of section */
2479 offset = s->data_offset;
2480
2481 if (0 == strcmp(sh_name, ".stab")) {
2482 stab_index = i;
2483 goto no_align;
2484 }
2485 if (0 == strcmp(sh_name, ".stabstr")) {
2486 stabstr_index = i;
2487 goto no_align;
2488 }
2489
2490 size = sh->sh_addralign - 1;
2491 offset = (offset + size) & ~size;
2492 if (sh->sh_addralign > s->sh_addralign)
2493 s->sh_addralign = sh->sh_addralign;
2494 s->data_offset = offset;
2495 no_align:
2496 sm_table[i].offset = offset;
2497 sm_table[i].s = s;
2498 /* concatenate sections */
2499 size = sh->sh_size;
2500 if (sh->sh_type != SHT_NOBITS) {
2501 unsigned char *ptr;
2502 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2503 ptr = section_ptr_add(s, size);
2504 read(fd, ptr, size);
2505 } else {
2506 s->data_offset += size;
2507 }
2508 next: ;
2509 }
2510
2511 /* //gr relocate stab strings */
2512 if (stab_index && stabstr_index) {
2513 Stab_Sym *a, *b;
2514 unsigned o;
2515 s = sm_table[stab_index].s;
2516 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2517 b = (Stab_Sym *)(s->data + s->data_offset);
2518 o = sm_table[stabstr_index].offset;
2519 while (a < b)
2520 a->n_strx += o, a++;
2521 }
2522
2523 /* second short pass to update sh_link and sh_info fields of new
2524 sections */
2525 for(i = 1; i < ehdr.e_shnum; i++) {
2526 s = sm_table[i].s;
2527 if (!s || !sm_table[i].new_section)
2528 continue;
2529 sh = &shdr[i];
2530 if (sh->sh_link > 0)
2531 s->link = sm_table[sh->sh_link].s;
2532 if (sh->sh_type == SHT_RELX) {
2533 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2534 /* update backward link */
2535 s1->sections[s->sh_info]->reloc = s;
2536 }
2537 }
2538 sm = sm_table;
2539
2540 /* resolve symbols */
2541 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2542
2543 sym = symtab + 1;
2544 for(i = 1; i < nb_syms; i++, sym++) {
2545 if (sym->st_shndx != SHN_UNDEF &&
2546 sym->st_shndx < SHN_LORESERVE) {
2547 sm = &sm_table[sym->st_shndx];
2548 if (sm->link_once) {
2549 /* if a symbol is in a link once section, we use the
2550 already defined symbol. It is very important to get
2551 correct relocations */
2552 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2553 name = strtab + sym->st_name;
2554 sym_index = find_elf_sym(symtab_section, name);
2555 if (sym_index)
2556 old_to_new_syms[i] = sym_index;
2557 }
2558 continue;
2559 }
2560 /* if no corresponding section added, no need to add symbol */
2561 if (!sm->s)
2562 continue;
2563 /* convert section number */
2564 sym->st_shndx = sm->s->sh_num;
2565 /* offset value */
2566 sym->st_value += sm->offset;
2567 }
2568 /* add symbol */
2569 name = strtab + sym->st_name;
2570 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2571 sym->st_info, sym->st_other,
2572 sym->st_shndx, name);
2573 old_to_new_syms[i] = sym_index;
2574 }
2575
2576 /* third pass to patch relocation entries */
2577 for(i = 1; i < ehdr.e_shnum; i++) {
2578 s = sm_table[i].s;
2579 if (!s)
2580 continue;
2581 sh = &shdr[i];
2582 offset = sm_table[i].offset;
2583 switch(s->sh_type) {
2584 case SHT_RELX:
2585 /* take relocation offset information */
2586 offseti = sm_table[sh->sh_info].offset;
2587 rel_end = (ElfW_Rel *)(s->data + s->data_offset);
2588 for(rel = (ElfW_Rel *)(s->data + offset);
2589 rel < rel_end;
2590 rel++) {
2591 int type;
2592 unsigned sym_index;
2593 /* convert symbol index */
2594 type = ELFW(R_TYPE)(rel->r_info);
2595 sym_index = ELFW(R_SYM)(rel->r_info);
2596 /* NOTE: only one symtab assumed */
2597 if (sym_index >= nb_syms)
2598 goto invalid_reloc;
2599 sym_index = old_to_new_syms[sym_index];
2600 /* ignore link_once in rel section. */
2601 if (!sym_index && !sm->link_once
2602 #ifdef TCC_TARGET_ARM
2603 && type != R_ARM_V4BX
2604 #endif
2605 ) {
2606 invalid_reloc:
2607 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2608 i, strsec + sh->sh_name, rel->r_offset);
2609 goto fail;
2610 }
2611 rel->r_info = ELFW(R_INFO)(sym_index, type);
2612 /* offset the relocation offset */
2613 rel->r_offset += offseti;
2614 #ifdef TCC_TARGET_ARM
2615 /* Jumps and branches from a Thumb code to a PLT entry need
2616 special handling since PLT entries are ARM code.
2617 Unconditional bl instructions referencing PLT entries are
2618 handled by converting these instructions into blx
2619 instructions. Other case of instructions referencing a PLT
2620 entry require to add a Thumb stub before the PLT entry to
2621 switch to ARM mode. We set bit 0 of the got offset of a
2622 symbol to indicate such a case. */
2623 if (type == R_ARM_THM_JUMP24)
2624 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
2625 #endif
2626 }
2627 break;
2628 default:
2629 break;
2630 }
2631 }
2632
2633 ret = 0;
2634 the_end:
2635 tcc_free(symtab);
2636 tcc_free(strtab);
2637 tcc_free(old_to_new_syms);
2638 tcc_free(sm_table);
2639 tcc_free(strsec);
2640 tcc_free(shdr);
2641 return ret;
2642 }
2643
2644 typedef struct ArchiveHeader {
2645 char ar_name[16]; /* name of this member */
2646 char ar_date[12]; /* file mtime */
2647 char ar_uid[6]; /* owner uid; printed as decimal */
2648 char ar_gid[6]; /* owner gid; printed as decimal */
2649 char ar_mode[8]; /* file mode, printed as octal */
2650 char ar_size[10]; /* file size, printed as decimal */
2651 char ar_fmag[2]; /* should contain ARFMAG */
2652 } ArchiveHeader;
2653
2654 static int get_be32(const uint8_t *b)
2655 {
2656 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
2657 }
2658
2659 /* load only the objects which resolve undefined symbols */
2660 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
2661 {
2662 int i, bound, nsyms, sym_index, off, ret;
2663 uint8_t *data;
2664 const char *ar_names, *p;
2665 const uint8_t *ar_index;
2666 ElfW(Sym) *sym;
2667
2668 data = tcc_malloc(size);
2669 if (read(fd, data, size) != size)
2670 goto fail;
2671 nsyms = get_be32(data);
2672 ar_index = data + 4;
2673 ar_names = ar_index + nsyms * 4;
2674
2675 do {
2676 bound = 0;
2677 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
2678 sym_index = find_elf_sym(symtab_section, p);
2679 if(sym_index) {
2680 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
2681 if(sym->st_shndx == SHN_UNDEF) {
2682 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
2683 #if 0
2684 printf("%5d\t%s\t%08x\n", i, p, sym->st_shndx);
2685 #endif
2686 ++bound;
2687 lseek(fd, off, SEEK_SET);
2688 if(tcc_load_object_file(s1, fd, off) < 0) {
2689 fail:
2690 ret = -1;
2691 goto the_end;
2692 }
2693 }
2694 }
2695 }
2696 } while(bound);
2697 ret = 0;
2698 the_end:
2699 tcc_free(data);
2700 return ret;
2701 }
2702
2703 /* load a '.a' file */
2704 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
2705 {
2706 ArchiveHeader hdr;
2707 char ar_size[11];
2708 char ar_name[17];
2709 char magic[8];
2710 int size, len, i;
2711 unsigned long file_offset;
2712
2713 /* skip magic which was already checked */
2714 read(fd, magic, sizeof(magic));
2715
2716 for(;;) {
2717 len = read(fd, &hdr, sizeof(hdr));
2718 if (len == 0)
2719 break;
2720 if (len != sizeof(hdr)) {
2721 tcc_error_noabort("invalid archive");
2722 return -1;
2723 }
2724 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
2725 ar_size[sizeof(hdr.ar_size)] = '\0';
2726 size = strtol(ar_size, NULL, 0);
2727 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
2728 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
2729 if (ar_name[i] != ' ')
2730 break;
2731 }
2732 ar_name[i + 1] = '\0';
2733 // printf("name='%s' size=%d %s\n", ar_name, size, ar_size);
2734 file_offset = lseek(fd, 0, SEEK_CUR);
2735 /* align to even */
2736 size = (size + 1) & ~1;
2737 if (!strcmp(ar_name, "/")) {
2738 /* coff symbol table : we handle it */
2739 if(s1->alacarte_link)
2740 return tcc_load_alacarte(s1, fd, size);
2741 } else if (!strcmp(ar_name, "//") ||
2742 !strcmp(ar_name, "__.SYMDEF") ||
2743 !strcmp(ar_name, "__.SYMDEF/") ||
2744 !strcmp(ar_name, "ARFILENAMES/")) {
2745 /* skip symbol table or archive names */
2746 } else {
2747 if (tcc_load_object_file(s1, fd, file_offset) < 0)
2748 return -1;
2749 }
2750 lseek(fd, file_offset + size, SEEK_SET);
2751 }
2752 return 0;
2753 }
2754
2755 #ifndef TCC_TARGET_PE
2756 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
2757 is referenced by the user (so it should be added as DT_NEEDED in
2758 the generated ELF file) */
2759 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
2760 {
2761 ElfW(Ehdr) ehdr;
2762 ElfW(Shdr) *shdr, *sh, *sh1;
2763 int i, j, nb_syms, nb_dts, sym_bind, ret;
2764 ElfW(Sym) *sym, *dynsym;
2765 ElfW(Dyn) *dt, *dynamic;
2766 unsigned char *dynstr;
2767 const char *name, *soname;
2768 DLLReference *dllref;
2769
2770 read(fd, &ehdr, sizeof(ehdr));
2771
2772 /* test CPU specific stuff */
2773 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2774 ehdr.e_machine != EM_TCC_TARGET) {
2775 tcc_error_noabort("bad architecture");
2776 return -1;
2777 }
2778
2779 /* read sections */
2780 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2781
2782 /* load dynamic section and dynamic symbols */
2783 nb_syms = 0;
2784 nb_dts = 0;
2785 dynamic = NULL;
2786 dynsym = NULL; /* avoid warning */
2787 dynstr = NULL; /* avoid warning */
2788 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
2789 switch(sh->sh_type) {
2790 case SHT_DYNAMIC:
2791 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
2792 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
2793 break;
2794 case SHT_DYNSYM:
2795 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2796 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
2797 sh1 = &shdr[sh->sh_link];
2798 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
2799 break;
2800 default:
2801 break;
2802 }
2803 }
2804
2805 /* compute the real library name */
2806 soname = tcc_basename(filename);
2807
2808 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2809 if (dt->d_tag == DT_SONAME) {
2810 soname = dynstr + dt->d_un.d_val;
2811 }
2812 }
2813
2814 /* if the dll is already loaded, do not load it */
2815 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2816 dllref = s1->loaded_dlls[i];
2817 if (!strcmp(soname, dllref->name)) {
2818 /* but update level if needed */
2819 if (level < dllref->level)
2820 dllref->level = level;
2821 ret = 0;
2822 goto the_end;
2823 }
2824 }
2825
2826 // printf("loading dll '%s'\n", soname);
2827
2828 /* add the dll and its level */
2829 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
2830 dllref->level = level;
2831 strcpy(dllref->name, soname);
2832 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
2833
2834 /* add dynamic symbols in dynsym_section */
2835 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
2836 sym_bind = ELFW(ST_BIND)(sym->st_info);
2837 if (sym_bind == STB_LOCAL)
2838 continue;
2839 name = dynstr + sym->st_name;
2840 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
2841 sym->st_info, sym->st_other, sym->st_shndx, name);
2842 }
2843
2844 /* load all referenced DLLs */
2845 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2846 switch(dt->d_tag) {
2847 case DT_NEEDED:
2848 name = dynstr + dt->d_un.d_val;
2849 for(j = 0; j < s1->nb_loaded_dlls; j++) {
2850 dllref = s1->loaded_dlls[j];
2851 if (!strcmp(name, dllref->name))
2852 goto already_loaded;
2853 }
2854 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
2855 tcc_error_noabort("referenced dll '%s' not found", name);
2856 ret = -1;
2857 goto the_end;
2858 }
2859 already_loaded:
2860 break;
2861 }
2862 }
2863 ret = 0;
2864 the_end:
2865 tcc_free(dynstr);
2866 tcc_free(dynsym);
2867 tcc_free(dynamic);
2868 tcc_free(shdr);
2869 return ret;
2870 }
2871
2872 #define LD_TOK_NAME 256
2873 #define LD_TOK_EOF (-1)
2874
2875 /* return next ld script token */
2876 static int ld_next(TCCState *s1, char *name, int name_size)
2877 {
2878 int c;
2879 char *q;
2880
2881 redo:
2882 switch(ch) {
2883 case ' ':
2884 case '\t':
2885 case '\f':
2886 case '\v':
2887 case '\r':
2888 case '\n':
2889 inp();
2890 goto redo;
2891 case '/':
2892 minp();
2893 if (ch == '*') {
2894 file->buf_ptr = parse_comment(file->buf_ptr);
2895 ch = file->buf_ptr[0];
2896 goto redo;
2897 } else {
2898 q = name;
2899 *q++ = '/';
2900 goto parse_name;
2901 }
2902 break;
2903 /* case 'a' ... 'z': */
2904 case 'a':
2905 case 'b':
2906 case 'c':
2907 case 'd':
2908 case 'e':
2909 case 'f':
2910 case 'g':
2911 case 'h':
2912 case 'i':
2913 case 'j':
2914 case 'k':
2915 case 'l':
2916 case 'm':
2917 case 'n':
2918 case 'o':
2919 case 'p':
2920 case 'q':
2921 case 'r':
2922 case 's':
2923 case 't':
2924 case 'u':
2925 case 'v':
2926 case 'w':
2927 case 'x':
2928 case 'y':
2929 case 'z':
2930 /* case 'A' ... 'z': */
2931 case 'A':
2932 case 'B':
2933 case 'C':
2934 case 'D':
2935 case 'E':
2936 case 'F':
2937 case 'G':
2938 case 'H':
2939 case 'I':
2940 case 'J':
2941 case 'K':
2942 case 'L':
2943 case 'M':
2944 case 'N':
2945 case 'O':
2946 case 'P':
2947 case 'Q':
2948 case 'R':
2949 case 'S':
2950 case 'T':
2951 case 'U':
2952 case 'V':
2953 case 'W':
2954 case 'X':
2955 case 'Y':
2956 case 'Z':
2957 case '_':
2958 case '\\':
2959 case '.':
2960 case '$':
2961 case '~':
2962 q = name;
2963 parse_name:
2964 for(;;) {
2965 if (!((ch >= 'a' && ch <= 'z') ||
2966 (ch >= 'A' && ch <= 'Z') ||
2967 (ch >= '0' && ch <= '9') ||
2968 strchr("/.-_+=$:\\,~", ch)))
2969 break;
2970 if ((q - name) < name_size - 1) {
2971 *q++ = ch;
2972 }
2973 minp();
2974 }
2975 *q = '\0';
2976 c = LD_TOK_NAME;
2977 break;
2978 case CH_EOF:
2979 c = LD_TOK_EOF;
2980 break;
2981 default:
2982 c = ch;
2983 inp();
2984 break;
2985 }
2986 #if 0
2987 printf("tok=%c %d\n", c, c);
2988 if (c == LD_TOK_NAME)
2989 printf(" name=%s\n", name);
2990 #endif
2991 return c;
2992 }
2993
2994 static int ld_add_file(TCCState *s1, const char filename[])
2995 {
2996 int ret;
2997
2998 ret = tcc_add_file_internal(s1, filename, 0);
2999 if (ret)
3000 ret = tcc_add_dll(s1, filename, 0);
3001 return ret;
3002 }
3003
3004 static inline int new_undef_syms(void)
3005 {
3006 int ret = 0;
3007 ret = new_undef_sym;
3008 new_undef_sym = 0;
3009 return ret;
3010 }
3011
3012 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3013 {
3014 char filename[1024], libname[1024];
3015 int t, group, nblibs = 0, ret = 0;
3016 char **libs = NULL;
3017
3018 group = !strcmp(cmd, "GROUP");
3019 if (!as_needed)
3020 new_undef_syms();
3021 t = ld_next(s1, filename, sizeof(filename));
3022 if (t != '(')
3023 expect("(");
3024 t = ld_next(s1, filename, sizeof(filename));
3025 for(;;) {
3026 libname[0] = '\0';
3027 if (t == LD_TOK_EOF) {
3028 tcc_error_noabort("unexpected end of file");
3029 ret = -1;
3030 goto lib_parse_error;
3031 } else if (t == ')') {
3032 break;
3033 } else if (t == '-') {
3034 t = ld_next(s1, filename, sizeof(filename));
3035 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3036 tcc_error_noabort("library name expected");
3037 ret = -1;
3038 goto lib_parse_error;
3039 }
3040 pstrcpy(libname, sizeof libname, &filename[1]);
3041 if (s1->static_link) {
3042 snprintf(filename, sizeof filename, "lib%s.a", libname);
3043 } else {
3044 snprintf(filename, sizeof filename, "lib%s.so", libname);
3045 }
3046 } else if (t != LD_TOK_NAME) {
3047 tcc_error_noabort("filename expected");
3048 ret = -1;
3049 goto lib_parse_error;
3050 }
3051 if (!strcmp(filename, "AS_NEEDED")) {
3052 ret = ld_add_file_list(s1, cmd, 1);
3053 if (ret)
3054 goto lib_parse_error;
3055 } else {
3056 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3057 if (!as_needed) {
3058 ret = ld_add_file(s1, filename);
3059 if (ret)
3060 goto lib_parse_error;
3061 if (group) {
3062 /* Add the filename *and* the libname to avoid future conversions */
3063 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3064 if (libname[0] != '\0')
3065 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3066 }
3067 }
3068 }
3069 t = ld_next(s1, filename, sizeof(filename));
3070 if (t == ',') {
3071 t = ld_next(s1, filename, sizeof(filename));
3072 }
3073 }
3074 if (group && !as_needed) {
3075 while (new_undef_syms()) {
3076 int i;
3077
3078 for (i = 0; i < nblibs; i ++)
3079 ld_add_file(s1, libs[i]);
3080 }
3081 }
3082 lib_parse_error:
3083 dynarray_reset(&libs, &nblibs);
3084 return ret;
3085 }
3086
3087 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3088 files */
3089 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3090 {
3091 char cmd[64];
3092 char filename[1024];
3093 int t, ret;
3094
3095 ch = file->buf_ptr[0];
3096 ch = handle_eob();
3097 for(;;) {
3098 t = ld_next(s1, cmd, sizeof(cmd));
3099 if (t == LD_TOK_EOF)
3100 return 0;
3101 else if (t != LD_TOK_NAME)
3102 return -1;
3103 if (!strcmp(cmd, "INPUT") ||
3104 !strcmp(cmd, "GROUP")) {
3105 ret = ld_add_file_list(s1, cmd, 0);
3106 if (ret)
3107 return ret;
3108 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3109 !strcmp(cmd, "TARGET")) {
3110 /* ignore some commands */
3111 t = ld_next(s1, cmd, sizeof(cmd));
3112 if (t != '(')
3113 expect("(");
3114 for(;;) {
3115 t = ld_next(s1, filename, sizeof(filename));
3116 if (t == LD_TOK_EOF) {
3117 tcc_error_noabort("unexpected end of file");
3118 return -1;
3119 } else if (t == ')') {
3120 break;
3121 }
3122 }
3123 } else {
3124 return -1;
3125 }
3126 }
3127 return 0;
3128 }
3129 #endif /* ndef TCC_TARGET_PE */
Tiny C Compiler main repository