?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
configure: prefer here-documents
[tinycc.git] / tccelf.c
blobda81d035a21eea91385e2465a991154c9a8db0ad
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, uplong 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 static void *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 NULL;
171 }
172 return (void*)(uplong)sym->st_value;
173 }
174
175 /* return elf symbol value */
176 LIBTCCAPI void *tcc_get_symbol(TCCState *s, const char *name)
177 {
178 return get_elf_sym_addr(s, name, 0);
179 }
180
181 /* return elf symbol value or error */
182 ST_FUNC void *tcc_get_symbol_err(TCCState *s, const char *name)
183 {
184 return get_elf_sym_addr(s, name, 1);
185 }
186
187 /* add an elf symbol : check if it is already defined and patch
188 it. Return symbol index. NOTE that sh_num can be SHN_UNDEF. */
189 ST_FUNC int add_elf_sym(Section *s, uplong value, unsigned long size,
190 int info, int other, int sh_num, const char *name)
191 {
192 ElfW(Sym) *esym;
193 int sym_bind, sym_index, sym_type, esym_bind;
194 unsigned char sym_vis, esym_vis, new_vis;
195
196 sym_bind = ELFW(ST_BIND)(info);
197 sym_type = ELFW(ST_TYPE)(info);
198 sym_vis = ELFW(ST_VISIBILITY)(other);
199
200 if (sym_bind != STB_LOCAL) {
201 /* we search global or weak symbols */
202 sym_index = find_elf_sym(s, name);
203 if (!sym_index)
204 goto do_def;
205 esym = &((ElfW(Sym) *)s->data)[sym_index];
206 if (esym->st_shndx != SHN_UNDEF) {
207 esym_bind = ELFW(ST_BIND)(esym->st_info);
208 /* propagate the most constraining visibility */
209 /* STV_DEFAULT(0)<STV_PROTECTED(3)<STV_HIDDEN(2)<STV_INTERNAL(1) */
210 esym_vis = ELFW(ST_VISIBILITY)(esym->st_other);
211 if (esym_vis == STV_DEFAULT) {
212 new_vis = sym_vis;
213 } else if (sym_vis == STV_DEFAULT) {
214 new_vis = esym_vis;
215 } else {
216 new_vis = (esym_vis < sym_vis) ? esym_vis : sym_vis;
217 }
218 esym->st_other = (esym->st_other & ~ELFW(ST_VISIBILITY)(-1))
219 | new_vis;
220 other = esym->st_other; /* in case we have to patch esym */
221 if (sh_num == SHN_UNDEF) {
222 /* ignore adding of undefined symbol if the
223 corresponding symbol is already defined */
224 } else if (sym_bind == STB_GLOBAL && esym_bind == STB_WEAK) {
225 /* global overrides weak, so patch */
226 goto do_patch;
227 } else if (sym_bind == STB_WEAK && esym_bind == STB_GLOBAL) {
228 /* weak is ignored if already global */
229 } else if (sym_bind == STB_WEAK && esym_bind == STB_WEAK) {
230 /* keep first-found weak definition, ignore subsequents */
231 } else if (sym_vis == STV_HIDDEN || sym_vis == STV_INTERNAL) {
232 /* ignore hidden symbols after */
233 } else if (esym->st_shndx == SHN_COMMON
234 && (sh_num < SHN_LORESERVE || sh_num == SHN_COMMON)) {
235 /* gr: Happens with 'tcc ... -static tcctest.c' on e.g. Ubuntu 6.01
236 No idea if this is the correct solution ... */
237 goto do_patch;
238 } else if (s == tcc_state->dynsymtab_section) {
239 /* we accept that two DLL define the same symbol */
240 } else {
241 #if 1
242 printf("new_bind=%x new_shndx=%x new_vis=%x old_bind=%x old_shndx=%x old_vis=%x\n",
243 sym_bind, sh_num, new_vis, esym_bind, esym->st_shndx, esym_vis);
244 #endif
245 tcc_error_noabort("'%s' defined twice", name);
246 }
247 } else {
248 do_patch:
249 esym->st_info = ELFW(ST_INFO)(sym_bind, sym_type);
250 esym->st_shndx = sh_num;
251 new_undef_sym = 1;
252 esym->st_value = value;
253 esym->st_size = size;
254 esym->st_other = other;
255 }
256 } else {
257 do_def:
258 sym_index = put_elf_sym(s, value, size,
259 ELFW(ST_INFO)(sym_bind, sym_type), other,
260 sh_num, name);
261 }
262 return sym_index;
263 }
264
265 /* put relocation */
266 ST_FUNC void put_elf_reloc(Section *symtab, Section *s, unsigned long offset,
267 int type, int symbol)
268 {
269 char buf[256];
270 Section *sr;
271 ElfW_Rel *rel;
272
273 sr = s->reloc;
274 if (!sr) {
275 /* if no relocation section, create it */
276 snprintf(buf, sizeof(buf), REL_SECTION_FMT, s->name);
277 /* if the symtab is allocated, then we consider the relocation
278 are also */
279 sr = new_section(tcc_state, buf, SHT_RELX, symtab->sh_flags);
280 sr->sh_entsize = sizeof(ElfW_Rel);
281 sr->link = symtab;
282 sr->sh_info = s->sh_num;
283 s->reloc = sr;
284 }
285 rel = section_ptr_add(sr, sizeof(ElfW_Rel));
286 rel->r_offset = offset;
287 rel->r_info = ELFW(R_INFO)(symbol, type);
288 #ifdef TCC_TARGET_X86_64
289 rel->r_addend = 0;
290 #endif
291 }
292
293 /* put stab debug information */
294
295 ST_FUNC void put_stabs(const char *str, int type, int other, int desc,
296 unsigned long value)
297 {
298 Stab_Sym *sym;
299
300 sym = section_ptr_add(stab_section, sizeof(Stab_Sym));
301 if (str) {
302 sym->n_strx = put_elf_str(stabstr_section, str);
303 } else {
304 sym->n_strx = 0;
305 }
306 sym->n_type = type;
307 sym->n_other = other;
308 sym->n_desc = desc;
309 sym->n_value = value;
310 }
311
312 ST_FUNC void put_stabs_r(const char *str, int type, int other, int desc,
313 unsigned long value, Section *sec, int sym_index)
314 {
315 put_stabs(str, type, other, desc, value);
316 put_elf_reloc(symtab_section, stab_section,
317 stab_section->data_offset - sizeof(unsigned int),
318 R_DATA_32, sym_index);
319 }
320
321 ST_FUNC void put_stabn(int type, int other, int desc, int value)
322 {
323 put_stabs(NULL, type, other, desc, value);
324 }
325
326 ST_FUNC void put_stabd(int type, int other, int desc)
327 {
328 put_stabs(NULL, type, other, desc, 0);
329 }
330
331 /* In an ELF file symbol table, the local symbols must appear below
332 the global and weak ones. Since TCC cannot sort it while generating
333 the code, we must do it after. All the relocation tables are also
334 modified to take into account the symbol table sorting */
335 static void sort_syms(TCCState *s1, Section *s)
336 {
337 int *old_to_new_syms;
338 ElfW(Sym) *new_syms;
339 int nb_syms, i;
340 ElfW(Sym) *p, *q;
341 ElfW_Rel *rel, *rel_end;
342 Section *sr;
343 int type, sym_index;
344
345 nb_syms = s->data_offset / sizeof(ElfW(Sym));
346 new_syms = tcc_malloc(nb_syms * sizeof(ElfW(Sym)));
347 old_to_new_syms = tcc_malloc(nb_syms * sizeof(int));
348
349 /* first pass for local symbols */
350 p = (ElfW(Sym) *)s->data;
351 q = new_syms;
352 for(i = 0; i < nb_syms; i++) {
353 if (ELFW(ST_BIND)(p->st_info) == STB_LOCAL) {
354 old_to_new_syms[i] = q - new_syms;
355 *q++ = *p;
356 }
357 p++;
358 }
359 /* save the number of local symbols in section header */
360 s->sh_info = q - new_syms;
361
362 /* then second pass for non local symbols */
363 p = (ElfW(Sym) *)s->data;
364 for(i = 0; i < nb_syms; i++) {
365 if (ELFW(ST_BIND)(p->st_info) != STB_LOCAL) {
366 old_to_new_syms[i] = q - new_syms;
367 *q++ = *p;
368 }
369 p++;
370 }
371
372 /* we copy the new symbols to the old */
373 memcpy(s->data, new_syms, nb_syms * sizeof(ElfW(Sym)));
374 tcc_free(new_syms);
375
376 /* now we modify all the relocations */
377 for(i = 1; i < s1->nb_sections; i++) {
378 sr = s1->sections[i];
379 if (sr->sh_type == SHT_RELX && sr->link == s) {
380 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
381 for(rel = (ElfW_Rel *)sr->data;
382 rel < rel_end;
383 rel++) {
384 sym_index = ELFW(R_SYM)(rel->r_info);
385 type = ELFW(R_TYPE)(rel->r_info);
386 sym_index = old_to_new_syms[sym_index];
387 rel->r_info = ELFW(R_INFO)(sym_index, type);
388 }
389 }
390 }
391
392 tcc_free(old_to_new_syms);
393 }
394
395 /* relocate common symbols in the .bss section */
396 ST_FUNC void relocate_common_syms(void)
397 {
398 ElfW(Sym) *sym, *sym_end;
399 unsigned long offset, align;
400
401 sym_end = (ElfW(Sym) *)(symtab_section->data + symtab_section->data_offset);
402 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
403 sym < sym_end;
404 sym++) {
405 if (sym->st_shndx == SHN_COMMON) {
406 /* align symbol */
407 align = sym->st_value;
408 offset = bss_section->data_offset;
409 offset = (offset + align - 1) & -align;
410 sym->st_value = offset;
411 sym->st_shndx = bss_section->sh_num;
412 offset += sym->st_size;
413 bss_section->data_offset = offset;
414 }
415 }
416 }
417
418 /* relocate symbol table, resolve undefined symbols if do_resolve is
419 true and output error if undefined symbol. */
420 ST_FUNC void relocate_syms(TCCState *s1, int do_resolve)
421 {
422 ElfW(Sym) *sym, *esym, *sym_end;
423 int sym_bind, sh_num, sym_index;
424 const char *name;
425
426 sym_end = (ElfW(Sym) *)(symtab_section->data + symtab_section->data_offset);
427 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
428 sym < sym_end;
429 sym++) {
430 sh_num = sym->st_shndx;
431 if (sh_num == SHN_UNDEF) {
432 name = strtab_section->data + sym->st_name;
433 if (do_resolve) {
434 #if !defined TCC_TARGET_PE || !defined _WIN32
435 void *addr;
436 name = symtab_section->link->data + sym->st_name;
437 addr = resolve_sym(s1, name);
438 if (addr) {
439 sym->st_value = (uplong)addr;
440 goto found;
441 }
442 #endif
443 } else if (s1->dynsym) {
444 /* if dynamic symbol exist, then use it */
445 sym_index = find_elf_sym(s1->dynsym, name);
446 if (sym_index) {
447 esym = &((ElfW(Sym) *)s1->dynsym->data)[sym_index];
448 sym->st_value = esym->st_value;
449 goto found;
450 }
451 }
452 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
453 it */
454 if (!strcmp(name, "_fp_hw"))
455 goto found;
456 /* only weak symbols are accepted to be undefined. Their
457 value is zero */
458 sym_bind = ELFW(ST_BIND)(sym->st_info);
459 if (sym_bind == STB_WEAK) {
460 sym->st_value = 0;
461 } else {
462 tcc_error_noabort("undefined symbol '%s'", name);
463 }
464 } else if (sh_num < SHN_LORESERVE) {
465 /* add section base */
466 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
467 }
468 found: ;
469 }
470 }
471
472 #ifndef TCC_TARGET_PE
473 #ifdef TCC_TARGET_X86_64
474 #define JMP_TABLE_ENTRY_SIZE 14
475 static uplong add_jmp_table(TCCState *s1, uplong val)
476 {
477 char *p = s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset;
478 s1->runtime_plt_and_got_offset += JMP_TABLE_ENTRY_SIZE;
479 /* jmp *0x0(%rip) */
480 p[0] = 0xff;
481 p[1] = 0x25;
482 *(int *)(p + 2) = 0;
483 *(uplong *)(p + 6) = val;
484 return (uplong)p;
485 }
486
487 static uplong add_got_table(TCCState *s1, uplong val)
488 {
489 uplong *p = (uplong *)(s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset);
490 s1->runtime_plt_and_got_offset += sizeof(uplong);
491 *p = val;
492 return (uplong)p;
493 }
494 #elif defined TCC_TARGET_ARM
495 #define JMP_TABLE_ENTRY_SIZE 8
496 static uplong add_jmp_table(TCCState *s1, int val)
497 {
498 uint32_t *p = (uint32_t *)(s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset);
499 s1->runtime_plt_and_got_offset += JMP_TABLE_ENTRY_SIZE;
500 /* ldr pc, [pc, #-4] */
501 p[0] = 0xE51FF004;
502 p[1] = val;
503 return (uplong)p;
504 }
505 #endif
506 #endif
507
508 /* relocate a given section (CPU dependent) */
509 ST_FUNC void relocate_section(TCCState *s1, Section *s)
510 {
511 Section *sr;
512 ElfW_Rel *rel, *rel_end, *qrel;
513 ElfW(Sym) *sym;
514 int type, sym_index;
515 unsigned char *ptr;
516 uplong val, addr;
517 #if defined TCC_TARGET_I386 || defined TCC_TARGET_X86_64
518 int esym_index;
519 #endif
520
521 sr = s->reloc;
522 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
523 qrel = (ElfW_Rel *)sr->data;
524 for(rel = qrel;
525 rel < rel_end;
526 rel++) {
527 ptr = s->data + rel->r_offset;
528
529 sym_index = ELFW(R_SYM)(rel->r_info);
530 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
531 val = sym->st_value;
532 #ifdef TCC_TARGET_X86_64
533 val += rel->r_addend;
534 #endif
535 type = ELFW(R_TYPE)(rel->r_info);
536 addr = s->sh_addr + rel->r_offset;
537
538 /* CPU specific */
539 switch(type) {
540 #if defined(TCC_TARGET_I386)
541 case R_386_32:
542 if (s1->output_type == TCC_OUTPUT_DLL) {
543 esym_index = s1->symtab_to_dynsym[sym_index];
544 qrel->r_offset = rel->r_offset;
545 if (esym_index) {
546 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_32);
547 qrel++;
548 break;
549 } else {
550 qrel->r_info = ELFW(R_INFO)(0, R_386_RELATIVE);
551 qrel++;
552 }
553 }
554 *(int *)ptr += val;
555 break;
556 case R_386_PC32:
557 if (s1->output_type == TCC_OUTPUT_DLL) {
558 /* DLL relocation */
559 esym_index = s1->symtab_to_dynsym[sym_index];
560 if (esym_index) {
561 qrel->r_offset = rel->r_offset;
562 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_PC32);
563 qrel++;
564 break;
565 }
566 }
567 *(int *)ptr += val - addr;
568 break;
569 case R_386_PLT32:
570 *(int *)ptr += val - addr;
571 break;
572 case R_386_GLOB_DAT:
573 case R_386_JMP_SLOT:
574 *(int *)ptr = val;
575 break;
576 case R_386_GOTPC:
577 *(int *)ptr += s1->got->sh_addr - addr;
578 break;
579 case R_386_GOTOFF:
580 *(int *)ptr += val - s1->got->sh_addr;
581 break;
582 case R_386_GOT32:
583 /* we load the got offset */
584 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
585 break;
586 case R_386_16:
587 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY) {
588 output_file:
589 tcc_error("can only produce 16-bit binary files");
590 }
591 *(short *)ptr += val;
592 break;
593 case R_386_PC16:
594 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY)
595 goto output_file;
596 *(short *)ptr += val - addr;
597 break;
598 #elif defined(TCC_TARGET_ARM)
599 case R_ARM_PC24:
600 case R_ARM_CALL:
601 case R_ARM_JUMP24:
602 case R_ARM_PLT32:
603 {
604 int x, is_thumb, is_call, h, blx_avail;
605 x = (*(int *)ptr)&0xffffff;
606 (*(int *)ptr) &= 0xff000000;
607 if (x & 0x800000)
608 x -= 0x1000000;
609 x <<= 2;
610 blx_avail = (TCC_ARM_VERSION >= 5);
611 is_thumb = val & 1;
612 is_call = (type == R_ARM_CALL);
613 x += val - addr;
614 h = x & 2;
615 #ifndef TCC_TARGET_PE
616 if ((x & 3) || x >= 0x4000000 || x < -0x4000000)
617 if (!(x & 3) || !blx_avail || !is_call)
618 if (s1->output_type == TCC_OUTPUT_MEMORY) {
619 x += add_jmp_table(s1, val) - val; /* add veneer */
620 is_thumb = 0; /* Veneer uses ARM instructions */
621 }
622 #endif
623 if ((x & 3) || x >= 0x4000000 || x < -0x4000000)
624 if (!(x & 3) || !blx_avail || !is_call)
625 tcc_error("can't relocate value at %x",addr);
626 x >>= 2;
627 x &= 0xffffff;
628 /* Only reached if blx is avail and it is a call */
629 if (is_thumb) {
630 x |= h << 24;
631 (*(int *)ptr) = 0xfa << 24; /* bl -> blx */
632 }
633 (*(int *)ptr) |= x;
634 }
635 break;
636 /* Since these relocations only concern Thumb-2 and blx instruction was
637 introduced before Thumb-2, we can assume blx is available and not
638 guard its use */
639 case R_ARM_THM_CALL:
640 case R_ARM_THM_JUMP24:
641 {
642 int x, hi, lo, s, j1, j2, i1, i2, imm10, imm11;
643 int to_thumb, is_call, to_plt, blx_bit = 1 << 12;
644 Section *plt;
645
646 /* weak reference */
647 if (sym->st_shndx == SHN_UNDEF &&
648 ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
649 break;
650
651 /* Get initial offset */
652 hi = (*(uint16_t *)ptr);
653 lo = (*(uint16_t *)(ptr+2));
654 s = (hi >> 10) & 1;
655 j1 = (lo >> 13) & 1;
656 j2 = (lo >> 11) & 1;
657 i1 = (j1 ^ s) ^ 1;
658 i2 = (j2 ^ s) ^ 1;
659 imm10 = hi & 0x3ff;
660 imm11 = lo & 0x7ff;
661 x = (s << 24) | (i1 << 23) | (i2 << 22) |
662 (imm10 << 12) | (imm11 << 1);
663 if (x & 0x01000000)
664 x -= 0x02000000;
665
666 /* Relocation infos */
667 to_thumb = val & 1;
668 plt = s1->plt;
669 to_plt = (val >= plt->sh_addr) &&
670 (val < plt->sh_addr + plt->data_offset);
671 is_call = (type == R_ARM_THM_CALL);
672
673 /* Compute final offset */
674 if (to_plt && !is_call) /* Point to 1st instr of Thumb stub */
675 x -= 4;
676 x += val - addr;
677 if (!to_thumb && is_call) {
678 blx_bit = 0; /* bl -> blx */
679 x = (x + 3) & -4; /* Compute offset from aligned PC */
680 }
681
682 /* Check that relocation is possible
683 * offset must not be out of range
684 * if target is to be entered in arm mode:
685 - bit 1 must not set
686 - instruction must be a call (bl) or a jump to PLT */
687 if (!to_thumb || x >= 0x1000000 || x < -0x1000000)
688 if (to_thumb || (val & 2) || (!is_call && !to_plt))
689 tcc_error("can't relocate value at %x",addr);
690
691 /* Compute and store final offset */
692 s = (x >> 24) & 1;
693 i1 = (x >> 23) & 1;
694 i2 = (x >> 22) & 1;
695 j1 = s ^ (i1 ^ 1);
696 j2 = s ^ (i2 ^ 1);
697 imm10 = (x >> 12) & 0x3ff;
698 imm11 = (x >> 1) & 0x7ff;
699 (*(uint16_t *)ptr) = (uint16_t) ((hi & 0xf800) |
700 (s << 10) | imm10);
701 (*(uint16_t *)(ptr+2)) = (uint16_t) ((lo & 0xc000) |
702 (j1 << 13) | blx_bit | (j2 << 11) |
703 imm11);
704 }
705 break;
706 case R_ARM_MOVT_ABS:
707 case R_ARM_MOVW_ABS_NC:
708 {
709 int x, imm4, imm12;
710 if (type == R_ARM_MOVT_ABS)
711 val >>= 16;
712 imm12 = val & 0xfff;
713 imm4 = (val >> 12) & 0xf;
714 x = (imm4 << 16) | imm12;
715 if (type == R_ARM_THM_MOVT_ABS)
716 *(int *)ptr |= x;
717 else
718 *(int *)ptr += x;
719 }
720 break;
721 case R_ARM_THM_MOVT_ABS:
722 case R_ARM_THM_MOVW_ABS_NC:
723 {
724 int x, i, imm4, imm3, imm8;
725 if (type == R_ARM_THM_MOVT_ABS)
726 val >>= 16;
727 imm8 = val & 0xff;
728 imm3 = (val >> 8) & 0x7;
729 i = (val >> 11) & 1;
730 imm4 = (val >> 12) & 0xf;
731 x = (imm3 << 28) | (imm8 << 16) | (i << 10) | imm4;
732 if (type == R_ARM_THM_MOVT_ABS)
733 *(int *)ptr |= x;
734 else
735 *(int *)ptr += x;
736 }
737 break;
738 case R_ARM_PREL31:
739 {
740 int x;
741 x = (*(int *)ptr) & 0x7fffffff;
742 (*(int *)ptr) &= 0x80000000;
743 x = (x * 2) / 2;
744 x += val - addr;
745 if((x^(x>>1))&0x40000000)
746 tcc_error("can't relocate value at %x",addr);
747 (*(int *)ptr) |= x & 0x7fffffff;
748 }
749 case R_ARM_ABS32:
750 *(int *)ptr += val;
751 break;
752 case R_ARM_REL32:
753 *(int *)ptr += val - addr;
754 break;
755 case R_ARM_BASE_PREL:
756 *(int *)ptr += s1->got->sh_addr - addr;
757 break;
758 case R_ARM_GOTOFF32:
759 *(int *)ptr += val - s1->got->sh_addr;
760 break;
761 case R_ARM_GOT_BREL:
762 /* we load the got offset */
763 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
764 break;
765 case R_ARM_COPY:
766 break;
767 case R_ARM_V4BX:
768 /* trade Thumb support for ARMv4 support */
769 if ((0x0ffffff0 & *(int*)ptr) == 0x012FFF10)
770 *(int*)ptr ^= 0xE12FFF10 ^ 0xE1A0F000; /* BX Rm -> MOV PC, Rm */
771 break;
772 default:
773 fprintf(stderr,"FIXME: handle reloc type %x at %x [%.8x] to %x\n",
774 type, (unsigned)addr, (unsigned)(uplong)ptr, (unsigned)val);
775 break;
776 #elif defined(TCC_TARGET_C67)
777 case R_C60_32:
778 *(int *)ptr += val;
779 break;
780 case R_C60LO16:
781 {
782 uint32_t orig;
783
784 /* put the low 16 bits of the absolute address */
785 // add to what is already there
786
787 orig = ((*(int *)(ptr )) >> 7) & 0xffff;
788 orig |= (((*(int *)(ptr+4)) >> 7) & 0xffff) << 16;
789
790 //patch both at once - assumes always in pairs Low - High
791
792 *(int *) ptr = (*(int *) ptr & (~(0xffff << 7)) ) | (((val+orig) & 0xffff) << 7);
793 *(int *)(ptr+4) = (*(int *)(ptr+4) & (~(0xffff << 7)) ) | ((((val+orig)>>16) & 0xffff) << 7);
794 }
795 break;
796 case R_C60HI16:
797 break;
798 default:
799 fprintf(stderr,"FIXME: handle reloc type %x at %x [%.8x] to %x\n",
800 type, (unsigned)addr, (unsigned)(uplong)ptr, (unsigned)val);
801 break;
802 #elif defined(TCC_TARGET_X86_64)
803 case R_X86_64_64:
804 if (s1->output_type == TCC_OUTPUT_DLL) {
805 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
806 qrel->r_addend = *(long long *)ptr + val;
807 qrel++;
808 }
809 *(long long *)ptr += val;
810 break;
811 case R_X86_64_32:
812 case R_X86_64_32S:
813 if (s1->output_type == TCC_OUTPUT_DLL) {
814 /* XXX: this logic may depend on TCC's codegen
815 now TCC uses R_X86_64_32 even for a 64bit pointer */
816 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
817 qrel->r_addend = *(int *)ptr + val;
818 qrel++;
819 }
820 *(int *)ptr += val;
821 break;
822
823 case R_X86_64_PC32:
824 if (s1->output_type == TCC_OUTPUT_DLL) {
825 /* DLL relocation */
826 esym_index = s1->symtab_to_dynsym[sym_index];
827 if (esym_index) {
828 qrel->r_offset = rel->r_offset;
829 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_PC32);
830 qrel->r_addend = *(int *)ptr;
831 qrel++;
832 break;
833 }
834 }
835 /* fall through */
836 case R_X86_64_PLT32: {
837 long long diff;
838 diff = (long long)val - addr;
839 if (diff <= -2147483647 || diff > 2147483647) {
840 #ifndef TCC_TARGET_PE
841 /* XXX: naive support for over 32bit jump */
842 if (s1->output_type == TCC_OUTPUT_MEMORY) {
843 val = (add_jmp_table(s1, val - rel->r_addend) +
844 rel->r_addend);
845 diff = val - addr;
846 }
847 #endif
848 if (diff <= -2147483647 || diff > 2147483647) {
849 tcc_error("internal error: relocation failed");
850 }
851 }
852 *(int *)ptr += diff;
853 }
854 break;
855 case R_X86_64_GLOB_DAT:
856 case R_X86_64_JUMP_SLOT:
857 /* They don't need addend */
858 *(int *)ptr = val - rel->r_addend;
859 break;
860 case R_X86_64_GOTPCREL:
861 #ifndef TCC_TARGET_PE
862 if (s1->output_type == TCC_OUTPUT_MEMORY) {
863 val = add_got_table(s1, val - rel->r_addend) + rel->r_addend;
864 *(int *)ptr += val - addr;
865 break;
866 }
867 #endif
868 *(int *)ptr += (s1->got->sh_addr - addr +
869 s1->sym_attrs[sym_index].got_offset - 4);
870 break;
871 case R_X86_64_GOTTPOFF:
872 *(int *)ptr += val - s1->got->sh_addr;
873 break;
874 case R_X86_64_GOT32:
875 /* we load the got offset */
876 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
877 break;
878 #else
879 #error unsupported processor
880 #endif
881 }
882 }
883 /* if the relocation is allocated, we change its symbol table */
884 if (sr->sh_flags & SHF_ALLOC)
885 sr->link = s1->dynsym;
886 }
887
888 /* relocate relocation table in 'sr' */
889 static void relocate_rel(TCCState *s1, Section *sr)
890 {
891 Section *s;
892 ElfW_Rel *rel, *rel_end;
893
894 s = s1->sections[sr->sh_info];
895 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
896 for(rel = (ElfW_Rel *)sr->data;
897 rel < rel_end;
898 rel++) {
899 rel->r_offset += s->sh_addr;
900 }
901 }
902
903 /* count the number of dynamic relocations so that we can reserve
904 their space */
905 static int prepare_dynamic_rel(TCCState *s1, Section *sr)
906 {
907 ElfW_Rel *rel, *rel_end;
908 int sym_index, esym_index, type, count;
909
910 count = 0;
911 rel_end = (ElfW_Rel *)(sr->data + sr->data_offset);
912 for(rel = (ElfW_Rel *)sr->data; rel < rel_end; rel++) {
913 sym_index = ELFW(R_SYM)(rel->r_info);
914 type = ELFW(R_TYPE)(rel->r_info);
915 switch(type) {
916 #if defined(TCC_TARGET_I386)
917 case R_386_32:
918 #elif defined(TCC_TARGET_X86_64)
919 case R_X86_64_32:
920 case R_X86_64_32S:
921 case R_X86_64_64:
922 #endif
923 count++;
924 break;
925 #if defined(TCC_TARGET_I386)
926 case R_386_PC32:
927 #elif defined(TCC_TARGET_X86_64)
928 case R_X86_64_PC32:
929 #endif
930 esym_index = s1->symtab_to_dynsym[sym_index];
931 if (esym_index)
932 count++;
933 break;
934 default:
935 break;
936 }
937 }
938 if (count) {
939 /* allocate the section */
940 sr->sh_flags |= SHF_ALLOC;
941 sr->sh_size = count * sizeof(ElfW_Rel);
942 }
943 return count;
944 }
945
946 static struct sym_attr *alloc_sym_attr(TCCState *s1, int index)
947 {
948 int n;
949 struct sym_attr *tab;
950
951 if (index >= s1->nb_sym_attrs) {
952 /* find immediately bigger power of 2 and reallocate array */
953 n = 1;
954 while (index >= n)
955 n *= 2;
956 tab = tcc_realloc(s1->sym_attrs, n * sizeof(*s1->sym_attrs));
957 s1->sym_attrs = tab;
958 memset(s1->sym_attrs + s1->nb_sym_attrs, 0,
959 (n - s1->nb_sym_attrs) * sizeof(*s1->sym_attrs));
960 s1->nb_sym_attrs = n;
961 }
962 return &s1->sym_attrs[index];
963 }
964
965 /* XXX: suppress that */
966 static void put32(unsigned char *p, uint32_t val)
967 {
968 p[0] = val;
969 p[1] = val >> 8;
970 p[2] = val >> 16;
971 p[3] = val >> 24;
972 }
973
974 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_ARM) || \
975 defined(TCC_TARGET_X86_64)
976 static uint32_t get32(unsigned char *p)
977 {
978 return p[0] | (p[1] << 8) | (p[2] << 16) | (p[3] << 24);
979 }
980 #endif
981
982 static void build_got(TCCState *s1)
983 {
984 unsigned char *ptr;
985
986 /* if no got, then create it */
987 s1->got = new_section(s1, ".got", SHT_PROGBITS, SHF_ALLOC | SHF_WRITE);
988 s1->got->sh_entsize = 4;
989 add_elf_sym(symtab_section, 0, 4, ELFW(ST_INFO)(STB_GLOBAL, STT_OBJECT),
990 0, s1->got->sh_num, "_GLOBAL_OFFSET_TABLE_");
991 ptr = section_ptr_add(s1->got, 3 * PTR_SIZE);
992 #if PTR_SIZE == 4
993 /* keep space for _DYNAMIC pointer, if present */
994 put32(ptr, 0);
995 /* two dummy got entries */
996 put32(ptr + 4, 0);
997 put32(ptr + 8, 0);
998 #else
999 /* keep space for _DYNAMIC pointer, if present */
1000 put32(ptr, 0);
1001 put32(ptr + 4, 0);
1002 /* two dummy got entries */
1003 put32(ptr + 8, 0);
1004 put32(ptr + 12, 0);
1005 put32(ptr + 16, 0);
1006 put32(ptr + 20, 0);
1007 #endif
1008 }
1009
1010 /* put a got entry corresponding to a symbol in symtab_section. 'size'
1011 and 'info' can be modifed if more precise info comes from the DLL */
1012 static void put_got_entry(TCCState *s1,
1013 int reloc_type, unsigned long size, int info,
1014 int sym_index)
1015 {
1016 int index;
1017 const char *name;
1018 ElfW(Sym) *sym;
1019 unsigned long offset;
1020 int *ptr;
1021
1022 if (!s1->got)
1023 build_got(s1);
1024
1025 /* if a got entry already exists for that symbol, no need to add one */
1026 if (sym_index < s1->nb_sym_attrs &&
1027 s1->sym_attrs[sym_index].got_offset)
1028 return;
1029
1030 alloc_sym_attr(s1, sym_index)->got_offset = s1->got->data_offset;
1031
1032 if (s1->dynsym) {
1033 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1034 name = symtab_section->link->data + sym->st_name;
1035 offset = sym->st_value;
1036 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_X86_64)
1037 if (reloc_type ==
1038 #ifdef TCC_TARGET_X86_64
1039 R_X86_64_JUMP_SLOT
1040 #else
1041 R_386_JMP_SLOT
1042 #endif
1043 ) {
1044 Section *plt;
1045 uint8_t *p;
1046 int modrm;
1047
1048 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1049 modrm = 0x25;
1050 #else
1051 /* if we build a DLL, we add a %ebx offset */
1052 if (s1->output_type == TCC_OUTPUT_DLL)
1053 modrm = 0xa3;
1054 else
1055 modrm = 0x25;
1056 #endif
1057
1058 /* add a PLT entry */
1059 plt = s1->plt;
1060 if (plt->data_offset == 0) {
1061 /* first plt entry */
1062 p = section_ptr_add(plt, 16);
1063 p[0] = 0xff; /* pushl got + PTR_SIZE */
1064 p[1] = modrm + 0x10;
1065 put32(p + 2, PTR_SIZE);
1066 p[6] = 0xff; /* jmp *(got + PTR_SIZE * 2) */
1067 p[7] = modrm;
1068 put32(p + 8, PTR_SIZE * 2);
1069 }
1070
1071 p = section_ptr_add(plt, 16);
1072 p[0] = 0xff; /* jmp *(got + x) */
1073 p[1] = modrm;
1074 put32(p + 2, s1->got->data_offset);
1075 p[6] = 0x68; /* push $xxx */
1076 put32(p + 7, (plt->data_offset - 32) >> 1);
1077 p[11] = 0xe9; /* jmp plt_start */
1078 put32(p + 12, -(plt->data_offset));
1079
1080 /* the symbol is modified so that it will be relocated to
1081 the PLT */
1082 #if !defined(TCC_OUTPUT_DLL_WITH_PLT)
1083 if (s1->output_type == TCC_OUTPUT_EXE)
1084 #endif
1085 offset = plt->data_offset - 16;
1086 }
1087 #elif defined(TCC_TARGET_ARM)
1088 if (reloc_type == R_ARM_JUMP_SLOT) {
1089 Section *plt;
1090 uint8_t *p;
1091
1092 /* if we build a DLL, we add a %ebx offset */
1093 if (s1->output_type == TCC_OUTPUT_DLL)
1094 tcc_error("DLLs unimplemented!");
1095
1096 /* add a PLT entry */
1097 plt = s1->plt;
1098 if (plt->data_offset == 0) {
1099 /* first plt entry */
1100 p = section_ptr_add(plt, 16);
1101 put32(p , 0xe52de004);
1102 put32(p + 4, 0xe59fe010);
1103 put32(p + 8, 0xe08fe00e);
1104 put32(p + 12, 0xe5bef008);
1105 }
1106
1107 if (s1->sym_attrs[sym_index].plt_thumb_stub) {
1108 p = section_ptr_add(plt, 20);
1109 put32(p , 0x4778); // bx pc
1110 put32(p+2, 0x46c0); // nop
1111 p += 4;
1112 } else
1113 p = section_ptr_add(plt, 16);
1114 put32(p , 0xe59fc004); // ldr ip, [pc, #4] // offset in GOT
1115 put32(p+4, 0xe08fc00c); // add ip, pc, ip // absolute address or offset
1116 put32(p+8, 0xe59cf000); // ldr pc, [ip] // load absolute address or load offset
1117 put32(p+12, s1->got->data_offset);
1118
1119 /* the symbol is modified so that it will be relocated to
1120 the PLT */
1121 if (s1->output_type == TCC_OUTPUT_EXE)
1122 offset = plt->data_offset - 16;
1123 }
1124 #elif defined(TCC_TARGET_C67)
1125 tcc_error("C67 got not implemented");
1126 #else
1127 #error unsupported CPU
1128 #endif
1129 index = put_elf_sym(s1->dynsym, offset,
1130 size, info, 0, sym->st_shndx, name);
1131 /* put a got entry */
1132 put_elf_reloc(s1->dynsym, s1->got,
1133 s1->got->data_offset,
1134 reloc_type, index);
1135 }
1136 ptr = section_ptr_add(s1->got, PTR_SIZE);
1137 *ptr = 0;
1138 }
1139
1140 /* build GOT and PLT entries */
1141 ST_FUNC void build_got_entries(TCCState *s1)
1142 {
1143 Section *s;
1144 ElfW_Rel *rel, *rel_end;
1145 ElfW(Sym) *sym;
1146 int i, type, reloc_type, sym_index;
1147
1148 for(i = 1; i < s1->nb_sections; i++) {
1149 s = s1->sections[i];
1150 if (s->sh_type != SHT_RELX)
1151 continue;
1152 /* no need to handle got relocations */
1153 if (s->link != symtab_section)
1154 continue;
1155 rel_end = (ElfW_Rel *)(s->data + s->data_offset);
1156 for(rel = (ElfW_Rel *)s->data;
1157 rel < rel_end;
1158 rel++) {
1159 type = ELFW(R_TYPE)(rel->r_info);
1160 switch(type) {
1161 #if defined(TCC_TARGET_I386)
1162 case R_386_GOT32:
1163 case R_386_GOTOFF:
1164 case R_386_GOTPC:
1165 case R_386_PLT32:
1166 if (!s1->got)
1167 build_got(s1);
1168 if (type == R_386_GOT32 || type == R_386_PLT32) {
1169 sym_index = ELFW(R_SYM)(rel->r_info);
1170 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1171 /* look at the symbol got offset. If none, then add one */
1172 if (type == R_386_GOT32)
1173 reloc_type = R_386_GLOB_DAT;
1174 else
1175 reloc_type = R_386_JMP_SLOT;
1176 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1177 sym_index);
1178 }
1179 break;
1180 #elif defined(TCC_TARGET_ARM)
1181 case R_ARM_GOT_BREL:
1182 case R_ARM_GOTOFF32:
1183 case R_ARM_BASE_PREL:
1184 case R_ARM_PLT32:
1185 if (!s1->got)
1186 build_got(s1);
1187 if (type == R_ARM_GOT_BREL || type == R_ARM_PLT32) {
1188 sym_index = ELFW(R_SYM)(rel->r_info);
1189 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1190 /* look at the symbol got offset. If none, then add one */
1191 if (type == R_ARM_GOT_BREL)
1192 reloc_type = R_ARM_GLOB_DAT;
1193 else
1194 reloc_type = R_ARM_JUMP_SLOT;
1195 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1196 sym_index);
1197 }
1198 break;
1199 #elif defined(TCC_TARGET_C67)
1200 case R_C60_GOT32:
1201 case R_C60_GOTOFF:
1202 case R_C60_GOTPC:
1203 case R_C60_PLT32:
1204 if (!s1->got)
1205 build_got(s1);
1206 if (type == R_C60_GOT32 || type == R_C60_PLT32) {
1207 sym_index = ELFW(R_SYM)(rel->r_info);
1208 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1209 /* look at the symbol got offset. If none, then add one */
1210 if (type == R_C60_GOT32)
1211 reloc_type = R_C60_GLOB_DAT;
1212 else
1213 reloc_type = R_C60_JMP_SLOT;
1214 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1215 sym_index);
1216 }
1217 break;
1218 #elif defined(TCC_TARGET_X86_64)
1219 case R_X86_64_GOT32:
1220 case R_X86_64_GOTTPOFF:
1221 case R_X86_64_GOTPCREL:
1222 case R_X86_64_PLT32:
1223 if (!s1->got)
1224 build_got(s1);
1225 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL ||
1226 type == R_X86_64_PLT32) {
1227 sym_index = ELFW(R_SYM)(rel->r_info);
1228 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1229 /* look at the symbol got offset. If none, then add one */
1230 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL)
1231 reloc_type = R_X86_64_GLOB_DAT;
1232 else
1233 reloc_type = R_X86_64_JUMP_SLOT;
1234 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1235 sym_index);
1236 }
1237 break;
1238 #else
1239 #error unsupported CPU
1240 #endif
1241 default:
1242 break;
1243 }
1244 }
1245 }
1246 }
1247
1248 ST_FUNC Section *new_symtab(TCCState *s1,
1249 const char *symtab_name, int sh_type, int sh_flags,
1250 const char *strtab_name,
1251 const char *hash_name, int hash_sh_flags)
1252 {
1253 Section *symtab, *strtab, *hash;
1254 int *ptr, nb_buckets;
1255
1256 symtab = new_section(s1, symtab_name, sh_type, sh_flags);
1257 symtab->sh_entsize = sizeof(ElfW(Sym));
1258 strtab = new_section(s1, strtab_name, SHT_STRTAB, sh_flags);
1259 put_elf_str(strtab, "");
1260 symtab->link = strtab;
1261 put_elf_sym(symtab, 0, 0, 0, 0, 0, NULL);
1262
1263 nb_buckets = 1;
1264
1265 hash = new_section(s1, hash_name, SHT_HASH, hash_sh_flags);
1266 hash->sh_entsize = sizeof(int);
1267 symtab->hash = hash;
1268 hash->link = symtab;
1269
1270 ptr = section_ptr_add(hash, (2 + nb_buckets + 1) * sizeof(int));
1271 ptr[0] = nb_buckets;
1272 ptr[1] = 1;
1273 memset(ptr + 2, 0, (nb_buckets + 1) * sizeof(int));
1274 return symtab;
1275 }
1276
1277 /* put dynamic tag */
1278 static void put_dt(Section *dynamic, int dt, uplong val)
1279 {
1280 ElfW(Dyn) *dyn;
1281 dyn = section_ptr_add(dynamic, sizeof(ElfW(Dyn)));
1282 dyn->d_tag = dt;
1283 dyn->d_un.d_val = val;
1284 }
1285
1286 static void add_init_array_defines(TCCState *s1, const char *section_name)
1287 {
1288 Section *s;
1289 long end_offset;
1290 char sym_start[1024];
1291 char sym_end[1024];
1292
1293 snprintf(sym_start, sizeof(sym_start), "__%s_start", section_name + 1);
1294 snprintf(sym_end, sizeof(sym_end), "__%s_end", section_name + 1);
1295
1296 s = find_section(s1, section_name);
1297 if (!s) {
1298 end_offset = 0;
1299 s = data_section;
1300 } else {
1301 end_offset = s->data_offset;
1302 }
1303
1304 add_elf_sym(symtab_section,
1305 0, 0,
1306 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1307 s->sh_num, sym_start);
1308 add_elf_sym(symtab_section,
1309 end_offset, 0,
1310 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1311 s->sh_num, sym_end);
1312 }
1313
1314 static int tcc_add_support(TCCState *s1, const char *filename)
1315 {
1316 char buf[1024];
1317 snprintf(buf, sizeof(buf), "%s/%s", s1->tcc_lib_path, filename);
1318 return tcc_add_file(s1, buf);
1319 }
1320
1321 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1322 {
1323 #ifdef CONFIG_TCC_BCHECK
1324 unsigned long *ptr;
1325 Section *init_section;
1326 unsigned char *pinit;
1327 int sym_index;
1328
1329 if (0 == s1->do_bounds_check)
1330 return;
1331
1332 /* XXX: add an object file to do that */
1333 ptr = section_ptr_add(bounds_section, sizeof(unsigned long));
1334 *ptr = 0;
1335 add_elf_sym(symtab_section, 0, 0,
1336 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1337 bounds_section->sh_num, "__bounds_start");
1338 /* add bound check code */
1339 #ifndef TCC_TARGET_PE
1340 tcc_add_support(s1, "bcheck.o");
1341 #endif
1342 #ifdef TCC_TARGET_I386
1343 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1344 /* add 'call __bound_init()' in .init section */
1345 init_section = find_section(s1, ".init");
1346 pinit = section_ptr_add(init_section, 5);
1347 pinit[0] = 0xe8;
1348 put32(pinit + 1, -4);
1349 sym_index = find_elf_sym(symtab_section, "__bound_init");
1350 put_elf_reloc(symtab_section, init_section,
1351 init_section->data_offset - 4, R_386_PC32, sym_index);
1352 }
1353 #endif
1354 #endif
1355 }
1356
1357 /* add tcc runtime libraries */
1358 ST_FUNC void tcc_add_runtime(TCCState *s1)
1359 {
1360 tcc_add_bcheck(s1);
1361
1362 /* add libc */
1363 if (!s1->nostdlib) {
1364 tcc_add_library(s1, "c");
1365 #ifdef CONFIG_USE_LIBGCC
1366 tcc_add_file(s1, TCC_LIBGCC);
1367 #elif !defined WITHOUT_LIBTCC
1368 tcc_add_support(s1, "libtcc1.a");
1369 #endif
1370 /* add crt end if not memory output */
1371 if (s1->output_type != TCC_OUTPUT_MEMORY)
1372 tcc_add_crt(s1, "crtn.o");
1373 }
1374 }
1375
1376 /* add various standard linker symbols (must be done after the
1377 sections are filled (for example after allocating common
1378 symbols)) */
1379 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1380 {
1381 char buf[1024];
1382 int i;
1383 Section *s;
1384
1385 add_elf_sym(symtab_section,
1386 text_section->data_offset, 0,
1387 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1388 text_section->sh_num, "_etext");
1389 add_elf_sym(symtab_section,
1390 data_section->data_offset, 0,
1391 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1392 data_section->sh_num, "_edata");
1393 add_elf_sym(symtab_section,
1394 bss_section->data_offset, 0,
1395 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1396 bss_section->sh_num, "_end");
1397 /* horrible new standard ldscript defines */
1398 add_init_array_defines(s1, ".preinit_array");
1399 add_init_array_defines(s1, ".init_array");
1400 add_init_array_defines(s1, ".fini_array");
1401
1402 /* add start and stop symbols for sections whose name can be
1403 expressed in C */
1404 for(i = 1; i < s1->nb_sections; i++) {
1405 s = s1->sections[i];
1406 if (s->sh_type == SHT_PROGBITS &&
1407 (s->sh_flags & SHF_ALLOC)) {
1408 const char *p;
1409 int ch;
1410
1411 /* check if section name can be expressed in C */
1412 p = s->name;
1413 for(;;) {
1414 ch = *p;
1415 if (!ch)
1416 break;
1417 if (!isid(ch) && !isnum(ch))
1418 goto next_sec;
1419 p++;
1420 }
1421 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1422 add_elf_sym(symtab_section,
1423 0, 0,
1424 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1425 s->sh_num, buf);
1426 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1427 add_elf_sym(symtab_section,
1428 s->data_offset, 0,
1429 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1430 s->sh_num, buf);
1431 }
1432 next_sec: ;
1433 }
1434 }
1435
1436 static void tcc_output_binary(TCCState *s1, FILE *f,
1437 const int *section_order)
1438 {
1439 Section *s;
1440 int i, offset, size;
1441
1442 offset = 0;
1443 for(i=1;i<s1->nb_sections;i++) {
1444 s = s1->sections[section_order[i]];
1445 if (s->sh_type != SHT_NOBITS &&
1446 (s->sh_flags & SHF_ALLOC)) {
1447 while (offset < s->sh_offset) {
1448 fputc(0, f);
1449 offset++;
1450 }
1451 size = s->sh_size;
1452 fwrite(s->data, 1, size, f);
1453 offset += size;
1454 }
1455 }
1456 }
1457
1458 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1459 #define HAVE_PHDR 1
1460 #define EXTRA_RELITEMS 14
1461
1462 /* move the relocation value from .dynsym to .got */
1463 void patch_dynsym_undef(TCCState *s1, Section *s)
1464 {
1465 uint32_t *gotd = (void *)s1->got->data;
1466 ElfW(Sym) *sym, *sym_end;
1467
1468 gotd += 3; // dummy entries in .got
1469 /* relocate symbols in .dynsym */
1470 sym_end = (ElfW(Sym) *)(s->data + s->data_offset);
1471 for (sym = (ElfW(Sym) *)s->data + 1; sym < sym_end; sym++) {
1472 if (sym->st_shndx == SHN_UNDEF) {
1473 *gotd++ = sym->st_value + 6; // XXX 6 is magic ?
1474 sym->st_value = 0;
1475 }
1476 }
1477 }
1478 #else
1479 #define HAVE_PHDR 0
1480 #define EXTRA_RELITEMS 9
1481
1482 /* zero plt offsets of weak symbols in .dynsym */
1483 void patch_dynsym_undef(TCCState *s1, Section *s)
1484 {
1485 ElfW(Sym) *sym, *sym_end;
1486
1487 sym_end = (ElfW(Sym) *)(s->data + s->data_offset);
1488 for (sym = (ElfW(Sym) *)s->data + 1; sym < sym_end; sym++)
1489 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1490 sym->st_value = 0;
1491 }
1492 #endif
1493
1494 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1495 {
1496 int sym_index = ELFW(R_SYM) (rel->r_info);
1497 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1498 unsigned long offset;
1499
1500 if (sym_index >= s1->nb_sym_attrs)
1501 return;
1502 offset = s1->sym_attrs[sym_index].got_offset;
1503 section_reserve(s1->got, offset + PTR_SIZE);
1504 #ifdef TCC_TARGET_X86_64
1505 /* only works for x86-64 */
1506 put32(s1->got->data + offset + 4, sym->st_value >> 32);
1507 #endif
1508 put32(s1->got->data + offset, sym->st_value & 0xffffffff);
1509 }
1510
1511 ST_FUNC void fill_got(TCCState *s1)
1512 {
1513 Section *s;
1514 ElfW_Rel *rel, *rel_end;
1515 int i;
1516
1517 for(i = 1; i < s1->nb_sections; i++) {
1518 s = s1->sections[i];
1519 if (s->sh_type != SHT_RELX)
1520 continue;
1521 /* no need to handle got relocations */
1522 if (s->link != symtab_section)
1523 continue;
1524 rel_end = (ElfW_Rel *) (s->data + s->data_offset);
1525 for(rel = (ElfW_Rel *) s->data; rel < rel_end; rel++) {
1526 switch (ELFW(R_TYPE) (rel->r_info)) {
1527 case R_X86_64_GOT32:
1528 case R_X86_64_GOTPCREL:
1529 case R_X86_64_PLT32:
1530 fill_got_entry(s1, rel);
1531 break;
1532 }
1533 }
1534 }
1535 }
1536
1537
1538 /* output an ELF file */
1539 /* XXX: suppress unneeded sections */
1540 static int elf_output_file(TCCState *s1, const char *filename)
1541 {
1542 ElfW(Ehdr) ehdr;
1543 FILE *f;
1544 int fd, mode, ret;
1545 int *section_order;
1546 int shnum, i, phnum, file_offset, offset, size, j, sh_order_index, k;
1547 long long tmp;
1548 uplong addr;
1549 Section *strsec, *s;
1550 ElfW(Shdr) shdr, *sh;
1551 ElfW(Phdr) *phdr, *ph;
1552 Section *interp, *dynamic, *dynstr;
1553 unsigned long saved_dynamic_data_offset;
1554 ElfW(Sym) *sym;
1555 int type, file_type;
1556 uplong rel_addr, rel_size;
1557 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1558 uplong bss_addr, bss_size;
1559 #endif
1560
1561 file_type = s1->output_type;
1562 s1->nb_errors = 0;
1563
1564 if (file_type != TCC_OUTPUT_OBJ) {
1565 tcc_add_runtime(s1);
1566 }
1567
1568 phdr = NULL;
1569 section_order = NULL;
1570 interp = NULL;
1571 dynamic = NULL;
1572 dynstr = NULL; /* avoid warning */
1573 saved_dynamic_data_offset = 0; /* avoid warning */
1574
1575 if (file_type != TCC_OUTPUT_OBJ) {
1576 relocate_common_syms();
1577
1578 tcc_add_linker_symbols(s1);
1579
1580 if (!s1->static_link) {
1581 const char *name;
1582 int sym_index, index;
1583 ElfW(Sym) *esym, *sym_end;
1584
1585 if (file_type == TCC_OUTPUT_EXE) {
1586 char *ptr;
1587 /* allow override the dynamic loader */
1588 const char *elfint = getenv("LD_SO");
1589 if (elfint == NULL)
1590 elfint = CONFIG_TCC_ELFINTERP;
1591 /* add interpreter section only if executable */
1592 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
1593 interp->sh_addralign = 1;
1594 ptr = section_ptr_add(interp, 1+strlen(elfint));
1595 strcpy(ptr, elfint);
1596 }
1597
1598 /* add dynamic symbol table */
1599 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
1600 ".dynstr",
1601 ".hash", SHF_ALLOC);
1602 dynstr = s1->dynsym->link;
1603
1604 /* add dynamic section */
1605 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
1606 SHF_ALLOC | SHF_WRITE);
1607 dynamic->link = dynstr;
1608 dynamic->sh_entsize = sizeof(ElfW(Dyn));
1609
1610 /* add PLT */
1611 s1->plt = new_section(s1, ".plt", SHT_PROGBITS,
1612 SHF_ALLOC | SHF_EXECINSTR);
1613 s1->plt->sh_entsize = 4;
1614
1615 build_got(s1);
1616
1617 /* scan for undefined symbols and see if they are in the
1618 dynamic symbols. If a symbol STT_FUNC or STT_GNU_IFUNC
1619 is found, then we add it in the PLT. If a symbol
1620 STT_OBJECT is found, we add it in the .bss section with
1621 a suitable relocation */
1622 sym_end = (ElfW(Sym) *)(symtab_section->data +
1623 symtab_section->data_offset);
1624 if (file_type == TCC_OUTPUT_EXE) {
1625 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
1626 sym < sym_end;
1627 sym++) {
1628 if (sym->st_shndx == SHN_UNDEF) {
1629 name = symtab_section->link->data + sym->st_name;
1630 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1631 if (sym_index) {
1632 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1633 type = ELFW(ST_TYPE)(esym->st_info);
1634 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1635 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1636 ELFW(ST_INFO)(STB_GLOBAL,type),
1637 sym - (ElfW(Sym) *)symtab_section->data);
1638 } else if (type == STT_OBJECT) {
1639 unsigned long offset;
1640 ElfW(Sym) *dynsym, *dynsym_end;
1641 offset = bss_section->data_offset;
1642 /* XXX: which alignment ? */
1643 offset = (offset + 16 - 1) & -16;
1644 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1645 esym->st_info, 0,
1646 bss_section->sh_num, name);
1647 // Ensure R_COPY works for weak symbol aliases
1648 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1649 dynsym_end = (ElfW(Sym) *)
1650 (s1->dynsymtab_section->data +
1651 s1->dynsymtab_section->data_offset);
1652 for(dynsym = (ElfW(Sym) *)s1->dynsymtab_section->data + 1;
1653 dynsym < dynsym_end; dynsym++) {
1654 if ((dynsym->st_value == esym->st_value)
1655 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1656 char *dynname;
1657 dynname = s1->dynsymtab_section->link->data
1658 + dynsym->st_name;
1659 put_elf_sym(s1->dynsym, offset,
1660 dynsym->st_size,
1661 dynsym->st_info, 0,
1662 bss_section->sh_num,
1663 dynname);
1664 break;
1665 }
1666 }
1667 }
1668 put_elf_reloc(s1->dynsym, bss_section,
1669 offset, R_COPY, index);
1670 offset += esym->st_size;
1671 bss_section->data_offset = offset;
1672 }
1673 } else {
1674 /* STB_WEAK undefined symbols are accepted */
1675 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
1676 it */
1677 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1678 !strcmp(name, "_fp_hw")) {
1679 } else {
1680 tcc_error_noabort("undefined symbol '%s'", name);
1681 }
1682 }
1683 } else if (s1->rdynamic &&
1684 ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1685 /* if -rdynamic option, then export all non
1686 local symbols */
1687 name = symtab_section->link->data + sym->st_name;
1688 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1689 sym->st_info, 0,
1690 sym->st_shndx, name);
1691 }
1692 }
1693
1694 if (s1->nb_errors)
1695 goto fail;
1696
1697 /* now look at unresolved dynamic symbols and export
1698 corresponding symbol */
1699 sym_end = (ElfW(Sym) *)(s1->dynsymtab_section->data +
1700 s1->dynsymtab_section->data_offset);
1701 for(esym = (ElfW(Sym) *)s1->dynsymtab_section->data + 1;
1702 esym < sym_end;
1703 esym++) {
1704 if (esym->st_shndx == SHN_UNDEF) {
1705 name = s1->dynsymtab_section->link->data + esym->st_name;
1706 sym_index = find_elf_sym(symtab_section, name);
1707 if (sym_index) {
1708 /* XXX: avoid adding a symbol if already
1709 present because of -rdynamic ? */
1710 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1711 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1712 sym->st_info, 0,
1713 sym->st_shndx, name);
1714 } else {
1715 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1716 /* weak symbols can stay undefined */
1717 } else {
1718 tcc_warning("undefined dynamic symbol '%s'", name);
1719 }
1720 }
1721 }
1722 }
1723 } else {
1724 int nb_syms;
1725 /* shared library case : we simply export all the global symbols */
1726 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1727 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1728 for(sym = (ElfW(Sym) *)symtab_section->data + 1;
1729 sym < sym_end;
1730 sym++) {
1731 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1732 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1733 if ((ELFW(ST_TYPE)(sym->st_info) == STT_FUNC ||
1734 ELFW(ST_TYPE)(sym->st_info) == STT_GNU_IFUNC)
1735 && sym->st_shndx == SHN_UNDEF) {
1736 put_got_entry(s1, R_JMP_SLOT, sym->st_size,
1737 sym->st_info,
1738 sym - (ElfW(Sym) *)symtab_section->data);
1739 }
1740 else if (ELFW(ST_TYPE)(sym->st_info) == STT_OBJECT) {
1741 put_got_entry(s1, R_X86_64_GLOB_DAT, sym->st_size,
1742 sym->st_info,
1743 sym - (ElfW(Sym) *)symtab_section->data);
1744 }
1745 else
1746 #endif
1747 {
1748 name = symtab_section->link->data + sym->st_name;
1749 index = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1750 sym->st_info, 0,
1751 sym->st_shndx, name);
1752 s1->symtab_to_dynsym[sym -
1753 (ElfW(Sym) *)symtab_section->data] =
1754 index;
1755 }
1756 }
1757 }
1758 }
1759
1760 build_got_entries(s1);
1761
1762 /* add a list of needed dlls */
1763 for(i = 0; i < s1->nb_loaded_dlls; i++) {
1764 DLLReference *dllref = s1->loaded_dlls[i];
1765 if (dllref->level == 0)
1766 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
1767 }
1768
1769 if (s1->rpath)
1770 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
1771
1772 /* XXX: currently, since we do not handle PIC code, we
1773 must relocate the readonly segments */
1774 if (file_type == TCC_OUTPUT_DLL) {
1775 if (s1->soname)
1776 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
1777 put_dt(dynamic, DT_TEXTREL, 0);
1778 }
1779
1780 if (s1->symbolic)
1781 put_dt(dynamic, DT_SYMBOLIC, 0);
1782
1783 /* add necessary space for other entries */
1784 saved_dynamic_data_offset = dynamic->data_offset;
1785 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
1786 } else {
1787 /* still need to build got entries in case of static link */
1788 build_got_entries(s1);
1789 }
1790 }
1791
1792 memset(&ehdr, 0, sizeof(ehdr));
1793
1794 /* we add a section for symbols */
1795 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
1796 put_elf_str(strsec, "");
1797
1798 /* compute number of sections */
1799 shnum = s1->nb_sections;
1800
1801 /* this array is used to reorder sections in the output file */
1802 section_order = tcc_malloc(sizeof(int) * shnum);
1803 section_order[0] = 0;
1804 sh_order_index = 1;
1805
1806 /* compute number of program headers */
1807 switch(file_type) {
1808 default:
1809 case TCC_OUTPUT_OBJ:
1810 phnum = 0;
1811 break;
1812 case TCC_OUTPUT_EXE:
1813 if (!s1->static_link)
1814 phnum = 4 + HAVE_PHDR;
1815 else
1816 phnum = 2;
1817 break;
1818 case TCC_OUTPUT_DLL:
1819 phnum = 3;
1820 break;
1821 }
1822
1823 /* allocate strings for section names and decide if an unallocated
1824 section should be output */
1825 /* NOTE: the strsec section comes last, so its size is also
1826 correct ! */
1827 for(i = 1; i < s1->nb_sections; i++) {
1828 s = s1->sections[i];
1829 s->sh_name = put_elf_str(strsec, s->name);
1830 #if 0 //gr
1831 printf("section: f=%08x t=%08x i=%08x %s %s\n",
1832 s->sh_flags,
1833 s->sh_type,
1834 s->sh_info,
1835 s->name,
1836 s->reloc ? s->reloc->name : "n"
1837 );
1838 #endif
1839 /* when generating a DLL, we include relocations but we may
1840 patch them */
1841 if (file_type == TCC_OUTPUT_DLL &&
1842 s->sh_type == SHT_RELX &&
1843 !(s->sh_flags & SHF_ALLOC)) {
1844 /* //gr: avoid bogus relocs for empty (debug) sections */
1845 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
1846 prepare_dynamic_rel(s1, s);
1847 else if (s1->do_debug)
1848 s->sh_size = s->data_offset;
1849 } else if (s1->do_debug ||
1850 file_type == TCC_OUTPUT_OBJ ||
1851 (s->sh_flags & SHF_ALLOC) ||
1852 i == (s1->nb_sections - 1)) {
1853 /* we output all sections if debug or object file */
1854 s->sh_size = s->data_offset;
1855 }
1856 }
1857
1858 /* allocate program segment headers */
1859 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
1860
1861 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
1862 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
1863 } else {
1864 file_offset = 0;
1865 }
1866 if (phnum > 0) {
1867 /* compute section to program header mapping */
1868 if (s1->has_text_addr) {
1869 int a_offset, p_offset;
1870 addr = s1->text_addr;
1871 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
1872 ELF_PAGE_SIZE */
1873 a_offset = (int) (addr & (s1->section_align - 1));
1874 p_offset = file_offset & (s1->section_align - 1);
1875 if (a_offset < p_offset)
1876 a_offset += s1->section_align;
1877 file_offset += (a_offset - p_offset);
1878 } else {
1879 if (file_type == TCC_OUTPUT_DLL)
1880 addr = 0;
1881 else
1882 addr = ELF_START_ADDR;
1883 /* compute address after headers */
1884 addr += (file_offset & (s1->section_align - 1));
1885 }
1886
1887 /* dynamic relocation table information, for .dynamic section */
1888 rel_size = 0;
1889 rel_addr = 0;
1890
1891 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1892 bss_addr = bss_size = 0;
1893 #endif
1894 /* leave one program header for the program interpreter */
1895 ph = &phdr[0];
1896 if (interp)
1897 ph += 1 + HAVE_PHDR;
1898
1899 for(j = 0; j < 2; j++) {
1900 ph->p_type = PT_LOAD;
1901 if (j == 0)
1902 ph->p_flags = PF_R | PF_X;
1903 else
1904 ph->p_flags = PF_R | PF_W;
1905 ph->p_align = s1->section_align;
1906
1907 /* we do the following ordering: interp, symbol tables,
1908 relocations, progbits, nobits */
1909 /* XXX: do faster and simpler sorting */
1910 for(k = 0; k < 5; k++) {
1911 for(i = 1; i < s1->nb_sections; i++) {
1912 s = s1->sections[i];
1913 /* compute if section should be included */
1914 if (j == 0) {
1915 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1916 SHF_ALLOC)
1917 continue;
1918 } else {
1919 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1920 (SHF_ALLOC | SHF_WRITE))
1921 continue;
1922 }
1923 if (s == interp) {
1924 if (k != 0)
1925 continue;
1926 } else if (s->sh_type == SHT_DYNSYM ||
1927 s->sh_type == SHT_STRTAB ||
1928 s->sh_type == SHT_HASH) {
1929 if (k != 1)
1930 continue;
1931 } else if (s->sh_type == SHT_RELX) {
1932 if (k != 2)
1933 continue;
1934 } else if (s->sh_type == SHT_NOBITS) {
1935 if (k != 4)
1936 continue;
1937 } else {
1938 if (k != 3)
1939 continue;
1940 }
1941 section_order[sh_order_index++] = i;
1942
1943 /* section matches: we align it and add its size */
1944 tmp = addr;
1945 addr = (addr + s->sh_addralign - 1) &
1946 ~(s->sh_addralign - 1);
1947 file_offset += (int) ( addr - tmp );
1948 s->sh_offset = file_offset;
1949 s->sh_addr = addr;
1950
1951 /* update program header infos */
1952 if (ph->p_offset == 0) {
1953 ph->p_offset = file_offset;
1954 ph->p_vaddr = addr;
1955 ph->p_paddr = ph->p_vaddr;
1956 }
1957 /* update dynamic relocation infos */
1958 if (s->sh_type == SHT_RELX) {
1959 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1960 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) { // rel_size == 0) {
1961 rel_addr = addr;
1962 rel_size += s->sh_size; // XXX only first rel.
1963 }
1964 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) { // rel_size == 0) {
1965 bss_addr = addr;
1966 bss_size = s->sh_size; // XXX only first rel.
1967 }
1968 #else
1969 if (rel_size == 0)
1970 rel_addr = addr;
1971 rel_size += s->sh_size;
1972 #endif
1973 }
1974 addr += s->sh_size;
1975 if (s->sh_type != SHT_NOBITS)
1976 file_offset += s->sh_size;
1977 }
1978 }
1979 ph->p_filesz = file_offset - ph->p_offset;
1980 ph->p_memsz = addr - ph->p_vaddr;
1981 ph++;
1982 if (j == 0) {
1983 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
1984 /* if in the middle of a page, we duplicate the page in
1985 memory so that one copy is RX and the other is RW */
1986 if ((addr & (s1->section_align - 1)) != 0)
1987 addr += s1->section_align;
1988 } else {
1989 addr = (addr + s1->section_align - 1) & ~(s1->section_align - 1);
1990 file_offset = (file_offset + s1->section_align - 1) &
1991 ~(s1->section_align - 1);
1992 }
1993 }
1994 }
1995
1996 /* if interpreter, then add corresponing program header */
1997 if (interp) {
1998 ph = &phdr[0];
1999
2000 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2001 {
2002 int len = phnum * sizeof(ElfW(Phdr));
2003
2004 ph->p_type = PT_PHDR;
2005 ph->p_offset = sizeof(ElfW(Ehdr));
2006 ph->p_vaddr = interp->sh_addr - len;
2007 ph->p_paddr = ph->p_vaddr;
2008 ph->p_filesz = ph->p_memsz = len;
2009 ph->p_flags = PF_R | PF_X;
2010 ph->p_align = 4; // interp->sh_addralign;
2011 ph++;
2012 }
2013 #endif
2014
2015 ph->p_type = PT_INTERP;
2016 ph->p_offset = interp->sh_offset;
2017 ph->p_vaddr = interp->sh_addr;
2018 ph->p_paddr = ph->p_vaddr;
2019 ph->p_filesz = interp->sh_size;
2020 ph->p_memsz = interp->sh_size;
2021 ph->p_flags = PF_R;
2022 ph->p_align = interp->sh_addralign;
2023 }
2024
2025 /* if dynamic section, then add corresponing program header */
2026 if (dynamic) {
2027 ElfW(Sym) *sym_end;
2028
2029 ph = &phdr[phnum - 1];
2030
2031 ph->p_type = PT_DYNAMIC;
2032 ph->p_offset = dynamic->sh_offset;
2033 ph->p_vaddr = dynamic->sh_addr;
2034 ph->p_paddr = ph->p_vaddr;
2035 ph->p_filesz = dynamic->sh_size;
2036 ph->p_memsz = dynamic->sh_size;
2037 ph->p_flags = PF_R | PF_W;
2038 ph->p_align = dynamic->sh_addralign;
2039
2040 /* put GOT dynamic section address */
2041 put32(s1->got->data, dynamic->sh_addr);
2042
2043 /* relocate the PLT */
2044 if (file_type == TCC_OUTPUT_EXE
2045 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2046 || file_type == TCC_OUTPUT_DLL
2047 #endif
2048 ) {
2049 uint8_t *p, *p_end;
2050
2051 p = s1->plt->data;
2052 p_end = p + s1->plt->data_offset;
2053 if (p < p_end) {
2054 #if defined(TCC_TARGET_I386)
2055 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
2056 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
2057 p += 16;
2058 while (p < p_end) {
2059 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
2060 p += 16;
2061 }
2062 #elif defined(TCC_TARGET_X86_64)
2063 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
2064 put32(p + 2, get32(p + 2) + x);
2065 put32(p + 8, get32(p + 8) + x - 6);
2066 p += 16;
2067 while (p < p_end) {
2068 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
2069 p += 16;
2070 }
2071 #elif defined(TCC_TARGET_ARM)
2072 int x;
2073 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
2074 p += 16;
2075 while (p < p_end) {
2076 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
2077 p += 4;
2078 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
2079 p += 16;
2080 }
2081 #elif defined(TCC_TARGET_C67)
2082 /* XXX: TODO */
2083 #else
2084 #error unsupported CPU
2085 #endif
2086 }
2087 }
2088
2089 /* relocate symbols in .dynsym */
2090 sym_end = (ElfW(Sym) *)(s1->dynsym->data + s1->dynsym->data_offset);
2091 for(sym = (ElfW(Sym) *)s1->dynsym->data + 1;
2092 sym < sym_end;
2093 sym++) {
2094 if (sym->st_shndx == SHN_UNDEF) {
2095 /* relocate to the PLT if the symbol corresponds
2096 to a PLT entry */
2097 if (sym->st_value)
2098 sym->st_value += s1->plt->sh_addr;
2099 } else if (sym->st_shndx < SHN_LORESERVE) {
2100 /* do symbol relocation */
2101 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2102 }
2103 }
2104
2105 /* put dynamic section entries */
2106 dynamic->data_offset = saved_dynamic_data_offset;
2107 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2108 put_dt(dynamic, DT_STRTAB, dynstr->sh_addr);
2109 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2110 put_dt(dynamic, DT_STRSZ, dynstr->data_offset);
2111 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2112 #ifdef TCC_TARGET_X86_64
2113 put_dt(dynamic, DT_RELA, rel_addr);
2114 put_dt(dynamic, DT_RELASZ, rel_size);
2115 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2116 #else
2117 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2118 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2119 put_dt(dynamic, DT_PLTRELSZ, rel_size);
2120 put_dt(dynamic, DT_JMPREL, rel_addr);
2121 put_dt(dynamic, DT_PLTREL, DT_REL);
2122 put_dt(dynamic, DT_REL, bss_addr);
2123 put_dt(dynamic, DT_RELSZ, bss_size);
2124 #else
2125 put_dt(dynamic, DT_REL, rel_addr);
2126 put_dt(dynamic, DT_RELSZ, rel_size);
2127 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2128 #endif
2129 #endif
2130 if (s1->do_debug)
2131 put_dt(dynamic, DT_DEBUG, 0);
2132 put_dt(dynamic, DT_NULL, 0);
2133 }
2134
2135 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2136 ehdr.e_phnum = phnum;
2137 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2138 }
2139
2140 /* all other sections come after */
2141 for(i = 1; i < s1->nb_sections; i++) {
2142 s = s1->sections[i];
2143 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
2144 continue;
2145 section_order[sh_order_index++] = i;
2146
2147 file_offset = (file_offset + s->sh_addralign - 1) &
2148 ~(s->sh_addralign - 1);
2149 s->sh_offset = file_offset;
2150 if (s->sh_type != SHT_NOBITS)
2151 file_offset += s->sh_size;
2152 }
2153
2154 /* if building executable or DLL, then relocate each section
2155 except the GOT which is already relocated */
2156 if (file_type != TCC_OUTPUT_OBJ) {
2157 relocate_syms(s1, 0);
2158
2159 if (s1->nb_errors != 0) {
2160 fail:
2161 ret = -1;
2162 goto the_end;
2163 }
2164
2165 /* relocate sections */
2166 /* XXX: ignore sections with allocated relocations ? */
2167 for(i = 1; i < s1->nb_sections; i++) {
2168 s = s1->sections[i];
2169 if (s->reloc && s != s1->got && (s->sh_flags & SHF_ALLOC)) //gr
2170 relocate_section(s1, s);
2171 }
2172
2173 /* relocate relocation entries if the relocation tables are
2174 allocated in the executable */
2175 for(i = 1; i < s1->nb_sections; i++) {
2176 s = s1->sections[i];
2177 if ((s->sh_flags & SHF_ALLOC) &&
2178 s->sh_type == SHT_RELX) {
2179 relocate_rel(s1, s);
2180 }
2181 }
2182
2183 /* get entry point address */
2184 if (file_type == TCC_OUTPUT_EXE)
2185 ehdr.e_entry = (uplong)tcc_get_symbol_err(s1, "_start");
2186 else
2187 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2188 }
2189 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2190 fill_got(s1);
2191
2192 /* write elf file */
2193 if (file_type == TCC_OUTPUT_OBJ)
2194 mode = 0666;
2195 else
2196 mode = 0777;
2197 unlink(filename);
2198 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2199 if (fd < 0) {
2200 tcc_error_noabort("could not write '%s'", filename);
2201 goto fail;
2202 }
2203 f = fdopen(fd, "wb");
2204 if (s1->verbose)
2205 printf("<- %s\n", filename);
2206
2207 #ifdef TCC_TARGET_COFF
2208 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF) {
2209 tcc_output_coff(s1, f);
2210 } else
2211 #endif
2212 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
2213 sort_syms(s1, symtab_section);
2214
2215 /* align to 4 */
2216 file_offset = (file_offset + 3) & -4;
2217
2218 /* fill header */
2219 ehdr.e_ident[0] = ELFMAG0;
2220 ehdr.e_ident[1] = ELFMAG1;
2221 ehdr.e_ident[2] = ELFMAG2;
2222 ehdr.e_ident[3] = ELFMAG3;
2223 ehdr.e_ident[4] = ELFCLASSW;
2224 ehdr.e_ident[5] = ELFDATA2LSB;
2225 ehdr.e_ident[6] = EV_CURRENT;
2226 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2227 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2228 #endif
2229 #ifdef TCC_TARGET_ARM
2230 #ifdef TCC_ARM_EABI
2231 ehdr.e_ident[EI_OSABI] = 0;
2232 ehdr.e_flags = 4 << 24;
2233 #else
2234 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2235 #endif
2236 #endif
2237 switch(file_type) {
2238 default:
2239 case TCC_OUTPUT_EXE:
2240 ehdr.e_type = ET_EXEC;
2241 break;
2242 case TCC_OUTPUT_DLL:
2243 ehdr.e_type = ET_DYN;
2244 break;
2245 case TCC_OUTPUT_OBJ:
2246 ehdr.e_type = ET_REL;
2247 break;
2248 }
2249 ehdr.e_machine = EM_TCC_TARGET;
2250 ehdr.e_version = EV_CURRENT;
2251 ehdr.e_shoff = file_offset;
2252 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2253 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2254 ehdr.e_shnum = shnum;
2255 ehdr.e_shstrndx = shnum - 1;
2256
2257 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2258 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2259 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2260
2261 for(i=1;i<s1->nb_sections;i++) {
2262 s = s1->sections[section_order[i]];
2263 if (s->sh_type != SHT_NOBITS) {
2264 if (s->sh_type == SHT_DYNSYM)
2265 patch_dynsym_undef(s1, s);
2266 while (offset < s->sh_offset) {
2267 fputc(0, f);
2268 offset++;
2269 }
2270 size = s->sh_size;
2271 fwrite(s->data, 1, size, f);
2272 offset += size;
2273 }
2274 }
2275
2276 /* output section headers */
2277 while (offset < ehdr.e_shoff) {
2278 fputc(0, f);
2279 offset++;
2280 }
2281
2282 for(i=0;i<s1->nb_sections;i++) {
2283 sh = &shdr;
2284 memset(sh, 0, sizeof(ElfW(Shdr)));
2285 s = s1->sections[i];
2286 if (s) {
2287 sh->sh_name = s->sh_name;
2288 sh->sh_type = s->sh_type;
2289 sh->sh_flags = s->sh_flags;
2290 sh->sh_entsize = s->sh_entsize;
2291 sh->sh_info = s->sh_info;
2292 if (s->link)
2293 sh->sh_link = s->link->sh_num;
2294 sh->sh_addralign = s->sh_addralign;
2295 sh->sh_addr = s->sh_addr;
2296 sh->sh_offset = s->sh_offset;
2297 sh->sh_size = s->sh_size;
2298 }
2299 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2300 }
2301 } else {
2302 tcc_output_binary(s1, f, section_order);
2303 }
2304 fclose(f);
2305
2306 ret = 0;
2307 the_end:
2308 tcc_free(s1->symtab_to_dynsym);
2309 tcc_free(section_order);
2310 tcc_free(phdr);
2311 tcc_free(s1->sym_attrs);
2312 return ret;
2313 }
2314
2315 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2316 {
2317 int ret;
2318 #ifdef TCC_TARGET_PE
2319 if (s->output_type != TCC_OUTPUT_OBJ) {
2320 ret = pe_output_file(s, filename);
2321 } else
2322 #endif
2323 {
2324 ret = elf_output_file(s, filename);
2325 }
2326 return ret;
2327 }
2328
2329 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2330 {
2331 void *data;
2332
2333 data = tcc_malloc(size);
2334 lseek(fd, file_offset, SEEK_SET);
2335 read(fd, data, size);
2336 return data;
2337 }
2338
2339 typedef struct SectionMergeInfo {
2340 Section *s; /* corresponding existing section */
2341 unsigned long offset; /* offset of the new section in the existing section */
2342 uint8_t new_section; /* true if section 's' was added */
2343 uint8_t link_once; /* true if link once section */
2344 } SectionMergeInfo;
2345
2346 /* load an object file and merge it with current files */
2347 /* XXX: handle correctly stab (debug) info */
2348 ST_FUNC int tcc_load_object_file(TCCState *s1,
2349 int fd, unsigned long file_offset)
2350 {
2351 ElfW(Ehdr) ehdr;
2352 ElfW(Shdr) *shdr, *sh;
2353 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2354 unsigned char *strsec, *strtab;
2355 int *old_to_new_syms;
2356 char *sh_name, *name;
2357 SectionMergeInfo *sm_table, *sm;
2358 ElfW(Sym) *sym, *symtab;
2359 ElfW_Rel *rel, *rel_end;
2360 Section *s;
2361
2362 int stab_index;
2363 int stabstr_index;
2364
2365 stab_index = stabstr_index = 0;
2366
2367 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2368 goto fail1;
2369 if (ehdr.e_ident[0] != ELFMAG0 ||
2370 ehdr.e_ident[1] != ELFMAG1 ||
2371 ehdr.e_ident[2] != ELFMAG2 ||
2372 ehdr.e_ident[3] != ELFMAG3)
2373 goto fail1;
2374 /* test if object file */
2375 if (ehdr.e_type != ET_REL)
2376 goto fail1;
2377 /* test CPU specific stuff */
2378 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2379 ehdr.e_machine != EM_TCC_TARGET) {
2380 fail1:
2381 tcc_error_noabort("invalid object file");
2382 return -1;
2383 }
2384 /* read sections */
2385 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2386 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2387 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2388
2389 /* load section names */
2390 sh = &shdr[ehdr.e_shstrndx];
2391 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2392
2393 /* load symtab and strtab */
2394 old_to_new_syms = NULL;
2395 symtab = NULL;
2396 strtab = NULL;
2397 nb_syms = 0;
2398 for(i = 1; i < ehdr.e_shnum; i++) {
2399 sh = &shdr[i];
2400 if (sh->sh_type == SHT_SYMTAB) {
2401 if (symtab) {
2402 tcc_error_noabort("object must contain only one symtab");
2403 fail:
2404 ret = -1;
2405 goto the_end;
2406 }
2407 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2408 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2409 sm_table[i].s = symtab_section;
2410
2411 /* now load strtab */
2412 sh = &shdr[sh->sh_link];
2413 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2414 }
2415 }
2416
2417 /* now examine each section and try to merge its content with the
2418 ones in memory */
2419 for(i = 1; i < ehdr.e_shnum; i++) {
2420 /* no need to examine section name strtab */
2421 if (i == ehdr.e_shstrndx)
2422 continue;
2423 sh = &shdr[i];
2424 sh_name = strsec + sh->sh_name;
2425 /* ignore sections types we do not handle */
2426 if (sh->sh_type != SHT_PROGBITS &&
2427 sh->sh_type != SHT_RELX &&
2428 #ifdef TCC_ARM_EABI
2429 sh->sh_type != SHT_ARM_EXIDX &&
2430 #endif
2431 sh->sh_type != SHT_NOBITS &&
2432 sh->sh_type != SHT_PREINIT_ARRAY &&
2433 sh->sh_type != SHT_INIT_ARRAY &&
2434 sh->sh_type != SHT_FINI_ARRAY &&
2435 strcmp(sh_name, ".stabstr")
2436 )
2437 continue;
2438 if (sh->sh_addralign < 1)
2439 sh->sh_addralign = 1;
2440 /* find corresponding section, if any */
2441 for(j = 1; j < s1->nb_sections;j++) {
2442 s = s1->sections[j];
2443 if (!strcmp(s->name, sh_name)) {
2444 if (!strncmp(sh_name, ".gnu.linkonce",
2445 sizeof(".gnu.linkonce") - 1)) {
2446 /* if a 'linkonce' section is already present, we
2447 do not add it again. It is a little tricky as
2448 symbols can still be defined in
2449 it. */
2450 sm_table[i].link_once = 1;
2451 goto next;
2452 } else {
2453 goto found;
2454 }
2455 }
2456 }
2457 /* not found: create new section */
2458 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2459 /* take as much info as possible from the section. sh_link and
2460 sh_info will be updated later */
2461 s->sh_addralign = sh->sh_addralign;
2462 s->sh_entsize = sh->sh_entsize;
2463 sm_table[i].new_section = 1;
2464 found:
2465 if (sh->sh_type != s->sh_type) {
2466 tcc_error_noabort("invalid section type");
2467 goto fail;
2468 }
2469
2470 /* align start of section */
2471 offset = s->data_offset;
2472
2473 if (0 == strcmp(sh_name, ".stab")) {
2474 stab_index = i;
2475 goto no_align;
2476 }
2477 if (0 == strcmp(sh_name, ".stabstr")) {
2478 stabstr_index = i;
2479 goto no_align;
2480 }
2481
2482 size = sh->sh_addralign - 1;
2483 offset = (offset + size) & ~size;
2484 if (sh->sh_addralign > s->sh_addralign)
2485 s->sh_addralign = sh->sh_addralign;
2486 s->data_offset = offset;
2487 no_align:
2488 sm_table[i].offset = offset;
2489 sm_table[i].s = s;
2490 /* concatenate sections */
2491 size = sh->sh_size;
2492 if (sh->sh_type != SHT_NOBITS) {
2493 unsigned char *ptr;
2494 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2495 ptr = section_ptr_add(s, size);
2496 read(fd, ptr, size);
2497 } else {
2498 s->data_offset += size;
2499 }
2500 next: ;
2501 }
2502
2503 /* //gr relocate stab strings */
2504 if (stab_index && stabstr_index) {
2505 Stab_Sym *a, *b;
2506 unsigned o;
2507 s = sm_table[stab_index].s;
2508 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2509 b = (Stab_Sym *)(s->data + s->data_offset);
2510 o = sm_table[stabstr_index].offset;
2511 while (a < b)
2512 a->n_strx += o, a++;
2513 }
2514
2515 /* second short pass to update sh_link and sh_info fields of new
2516 sections */
2517 for(i = 1; i < ehdr.e_shnum; i++) {
2518 s = sm_table[i].s;
2519 if (!s || !sm_table[i].new_section)
2520 continue;
2521 sh = &shdr[i];
2522 if (sh->sh_link > 0)
2523 s->link = sm_table[sh->sh_link].s;
2524 if (sh->sh_type == SHT_RELX) {
2525 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2526 /* update backward link */
2527 s1->sections[s->sh_info]->reloc = s;
2528 }
2529 }
2530 sm = sm_table;
2531
2532 /* resolve symbols */
2533 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2534
2535 sym = symtab + 1;
2536 for(i = 1; i < nb_syms; i++, sym++) {
2537 if (sym->st_shndx != SHN_UNDEF &&
2538 sym->st_shndx < SHN_LORESERVE) {
2539 sm = &sm_table[sym->st_shndx];
2540 if (sm->link_once) {
2541 /* if a symbol is in a link once section, we use the
2542 already defined symbol. It is very important to get
2543 correct relocations */
2544 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2545 name = strtab + sym->st_name;
2546 sym_index = find_elf_sym(symtab_section, name);
2547 if (sym_index)
2548 old_to_new_syms[i] = sym_index;
2549 }
2550 continue;
2551 }
2552 /* if no corresponding section added, no need to add symbol */
2553 if (!sm->s)
2554 continue;
2555 /* convert section number */
2556 sym->st_shndx = sm->s->sh_num;
2557 /* offset value */
2558 sym->st_value += sm->offset;
2559 }
2560 /* add symbol */
2561 name = strtab + sym->st_name;
2562 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2563 sym->st_info, sym->st_other,
2564 sym->st_shndx, name);
2565 old_to_new_syms[i] = sym_index;
2566 }
2567
2568 /* third pass to patch relocation entries */
2569 for(i = 1; i < ehdr.e_shnum; i++) {
2570 s = sm_table[i].s;
2571 if (!s)
2572 continue;
2573 sh = &shdr[i];
2574 offset = sm_table[i].offset;
2575 switch(s->sh_type) {
2576 case SHT_RELX:
2577 /* take relocation offset information */
2578 offseti = sm_table[sh->sh_info].offset;
2579 rel_end = (ElfW_Rel *)(s->data + s->data_offset);
2580 for(rel = (ElfW_Rel *)(s->data + offset);
2581 rel < rel_end;
2582 rel++) {
2583 int type;
2584 unsigned sym_index;
2585 /* convert symbol index */
2586 type = ELFW(R_TYPE)(rel->r_info);
2587 sym_index = ELFW(R_SYM)(rel->r_info);
2588 /* NOTE: only one symtab assumed */
2589 if (sym_index >= nb_syms)
2590 goto invalid_reloc;
2591 sym_index = old_to_new_syms[sym_index];
2592 /* ignore link_once in rel section. */
2593 if (!sym_index && !sm->link_once
2594 #ifdef TCC_TARGET_ARM
2595 && type != R_ARM_V4BX
2596 #endif
2597 ) {
2598 invalid_reloc:
2599 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2600 i, strsec + sh->sh_name, rel->r_offset);
2601 goto fail;
2602 }
2603 rel->r_info = ELFW(R_INFO)(sym_index, type);
2604 /* offset the relocation offset */
2605 rel->r_offset += offseti;
2606 #ifdef TCC_TARGET_ARM
2607 /* Jumps and branches from a Thumb code to a PLT entry need
2608 special handling since PLT entries are ARM code.
2609 Unconditional bl instructions referencing PLT entries are
2610 handled by converting these instructions into blx
2611 instructions. Other case of instructions referencing a PLT
2612 entry require to add a Thumb stub before the PLT entry to
2613 switch to ARM mode. We set bit 0 of the got offset of a
2614 symbol to indicate such a case. */
2615 if (type == R_ARM_THM_JUMP24)
2616 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
2617 #endif
2618 }
2619 break;
2620 default:
2621 break;
2622 }
2623 }
2624
2625 ret = 0;
2626 the_end:
2627 tcc_free(symtab);
2628 tcc_free(strtab);
2629 tcc_free(old_to_new_syms);
2630 tcc_free(sm_table);
2631 tcc_free(strsec);
2632 tcc_free(shdr);
2633 return ret;
2634 }
2635
2636 typedef struct ArchiveHeader {
2637 char ar_name[16]; /* name of this member */
2638 char ar_date[12]; /* file mtime */
2639 char ar_uid[6]; /* owner uid; printed as decimal */
2640 char ar_gid[6]; /* owner gid; printed as decimal */
2641 char ar_mode[8]; /* file mode, printed as octal */
2642 char ar_size[10]; /* file size, printed as decimal */
2643 char ar_fmag[2]; /* should contain ARFMAG */
2644 } ArchiveHeader;
2645
2646 static int get_be32(const uint8_t *b)
2647 {
2648 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
2649 }
2650
2651 /* load only the objects which resolve undefined symbols */
2652 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
2653 {
2654 int i, bound, nsyms, sym_index, off, ret;
2655 uint8_t *data;
2656 const char *ar_names, *p;
2657 const uint8_t *ar_index;
2658 ElfW(Sym) *sym;
2659
2660 data = tcc_malloc(size);
2661 if (read(fd, data, size) != size)
2662 goto fail;
2663 nsyms = get_be32(data);
2664 ar_index = data + 4;
2665 ar_names = ar_index + nsyms * 4;
2666
2667 do {
2668 bound = 0;
2669 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
2670 sym_index = find_elf_sym(symtab_section, p);
2671 if(sym_index) {
2672 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
2673 if(sym->st_shndx == SHN_UNDEF) {
2674 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
2675 #if 0
2676 printf("%5d\t%s\t%08x\n", i, p, sym->st_shndx);
2677 #endif
2678 ++bound;
2679 lseek(fd, off, SEEK_SET);
2680 if(tcc_load_object_file(s1, fd, off) < 0) {
2681 fail:
2682 ret = -1;
2683 goto the_end;
2684 }
2685 }
2686 }
2687 }
2688 } while(bound);
2689 ret = 0;
2690 the_end:
2691 tcc_free(data);
2692 return ret;
2693 }
2694
2695 /* load a '.a' file */
2696 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
2697 {
2698 ArchiveHeader hdr;
2699 char ar_size[11];
2700 char ar_name[17];
2701 char magic[8];
2702 int size, len, i;
2703 unsigned long file_offset;
2704
2705 /* skip magic which was already checked */
2706 read(fd, magic, sizeof(magic));
2707
2708 for(;;) {
2709 len = read(fd, &hdr, sizeof(hdr));
2710 if (len == 0)
2711 break;
2712 if (len != sizeof(hdr)) {
2713 tcc_error_noabort("invalid archive");
2714 return -1;
2715 }
2716 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
2717 ar_size[sizeof(hdr.ar_size)] = '\0';
2718 size = strtol(ar_size, NULL, 0);
2719 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
2720 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
2721 if (ar_name[i] != ' ')
2722 break;
2723 }
2724 ar_name[i + 1] = '\0';
2725 // printf("name='%s' size=%d %s\n", ar_name, size, ar_size);
2726 file_offset = lseek(fd, 0, SEEK_CUR);
2727 /* align to even */
2728 size = (size + 1) & ~1;
2729 if (!strcmp(ar_name, "/")) {
2730 /* coff symbol table : we handle it */
2731 if(s1->alacarte_link)
2732 return tcc_load_alacarte(s1, fd, size);
2733 } else if (!strcmp(ar_name, "//") ||
2734 !strcmp(ar_name, "__.SYMDEF") ||
2735 !strcmp(ar_name, "__.SYMDEF/") ||
2736 !strcmp(ar_name, "ARFILENAMES/")) {
2737 /* skip symbol table or archive names */
2738 } else {
2739 if (tcc_load_object_file(s1, fd, file_offset) < 0)
2740 return -1;
2741 }
2742 lseek(fd, file_offset + size, SEEK_SET);
2743 }
2744 return 0;
2745 }
2746
2747 #ifndef TCC_TARGET_PE
2748 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
2749 is referenced by the user (so it should be added as DT_NEEDED in
2750 the generated ELF file) */
2751 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
2752 {
2753 ElfW(Ehdr) ehdr;
2754 ElfW(Shdr) *shdr, *sh, *sh1;
2755 int i, j, nb_syms, nb_dts, sym_bind, ret;
2756 ElfW(Sym) *sym, *dynsym;
2757 ElfW(Dyn) *dt, *dynamic;
2758 unsigned char *dynstr;
2759 const char *name, *soname;
2760 DLLReference *dllref;
2761
2762 read(fd, &ehdr, sizeof(ehdr));
2763
2764 /* test CPU specific stuff */
2765 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2766 ehdr.e_machine != EM_TCC_TARGET) {
2767 tcc_error_noabort("bad architecture");
2768 return -1;
2769 }
2770
2771 /* read sections */
2772 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2773
2774 /* load dynamic section and dynamic symbols */
2775 nb_syms = 0;
2776 nb_dts = 0;
2777 dynamic = NULL;
2778 dynsym = NULL; /* avoid warning */
2779 dynstr = NULL; /* avoid warning */
2780 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
2781 switch(sh->sh_type) {
2782 case SHT_DYNAMIC:
2783 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
2784 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
2785 break;
2786 case SHT_DYNSYM:
2787 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2788 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
2789 sh1 = &shdr[sh->sh_link];
2790 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
2791 break;
2792 default:
2793 break;
2794 }
2795 }
2796
2797 /* compute the real library name */
2798 soname = tcc_basename(filename);
2799
2800 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2801 if (dt->d_tag == DT_SONAME) {
2802 soname = dynstr + dt->d_un.d_val;
2803 }
2804 }
2805
2806 /* if the dll is already loaded, do not load it */
2807 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2808 dllref = s1->loaded_dlls[i];
2809 if (!strcmp(soname, dllref->name)) {
2810 /* but update level if needed */
2811 if (level < dllref->level)
2812 dllref->level = level;
2813 ret = 0;
2814 goto the_end;
2815 }
2816 }
2817
2818 // printf("loading dll '%s'\n", soname);
2819
2820 /* add the dll and its level */
2821 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
2822 dllref->level = level;
2823 strcpy(dllref->name, soname);
2824 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
2825
2826 /* add dynamic symbols in dynsym_section */
2827 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
2828 sym_bind = ELFW(ST_BIND)(sym->st_info);
2829 if (sym_bind == STB_LOCAL)
2830 continue;
2831 name = dynstr + sym->st_name;
2832 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
2833 sym->st_info, sym->st_other, sym->st_shndx, name);
2834 }
2835
2836 /* load all referenced DLLs */
2837 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2838 switch(dt->d_tag) {
2839 case DT_NEEDED:
2840 name = dynstr + dt->d_un.d_val;
2841 for(j = 0; j < s1->nb_loaded_dlls; j++) {
2842 dllref = s1->loaded_dlls[j];
2843 if (!strcmp(name, dllref->name))
2844 goto already_loaded;
2845 }
2846 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
2847 tcc_error_noabort("referenced dll '%s' not found", name);
2848 ret = -1;
2849 goto the_end;
2850 }
2851 already_loaded:
2852 break;
2853 }
2854 }
2855 ret = 0;
2856 the_end:
2857 tcc_free(dynstr);
2858 tcc_free(dynsym);
2859 tcc_free(dynamic);
2860 tcc_free(shdr);
2861 return ret;
2862 }
2863
2864 #define LD_TOK_NAME 256
2865 #define LD_TOK_EOF (-1)
2866
2867 /* return next ld script token */
2868 static int ld_next(TCCState *s1, char *name, int name_size)
2869 {
2870 int c;
2871 char *q;
2872
2873 redo:
2874 switch(ch) {
2875 case ' ':
2876 case '\t':
2877 case '\f':
2878 case '\v':
2879 case '\r':
2880 case '\n':
2881 inp();
2882 goto redo;
2883 case '/':
2884 minp();
2885 if (ch == '*') {
2886 file->buf_ptr = parse_comment(file->buf_ptr);
2887 ch = file->buf_ptr[0];
2888 goto redo;
2889 } else {
2890 q = name;
2891 *q++ = '/';
2892 goto parse_name;
2893 }
2894 break;
2895 /* case 'a' ... 'z': */
2896 case 'a':
2897 case 'b':
2898 case 'c':
2899 case 'd':
2900 case 'e':
2901 case 'f':
2902 case 'g':
2903 case 'h':
2904 case 'i':
2905 case 'j':
2906 case 'k':
2907 case 'l':
2908 case 'm':
2909 case 'n':
2910 case 'o':
2911 case 'p':
2912 case 'q':
2913 case 'r':
2914 case 's':
2915 case 't':
2916 case 'u':
2917 case 'v':
2918 case 'w':
2919 case 'x':
2920 case 'y':
2921 case 'z':
2922 /* case 'A' ... 'z': */
2923 case 'A':
2924 case 'B':
2925 case 'C':
2926 case 'D':
2927 case 'E':
2928 case 'F':
2929 case 'G':
2930 case 'H':
2931 case 'I':
2932 case 'J':
2933 case 'K':
2934 case 'L':
2935 case 'M':
2936 case 'N':
2937 case 'O':
2938 case 'P':
2939 case 'Q':
2940 case 'R':
2941 case 'S':
2942 case 'T':
2943 case 'U':
2944 case 'V':
2945 case 'W':
2946 case 'X':
2947 case 'Y':
2948 case 'Z':
2949 case '_':
2950 case '\\':
2951 case '.':
2952 case '$':
2953 case '~':
2954 q = name;
2955 parse_name:
2956 for(;;) {
2957 if (!((ch >= 'a' && ch <= 'z') ||
2958 (ch >= 'A' && ch <= 'Z') ||
2959 (ch >= '0' && ch <= '9') ||
2960 strchr("/.-_+=$:\\,~", ch)))
2961 break;
2962 if ((q - name) < name_size - 1) {
2963 *q++ = ch;
2964 }
2965 minp();
2966 }
2967 *q = '\0';
2968 c = LD_TOK_NAME;
2969 break;
2970 case CH_EOF:
2971 c = LD_TOK_EOF;
2972 break;
2973 default:
2974 c = ch;
2975 inp();
2976 break;
2977 }
2978 #if 0
2979 printf("tok=%c %d\n", c, c);
2980 if (c == LD_TOK_NAME)
2981 printf(" name=%s\n", name);
2982 #endif
2983 return c;
2984 }
2985
2986 /*
2987 * Extract the file name from the library name
2988 *
2989 * /!\ No test on filename capacity, be careful
2990 */
2991 static void libname_to_filename(TCCState *s1, const char libname[], char filename[])
2992 {
2993 if (!s1->static_link) {
2994 sprintf(filename, "lib%s.so", libname);
2995 } else {
2996 sprintf(filename, "lib%s.a", libname);
2997 }
2998 }
2999
3000 static int ld_add_file(TCCState *s1, const char filename[])
3001 {
3002 int ret;
3003
3004 ret = tcc_add_file_internal(s1, filename, 0);
3005 if (ret)
3006 ret = tcc_add_dll(s1, filename, 0);
3007 return ret;
3008 }
3009
3010 static inline int new_undef_syms(void)
3011 {
3012 int ret = 0;
3013 ret = new_undef_sym;
3014 new_undef_sym = 0;
3015 return ret;
3016 }
3017
3018 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3019 {
3020 char filename[1024], libname[1024];
3021 int t, group, nblibs = 0, ret = 0;
3022 char **libs = NULL;
3023
3024 group = !strcmp(cmd, "GROUP");
3025 if (!as_needed)
3026 new_undef_syms();
3027 t = ld_next(s1, filename, sizeof(filename));
3028 if (t != '(')
3029 expect("(");
3030 t = ld_next(s1, filename, sizeof(filename));
3031 for(;;) {
3032 libname[0] = '\0';
3033 if (t == LD_TOK_EOF) {
3034 tcc_error_noabort("unexpected end of file");
3035 ret = -1;
3036 goto lib_parse_error;
3037 } else if (t == ')') {
3038 break;
3039 } else if (t == '-') {
3040 t = ld_next(s1, filename, sizeof(filename));
3041 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3042 tcc_error_noabort("library name expected");
3043 ret = -1;
3044 goto lib_parse_error;
3045 }
3046 strcpy(libname, &filename[1]);
3047 libname_to_filename(s1, libname, filename);
3048 } else if (t != LD_TOK_NAME) {
3049 tcc_error_noabort("filename expected");
3050 ret = -1;
3051 goto lib_parse_error;
3052 }
3053 if (!strcmp(filename, "AS_NEEDED")) {
3054 ret = ld_add_file_list(s1, cmd, 1);
3055 if (ret)
3056 goto lib_parse_error;
3057 } else {
3058 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3059 if (!as_needed) {
3060 ret = ld_add_file(s1, filename);
3061 if (ret)
3062 goto lib_parse_error;
3063 if (group) {
3064 /* Add the filename *and* the libname to avoid future conversions */
3065 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3066 if (libname[0] != '\0')
3067 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3068 }
3069 }
3070 }
3071 t = ld_next(s1, filename, sizeof(filename));
3072 if (t == ',') {
3073 t = ld_next(s1, filename, sizeof(filename));
3074 }
3075 }
3076 if (group && !as_needed) {
3077 while (new_undef_syms()) {
3078 int i;
3079
3080 for (i = 0; i < nblibs; i ++)
3081 ld_add_file(s1, libs[i]);
3082 }
3083 }
3084 lib_parse_error:
3085 dynarray_reset(&libs, &nblibs);
3086 return ret;
3087 }
3088
3089 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3090 files */
3091 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3092 {
3093 char cmd[64];
3094 char filename[1024];
3095 int t, ret;
3096
3097 ch = file->buf_ptr[0];
3098 ch = handle_eob();
3099 for(;;) {
3100 t = ld_next(s1, cmd, sizeof(cmd));
3101 if (t == LD_TOK_EOF)
3102 return 0;
3103 else if (t != LD_TOK_NAME)
3104 return -1;
3105 if (!strcmp(cmd, "INPUT") ||
3106 !strcmp(cmd, "GROUP")) {
3107 ret = ld_add_file_list(s1, cmd, 0);
3108 if (ret)
3109 return ret;
3110 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3111 !strcmp(cmd, "TARGET")) {
3112 /* ignore some commands */
3113 t = ld_next(s1, cmd, sizeof(cmd));
3114 if (t != '(')
3115 expect("(");
3116 for(;;) {
3117 t = ld_next(s1, filename, sizeof(filename));
3118 if (t == LD_TOK_EOF) {
3119 tcc_error_noabort("unexpected end of file");
3120 return -1;
3121 } else if (t == ')') {
3122 break;
3123 }
3124 }
3125 } else {
3126 return -1;
3127 }
3128 }
3129 return 0;
3130 }
3131 #endif /* ndef TCC_TARGET_PE */
Tiny C Compiler main repository