?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
-std=c99 option for the tcc: allow to use a tcc as a reference compiler for "make...
[tinycc.git] / tccelf.c
blob419a0e19213c41ca9b4068f776103b4e560ed806
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 /* Define this to get some debug output during relocation processing. */
24 #undef DEBUG_RELOC
25
26 /* XXX: avoid static variable */
27 static int new_undef_sym = 0; /* Is there a new undefined sym since last new_undef_sym() */
28
29 ST_FUNC int put_elf_str(Section *s, const char *sym)
30 {
31 int offset, len;
32 char *ptr;
33
34 len = strlen(sym) + 1;
35 offset = s->data_offset;
36 ptr = section_ptr_add(s, len);
37 memcpy(ptr, sym, len);
38 return offset;
39 }
40
41 /* elf symbol hashing function */
42 static unsigned long elf_hash(const unsigned char *name)
43 {
44 unsigned long h = 0, g;
45
46 while (*name) {
47 h = (h << 4) + *name++;
48 g = h & 0xf0000000;
49 if (g)
50 h ^= g >> 24;
51 h &= ~g;
52 }
53 return h;
54 }
55
56 /* rebuild hash table of section s */
57 /* NOTE: we do factorize the hash table code to go faster */
58 static void rebuild_hash(Section *s, unsigned int nb_buckets)
59 {
60 ElfW(Sym) *sym;
61 int *ptr, *hash, nb_syms, sym_index, h;
62 unsigned char *strtab;
63
64 strtab = s->link->data;
65 nb_syms = s->data_offset / sizeof(ElfW(Sym));
66
67 s->hash->data_offset = 0;
68 ptr = section_ptr_add(s->hash, (2 + nb_buckets + nb_syms) * sizeof(int));
69 ptr[0] = nb_buckets;
70 ptr[1] = nb_syms;
71 ptr += 2;
72 hash = ptr;
73 memset(hash, 0, (nb_buckets + 1) * sizeof(int));
74 ptr += nb_buckets + 1;
75
76 sym = (ElfW(Sym) *)s->data + 1;
77 for(sym_index = 1; sym_index < nb_syms; sym_index++) {
78 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
79 h = elf_hash(strtab + sym->st_name) % nb_buckets;
80 *ptr = hash[h];
81 hash[h] = sym_index;
82 } else {
83 *ptr = 0;
84 }
85 ptr++;
86 sym++;
87 }
88 }
89
90 /* return the symbol number */
91 ST_FUNC int put_elf_sym(Section *s, addr_t value, unsigned long size,
92 int info, int other, int shndx, const char *name)
93 {
94 int name_offset, sym_index;
95 int nbuckets, h;
96 ElfW(Sym) *sym;
97 Section *hs;
98
99 sym = section_ptr_add(s, sizeof(ElfW(Sym)));
100 if (name)
101 name_offset = put_elf_str(s->link, name);
102 else
103 name_offset = 0;
104 /* XXX: endianness */
105 sym->st_name = name_offset;
106 sym->st_value = value;
107 sym->st_size = size;
108 sym->st_info = info;
109 sym->st_other = other;
110 sym->st_shndx = shndx;
111 sym_index = sym - (ElfW(Sym) *)s->data;
112 hs = s->hash;
113 if (hs) {
114 int *ptr, *base;
115 ptr = section_ptr_add(hs, sizeof(int));
116 base = (int *)hs->data;
117 /* only add global or weak symbols */
118 if (ELFW(ST_BIND)(info) != STB_LOCAL) {
119 /* add another hashing entry */
120 nbuckets = base[0];
121 h = elf_hash((unsigned char *) name) % nbuckets;
122 *ptr = base[2 + h];
123 base[2 + h] = sym_index;
124 base[1]++;
125 /* we resize the hash table */
126 hs->nb_hashed_syms++;
127 if (hs->nb_hashed_syms > 2 * nbuckets) {
128 rebuild_hash(s, 2 * nbuckets);
129 }
130 } else {
131 *ptr = 0;
132 base[1]++;
133 }
134 }
135 return sym_index;
136 }
137
138 /* find global ELF symbol 'name' and return its index. Return 0 if not
139 found. */
140 ST_FUNC int find_elf_sym(Section *s, const char *name)
141 {
142 ElfW(Sym) *sym;
143 Section *hs;
144 int nbuckets, sym_index, h;
145 const char *name1;
146
147 hs = s->hash;
148 if (!hs)
149 return 0;
150 nbuckets = ((int *)hs->data)[0];
151 h = elf_hash((unsigned char *) name) % nbuckets;
152 sym_index = ((int *)hs->data)[2 + h];
153 while (sym_index != 0) {
154 sym = &((ElfW(Sym) *)s->data)[sym_index];
155 name1 = (char *) s->link->data + sym->st_name;
156 if (!strcmp(name, name1))
157 return sym_index;
158 sym_index = ((int *)hs->data)[2 + nbuckets + sym_index];
159 }
160 return 0;
161 }
162
163 /* return elf symbol value, signal error if 'err' is nonzero */
164 ST_FUNC addr_t get_elf_sym_addr(TCCState *s, const char *name, int err)
165 {
166 int sym_index;
167 ElfW(Sym) *sym;
168
169 sym_index = find_elf_sym(s->symtab, name);
170 sym = &((ElfW(Sym) *)s->symtab->data)[sym_index];
171 if (!sym_index || sym->st_shndx == SHN_UNDEF) {
172 if (err)
173 tcc_error("%s not defined", name);
174 return 0;
175 }
176 return sym->st_value;
177 }
178
179 /* return elf symbol value */
180 LIBTCCAPI void *tcc_get_symbol(TCCState *s, const char *name)
181 {
182 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 0);
183 }
184
185 #if defined TCC_IS_NATIVE || defined TCC_TARGET_PE
186 /* return elf symbol value or error */
187 ST_FUNC void* tcc_get_symbol_err(TCCState *s, const char *name)
188 {
189 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 1);
190 }
191 #endif
192
193 /* add an elf symbol : check if it is already defined and patch
194 it. Return symbol index. NOTE that sh_num can be SHN_UNDEF. */
195 ST_FUNC int add_elf_sym(Section *s, addr_t value, unsigned long size,
196 int info, int other, int sh_num, const char *name)
197 {
198 ElfW(Sym) *esym;
199 int sym_bind, sym_index, sym_type, esym_bind;
200 unsigned char sym_vis, esym_vis, new_vis;
201
202 sym_bind = ELFW(ST_BIND)(info);
203 sym_type = ELFW(ST_TYPE)(info);
204 sym_vis = ELFW(ST_VISIBILITY)(other);
205
206 if (sym_bind != STB_LOCAL) {
207 /* we search global or weak symbols */
208 sym_index = find_elf_sym(s, name);
209 if (!sym_index)
210 goto do_def;
211 esym = &((ElfW(Sym) *)s->data)[sym_index];
212 if (esym->st_shndx != SHN_UNDEF) {
213 esym_bind = ELFW(ST_BIND)(esym->st_info);
214 /* propagate the most constraining visibility */
215 /* STV_DEFAULT(0)<STV_PROTECTED(3)<STV_HIDDEN(2)<STV_INTERNAL(1) */
216 esym_vis = ELFW(ST_VISIBILITY)(esym->st_other);
217 if (esym_vis == STV_DEFAULT) {
218 new_vis = sym_vis;
219 } else if (sym_vis == STV_DEFAULT) {
220 new_vis = esym_vis;
221 } else {
222 new_vis = (esym_vis < sym_vis) ? esym_vis : sym_vis;
223 }
224 esym->st_other = (esym->st_other & ~ELFW(ST_VISIBILITY)(-1))
225 | new_vis;
226 other = esym->st_other; /* in case we have to patch esym */
227 if (sh_num == SHN_UNDEF) {
228 /* ignore adding of undefined symbol if the
229 corresponding symbol is already defined */
230 } else if (sym_bind == STB_GLOBAL && esym_bind == STB_WEAK) {
231 /* global overrides weak, so patch */
232 goto do_patch;
233 } else if (sym_bind == STB_WEAK && esym_bind == STB_GLOBAL) {
234 /* weak is ignored if already global */
235 } else if (sym_bind == STB_WEAK && esym_bind == STB_WEAK) {
236 /* keep first-found weak definition, ignore subsequents */
237 } else if (sym_vis == STV_HIDDEN || sym_vis == STV_INTERNAL) {
238 /* ignore hidden symbols after */
239 } else if (esym->st_shndx == SHN_COMMON
240 && (sh_num < SHN_LORESERVE || sh_num == SHN_COMMON)) {
241 /* gr: Happens with 'tcc ... -static tcctest.c' on e.g. Ubuntu 6.01
242 No idea if this is the correct solution ... */
243 goto do_patch;
244 } else if (s == tcc_state->dynsymtab_section) {
245 /* we accept that two DLL define the same symbol */
246 } else {
247 #if 0
248 printf("new_bind=%x new_shndx=%x new_vis=%x old_bind=%x old_shndx=%x old_vis=%x\n",
249 sym_bind, sh_num, new_vis, esym_bind, esym->st_shndx, esym_vis);
250 #endif
251 tcc_error_noabort("'%s' defined twice", name);
252 }
253 } else {
254 do_patch:
255 esym->st_info = ELFW(ST_INFO)(sym_bind, sym_type);
256 esym->st_shndx = sh_num;
257 new_undef_sym = 1;
258 esym->st_value = value;
259 esym->st_size = size;
260 esym->st_other = other;
261 }
262 } else {
263 do_def:
264 sym_index = put_elf_sym(s, value, size,
265 ELFW(ST_INFO)(sym_bind, sym_type), other,
266 sh_num, name);
267 }
268 return sym_index;
269 }
270
271 /* put relocation */
272 ST_FUNC void put_elf_reloca(Section *symtab, Section *s, unsigned long offset,
273 int type, int symbol, unsigned long addend)
274 {
275 char buf[256];
276 Section *sr;
277 ElfW_Rel *rel;
278
279 sr = s->reloc;
280 if (!sr) {
281 /* if no relocation section, create it */
282 snprintf(buf, sizeof(buf), REL_SECTION_FMT, s->name);
283 /* if the symtab is allocated, then we consider the relocation
284 are also */
285 sr = new_section(tcc_state, buf, SHT_RELX, symtab->sh_flags);
286 sr->sh_entsize = sizeof(ElfW_Rel);
287 sr->link = symtab;
288 sr->sh_info = s->sh_num;
289 s->reloc = sr;
290 }
291 rel = section_ptr_add(sr, sizeof(ElfW_Rel));
292 rel->r_offset = offset;
293 rel->r_info = ELFW(R_INFO)(symbol, type);
294 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
295 rel->r_addend = addend;
296 #else
297 if (addend)
298 tcc_error("non-zero addend on REL architecture");
299 #endif
300 }
301
302 ST_FUNC void put_elf_reloc(Section *symtab, Section *s, unsigned long offset,
303 int type, int symbol)
304 {
305 put_elf_reloca(symtab, s, offset, type, symbol, 0);
306 }
307
308 /* put stab debug information */
309
310 ST_FUNC void put_stabs(const char *str, int type, int other, int desc,
311 unsigned long value)
312 {
313 Stab_Sym *sym;
314
315 sym = section_ptr_add(stab_section, sizeof(Stab_Sym));
316 if (str) {
317 sym->n_strx = put_elf_str(stabstr_section, str);
318 } else {
319 sym->n_strx = 0;
320 }
321 sym->n_type = type;
322 sym->n_other = other;
323 sym->n_desc = desc;
324 sym->n_value = value;
325 }
326
327 ST_FUNC void put_stabs_r(const char *str, int type, int other, int desc,
328 unsigned long value, Section *sec, int sym_index)
329 {
330 put_stabs(str, type, other, desc, value);
331 put_elf_reloc(symtab_section, stab_section,
332 stab_section->data_offset - sizeof(unsigned int),
333 R_DATA_32, sym_index);
334 }
335
336 ST_FUNC void put_stabn(int type, int other, int desc, int value)
337 {
338 put_stabs(NULL, type, other, desc, value);
339 }
340
341 ST_FUNC void put_stabd(int type, int other, int desc)
342 {
343 put_stabs(NULL, type, other, desc, 0);
344 }
345
346 /* Browse each elem of type <type> in section <sec> starting at elem <startoff>
347 using variable <elem> */
348 #define for_each_elem(sec, startoff, elem, type) \
349 for (elem = (type *) sec->data + startoff; \
350 elem < (type *) (sec->data + sec->data_offset); elem++)
351
352 /* In an ELF file symbol table, the local symbols must appear below
353 the global and weak ones. Since TCC cannot sort it while generating
354 the code, we must do it after. All the relocation tables are also
355 modified to take into account the symbol table sorting */
356 static void sort_syms(TCCState *s1, Section *s)
357 {
358 int *old_to_new_syms;
359 ElfW(Sym) *new_syms;
360 int nb_syms, i;
361 ElfW(Sym) *p, *q;
362 ElfW_Rel *rel;
363 Section *sr;
364 int type, sym_index;
365
366 nb_syms = s->data_offset / sizeof(ElfW(Sym));
367 new_syms = tcc_malloc(nb_syms * sizeof(ElfW(Sym)));
368 old_to_new_syms = tcc_malloc(nb_syms * sizeof(int));
369
370 /* first pass for local symbols */
371 p = (ElfW(Sym) *)s->data;
372 q = new_syms;
373 for(i = 0; i < nb_syms; i++) {
374 if (ELFW(ST_BIND)(p->st_info) == STB_LOCAL) {
375 old_to_new_syms[i] = q - new_syms;
376 *q++ = *p;
377 }
378 p++;
379 }
380 /* save the number of local symbols in section header */
381 s->sh_info = q - new_syms;
382
383 /* then second pass for non local symbols */
384 p = (ElfW(Sym) *)s->data;
385 for(i = 0; i < nb_syms; i++) {
386 if (ELFW(ST_BIND)(p->st_info) != STB_LOCAL) {
387 old_to_new_syms[i] = q - new_syms;
388 *q++ = *p;
389 }
390 p++;
391 }
392
393 /* we copy the new symbols to the old */
394 memcpy(s->data, new_syms, nb_syms * sizeof(ElfW(Sym)));
395 tcc_free(new_syms);
396
397 /* now we modify all the relocations */
398 for(i = 1; i < s1->nb_sections; i++) {
399 sr = s1->sections[i];
400 if (sr->sh_type == SHT_RELX && sr->link == s) {
401 for_each_elem(sr, 0, rel, ElfW_Rel) {
402 sym_index = ELFW(R_SYM)(rel->r_info);
403 type = ELFW(R_TYPE)(rel->r_info);
404 sym_index = old_to_new_syms[sym_index];
405 rel->r_info = ELFW(R_INFO)(sym_index, type);
406 }
407 }
408 }
409
410 tcc_free(old_to_new_syms);
411 }
412
413 /* relocate common symbols in the .bss section */
414 ST_FUNC void relocate_common_syms(void)
415 {
416 ElfW(Sym) *sym;
417 unsigned long offset, align;
418
419 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
420 if (sym->st_shndx == SHN_COMMON) {
421 /* align symbol */
422 align = sym->st_value;
423 offset = bss_section->data_offset;
424 offset = (offset + align - 1) & -align;
425 sym->st_value = offset;
426 sym->st_shndx = bss_section->sh_num;
427 offset += sym->st_size;
428 bss_section->data_offset = offset;
429 }
430 }
431 }
432
433 /* relocate symbol table, resolve undefined symbols if do_resolve is
434 true and output error if undefined symbol. */
435 ST_FUNC void relocate_syms(TCCState *s1, int do_resolve)
436 {
437 ElfW(Sym) *sym, *esym;
438 int sym_bind, sh_num, sym_index;
439 const char *name;
440
441 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
442 sh_num = sym->st_shndx;
443 if (sh_num == SHN_UNDEF) {
444 name = (char *) strtab_section->data + sym->st_name;
445 /* Use ld.so to resolve symbol for us (for tcc -run) */
446 if (do_resolve) {
447 #if defined TCC_IS_NATIVE && !defined _WIN32
448 void *addr;
449 name = (char *) symtab_section->link->data + sym->st_name;
450 addr = resolve_sym(s1, name);
451 if (addr) {
452 sym->st_value = (addr_t)addr;
453 #ifdef DEBUG_RELOC
454 printf ("relocate_sym: %s -> 0x%lx\n", name, sym->st_value);
455 #endif
456 goto found;
457 }
458 #endif
459 } else if (s1->dynsym) {
460 /* if dynamic symbol exist, then use it */
461 sym_index = find_elf_sym(s1->dynsym, name);
462 if (sym_index) {
463 esym = &((ElfW(Sym) *)s1->dynsym->data)[sym_index];
464 sym->st_value = esym->st_value;
465 goto found;
466 }
467 }
468 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
469 it */
470 if (!strcmp(name, "_fp_hw"))
471 goto found;
472 /* only weak symbols are accepted to be undefined. Their
473 value is zero */
474 sym_bind = ELFW(ST_BIND)(sym->st_info);
475 if (sym_bind == STB_WEAK) {
476 sym->st_value = 0;
477 } else {
478 tcc_error_noabort("undefined symbol '%s'", name);
479 }
480 } else if (sh_num < SHN_LORESERVE) {
481 /* add section base */
482 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
483 }
484 found: ;
485 }
486 }
487
488 /* relocate a given section (CPU dependent) by applying the relocations
489 in the associated relocation section */
490 ST_FUNC void relocate_section(TCCState *s1, Section *s)
491 {
492 Section *sr = s->reloc;
493 ElfW_Rel *rel;
494 ElfW(Sym) *sym;
495 int type, sym_index;
496 unsigned char *ptr;
497 addr_t val, addr;
498 #if defined TCC_TARGET_I386 || defined TCC_TARGET_X86_64
499 ElfW_Rel *qrel = (ElfW_Rel *) sr->data; /* ptr to next reloc entry reused */
500 int esym_index;
501 #endif
502
503 for_each_elem(sr, 0, rel, ElfW_Rel) {
504 ptr = s->data + rel->r_offset;
505
506 sym_index = ELFW(R_SYM)(rel->r_info);
507 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
508 val = sym->st_value;
509 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
510 val += rel->r_addend;
511 #endif
512 type = ELFW(R_TYPE)(rel->r_info);
513 addr = s->sh_addr + rel->r_offset;
514
515 /* CPU specific */
516 switch(type) {
517 #if defined(TCC_TARGET_I386)
518 case R_386_32:
519 if (s1->output_type == TCC_OUTPUT_DLL) {
520 esym_index = s1->symtab_to_dynsym[sym_index];
521 qrel->r_offset = rel->r_offset;
522 if (esym_index) {
523 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_32);
524 qrel++;
525 break;
526 } else {
527 qrel->r_info = ELFW(R_INFO)(0, R_386_RELATIVE);
528 qrel++;
529 }
530 }
531 *(int *)ptr += val;
532 break;
533 case R_386_PC32:
534 if (s1->output_type == TCC_OUTPUT_DLL) {
535 /* DLL relocation */
536 esym_index = s1->symtab_to_dynsym[sym_index];
537 if (esym_index) {
538 qrel->r_offset = rel->r_offset;
539 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_PC32);
540 qrel++;
541 break;
542 }
543 }
544 *(int *)ptr += val - addr;
545 break;
546 case R_386_PLT32:
547 *(int *)ptr += val - addr;
548 break;
549 case R_386_GLOB_DAT:
550 case R_386_JMP_SLOT:
551 *(int *)ptr = val;
552 break;
553 case R_386_GOTPC:
554 *(int *)ptr += s1->got->sh_addr - addr;
555 break;
556 case R_386_GOTOFF:
557 *(int *)ptr += val - s1->got->sh_addr;
558 break;
559 case R_386_GOT32:
560 /* we load the got offset */
561 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
562 break;
563 case R_386_16:
564 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY) {
565 output_file:
566 tcc_error("can only produce 16-bit binary files");
567 }
568 *(short *)ptr += val;
569 break;
570 case R_386_PC16:
571 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY)
572 goto output_file;
573 *(short *)ptr += val - addr;
574 break;
575 #elif defined(TCC_TARGET_ARM)
576 case R_ARM_PC24:
577 case R_ARM_CALL:
578 case R_ARM_JUMP24:
579 case R_ARM_PLT32:
580 {
581 int x, is_thumb, is_call, h, blx_avail, is_bl, th_ko;
582 x = (*(int *) ptr) & 0xffffff;
583 if (sym->st_shndx == SHN_UNDEF)
584 val = s1->plt->sh_addr;
585 #ifdef DEBUG_RELOC
586 printf ("reloc %d: x=0x%x val=0x%x ", type, x, val);
587 #endif
588 (*(int *)ptr) &= 0xff000000;
589 if (x & 0x800000)
590 x -= 0x1000000;
591 x <<= 2;
592 blx_avail = (TCC_ARM_VERSION >= 5);
593 is_thumb = val & 1;
594 is_bl = (*(unsigned *) ptr) >> 24 == 0xeb;
595 is_call = (type == R_ARM_CALL || (type == R_ARM_PC24 && is_bl));
596 x += val - addr;
597 #ifdef DEBUG_RELOC
598 printf (" newx=0x%x name=%s\n", x,
599 (char *) symtab_section->link->data + sym->st_name);
600 #endif
601 h = x & 2;
602 th_ko = (x & 3) && (!blx_avail || !is_call);
603 if (th_ko || x >= 0x2000000 || x < -0x2000000)
604 tcc_error("can't relocate value at %x,%d",addr, type);
605 x >>= 2;
606 x &= 0xffffff;
607 /* Only reached if blx is avail and it is a call */
608 if (is_thumb) {
609 x |= h << 24;
610 (*(int *)ptr) = 0xfa << 24; /* bl -> blx */
611 }
612 (*(int *) ptr) |= x;
613 }
614 break;
615 /* Since these relocations only concern Thumb-2 and blx instruction was
616 introduced before Thumb-2, we can assume blx is available and not
617 guard its use */
618 case R_ARM_THM_PC22:
619 case R_ARM_THM_JUMP24:
620 {
621 int x, hi, lo, s, j1, j2, i1, i2, imm10, imm11;
622 int to_thumb, is_call, to_plt, blx_bit = 1 << 12;
623 Section *plt;
624
625 /* weak reference */
626 if (sym->st_shndx == SHN_UNDEF &&
627 ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
628 break;
629
630 /* Get initial offset */
631 hi = (*(uint16_t *)ptr);
632 lo = (*(uint16_t *)(ptr+2));
633 s = (hi >> 10) & 1;
634 j1 = (lo >> 13) & 1;
635 j2 = (lo >> 11) & 1;
636 i1 = (j1 ^ s) ^ 1;
637 i2 = (j2 ^ s) ^ 1;
638 imm10 = hi & 0x3ff;
639 imm11 = lo & 0x7ff;
640 x = (s << 24) | (i1 << 23) | (i2 << 22) |
641 (imm10 << 12) | (imm11 << 1);
642 if (x & 0x01000000)
643 x -= 0x02000000;
644
645 /* Relocation infos */
646 to_thumb = val & 1;
647 plt = s1->plt;
648 to_plt = (val >= plt->sh_addr) &&
649 (val < plt->sh_addr + plt->data_offset);
650 is_call = (type == R_ARM_THM_PC22);
651
652 /* Compute final offset */
653 if (to_plt && !is_call) /* Point to 1st instr of Thumb stub */
654 x -= 4;
655 x += val - addr;
656 if (!to_thumb && is_call) {
657 blx_bit = 0; /* bl -> blx */
658 x = (x + 3) & -4; /* Compute offset from aligned PC */
659 }
660
661 /* Check that relocation is possible
662 * offset must not be out of range
663 * if target is to be entered in arm mode:
664 - bit 1 must not set
665 - instruction must be a call (bl) or a jump to PLT */
666 if (!to_thumb || x >= 0x1000000 || x < -0x1000000)
667 if (to_thumb || (val & 2) || (!is_call && !to_plt))
668 tcc_error("can't relocate value at %x,%d",addr, type);
669
670 /* Compute and store final offset */
671 s = (x >> 24) & 1;
672 i1 = (x >> 23) & 1;
673 i2 = (x >> 22) & 1;
674 j1 = s ^ (i1 ^ 1);
675 j2 = s ^ (i2 ^ 1);
676 imm10 = (x >> 12) & 0x3ff;
677 imm11 = (x >> 1) & 0x7ff;
678 (*(uint16_t *)ptr) = (uint16_t) ((hi & 0xf800) |
679 (s << 10) | imm10);
680 (*(uint16_t *)(ptr+2)) = (uint16_t) ((lo & 0xc000) |
681 (j1 << 13) | blx_bit | (j2 << 11) |
682 imm11);
683 }
684 break;
685 case R_ARM_MOVT_ABS:
686 case R_ARM_MOVW_ABS_NC:
687 {
688 int x, imm4, imm12;
689 if (type == R_ARM_MOVT_ABS)
690 val >>= 16;
691 imm12 = val & 0xfff;
692 imm4 = (val >> 12) & 0xf;
693 x = (imm4 << 16) | imm12;
694 if (type == R_ARM_THM_MOVT_ABS)
695 *(int *)ptr |= x;
696 else
697 *(int *)ptr += x;
698 }
699 break;
700 case R_ARM_THM_MOVT_ABS:
701 case R_ARM_THM_MOVW_ABS_NC:
702 {
703 int x, i, imm4, imm3, imm8;
704 if (type == R_ARM_THM_MOVT_ABS)
705 val >>= 16;
706 imm8 = val & 0xff;
707 imm3 = (val >> 8) & 0x7;
708 i = (val >> 11) & 1;
709 imm4 = (val >> 12) & 0xf;
710 x = (imm3 << 28) | (imm8 << 16) | (i << 10) | imm4;
711 if (type == R_ARM_THM_MOVT_ABS)
712 *(int *)ptr |= x;
713 else
714 *(int *)ptr += x;
715 }
716 break;
717 case R_ARM_PREL31:
718 {
719 int x;
720 x = (*(int *)ptr) & 0x7fffffff;
721 (*(int *)ptr) &= 0x80000000;
722 x = (x * 2) / 2;
723 x += val - addr;
724 if((x^(x>>1))&0x40000000)
725 tcc_error("can't relocate value at %x,%d",addr, type);
726 (*(int *)ptr) |= x & 0x7fffffff;
727 }
728 case R_ARM_ABS32:
729 *(int *)ptr += val;
730 break;
731 case R_ARM_REL32:
732 *(int *)ptr += val - addr;
733 break;
734 case R_ARM_GOTPC:
735 *(int *)ptr += s1->got->sh_addr - addr;
736 break;
737 case R_ARM_GOTOFF:
738 *(int *)ptr += val - s1->got->sh_addr;
739 break;
740 case R_ARM_GOT32:
741 /* we load the got offset */
742 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
743 break;
744 case R_ARM_COPY:
745 break;
746 case R_ARM_V4BX:
747 /* trade Thumb support for ARMv4 support */
748 if ((0x0ffffff0 & *(int*)ptr) == 0x012FFF10)
749 *(int*)ptr ^= 0xE12FFF10 ^ 0xE1A0F000; /* BX Rm -> MOV PC, Rm */
750 break;
751 case R_ARM_GLOB_DAT:
752 case R_ARM_JUMP_SLOT:
753 *(addr_t *)ptr = val;
754 break;
755 case R_ARM_NONE:
756 /* Nothing to do. Normally used to indicate a dependency
757 on a certain symbol (like for exception handling under EABI). */
758 break;
759 default:
760 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
761 type, (unsigned)addr, ptr, (unsigned)val);
762 break;
763 #elif defined(TCC_TARGET_ARM64)
764 case R_AARCH64_ABS64:
765 *(uint64_t *)ptr = val;
766 break;
767 case R_AARCH64_ABS32:
768 *(uint32_t *)ptr = val;
769 break;
770 case R_AARCH64_MOVW_UABS_G0_NC:
771 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0xffe0001f) |
772 (val & 0xffff) << 5;
773 break;
774 case R_AARCH64_MOVW_UABS_G1_NC:
775 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0xffe0001f) |
776 (val >> 16 & 0xffff) << 5;
777 break;
778 case R_AARCH64_MOVW_UABS_G2_NC:
779 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0xffe0001f) |
780 (val >> 32 & 0xffff) << 5;
781 break;
782 case R_AARCH64_MOVW_UABS_G3:
783 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0xffe0001f) |
784 (val >> 48 & 0xffff) << 5;
785 break;
786 case R_AARCH64_ADR_PREL_PG_HI21: {
787 uint64_t off = (val >> 12) - (addr >> 12);
788 if ((off + ((uint64_t)1 << 20)) >> 21)
789 tcc_error("R_AARCH64_ADR_PREL_PG_HI21 relocation failed");
790 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0x9f00001f) |
791 (off & 0x1ffffc) << 3 | (off & 3) << 29;
792 break;
793 }
794 case R_AARCH64_ADD_ABS_LO12_NC:
795 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0xffc003ff) |
796 (val & 0xfff) << 10;
797 break;
798 case R_AARCH64_JUMP26:
799 case R_AARCH64_CALL26:
800 /* This check must match the one in build_got_entries, testing
801 if we really need a PLT slot. */
802 if (sym->st_shndx == SHN_UNDEF)
803 /* We've put the PLT slot offset into r_addend when generating
804 it, and that's what we must use as relocation value (adjusted
805 by section offset of course). */
806 val = s1->plt->sh_addr + rel->r_addend;
807 #ifdef DEBUG_RELOC
808 printf ("reloc %d @ 0x%lx: val=0x%lx name=%s\n", type, addr, val,
809 (char *) symtab_section->link->data + sym->st_name);
810 #endif
811 if (((val - addr) + ((uint64_t)1 << 27)) & ~(uint64_t)0xffffffc)
812 {
813 tcc_error("R_AARCH64_(JUMP|CALL)26 relocation failed (val=%lx, addr=%lx)", addr, val);
814 }
815 *(uint32_t *)ptr = 0x14000000 | (type == R_AARCH64_CALL26) << 31 |
816 ((val - addr) >> 2 & 0x3ffffff);
817 break;
818 case R_AARCH64_ADR_GOT_PAGE: {
819 uint64_t off =
820 (((s1->got->sh_addr +
821 s1->sym_attrs[sym_index].got_offset) >> 12) - (addr >> 12));
822 if ((off + ((uint64_t)1 << 20)) >> 21)
823 tcc_error("R_AARCH64_ADR_GOT_PAGE relocation failed");
824 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0x9f00001f) |
825 (off & 0x1ffffc) << 3 | (off & 3) << 29;
826 break;
827 }
828 case R_AARCH64_LD64_GOT_LO12_NC:
829 *(uint32_t *)ptr = (*(uint32_t *)ptr & 0xfff803ff) |
830 ((s1->got->sh_addr + s1->sym_attrs[sym_index].got_offset)
831 & 0xff8) << 7;
832 break;
833 case R_AARCH64_COPY:
834 break;
835 case R_AARCH64_GLOB_DAT:
836 case R_AARCH64_JUMP_SLOT:
837 /* They don't need addend */
838 #ifdef DEBUG_RELOC
839 printf ("reloc %d @ 0x%lx: val=0x%lx name=%s\n", type, addr,
840 val - rel->r_addend,
841 (char *) symtab_section->link->data + sym->st_name);
842 #endif
843 *(addr_t *)ptr = val - rel->r_addend;
844 break;
845 default:
846 fprintf(stderr, "FIXME: handle reloc type %x at %x [%p] to %x\n",
847 type, (unsigned)addr, ptr, (unsigned)val);
848 break;
849 #elif defined(TCC_TARGET_C67)
850 case R_C60_32:
851 *(int *)ptr += val;
852 break;
853 case R_C60LO16:
854 {
855 uint32_t orig;
856
857 /* put the low 16 bits of the absolute address
858 add to what is already there */
859
860 orig = ((*(int *)(ptr )) >> 7) & 0xffff;
861 orig |= (((*(int *)(ptr+4)) >> 7) & 0xffff) << 16;
862
863 /* patch both at once - assumes always in pairs Low - High */
864
865 *(int *) ptr = (*(int *) ptr & (~(0xffff << 7)) ) | (((val+orig) & 0xffff) << 7);
866 *(int *)(ptr+4) = (*(int *)(ptr+4) & (~(0xffff << 7)) ) | ((((val+orig)>>16) & 0xffff) << 7);
867 }
868 break;
869 case R_C60HI16:
870 break;
871 default:
872 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
873 type, (unsigned)addr, ptr, (unsigned)val);
874 break;
875 #elif defined(TCC_TARGET_X86_64)
876 case R_X86_64_64:
877 if (s1->output_type == TCC_OUTPUT_DLL) {
878 esym_index = s1->symtab_to_dynsym[sym_index];
879 qrel->r_offset = rel->r_offset;
880 if (esym_index) {
881 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_64);
882 qrel->r_addend = rel->r_addend;
883 qrel++;
884 break;
885 } else {
886 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
887 qrel->r_addend = *(long long *)ptr + val;
888 qrel++;
889 }
890 }
891 *(long long *)ptr += val;
892 break;
893 case R_X86_64_32:
894 case R_X86_64_32S:
895 if (s1->output_type == TCC_OUTPUT_DLL) {
896 /* XXX: this logic may depend on TCC's codegen
897 now TCC uses R_X86_64_32 even for a 64bit pointer */
898 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
899 qrel->r_addend = *(int *)ptr + val;
900 qrel++;
901 }
902 *(int *)ptr += val;
903 break;
904
905 case R_X86_64_PC32:
906 if (s1->output_type == TCC_OUTPUT_DLL) {
907 /* DLL relocation */
908 esym_index = s1->symtab_to_dynsym[sym_index];
909 if (esym_index) {
910 qrel->r_offset = rel->r_offset;
911 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_PC32);
912 qrel->r_addend = *(int *)ptr;
913 qrel++;
914 break;
915 }
916 }
917 goto plt32pc32;
918
919 case R_X86_64_PLT32:
920 /* We've put the PLT slot offset into r_addend when generating
921 it, and that's what we must use as relocation value (adjusted
922 by section offset of course). */
923 val = s1->plt->sh_addr + rel->r_addend;
924 /* fallthrough. */
925
926 plt32pc32:
927 {
928 long long diff;
929 diff = (long long)val - addr;
930 if (diff <= -2147483647 || diff > 2147483647) {
931 tcc_error("internal error: relocation failed");
932 }
933 *(int *)ptr += diff;
934 }
935 break;
936 case R_X86_64_GLOB_DAT:
937 case R_X86_64_JUMP_SLOT:
938 /* They don't need addend */
939 *(addr_t *)ptr = val - rel->r_addend;
940 break;
941 case R_X86_64_GOTPCREL:
942 *(int *)ptr += (s1->got->sh_addr - addr +
943 s1->sym_attrs[sym_index].got_offset - 4);
944 break;
945 case R_X86_64_GOTTPOFF:
946 *(int *)ptr += val - s1->got->sh_addr;
947 break;
948 case R_X86_64_GOT32:
949 /* we load the got offset */
950 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
951 break;
952 #else
953 #error unsupported processor
954 #endif
955 }
956 }
957 /* if the relocation is allocated, we change its symbol table */
958 if (sr->sh_flags & SHF_ALLOC)
959 sr->link = s1->dynsym;
960 }
961
962 /* relocate relocation table in 'sr' */
963 static void relocate_rel(TCCState *s1, Section *sr)
964 {
965 Section *s;
966 ElfW_Rel *rel;
967
968 s = s1->sections[sr->sh_info];
969 for_each_elem(sr, 0, rel, ElfW_Rel)
970 rel->r_offset += s->sh_addr;
971 }
972
973 /* count the number of dynamic relocations so that we can reserve
974 their space */
975 static int prepare_dynamic_rel(TCCState *s1, Section *sr)
976 {
977 ElfW_Rel *rel;
978 int sym_index, esym_index, type, count;
979
980 count = 0;
981 for_each_elem(sr, 0, rel, ElfW_Rel) {
982 sym_index = ELFW(R_SYM)(rel->r_info);
983 type = ELFW(R_TYPE)(rel->r_info);
984 switch(type) {
985 #if defined(TCC_TARGET_I386)
986 case R_386_32:
987 #elif defined(TCC_TARGET_X86_64)
988 case R_X86_64_32:
989 case R_X86_64_32S:
990 case R_X86_64_64:
991 #endif
992 count++;
993 break;
994 #if defined(TCC_TARGET_I386)
995 case R_386_PC32:
996 #elif defined(TCC_TARGET_X86_64)
997 case R_X86_64_PC32:
998 #endif
999 esym_index = s1->symtab_to_dynsym[sym_index];
1000 if (esym_index)
1001 count++;
1002 break;
1003 default:
1004 break;
1005 }
1006 }
1007 if (count) {
1008 /* allocate the section */
1009 sr->sh_flags |= SHF_ALLOC;
1010 sr->sh_size = count * sizeof(ElfW_Rel);
1011 }
1012 return count;
1013 }
1014
1015 static struct sym_attr *alloc_sym_attr(TCCState *s1, int index)
1016 {
1017 int n;
1018 struct sym_attr *tab;
1019
1020 if (index >= s1->nb_sym_attrs) {
1021 /* find immediately bigger power of 2 and reallocate array */
1022 n = 1;
1023 while (index >= n)
1024 n *= 2;
1025 tab = tcc_realloc(s1->sym_attrs, n * sizeof(*s1->sym_attrs));
1026 s1->sym_attrs = tab;
1027 memset(s1->sym_attrs + s1->nb_sym_attrs, 0,
1028 (n - s1->nb_sym_attrs) * sizeof(*s1->sym_attrs));
1029 s1->nb_sym_attrs = n;
1030 }
1031 return &s1->sym_attrs[index];
1032 }
1033
1034 /* XXX: suppress that */
1035 static void put32(unsigned char *p, uint32_t val)
1036 {
1037 p[0] = val;
1038 p[1] = val >> 8;
1039 p[2] = val >> 16;
1040 p[3] = val >> 24;
1041 }
1042
1043 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_ARM) || \
1044 defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
1045 static uint32_t get32(unsigned char *p)
1046 {
1047 return p[0] | (p[1] << 8) | (p[2] << 16) | (p[3] << 24);
1048 }
1049 #endif
1050
1051 static void build_got(TCCState *s1)
1052 {
1053 unsigned char *ptr;
1054
1055 /* if no got, then create it */
1056 s1->got = new_section(s1, ".got", SHT_PROGBITS, SHF_ALLOC | SHF_WRITE);
1057 s1->got->sh_entsize = 4;
1058 add_elf_sym(symtab_section, 0, 4, ELFW(ST_INFO)(STB_GLOBAL, STT_OBJECT),
1059 0, s1->got->sh_num, "_GLOBAL_OFFSET_TABLE_");
1060 ptr = section_ptr_add(s1->got, 3 * PTR_SIZE);
1061 #if PTR_SIZE == 4
1062 /* keep space for _DYNAMIC pointer, if present */
1063 put32(ptr, 0);
1064 /* two dummy got entries */
1065 put32(ptr + 4, 0);
1066 put32(ptr + 8, 0);
1067 #else
1068 /* keep space for _DYNAMIC pointer, if present */
1069 put32(ptr, 0);
1070 put32(ptr + 4, 0);
1071 /* two dummy got entries */
1072 put32(ptr + 8, 0);
1073 put32(ptr + 12, 0);
1074 put32(ptr + 16, 0);
1075 put32(ptr + 20, 0);
1076 #endif
1077 }
1078
1079 /* put a got or plt entry corresponding to a symbol in symtab_section. 'size'
1080 and 'info' can be modifed if more precise info comes from the DLL.
1081 Returns offset of GOT or PLT slot. */
1082 static unsigned long put_got_entry(TCCState *s1,
1083 int reloc_type, unsigned long size, int info,
1084 int sym_index)
1085 {
1086 int index, need_plt_entry;
1087 const char *name;
1088 ElfW(Sym) *sym;
1089 unsigned long offset;
1090 int *ptr;
1091 struct sym_attr *symattr;
1092
1093 if (!s1->got)
1094 build_got(s1);
1095
1096 need_plt_entry =
1097 #ifdef TCC_TARGET_X86_64
1098 (reloc_type == R_X86_64_JUMP_SLOT);
1099 #elif defined(TCC_TARGET_I386)
1100 (reloc_type == R_386_JMP_SLOT);
1101 #elif defined(TCC_TARGET_ARM)
1102 (reloc_type == R_ARM_JUMP_SLOT);
1103 #elif defined(TCC_TARGET_ARM64)
1104 (reloc_type == R_AARCH64_JUMP_SLOT);
1105 #else
1106 0;
1107 #endif
1108
1109 if (need_plt_entry && !s1->plt) {
1110 /* add PLT */
1111 s1->plt = new_section(s1, ".plt", SHT_PROGBITS,
1112 SHF_ALLOC | SHF_EXECINSTR);
1113 s1->plt->sh_entsize = 4;
1114 }
1115
1116 /* If a got/plt entry already exists for that symbol, no need to add one */
1117 if (sym_index < s1->nb_sym_attrs) {
1118 if (need_plt_entry && s1->sym_attrs[sym_index].plt_offset)
1119 return s1->sym_attrs[sym_index].plt_offset;
1120 else if (!need_plt_entry && s1->sym_attrs[sym_index].got_offset)
1121 return s1->sym_attrs[sym_index].got_offset;
1122 }
1123
1124 symattr = alloc_sym_attr(s1, sym_index);
1125
1126 /* Only store the GOT offset if it's not generated for the PLT entry. */
1127 if (!need_plt_entry)
1128 symattr->got_offset = s1->got->data_offset;
1129
1130 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1131 name = (char *) symtab_section->link->data + sym->st_name;
1132 offset = sym->st_value;
1133 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_X86_64)
1134 if (need_plt_entry) {
1135 Section *plt;
1136 uint8_t *p;
1137 int modrm;
1138 unsigned long relofs;
1139
1140 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1141 modrm = 0x25;
1142 #else
1143 /* if we build a DLL, we add a %ebx offset */
1144 if (s1->output_type == TCC_OUTPUT_DLL)
1145 modrm = 0xa3;
1146 else
1147 modrm = 0x25;
1148 #endif
1149
1150 /* add a PLT entry */
1151 plt = s1->plt;
1152 if (plt->data_offset == 0) {
1153 /* first plt entry */
1154 p = section_ptr_add(plt, 16);
1155 p[0] = 0xff; /* pushl got + PTR_SIZE */
1156 p[1] = modrm + 0x10;
1157 put32(p + 2, PTR_SIZE);
1158 p[6] = 0xff; /* jmp *(got + PTR_SIZE * 2) */
1159 p[7] = modrm;
1160 put32(p + 8, PTR_SIZE * 2);
1161 }
1162
1163 /* The PLT slot refers to the relocation entry it needs
1164 via offset. The reloc entry is created below, so its
1165 offset is the current data_offset. */
1166 relofs = s1->got->reloc ? s1->got->reloc->data_offset : 0;
1167 symattr->plt_offset = plt->data_offset;
1168 p = section_ptr_add(plt, 16);
1169 p[0] = 0xff; /* jmp *(got + x) */
1170 p[1] = modrm;
1171 put32(p + 2, s1->got->data_offset);
1172 p[6] = 0x68; /* push $xxx */
1173 #ifdef TCC_TARGET_X86_64
1174 /* On x86-64, the relocation is referred to by _index_. */
1175 put32(p + 7, relofs / sizeof (ElfW_Rel));
1176 #else
1177 put32(p + 7, relofs);
1178 #endif
1179 p[11] = 0xe9; /* jmp plt_start */
1180 put32(p + 12, -(plt->data_offset));
1181
1182 /* If this was an UNDEF symbol set the offset in the
1183 dynsymtab to the PLT slot, so that PC32 relocs to it
1184 can be resolved. */
1185 if (sym->st_shndx == SHN_UNDEF)
1186 offset = plt->data_offset - 16;
1187 }
1188 #elif defined(TCC_TARGET_ARM)
1189 if (need_plt_entry) {
1190 Section *plt;
1191 uint8_t *p;
1192
1193 /* if we build a DLL, we add a %ebx offset */
1194 if (s1->output_type == TCC_OUTPUT_DLL)
1195 tcc_error("DLLs unimplemented!");
1196
1197 /* add a PLT entry */
1198 plt = s1->plt;
1199 if (plt->data_offset == 0) {
1200 /* first plt entry */
1201 p = section_ptr_add(plt, 16);
1202 put32(p, 0xe52de004); /* push {lr} */
1203 put32(p+4, 0xe59fe010); /* ldr lr, [pc, #16] */
1204 put32(p+8, 0xe08fe00e); /* add lr, pc, lr */
1205 put32(p+12, 0xe5bef008); /* ldr pc, [lr, #8]! */
1206 }
1207
1208 symattr->plt_offset = plt->data_offset;
1209 if (symattr->plt_thumb_stub) {
1210 p = section_ptr_add(plt, 20);
1211 put32(p, 0x4778); /* bx pc */
1212 put32(p+2, 0x46c0); /* nop */
1213 p += 4;
1214 } else
1215 p = section_ptr_add(plt, 16);
1216 put32(p, 0xe59fc004); /* ldr ip, [pc, #4] ; GOT entry offset */
1217 put32(p+4, 0xe08fc00c); /* add ip, pc, ip ; addr of GOT entry */
1218 put32(p+8, 0xe59cf000); /* ldr pc, [ip] ; jump to GOT entry */
1219 put32(p+12, s1->got->data_offset); /* GOT entry off once patched */
1220
1221 /* the symbol is modified so that it will be relocated to
1222 the PLT */
1223 if (sym->st_shndx == SHN_UNDEF)
1224 offset = plt->data_offset - 16;
1225 }
1226 #elif defined(TCC_TARGET_ARM64)
1227 if (need_plt_entry) {
1228 Section *plt;
1229 uint8_t *p;
1230
1231 if (s1->output_type == TCC_OUTPUT_DLL)
1232 tcc_error("DLLs unimplemented!");
1233
1234 plt = s1->plt;
1235 if (plt->data_offset == 0)
1236 section_ptr_add(plt, 32);
1237 symattr->plt_offset = plt->data_offset;
1238 p = section_ptr_add(plt, 16);
1239 put32(p, s1->got->data_offset);
1240 put32(p + 4, (uint64_t)s1->got->data_offset >> 32);
1241
1242 if (sym->st_shndx == SHN_UNDEF)
1243 offset = plt->data_offset - 16;
1244 }
1245 #elif defined(TCC_TARGET_C67)
1246 if (s1->dynsym) {
1247 tcc_error("C67 got not implemented");
1248 }
1249 #else
1250 #error unsupported CPU
1251 #endif
1252 if (s1->dynsym) {
1253 /* XXX This might generate multiple syms for name. */
1254 index = put_elf_sym(s1->dynsym, offset,
1255 size, info, 0, sym->st_shndx, name);
1256 /* Create the relocation (it's against the GOT for PLT
1257 and GOT relocs). */
1258 put_elf_reloc(s1->dynsym, s1->got,
1259 s1->got->data_offset,
1260 reloc_type, index);
1261 } else {
1262 /* Without .dynsym (i.e. static link or memory output) we
1263 still need relocs against the generated got, so as to fill
1264 the entries with the symbol values (determined later). */
1265 put_elf_reloc(symtab_section, s1->got,
1266 s1->got->data_offset,
1267 reloc_type, sym_index);
1268 }
1269 /* And now create the GOT slot itself. */
1270 ptr = section_ptr_add(s1->got, PTR_SIZE);
1271 *ptr = 0;
1272 if (need_plt_entry)
1273 return symattr->plt_offset;
1274 else
1275 return symattr->got_offset;
1276 }
1277
1278 /* build GOT and PLT entries */
1279 ST_FUNC void build_got_entries(TCCState *s1)
1280 {
1281 Section *s;
1282 ElfW_Rel *rel;
1283 ElfW(Sym) *sym;
1284 int i, type, reloc_type, sym_index;
1285
1286 for(i = 1; i < s1->nb_sections; i++) {
1287 s = s1->sections[i];
1288 if (s->sh_type != SHT_RELX)
1289 continue;
1290 /* no need to handle got relocations */
1291 if (s->link != symtab_section)
1292 continue;
1293 for_each_elem(s, 0, rel, ElfW_Rel) {
1294 type = ELFW(R_TYPE)(rel->r_info);
1295 switch(type) {
1296 #if defined(TCC_TARGET_I386)
1297 case R_386_GOT32:
1298 case R_386_GOTOFF:
1299 case R_386_GOTPC:
1300 case R_386_PLT32:
1301 if (!s1->got)
1302 build_got(s1);
1303 if (type == R_386_GOT32 || type == R_386_PLT32) {
1304 sym_index = ELFW(R_SYM)(rel->r_info);
1305 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1306 /* look at the symbol got offset. If none, then add one */
1307 if (type == R_386_GOT32)
1308 reloc_type = R_386_GLOB_DAT;
1309 else
1310 reloc_type = R_386_JMP_SLOT;
1311 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1312 sym_index);
1313 }
1314 break;
1315 #elif defined(TCC_TARGET_ARM)
1316 case R_ARM_PC24:
1317 case R_ARM_CALL:
1318 case R_ARM_JUMP24:
1319 case R_ARM_GOT32:
1320 case R_ARM_GOTOFF:
1321 case R_ARM_GOTPC:
1322 case R_ARM_PLT32:
1323 if (!s1->got)
1324 build_got(s1);
1325 sym_index = ELFW(R_SYM)(rel->r_info);
1326 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1327 if (type != R_ARM_GOTOFF && type != R_ARM_GOTPC
1328 && sym->st_shndx == SHN_UNDEF) {
1329 unsigned long ofs;
1330 /* look at the symbol got offset. If none, then add one */
1331 if (type == R_ARM_GOT32)
1332 reloc_type = R_ARM_GLOB_DAT;
1333 else
1334 reloc_type = R_ARM_JUMP_SLOT;
1335 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1336 sym->st_info, sym_index);
1337 #ifdef DEBUG_RELOC
1338 printf ("maybegot: %s, %d, %d --> ofs=0x%x\n",
1339 (char *) symtab_section->link->data + sym->st_name,
1340 type, sym->st_shndx, ofs);
1341 #endif
1342 if (type != R_ARM_GOT32) {
1343 addr_t *ptr = (addr_t*)(s1->sections[s->sh_info]->data
1344 + rel->r_offset);
1345 /* x must be signed! */
1346 int x = *ptr & 0xffffff;
1347 x = (x << 8) >> 8;
1348 x <<= 2;
1349 x += ofs;
1350 x >>= 2;
1351 #ifdef DEBUG_RELOC
1352 printf ("insn=0x%x --> 0x%x (x==0x%x)\n", *ptr,
1353 (*ptr & 0xff000000) | x, x);
1354 #endif
1355 *ptr = (*ptr & 0xff000000) | x;
1356 }
1357 }
1358 break;
1359 case R_ARM_THM_JUMP24:
1360 sym_index = ELFW(R_SYM)(rel->r_info);
1361 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1362 /* We are relocating a jump from thumb code to arm code */
1363 if (sym->st_shndx != SHN_UNDEF && !(sym->st_value & 1)) {
1364 int index;
1365 uint8_t *p;
1366 char *name, buf[1024];
1367 Section *text_section;
1368
1369 name = (char *) symtab_section->link->data + sym->st_name;
1370 text_section = s1->sections[sym->st_shndx];
1371 /* Modify reloc to target a thumb stub to switch to ARM */
1372 snprintf(buf, sizeof(buf), "%s_from_thumb", name);
1373 index = put_elf_sym(symtab_section,
1374 text_section->data_offset + 1,
1375 sym->st_size, sym->st_info, 0,
1376 sym->st_shndx, buf);
1377 rel->r_info = ELFW(R_INFO)(index, type);
1378 /* Create a thumb stub fonction to switch to ARM mode */
1379 put_elf_reloc(symtab_section, text_section,
1380 text_section->data_offset + 4, R_ARM_JUMP24,
1381 sym_index);
1382 p = section_ptr_add(text_section, 8);
1383 put32(p, 0x4778); /* bx pc */
1384 put32(p+2, 0x46c0); /* nop */
1385 put32(p+4, 0xeafffffe); /* b $sym */
1386 }
1387 #elif defined(TCC_TARGET_ARM64)
1388 //xx Other cases may be required here:
1389 case R_AARCH64_ADR_GOT_PAGE:
1390 case R_AARCH64_LD64_GOT_LO12_NC:
1391 if (!s1->got)
1392 build_got(s1);
1393 sym_index = ELFW(R_SYM)(rel->r_info);
1394 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1395 reloc_type = R_AARCH64_GLOB_DAT;
1396 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1397 sym_index);
1398 break;
1399
1400 case R_AARCH64_JUMP26:
1401 case R_AARCH64_CALL26:
1402 if (!s1->got)
1403 build_got(s1);
1404 sym_index = ELFW(R_SYM)(rel->r_info);
1405 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1406 if (sym->st_shndx == SHN_UNDEF) {
1407 unsigned long ofs;
1408 reloc_type = R_AARCH64_JUMP_SLOT;
1409 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1410 sym->st_info, sym_index);
1411 /* We store the place of the generated PLT slot
1412 in our addend. */
1413 rel->r_addend += ofs;
1414 }
1415 break;
1416 #elif defined(TCC_TARGET_C67)
1417 case R_C60_GOT32:
1418 case R_C60_GOTOFF:
1419 case R_C60_GOTPC:
1420 case R_C60_PLT32:
1421 if (!s1->got)
1422 build_got(s1);
1423 if (type == R_C60_GOT32 || type == R_C60_PLT32) {
1424 sym_index = ELFW(R_SYM)(rel->r_info);
1425 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1426 /* look at the symbol got offset. If none, then add one */
1427 if (type == R_C60_GOT32)
1428 reloc_type = R_C60_GLOB_DAT;
1429 else
1430 reloc_type = R_C60_JMP_SLOT;
1431 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1432 sym_index);
1433 }
1434 break;
1435 #elif defined(TCC_TARGET_X86_64)
1436 case R_X86_64_GOT32:
1437 case R_X86_64_GOTTPOFF:
1438 case R_X86_64_GOTPCREL:
1439 case R_X86_64_PLT32:
1440 sym_index = ELFW(R_SYM)(rel->r_info);
1441 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1442 if (type == R_X86_64_PLT32 &&
1443 ELFW(ST_VISIBILITY)(sym->st_other) != STV_DEFAULT)
1444 {
1445 rel->r_info = ELFW(R_INFO)(sym_index, R_X86_64_PC32);
1446 break;
1447 }
1448
1449 if (!s1->got)
1450 build_got(s1);
1451 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL ||
1452 type == R_X86_64_PLT32) {
1453 unsigned long ofs;
1454 /* look at the symbol got offset. If none, then add one */
1455 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL)
1456 reloc_type = R_X86_64_GLOB_DAT;
1457 else
1458 reloc_type = R_X86_64_JUMP_SLOT;
1459 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1460 sym->st_info, sym_index);
1461 if (type == R_X86_64_PLT32)
1462 /* We store the place of the generated PLT slot
1463 in our addend. */
1464 rel->r_addend += ofs;
1465 }
1466 break;
1467 #else
1468 #error unsupported CPU
1469 #endif
1470 default:
1471 break;
1472 }
1473 }
1474 }
1475 }
1476
1477 ST_FUNC Section *new_symtab(TCCState *s1,
1478 const char *symtab_name, int sh_type, int sh_flags,
1479 const char *strtab_name,
1480 const char *hash_name, int hash_sh_flags)
1481 {
1482 Section *symtab, *strtab, *hash;
1483 int *ptr, nb_buckets;
1484
1485 symtab = new_section(s1, symtab_name, sh_type, sh_flags);
1486 symtab->sh_entsize = sizeof(ElfW(Sym));
1487 strtab = new_section(s1, strtab_name, SHT_STRTAB, sh_flags);
1488 put_elf_str(strtab, "");
1489 symtab->link = strtab;
1490 put_elf_sym(symtab, 0, 0, 0, 0, 0, NULL);
1491
1492 nb_buckets = 1;
1493
1494 hash = new_section(s1, hash_name, SHT_HASH, hash_sh_flags);
1495 hash->sh_entsize = sizeof(int);
1496 symtab->hash = hash;
1497 hash->link = symtab;
1498
1499 ptr = section_ptr_add(hash, (2 + nb_buckets + 1) * sizeof(int));
1500 ptr[0] = nb_buckets;
1501 ptr[1] = 1;
1502 memset(ptr + 2, 0, (nb_buckets + 1) * sizeof(int));
1503 return symtab;
1504 }
1505
1506 /* put dynamic tag */
1507 static void put_dt(Section *dynamic, int dt, addr_t val)
1508 {
1509 ElfW(Dyn) *dyn;
1510 dyn = section_ptr_add(dynamic, sizeof(ElfW(Dyn)));
1511 dyn->d_tag = dt;
1512 dyn->d_un.d_val = val;
1513 }
1514
1515 static void add_init_array_defines(TCCState *s1, const char *section_name)
1516 {
1517 Section *s;
1518 long end_offset;
1519 char sym_start[1024];
1520 char sym_end[1024];
1521
1522 snprintf(sym_start, sizeof(sym_start), "__%s_start", section_name + 1);
1523 snprintf(sym_end, sizeof(sym_end), "__%s_end", section_name + 1);
1524
1525 s = find_section(s1, section_name);
1526 if (!s) {
1527 end_offset = 0;
1528 s = data_section;
1529 } else {
1530 end_offset = s->data_offset;
1531 }
1532
1533 add_elf_sym(symtab_section,
1534 0, 0,
1535 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1536 s->sh_num, sym_start);
1537 add_elf_sym(symtab_section,
1538 end_offset, 0,
1539 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1540 s->sh_num, sym_end);
1541 }
1542
1543 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1544 {
1545 #ifdef CONFIG_TCC_BCHECK
1546 unsigned long *ptr;
1547 Section *init_section;
1548 unsigned char *pinit;
1549 int sym_index;
1550
1551 if (0 == s1->do_bounds_check)
1552 return;
1553
1554 /* XXX: add an object file to do that */
1555 ptr = section_ptr_add(bounds_section, sizeof(unsigned long));
1556 *ptr = 0;
1557 add_elf_sym(symtab_section, 0, 0,
1558 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1559 bounds_section->sh_num, "__bounds_start");
1560 #ifdef TCC_TARGET_I386
1561 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1562 /* add 'call __bound_init()' in .init section */
1563 init_section = find_section(s1, ".init");
1564 pinit = section_ptr_add(init_section, 5);
1565 pinit[0] = 0xe8;
1566 put32(pinit + 1, -4);
1567 sym_index = find_elf_sym(symtab_section, "__bound_init");
1568 put_elf_reloc(symtab_section, init_section,
1569 init_section->data_offset - 4, R_386_PC32, sym_index);
1570 }
1571 #endif
1572 #endif
1573 }
1574
1575 static inline int tcc_add_support(TCCState *s1, const char *filename)
1576 {
1577 char buf[1024];
1578 snprintf(buf, sizeof(buf), "%s/%s", s1->tcc_lib_path, filename);
1579 return tcc_add_file(s1, buf);
1580 }
1581
1582 /* add tcc runtime libraries */
1583 ST_FUNC void tcc_add_runtime(TCCState *s1)
1584 {
1585 /* add libc */
1586 if (!s1->nostdlib) {
1587 tcc_add_library(s1, "c");
1588 #ifdef CONFIG_USE_LIBGCC
1589 if (!s1->static_link) {
1590 tcc_add_file(s1, TCC_LIBGCC);
1591 tcc_add_support(s1, "libtcc1.a");
1592 } else
1593 tcc_add_support(s1, "libtcc1.a");
1594 #else
1595 tcc_add_support(s1, "libtcc1.a");
1596 #endif
1597 }
1598
1599 /* tcc_add_bcheck tries to relocate a call to __bound_init in _init so
1600 libtcc1.a must be loaded before for __bound_init to be defined and
1601 crtn.o must be loaded after to not finalize _init too early. */
1602 tcc_add_bcheck(s1);
1603
1604 if (!s1->nostdlib) {
1605 /* add crt end if not memory output */
1606 if (s1->output_type != TCC_OUTPUT_MEMORY)
1607 tcc_add_crt(s1, "crtn.o");
1608 }
1609 }
1610
1611 /* add various standard linker symbols (must be done after the
1612 sections are filled (for example after allocating common
1613 symbols)) */
1614 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1615 {
1616 char buf[1024];
1617 int i;
1618 Section *s;
1619
1620 add_elf_sym(symtab_section,
1621 text_section->data_offset, 0,
1622 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1623 text_section->sh_num, "_etext");
1624 add_elf_sym(symtab_section,
1625 data_section->data_offset, 0,
1626 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1627 data_section->sh_num, "_edata");
1628 add_elf_sym(symtab_section,
1629 bss_section->data_offset, 0,
1630 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1631 bss_section->sh_num, "_end");
1632 /* horrible new standard ldscript defines */
1633 add_init_array_defines(s1, ".preinit_array");
1634 add_init_array_defines(s1, ".init_array");
1635 add_init_array_defines(s1, ".fini_array");
1636
1637 /* add start and stop symbols for sections whose name can be
1638 expressed in C */
1639 for(i = 1; i < s1->nb_sections; i++) {
1640 s = s1->sections[i];
1641 if (s->sh_type == SHT_PROGBITS &&
1642 (s->sh_flags & SHF_ALLOC)) {
1643 const char *p;
1644 int ch;
1645
1646 /* check if section name can be expressed in C */
1647 p = s->name;
1648 for(;;) {
1649 ch = *p;
1650 if (!ch)
1651 break;
1652 if (!isid(ch) && !isnum(ch))
1653 goto next_sec;
1654 p++;
1655 }
1656 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1657 add_elf_sym(symtab_section,
1658 0, 0,
1659 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1660 s->sh_num, buf);
1661 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1662 add_elf_sym(symtab_section,
1663 s->data_offset, 0,
1664 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1665 s->sh_num, buf);
1666 }
1667 next_sec: ;
1668 }
1669 }
1670
1671 static void tcc_output_binary(TCCState *s1, FILE *f,
1672 const int *sec_order)
1673 {
1674 Section *s;
1675 int i, offset, size;
1676
1677 offset = 0;
1678 for(i=1;i<s1->nb_sections;i++) {
1679 s = s1->sections[sec_order[i]];
1680 if (s->sh_type != SHT_NOBITS &&
1681 (s->sh_flags & SHF_ALLOC)) {
1682 while (offset < s->sh_offset) {
1683 fputc(0, f);
1684 offset++;
1685 }
1686 size = s->sh_size;
1687 fwrite(s->data, 1, size, f);
1688 offset += size;
1689 }
1690 }
1691 }
1692
1693 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1694 #define HAVE_PHDR 1
1695 #define EXTRA_RELITEMS 14
1696
1697 /* move the relocation value from .dynsym to .got */
1698 void patch_dynsym_undef(TCCState *s1, Section *s)
1699 {
1700 uint32_t *gotd = (void *)s1->got->data;
1701 ElfW(Sym) *sym;
1702
1703 gotd += 3; /* dummy entries in .got */
1704 /* relocate symbols in .dynsym */
1705 for_each_elem(s, 1, sym, ElfW(Sym)) {
1706 if (sym->st_shndx == SHN_UNDEF) {
1707 *gotd++ = sym->st_value + 6; /* XXX 6 is magic ? */
1708 sym->st_value = 0;
1709 }
1710 }
1711 }
1712 #else
1713 #define HAVE_PHDR 1
1714 #define EXTRA_RELITEMS 9
1715
1716 /* zero plt offsets of weak symbols in .dynsym */
1717 void patch_dynsym_undef(TCCState *s1, Section *s)
1718 {
1719 ElfW(Sym) *sym;
1720
1721 for_each_elem(s, 1, sym, ElfW(Sym))
1722 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1723 sym->st_value = 0;
1724 }
1725 #endif
1726
1727 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1728 {
1729 int sym_index = ELFW(R_SYM) (rel->r_info);
1730 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1731 unsigned long offset;
1732
1733 if (sym_index >= s1->nb_sym_attrs)
1734 return;
1735 offset = s1->sym_attrs[sym_index].got_offset;
1736 section_reserve(s1->got, offset + PTR_SIZE);
1737 #ifdef TCC_TARGET_X86_64
1738 /* only works for x86-64 */
1739 put32(s1->got->data + offset + 4, sym->st_value >> 32);
1740 #endif
1741 put32(s1->got->data + offset, sym->st_value & 0xffffffff);
1742 }
1743
1744 /* Perform relocation to GOT or PLT entries */
1745 ST_FUNC void fill_got(TCCState *s1)
1746 {
1747 Section *s;
1748 ElfW_Rel *rel;
1749 int i;
1750
1751 for(i = 1; i < s1->nb_sections; i++) {
1752 s = s1->sections[i];
1753 if (s->sh_type != SHT_RELX)
1754 continue;
1755 /* no need to handle got relocations */
1756 if (s->link != symtab_section)
1757 continue;
1758 for_each_elem(s, 0, rel, ElfW_Rel) {
1759 switch (ELFW(R_TYPE) (rel->r_info)) {
1760 case R_X86_64_GOT32:
1761 case R_X86_64_GOTPCREL:
1762 case R_X86_64_PLT32:
1763 fill_got_entry(s1, rel);
1764 break;
1765 }
1766 }
1767 }
1768 }
1769
1770 /* Bind symbols of executable: resolve undefined symbols from exported symbols
1771 in shared libraries and export non local defined symbols to shared libraries
1772 if -rdynamic switch was given on command line */
1773 static void bind_exe_dynsyms(TCCState *s1)
1774 {
1775 const char *name;
1776 int sym_index, index;
1777 ElfW(Sym) *sym, *esym;
1778 int type;
1779
1780 /* Resolve undefined symbols from dynamic symbols. When there is a match:
1781 - if STT_FUNC or STT_GNU_IFUNC symbol -> add it in PLT
1782 - if STT_OBJECT symbol -> add it in .bss section with suitable reloc */
1783 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1784 if (sym->st_shndx == SHN_UNDEF) {
1785 name = (char *) symtab_section->link->data + sym->st_name;
1786 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1787 if (sym_index) {
1788 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1789 type = ELFW(ST_TYPE)(esym->st_info);
1790 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1791 /* Indirect functions shall have STT_FUNC type in executable
1792 * dynsym section. Indeed, a dlsym call following a lazy
1793 * resolution would pick the symbol value from the
1794 * executable dynsym entry which would contain the address
1795 * of the function wanted by the caller of dlsym instead of
1796 * the address of the function that would return that
1797 * address */
1798 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1799 ELFW(ST_INFO)(STB_GLOBAL,STT_FUNC),
1800 sym - (ElfW(Sym) *)symtab_section->data);
1801 } else if (type == STT_OBJECT) {
1802 unsigned long offset;
1803 ElfW(Sym) *dynsym;
1804 offset = bss_section->data_offset;
1805 /* XXX: which alignment ? */
1806 offset = (offset + 16 - 1) & -16;
1807 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1808 esym->st_info, 0, bss_section->sh_num,
1809 name);
1810 /* Ensure R_COPY works for weak symbol aliases */
1811 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1812 for_each_elem(s1->dynsymtab_section, 1, dynsym, ElfW(Sym)) {
1813 if ((dynsym->st_value == esym->st_value)
1814 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1815 char *dynname = (char *) s1->dynsymtab_section->link->data
1816 + dynsym->st_name;
1817 put_elf_sym(s1->dynsym, offset, dynsym->st_size,
1818 dynsym->st_info, 0,
1819 bss_section->sh_num, dynname);
1820 break;
1821 }
1822 }
1823 }
1824 put_elf_reloc(s1->dynsym, bss_section,
1825 offset, R_COPY, index);
1826 offset += esym->st_size;
1827 bss_section->data_offset = offset;
1828 }
1829 } else {
1830 /* STB_WEAK undefined symbols are accepted */
1831 /* XXX: _fp_hw seems to be part of the ABI, so we ignore it */
1832 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1833 !strcmp(name, "_fp_hw")) {
1834 } else {
1835 tcc_error_noabort("undefined symbol '%s'", name);
1836 }
1837 }
1838 } else if (s1->rdynamic && ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1839 /* if -rdynamic option, then export all non local symbols */
1840 name = (char *) symtab_section->link->data + sym->st_name;
1841 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size, sym->st_info,
1842 0, sym->st_shndx, name);
1843 }
1844 }
1845 }
1846
1847 /* Bind symbols of libraries: export non local symbols of executable that
1848 resolve undefined symbols of shared libraries */
1849 static void bind_libs_dynsyms(TCCState *s1)
1850 {
1851 const char *name;
1852 int sym_index;
1853 ElfW(Sym) *sym, *esym;
1854
1855 /* now look at unresolved dynamic symbols and export
1856 corresponding symbol */
1857 for_each_elem(s1->dynsymtab_section, 1, esym, ElfW(Sym)) {
1858 if (esym->st_shndx == SHN_UNDEF) {
1859 name = (char *) s1->dynsymtab_section->link->data + esym->st_name;
1860 sym_index = find_elf_sym(symtab_section, name);
1861 if (sym_index) {
1862 /* XXX: avoid adding a symbol if already present because of
1863 -rdynamic ? */
1864 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1865 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1866 sym->st_info, 0, sym->st_shndx, name);
1867 } else {
1868 /* weak symbols can stay undefined */
1869 if (ELFW(ST_BIND)(esym->st_info) != STB_WEAK)
1870 tcc_warning("undefined dynamic symbol '%s'", name);
1871 }
1872 }
1873 }
1874 }
1875
1876 /* Export all non local symbols (for shared libraries) */
1877 static void export_global_syms(TCCState *s1)
1878 {
1879 int nb_syms, dynindex, index;
1880 const char *name;
1881 ElfW(Sym) *sym;
1882
1883 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1884 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1885 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1886 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1887 name = (char *) symtab_section->link->data + sym->st_name;
1888 dynindex = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1889 sym->st_info, 0, sym->st_shndx, name);
1890 index = sym - (ElfW(Sym) *) symtab_section->data;
1891 s1->symtab_to_dynsym[index] = dynindex;
1892 }
1893 }
1894 }
1895
1896 /* relocate the PLT: compute addresses and offsets in the PLT now that final
1897 address for PLT and GOT are known (see fill_program_header) */
1898 ST_FUNC void relocate_plt(TCCState *s1)
1899 {
1900 uint8_t *p, *p_end;
1901
1902 if (!s1->plt)
1903 return;
1904
1905 p = s1->plt->data;
1906 p_end = p + s1->plt->data_offset;
1907 if (p < p_end) {
1908 #if defined(TCC_TARGET_I386)
1909 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1910 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
1911 p += 16;
1912 while (p < p_end) {
1913 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1914 p += 16;
1915 }
1916 #elif defined(TCC_TARGET_X86_64)
1917 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
1918 put32(p + 2, get32(p + 2) + x);
1919 put32(p + 8, get32(p + 8) + x - 6);
1920 p += 16;
1921 while (p < p_end) {
1922 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
1923 p += 16;
1924 }
1925 #elif defined(TCC_TARGET_ARM)
1926 int x;
1927 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
1928 p += 16;
1929 while (p < p_end) {
1930 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
1931 p += 4;
1932 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
1933 p += 16;
1934 }
1935 #elif defined(TCC_TARGET_ARM64)
1936 uint64_t plt = s1->plt->sh_addr;
1937 uint64_t got = s1->got->sh_addr;
1938 uint64_t off = (got >> 12) - (plt >> 12);
1939 if ((off + ((uint32_t)1 << 20)) >> 21)
1940 tcc_error("Failed relocating PLT (off=0x%lx, got=0x%lx, plt=0x%lx)", off, got, plt);
1941 put32(p, 0xa9bf7bf0); // stp x16,x30,[sp,#-16]!
1942 put32(p + 4, (0x90000010 | // adrp x16,...
1943 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1944 put32(p + 8, (0xf9400211 | // ldr x17,[x16,#...]
1945 (got & 0xff8) << 7));
1946 put32(p + 12, (0x91000210 | // add x16,x16,#...
1947 (got & 0xfff) << 10));
1948 put32(p + 16, 0xd61f0220); // br x17
1949 put32(p + 20, 0xd503201f); // nop
1950 put32(p + 24, 0xd503201f); // nop
1951 put32(p + 28, 0xd503201f); // nop
1952 p += 32;
1953 while (p < p_end) {
1954 uint64_t pc = plt + (p - s1->plt->data);
1955 uint64_t addr = got +
1956 (get32(p) | (uint64_t)get32(p + 4) << 32);
1957 uint32_t off = (addr >> 12) - (pc >> 12);
1958 if ((off + ((uint32_t)1 << 20)) >> 21)
1959 tcc_error("Failed relocating PLT (off=0x%lx, addr=0x%lx, pc=0x%lx)", off, addr, pc);
1960 put32(p, (0x90000010 | // adrp x16,...
1961 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1962 put32(p + 4, (0xf9400211 | // ldr x17,[x16,#...]
1963 (addr & 0xff8) << 7));
1964 put32(p + 8, (0x91000210 | // add x16,x16,#...
1965 (addr & 0xfff) << 10));
1966 put32(p + 12, 0xd61f0220); // br x17
1967 p += 16;
1968 }
1969 #elif defined(TCC_TARGET_C67)
1970 /* XXX: TODO */
1971 #else
1972 #error unsupported CPU
1973 #endif
1974 }
1975 }
1976
1977 /* Allocate strings for section names and decide if an unallocated section
1978 should be output.
1979
1980 NOTE: the strsec section comes last, so its size is also correct ! */
1981 static void alloc_sec_names(TCCState *s1, int file_type, Section *strsec)
1982 {
1983 int i;
1984 Section *s;
1985
1986 /* Allocate strings for section names */
1987 for(i = 1; i < s1->nb_sections; i++) {
1988 s = s1->sections[i];
1989 s->sh_name = put_elf_str(strsec, s->name);
1990 /* when generating a DLL, we include relocations but we may
1991 patch them */
1992 if (file_type == TCC_OUTPUT_DLL &&
1993 s->sh_type == SHT_RELX &&
1994 !(s->sh_flags & SHF_ALLOC)) {
1995 /* gr: avoid bogus relocs for empty (debug) sections */
1996 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
1997 prepare_dynamic_rel(s1, s);
1998 else if (s1->do_debug)
1999 s->sh_size = s->data_offset;
2000 } else if (s1->do_debug ||
2001 file_type == TCC_OUTPUT_OBJ ||
2002 (s->sh_flags & SHF_ALLOC) ||
2003 i == (s1->nb_sections - 1)) {
2004 /* we output all sections if debug or object file */
2005 s->sh_size = s->data_offset;
2006 }
2007 }
2008 }
2009
2010 /* Info to be copied in dynamic section */
2011 struct dyn_inf {
2012 Section *dynamic;
2013 Section *dynstr;
2014 unsigned long dyn_rel_off;
2015 addr_t rel_addr;
2016 addr_t rel_size;
2017 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2018 addr_t bss_addr;
2019 addr_t bss_size;
2020 #endif
2021 };
2022
2023 /* Assign sections to segments and decide how are sections laid out when loaded
2024 in memory. This function also fills corresponding program headers. */
2025 static int layout_sections(TCCState *s1, ElfW(Phdr) *phdr, int phnum,
2026 Section *interp, Section* strsec,
2027 struct dyn_inf *dyninf, int *sec_order)
2028 {
2029 int i, j, k, file_type, sh_order_index, file_offset;
2030 unsigned long s_align;
2031 long long tmp;
2032 addr_t addr;
2033 ElfW(Phdr) *ph;
2034 Section *s;
2035
2036 file_type = s1->output_type;
2037 sh_order_index = 1;
2038 file_offset = 0;
2039 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2040 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2041 s_align = ELF_PAGE_SIZE;
2042 if (s1->section_align)
2043 s_align = s1->section_align;
2044
2045 if (phnum > 0) {
2046 if (s1->has_text_addr) {
2047 int a_offset, p_offset;
2048 addr = s1->text_addr;
2049 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
2050 ELF_PAGE_SIZE */
2051 a_offset = (int) (addr & (s_align - 1));
2052 p_offset = file_offset & (s_align - 1);
2053 if (a_offset < p_offset)
2054 a_offset += s_align;
2055 file_offset += (a_offset - p_offset);
2056 } else {
2057 if (file_type == TCC_OUTPUT_DLL)
2058 addr = 0;
2059 else
2060 addr = ELF_START_ADDR;
2061 /* compute address after headers */
2062 addr += (file_offset & (s_align - 1));
2063 }
2064
2065 ph = &phdr[0];
2066 /* Leave one program headers for the program interpreter and one for
2067 the program header table itself if needed. These are done later as
2068 they require section layout to be done first. */
2069 if (interp)
2070 ph += 1 + HAVE_PHDR;
2071
2072 /* dynamic relocation table information, for .dynamic section */
2073 dyninf->rel_addr = dyninf->rel_size = 0;
2074 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2075 dyninf->bss_addr = dyninf->bss_size = 0;
2076 #endif
2077
2078 for(j = 0; j < 2; j++) {
2079 ph->p_type = PT_LOAD;
2080 if (j == 0)
2081 ph->p_flags = PF_R | PF_X;
2082 else
2083 ph->p_flags = PF_R | PF_W;
2084 ph->p_align = s_align;
2085
2086 /* Decide the layout of sections loaded in memory. This must
2087 be done before program headers are filled since they contain
2088 info about the layout. We do the following ordering: interp,
2089 symbol tables, relocations, progbits, nobits */
2090 /* XXX: do faster and simpler sorting */
2091 for(k = 0; k < 5; k++) {
2092 for(i = 1; i < s1->nb_sections; i++) {
2093 s = s1->sections[i];
2094 /* compute if section should be included */
2095 if (j == 0) {
2096 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2097 SHF_ALLOC)
2098 continue;
2099 } else {
2100 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2101 (SHF_ALLOC | SHF_WRITE))
2102 continue;
2103 }
2104 if (s == interp) {
2105 if (k != 0)
2106 continue;
2107 } else if (s->sh_type == SHT_DYNSYM ||
2108 s->sh_type == SHT_STRTAB ||
2109 s->sh_type == SHT_HASH) {
2110 if (k != 1)
2111 continue;
2112 } else if (s->sh_type == SHT_RELX) {
2113 if (k != 2)
2114 continue;
2115 } else if (s->sh_type == SHT_NOBITS) {
2116 if (k != 4)
2117 continue;
2118 } else {
2119 if (k != 3)
2120 continue;
2121 }
2122 sec_order[sh_order_index++] = i;
2123
2124 /* section matches: we align it and add its size */
2125 tmp = addr;
2126 addr = (addr + s->sh_addralign - 1) &
2127 ~(s->sh_addralign - 1);
2128 file_offset += (int) ( addr - tmp );
2129 s->sh_offset = file_offset;
2130 s->sh_addr = addr;
2131
2132 /* update program header infos */
2133 if (ph->p_offset == 0) {
2134 ph->p_offset = file_offset;
2135 ph->p_vaddr = addr;
2136 ph->p_paddr = ph->p_vaddr;
2137 }
2138 /* update dynamic relocation infos */
2139 if (s->sh_type == SHT_RELX) {
2140 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2141 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) {
2142 dyninf->rel_addr = addr;
2143 dyninf->rel_size += s->sh_size; /* XXX only first rel. */
2144 }
2145 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) {
2146 dyninf->bss_addr = addr;
2147 dyninf->bss_size = s->sh_size; /* XXX only first rel. */
2148 }
2149 #else
2150 if (dyninf->rel_size == 0)
2151 dyninf->rel_addr = addr;
2152 dyninf->rel_size += s->sh_size;
2153 #endif
2154 }
2155 addr += s->sh_size;
2156 if (s->sh_type != SHT_NOBITS)
2157 file_offset += s->sh_size;
2158 }
2159 }
2160 if (j == 0) {
2161 /* Make the first PT_LOAD segment include the program
2162 headers itself (and the ELF header as well), it'll
2163 come out with same memory use but will make various
2164 tools like binutils strip work better. */
2165 ph->p_offset &= ~(ph->p_align - 1);
2166 ph->p_vaddr &= ~(ph->p_align - 1);
2167 ph->p_paddr &= ~(ph->p_align - 1);
2168 }
2169 ph->p_filesz = file_offset - ph->p_offset;
2170 ph->p_memsz = addr - ph->p_vaddr;
2171 ph++;
2172 if (j == 0) {
2173 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
2174 /* if in the middle of a page, we duplicate the page in
2175 memory so that one copy is RX and the other is RW */
2176 if ((addr & (s_align - 1)) != 0)
2177 addr += s_align;
2178 } else {
2179 addr = (addr + s_align - 1) & ~(s_align - 1);
2180 file_offset = (file_offset + s_align - 1) & ~(s_align - 1);
2181 }
2182 }
2183 }
2184 }
2185
2186 /* all other sections come after */
2187 for(i = 1; i < s1->nb_sections; i++) {
2188 s = s1->sections[i];
2189 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
2190 continue;
2191 sec_order[sh_order_index++] = i;
2192
2193 file_offset = (file_offset + s->sh_addralign - 1) &
2194 ~(s->sh_addralign - 1);
2195 s->sh_offset = file_offset;
2196 if (s->sh_type != SHT_NOBITS)
2197 file_offset += s->sh_size;
2198 }
2199
2200 return file_offset;
2201 }
2202
2203 static void fill_unloadable_phdr(ElfW(Phdr) *phdr, int phnum, Section *interp,
2204 Section *dynamic)
2205 {
2206 ElfW(Phdr) *ph;
2207
2208 /* if interpreter, then add corresponding program header */
2209 if (interp) {
2210 ph = &phdr[0];
2211
2212 if (HAVE_PHDR)
2213 {
2214 int len = phnum * sizeof(ElfW(Phdr));
2215
2216 ph->p_type = PT_PHDR;
2217 ph->p_offset = sizeof(ElfW(Ehdr));
2218 ph->p_vaddr = interp->sh_addr - len;
2219 ph->p_paddr = ph->p_vaddr;
2220 ph->p_filesz = ph->p_memsz = len;
2221 ph->p_flags = PF_R | PF_X;
2222 ph->p_align = 4; /* interp->sh_addralign; */
2223 ph++;
2224 }
2225
2226 ph->p_type = PT_INTERP;
2227 ph->p_offset = interp->sh_offset;
2228 ph->p_vaddr = interp->sh_addr;
2229 ph->p_paddr = ph->p_vaddr;
2230 ph->p_filesz = interp->sh_size;
2231 ph->p_memsz = interp->sh_size;
2232 ph->p_flags = PF_R;
2233 ph->p_align = interp->sh_addralign;
2234 }
2235
2236 /* if dynamic section, then add corresponding program header */
2237 if (dynamic) {
2238 ph = &phdr[phnum - 1];
2239
2240 ph->p_type = PT_DYNAMIC;
2241 ph->p_offset = dynamic->sh_offset;
2242 ph->p_vaddr = dynamic->sh_addr;
2243 ph->p_paddr = ph->p_vaddr;
2244 ph->p_filesz = dynamic->sh_size;
2245 ph->p_memsz = dynamic->sh_size;
2246 ph->p_flags = PF_R | PF_W;
2247 ph->p_align = dynamic->sh_addralign;
2248 }
2249 }
2250
2251 /* Fill the dynamic section with tags describing the address and size of
2252 sections */
2253 static void fill_dynamic(TCCState *s1, struct dyn_inf *dyninf)
2254 {
2255 Section *dynamic;
2256
2257 dynamic = dyninf->dynamic;
2258
2259 /* put dynamic section entries */
2260 dynamic->data_offset = dyninf->dyn_rel_off;
2261 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2262 put_dt(dynamic, DT_STRTAB, dyninf->dynstr->sh_addr);
2263 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2264 put_dt(dynamic, DT_STRSZ, dyninf->dynstr->data_offset);
2265 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2266 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
2267 put_dt(dynamic, DT_RELA, dyninf->rel_addr);
2268 put_dt(dynamic, DT_RELASZ, dyninf->rel_size);
2269 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2270 #else
2271 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2272 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2273 put_dt(dynamic, DT_PLTRELSZ, dyninf->rel_size);
2274 put_dt(dynamic, DT_JMPREL, dyninf->rel_addr);
2275 put_dt(dynamic, DT_PLTREL, DT_REL);
2276 put_dt(dynamic, DT_REL, dyninf->bss_addr);
2277 put_dt(dynamic, DT_RELSZ, dyninf->bss_size);
2278 #else
2279 put_dt(dynamic, DT_REL, dyninf->rel_addr);
2280 put_dt(dynamic, DT_RELSZ, dyninf->rel_size);
2281 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2282 #endif
2283 #endif
2284 if (s1->do_debug)
2285 put_dt(dynamic, DT_DEBUG, 0);
2286 put_dt(dynamic, DT_NULL, 0);
2287 }
2288
2289 /* Relocate remaining sections and symbols (that is those not related to
2290 dynamic linking) */
2291 static int final_sections_reloc(TCCState *s1)
2292 {
2293 int i;
2294 Section *s;
2295
2296 relocate_syms(s1, 0);
2297
2298 if (s1->nb_errors != 0)
2299 return -1;
2300
2301 /* relocate sections */
2302 /* XXX: ignore sections with allocated relocations ? */
2303 for(i = 1; i < s1->nb_sections; i++) {
2304 s = s1->sections[i];
2305 if (s->reloc && s != s1->got && (s->sh_flags & SHF_ALLOC)) //gr
2306 relocate_section(s1, s);
2307 }
2308
2309 /* relocate relocation entries if the relocation tables are
2310 allocated in the executable */
2311 for(i = 1; i < s1->nb_sections; i++) {
2312 s = s1->sections[i];
2313 if ((s->sh_flags & SHF_ALLOC) &&
2314 s->sh_type == SHT_RELX) {
2315 relocate_rel(s1, s);
2316 }
2317 }
2318 return 0;
2319 }
2320
2321 /* Create an ELF file on disk.
2322 This function handle ELF specific layout requirements */
2323 static void tcc_output_elf(TCCState *s1, FILE *f, int phnum, ElfW(Phdr) *phdr,
2324 int file_offset, int *sec_order)
2325 {
2326 int i, shnum, offset, size, file_type;
2327 Section *s;
2328 ElfW(Ehdr) ehdr;
2329 ElfW(Shdr) shdr, *sh;
2330
2331 file_type = s1->output_type;
2332 shnum = s1->nb_sections;
2333
2334 memset(&ehdr, 0, sizeof(ehdr));
2335
2336 if (phnum > 0) {
2337 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2338 ehdr.e_phnum = phnum;
2339 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2340 }
2341
2342 /* align to 4 */
2343 file_offset = (file_offset + 3) & -4;
2344
2345 /* fill header */
2346 ehdr.e_ident[0] = ELFMAG0;
2347 ehdr.e_ident[1] = ELFMAG1;
2348 ehdr.e_ident[2] = ELFMAG2;
2349 ehdr.e_ident[3] = ELFMAG3;
2350 ehdr.e_ident[4] = ELFCLASSW;
2351 ehdr.e_ident[5] = ELFDATA2LSB;
2352 ehdr.e_ident[6] = EV_CURRENT;
2353 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2354 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2355 #endif
2356 #ifdef TCC_TARGET_ARM
2357 #ifdef TCC_ARM_EABI
2358 ehdr.e_ident[EI_OSABI] = 0;
2359 ehdr.e_flags = EF_ARM_EABI_VER4;
2360 if (file_type == TCC_OUTPUT_EXE || file_type == TCC_OUTPUT_DLL)
2361 ehdr.e_flags |= EF_ARM_HASENTRY;
2362 if (s1->float_abi == ARM_HARD_FLOAT)
2363 ehdr.e_flags |= EF_ARM_VFP_FLOAT;
2364 else
2365 ehdr.e_flags |= EF_ARM_SOFT_FLOAT;
2366 #else
2367 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2368 #endif
2369 #endif
2370 switch(file_type) {
2371 default:
2372 case TCC_OUTPUT_EXE:
2373 ehdr.e_type = ET_EXEC;
2374 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2375 break;
2376 case TCC_OUTPUT_DLL:
2377 ehdr.e_type = ET_DYN;
2378 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2379 break;
2380 case TCC_OUTPUT_OBJ:
2381 ehdr.e_type = ET_REL;
2382 break;
2383 }
2384 ehdr.e_machine = EM_TCC_TARGET;
2385 ehdr.e_version = EV_CURRENT;
2386 ehdr.e_shoff = file_offset;
2387 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2388 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2389 ehdr.e_shnum = shnum;
2390 ehdr.e_shstrndx = shnum - 1;
2391
2392 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2393 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2394 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2395
2396 sort_syms(s1, symtab_section);
2397 for(i = 1; i < s1->nb_sections; i++) {
2398 s = s1->sections[sec_order[i]];
2399 if (s->sh_type != SHT_NOBITS) {
2400 if (s->sh_type == SHT_DYNSYM)
2401 patch_dynsym_undef(s1, s);
2402 while (offset < s->sh_offset) {
2403 fputc(0, f);
2404 offset++;
2405 }
2406 size = s->sh_size;
2407 fwrite(s->data, 1, size, f);
2408 offset += size;
2409 }
2410 }
2411
2412 /* output section headers */
2413 while (offset < ehdr.e_shoff) {
2414 fputc(0, f);
2415 offset++;
2416 }
2417
2418 for(i = 0; i < s1->nb_sections; i++) {
2419 sh = &shdr;
2420 memset(sh, 0, sizeof(ElfW(Shdr)));
2421 s = s1->sections[i];
2422 if (s) {
2423 sh->sh_name = s->sh_name;
2424 sh->sh_type = s->sh_type;
2425 sh->sh_flags = s->sh_flags;
2426 sh->sh_entsize = s->sh_entsize;
2427 sh->sh_info = s->sh_info;
2428 if (s->link)
2429 sh->sh_link = s->link->sh_num;
2430 sh->sh_addralign = s->sh_addralign;
2431 sh->sh_addr = s->sh_addr;
2432 sh->sh_offset = s->sh_offset;
2433 sh->sh_size = s->sh_size;
2434 }
2435 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2436 }
2437 }
2438
2439 /* Write an elf, coff or "binary" file */
2440 static int tcc_write_elf_file(TCCState *s1, const char *filename, int phnum,
2441 ElfW(Phdr) *phdr, int file_offset, int *sec_order)
2442 {
2443 int fd, mode, file_type;
2444 FILE *f;
2445
2446 file_type = s1->output_type;
2447 if (file_type == TCC_OUTPUT_OBJ)
2448 mode = 0666;
2449 else
2450 mode = 0777;
2451 unlink(filename);
2452 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2453 if (fd < 0) {
2454 tcc_error_noabort("could not write '%s'", filename);
2455 return -1;
2456 }
2457 f = fdopen(fd, "wb");
2458 if (s1->verbose)
2459 printf("<- %s\n", filename);
2460
2461 #ifdef TCC_TARGET_COFF
2462 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF)
2463 tcc_output_coff(s1, f);
2464 else
2465 #endif
2466 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2467 tcc_output_elf(s1, f, phnum, phdr, file_offset, sec_order);
2468 else
2469 tcc_output_binary(s1, f, sec_order);
2470 fclose(f);
2471
2472 return 0;
2473 }
2474
2475 /* Output an elf, coff or binary file */
2476 /* XXX: suppress unneeded sections */
2477 static int elf_output_file(TCCState *s1, const char *filename)
2478 {
2479 int i, ret, phnum, shnum, file_type, file_offset, *sec_order;
2480 struct dyn_inf dyninf;
2481 ElfW(Phdr) *phdr;
2482 ElfW(Sym) *sym;
2483 Section *strsec, *interp, *dynamic, *dynstr;
2484
2485 file_type = s1->output_type;
2486 s1->nb_errors = 0;
2487
2488 /* if linking, also link in runtime libraries (libc, libgcc, etc.) */
2489 if (file_type != TCC_OUTPUT_OBJ) {
2490 tcc_add_runtime(s1);
2491 }
2492
2493 phdr = NULL;
2494 sec_order = NULL;
2495 interp = dynamic = dynstr = NULL; /* avoid warning */
2496 dyninf.dyn_rel_off = 0; /* avoid warning */
2497
2498 if (file_type != TCC_OUTPUT_OBJ) {
2499 relocate_common_syms();
2500
2501 tcc_add_linker_symbols(s1);
2502
2503 if (!s1->static_link) {
2504 if (file_type == TCC_OUTPUT_EXE) {
2505 char *ptr;
2506 /* allow override the dynamic loader */
2507 const char *elfint = getenv("LD_SO");
2508 if (elfint == NULL)
2509 elfint = DEFAULT_ELFINTERP(s1);
2510 /* add interpreter section only if executable */
2511 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
2512 interp->sh_addralign = 1;
2513 ptr = section_ptr_add(interp, 1 + strlen(elfint));
2514 strcpy(ptr, elfint);
2515 }
2516
2517 /* add dynamic symbol table */
2518 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
2519 ".dynstr",
2520 ".hash", SHF_ALLOC);
2521 dynstr = s1->dynsym->link;
2522
2523 /* add dynamic section */
2524 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
2525 SHF_ALLOC | SHF_WRITE);
2526 dynamic->link = dynstr;
2527 dynamic->sh_entsize = sizeof(ElfW(Dyn));
2528
2529 build_got(s1);
2530
2531 if (file_type == TCC_OUTPUT_EXE) {
2532 bind_exe_dynsyms(s1);
2533
2534 if (s1->nb_errors) {
2535 ret = -1;
2536 goto the_end;
2537 }
2538
2539 bind_libs_dynsyms(s1);
2540 } else /* shared library case: simply export all global symbols */
2541 export_global_syms(s1);
2542
2543 build_got_entries(s1);
2544
2545 /* add a list of needed dlls */
2546 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2547 DLLReference *dllref = s1->loaded_dlls[i];
2548 if (dllref->level == 0)
2549 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
2550 }
2551
2552 if (s1->rpath)
2553 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
2554
2555 /* XXX: currently, since we do not handle PIC code, we
2556 must relocate the readonly segments */
2557 if (file_type == TCC_OUTPUT_DLL) {
2558 if (s1->soname)
2559 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
2560 put_dt(dynamic, DT_TEXTREL, 0);
2561 }
2562
2563 if (s1->symbolic)
2564 put_dt(dynamic, DT_SYMBOLIC, 0);
2565
2566 /* add necessary space for other entries */
2567 dyninf.dyn_rel_off = dynamic->data_offset;
2568 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
2569 } else {
2570 /* still need to build got entries in case of static link */
2571 build_got_entries(s1);
2572 }
2573 }
2574
2575 /* we add a section for symbols */
2576 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
2577 put_elf_str(strsec, "");
2578
2579 /* compute number of sections */
2580 shnum = s1->nb_sections;
2581
2582 /* this array is used to reorder sections in the output file */
2583 sec_order = tcc_malloc(sizeof(int) * shnum);
2584 sec_order[0] = 0;
2585
2586 /* compute number of program headers */
2587 switch(file_type) {
2588 default:
2589 case TCC_OUTPUT_OBJ:
2590 phnum = 0;
2591 break;
2592 case TCC_OUTPUT_EXE:
2593 if (!s1->static_link)
2594 phnum = 4 + HAVE_PHDR;
2595 else
2596 phnum = 2;
2597 break;
2598 case TCC_OUTPUT_DLL:
2599 phnum = 3;
2600 break;
2601 }
2602
2603 /* Allocate strings for section names */
2604 alloc_sec_names(s1, file_type, strsec);
2605
2606 /* allocate program segment headers */
2607 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
2608
2609 /* compute section to program header mapping */
2610 file_offset = layout_sections(s1, phdr, phnum, interp, strsec, &dyninf,
2611 sec_order);
2612
2613 /* Fill remaining program header and finalize relocation related to dynamic
2614 linking. */
2615 if (phnum > 0) {
2616 fill_unloadable_phdr(phdr, phnum, interp, dynamic);
2617 if (dynamic) {
2618 dyninf.dynamic = dynamic;
2619 dyninf.dynstr = dynstr;
2620
2621 fill_dynamic(s1, &dyninf);
2622
2623 /* put in GOT the dynamic section address and relocate PLT */
2624 put32(s1->got->data, dynamic->sh_addr);
2625 if (file_type == TCC_OUTPUT_EXE
2626 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2627 || file_type == TCC_OUTPUT_DLL
2628 #endif
2629 )
2630 relocate_plt(s1);
2631
2632 /* relocate symbols in .dynsym now that final addresses are known */
2633 for_each_elem(s1->dynsym, 1, sym, ElfW(Sym)) {
2634 if (sym->st_shndx == SHN_UNDEF) {
2635 /* relocate to PLT if symbol corresponds to a PLT entry,
2636 but not if it's a weak symbol */
2637 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
2638 sym->st_value = 0;
2639 else if (sym->st_value)
2640 sym->st_value += s1->plt->sh_addr;
2641 } else if (sym->st_shndx < SHN_LORESERVE) {
2642 /* do symbol relocation */
2643 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2644 }
2645 }
2646 }
2647 }
2648
2649 /* if building executable or DLL, then relocate each section
2650 except the GOT which is already relocated */
2651 if (file_type != TCC_OUTPUT_OBJ) {
2652 ret = final_sections_reloc(s1);
2653 if (ret)
2654 goto the_end;
2655 }
2656
2657 /* Perform relocation to GOT or PLT entries */
2658 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2659 fill_got(s1);
2660
2661 /* Create the ELF file with name 'filename' */
2662 ret = tcc_write_elf_file(s1, filename, phnum, phdr, file_offset, sec_order);
2663 the_end:
2664 tcc_free(s1->symtab_to_dynsym);
2665 tcc_free(sec_order);
2666 tcc_free(phdr);
2667 tcc_free(s1->sym_attrs);
2668 s1->sym_attrs = NULL;
2669 return ret;
2670 }
2671
2672 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2673 {
2674 int ret;
2675 #ifdef TCC_TARGET_PE
2676 if (s->output_type != TCC_OUTPUT_OBJ) {
2677 ret = pe_output_file(s, filename);
2678 } else
2679 #endif
2680 ret = elf_output_file(s, filename);
2681 return ret;
2682 }
2683
2684 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2685 {
2686 void *data;
2687
2688 data = tcc_malloc(size);
2689 lseek(fd, file_offset, SEEK_SET);
2690 read(fd, data, size);
2691 return data;
2692 }
2693
2694 typedef struct SectionMergeInfo {
2695 Section *s; /* corresponding existing section */
2696 unsigned long offset; /* offset of the new section in the existing section */
2697 uint8_t new_section; /* true if section 's' was added */
2698 uint8_t link_once; /* true if link once section */
2699 } SectionMergeInfo;
2700
2701 /* load an object file and merge it with current files */
2702 /* XXX: handle correctly stab (debug) info */
2703 ST_FUNC int tcc_load_object_file(TCCState *s1,
2704 int fd, unsigned long file_offset)
2705 {
2706 ElfW(Ehdr) ehdr;
2707 ElfW(Shdr) *shdr, *sh;
2708 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2709 unsigned char *strsec, *strtab;
2710 int *old_to_new_syms;
2711 char *sh_name, *name;
2712 SectionMergeInfo *sm_table, *sm;
2713 ElfW(Sym) *sym, *symtab;
2714 ElfW_Rel *rel;
2715 Section *s;
2716
2717 int stab_index;
2718 int stabstr_index;
2719
2720 stab_index = stabstr_index = 0;
2721
2722 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2723 goto fail1;
2724 if (ehdr.e_ident[0] != ELFMAG0 ||
2725 ehdr.e_ident[1] != ELFMAG1 ||
2726 ehdr.e_ident[2] != ELFMAG2 ||
2727 ehdr.e_ident[3] != ELFMAG3)
2728 goto fail1;
2729 /* test if object file */
2730 if (ehdr.e_type != ET_REL)
2731 goto fail1;
2732 /* test CPU specific stuff */
2733 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2734 ehdr.e_machine != EM_TCC_TARGET) {
2735 fail1:
2736 tcc_error_noabort("invalid object file");
2737 return -1;
2738 }
2739 /* read sections */
2740 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2741 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2742 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2743
2744 /* load section names */
2745 sh = &shdr[ehdr.e_shstrndx];
2746 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2747
2748 /* load symtab and strtab */
2749 old_to_new_syms = NULL;
2750 symtab = NULL;
2751 strtab = NULL;
2752 nb_syms = 0;
2753 for(i = 1; i < ehdr.e_shnum; i++) {
2754 sh = &shdr[i];
2755 if (sh->sh_type == SHT_SYMTAB) {
2756 if (symtab) {
2757 tcc_error_noabort("object must contain only one symtab");
2758 fail:
2759 ret = -1;
2760 goto the_end;
2761 }
2762 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2763 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2764 sm_table[i].s = symtab_section;
2765
2766 /* now load strtab */
2767 sh = &shdr[sh->sh_link];
2768 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2769 }
2770 }
2771
2772 /* now examine each section and try to merge its content with the
2773 ones in memory */
2774 for(i = 1; i < ehdr.e_shnum; i++) {
2775 /* no need to examine section name strtab */
2776 if (i == ehdr.e_shstrndx)
2777 continue;
2778 sh = &shdr[i];
2779 sh_name = (char *) strsec + sh->sh_name;
2780 /* ignore sections types we do not handle */
2781 if (sh->sh_type != SHT_PROGBITS &&
2782 sh->sh_type != SHT_RELX &&
2783 #ifdef TCC_ARM_EABI
2784 sh->sh_type != SHT_ARM_EXIDX &&
2785 #endif
2786 sh->sh_type != SHT_NOBITS &&
2787 sh->sh_type != SHT_PREINIT_ARRAY &&
2788 sh->sh_type != SHT_INIT_ARRAY &&
2789 sh->sh_type != SHT_FINI_ARRAY &&
2790 strcmp(sh_name, ".stabstr")
2791 )
2792 continue;
2793 if (sh->sh_addralign < 1)
2794 sh->sh_addralign = 1;
2795 /* find corresponding section, if any */
2796 for(j = 1; j < s1->nb_sections;j++) {
2797 s = s1->sections[j];
2798 if (!strcmp(s->name, sh_name)) {
2799 if (!strncmp(sh_name, ".gnu.linkonce",
2800 sizeof(".gnu.linkonce") - 1)) {
2801 /* if a 'linkonce' section is already present, we
2802 do not add it again. It is a little tricky as
2803 symbols can still be defined in
2804 it. */
2805 sm_table[i].link_once = 1;
2806 goto next;
2807 } else {
2808 goto found;
2809 }
2810 }
2811 }
2812 /* not found: create new section */
2813 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2814 /* take as much info as possible from the section. sh_link and
2815 sh_info will be updated later */
2816 s->sh_addralign = sh->sh_addralign;
2817 s->sh_entsize = sh->sh_entsize;
2818 sm_table[i].new_section = 1;
2819 found:
2820 if (sh->sh_type != s->sh_type) {
2821 tcc_error_noabort("invalid section type");
2822 goto fail;
2823 }
2824
2825 /* align start of section */
2826 offset = s->data_offset;
2827
2828 if (0 == strcmp(sh_name, ".stab")) {
2829 stab_index = i;
2830 goto no_align;
2831 }
2832 if (0 == strcmp(sh_name, ".stabstr")) {
2833 stabstr_index = i;
2834 goto no_align;
2835 }
2836
2837 size = sh->sh_addralign - 1;
2838 offset = (offset + size) & ~size;
2839 if (sh->sh_addralign > s->sh_addralign)
2840 s->sh_addralign = sh->sh_addralign;
2841 s->data_offset = offset;
2842 no_align:
2843 sm_table[i].offset = offset;
2844 sm_table[i].s = s;
2845 /* concatenate sections */
2846 size = sh->sh_size;
2847 if (sh->sh_type != SHT_NOBITS) {
2848 unsigned char *ptr;
2849 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2850 ptr = section_ptr_add(s, size);
2851 read(fd, ptr, size);
2852 } else {
2853 s->data_offset += size;
2854 }
2855 next: ;
2856 }
2857
2858 /* gr relocate stab strings */
2859 if (stab_index && stabstr_index) {
2860 Stab_Sym *a, *b;
2861 unsigned o;
2862 s = sm_table[stab_index].s;
2863 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2864 b = (Stab_Sym *)(s->data + s->data_offset);
2865 o = sm_table[stabstr_index].offset;
2866 while (a < b)
2867 a->n_strx += o, a++;
2868 }
2869
2870 /* second short pass to update sh_link and sh_info fields of new
2871 sections */
2872 for(i = 1; i < ehdr.e_shnum; i++) {
2873 s = sm_table[i].s;
2874 if (!s || !sm_table[i].new_section)
2875 continue;
2876 sh = &shdr[i];
2877 if (sh->sh_link > 0)
2878 s->link = sm_table[sh->sh_link].s;
2879 if (sh->sh_type == SHT_RELX) {
2880 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2881 /* update backward link */
2882 s1->sections[s->sh_info]->reloc = s;
2883 }
2884 }
2885 sm = sm_table;
2886
2887 /* resolve symbols */
2888 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2889
2890 sym = symtab + 1;
2891 for(i = 1; i < nb_syms; i++, sym++) {
2892 if (sym->st_shndx != SHN_UNDEF &&
2893 sym->st_shndx < SHN_LORESERVE) {
2894 sm = &sm_table[sym->st_shndx];
2895 if (sm->link_once) {
2896 /* if a symbol is in a link once section, we use the
2897 already defined symbol. It is very important to get
2898 correct relocations */
2899 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2900 name = (char *) strtab + sym->st_name;
2901 sym_index = find_elf_sym(symtab_section, name);
2902 if (sym_index)
2903 old_to_new_syms[i] = sym_index;
2904 }
2905 continue;
2906 }
2907 /* if no corresponding section added, no need to add symbol */
2908 if (!sm->s)
2909 continue;
2910 /* convert section number */
2911 sym->st_shndx = sm->s->sh_num;
2912 /* offset value */
2913 sym->st_value += sm->offset;
2914 }
2915 /* add symbol */
2916 name = (char *) strtab + sym->st_name;
2917 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2918 sym->st_info, sym->st_other,
2919 sym->st_shndx, name);
2920 old_to_new_syms[i] = sym_index;
2921 }
2922
2923 /* third pass to patch relocation entries */
2924 for(i = 1; i < ehdr.e_shnum; i++) {
2925 s = sm_table[i].s;
2926 if (!s)
2927 continue;
2928 sh = &shdr[i];
2929 offset = sm_table[i].offset;
2930 switch(s->sh_type) {
2931 case SHT_RELX:
2932 /* take relocation offset information */
2933 offseti = sm_table[sh->sh_info].offset;
2934 for_each_elem(s, (offset / sizeof(*rel)), rel, ElfW_Rel) {
2935 int type;
2936 unsigned sym_index;
2937 /* convert symbol index */
2938 type = ELFW(R_TYPE)(rel->r_info);
2939 sym_index = ELFW(R_SYM)(rel->r_info);
2940 /* NOTE: only one symtab assumed */
2941 if (sym_index >= nb_syms)
2942 goto invalid_reloc;
2943 sym_index = old_to_new_syms[sym_index];
2944 /* ignore link_once in rel section. */
2945 if (!sym_index && !sm->link_once
2946 #ifdef TCC_TARGET_ARM
2947 && type != R_ARM_V4BX
2948 #endif
2949 ) {
2950 invalid_reloc:
2951 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2952 i, strsec + sh->sh_name, rel->r_offset);
2953 goto fail;
2954 }
2955 rel->r_info = ELFW(R_INFO)(sym_index, type);
2956 /* offset the relocation offset */
2957 rel->r_offset += offseti;
2958 #ifdef TCC_TARGET_ARM
2959 /* Jumps and branches from a Thumb code to a PLT entry need
2960 special handling since PLT entries are ARM code.
2961 Unconditional bl instructions referencing PLT entries are
2962 handled by converting these instructions into blx
2963 instructions. Other case of instructions referencing a PLT
2964 entry require to add a Thumb stub before the PLT entry to
2965 switch to ARM mode. We set bit plt_thumb_stub of the
2966 attribute of a symbol to indicate such a case. */
2967 if (type == R_ARM_THM_JUMP24)
2968 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
2969 #endif
2970 }
2971 break;
2972 default:
2973 break;
2974 }
2975 }
2976
2977 ret = 0;
2978 the_end:
2979 tcc_free(symtab);
2980 tcc_free(strtab);
2981 tcc_free(old_to_new_syms);
2982 tcc_free(sm_table);
2983 tcc_free(strsec);
2984 tcc_free(shdr);
2985 return ret;
2986 }
2987
2988 typedef struct ArchiveHeader {
2989 char ar_name[16]; /* name of this member */
2990 char ar_date[12]; /* file mtime */
2991 char ar_uid[6]; /* owner uid; printed as decimal */
2992 char ar_gid[6]; /* owner gid; printed as decimal */
2993 char ar_mode[8]; /* file mode, printed as octal */
2994 char ar_size[10]; /* file size, printed as decimal */
2995 char ar_fmag[2]; /* should contain ARFMAG */
2996 } ArchiveHeader;
2997
2998 static int get_be32(const uint8_t *b)
2999 {
3000 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
3001 }
3002
3003 /* load only the objects which resolve undefined symbols */
3004 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
3005 {
3006 int i, bound, nsyms, sym_index, off, ret;
3007 uint8_t *data;
3008 const char *ar_names, *p;
3009 const uint8_t *ar_index;
3010 ElfW(Sym) *sym;
3011
3012 data = tcc_malloc(size);
3013 if (read(fd, data, size) != size)
3014 goto fail;
3015 nsyms = get_be32(data);
3016 ar_index = data + 4;
3017 ar_names = (char *) ar_index + nsyms * 4;
3018
3019 do {
3020 bound = 0;
3021 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
3022 sym_index = find_elf_sym(symtab_section, p);
3023 if(sym_index) {
3024 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
3025 if(sym->st_shndx == SHN_UNDEF) {
3026 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
3027 ++bound;
3028 lseek(fd, off, SEEK_SET);
3029 if(tcc_load_object_file(s1, fd, off) < 0) {
3030 fail:
3031 ret = -1;
3032 goto the_end;
3033 }
3034 }
3035 }
3036 }
3037 } while(bound);
3038 ret = 0;
3039 the_end:
3040 tcc_free(data);
3041 return ret;
3042 }
3043
3044 /* load a '.a' file */
3045 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
3046 {
3047 ArchiveHeader hdr;
3048 char ar_size[11];
3049 char ar_name[17];
3050 char magic[8];
3051 int size, len, i;
3052 unsigned long file_offset;
3053
3054 /* skip magic which was already checked */
3055 read(fd, magic, sizeof(magic));
3056
3057 for(;;) {
3058 len = read(fd, &hdr, sizeof(hdr));
3059 if (len == 0)
3060 break;
3061 if (len != sizeof(hdr)) {
3062 tcc_error_noabort("invalid archive");
3063 return -1;
3064 }
3065 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
3066 ar_size[sizeof(hdr.ar_size)] = '\0';
3067 size = strtol(ar_size, NULL, 0);
3068 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
3069 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
3070 if (ar_name[i] != ' ')
3071 break;
3072 }
3073 ar_name[i + 1] = '\0';
3074 file_offset = lseek(fd, 0, SEEK_CUR);
3075 /* align to even */
3076 size = (size + 1) & ~1;
3077 if (!strcmp(ar_name, "/")) {
3078 /* coff symbol table : we handle it */
3079 if(s1->alacarte_link)
3080 return tcc_load_alacarte(s1, fd, size);
3081 } else if (!strcmp(ar_name, "//") ||
3082 !strcmp(ar_name, "__.SYMDEF") ||
3083 !strcmp(ar_name, "__.SYMDEF/") ||
3084 !strcmp(ar_name, "ARFILENAMES/")) {
3085 /* skip symbol table or archive names */
3086 } else {
3087 if (tcc_load_object_file(s1, fd, file_offset) < 0)
3088 return -1;
3089 }
3090 lseek(fd, file_offset + size, SEEK_SET);
3091 }
3092 return 0;
3093 }
3094
3095 #ifndef TCC_TARGET_PE
3096 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
3097 is referenced by the user (so it should be added as DT_NEEDED in
3098 the generated ELF file) */
3099 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
3100 {
3101 ElfW(Ehdr) ehdr;
3102 ElfW(Shdr) *shdr, *sh, *sh1;
3103 int i, j, nb_syms, nb_dts, sym_bind, ret;
3104 ElfW(Sym) *sym, *dynsym;
3105 ElfW(Dyn) *dt, *dynamic;
3106 unsigned char *dynstr;
3107 const char *name, *soname;
3108 DLLReference *dllref;
3109
3110 read(fd, &ehdr, sizeof(ehdr));
3111
3112 /* test CPU specific stuff */
3113 if (ehdr.e_ident[5] != ELFDATA2LSB ||
3114 ehdr.e_machine != EM_TCC_TARGET) {
3115 tcc_error_noabort("bad architecture");
3116 return -1;
3117 }
3118
3119 /* read sections */
3120 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
3121
3122 /* load dynamic section and dynamic symbols */
3123 nb_syms = 0;
3124 nb_dts = 0;
3125 dynamic = NULL;
3126 dynsym = NULL; /* avoid warning */
3127 dynstr = NULL; /* avoid warning */
3128 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
3129 switch(sh->sh_type) {
3130 case SHT_DYNAMIC:
3131 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
3132 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
3133 break;
3134 case SHT_DYNSYM:
3135 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
3136 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
3137 sh1 = &shdr[sh->sh_link];
3138 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
3139 break;
3140 default:
3141 break;
3142 }
3143 }
3144
3145 /* compute the real library name */
3146 soname = tcc_basename(filename);
3147
3148 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3149 if (dt->d_tag == DT_SONAME) {
3150 soname = (char *) dynstr + dt->d_un.d_val;
3151 }
3152 }
3153
3154 /* if the dll is already loaded, do not load it */
3155 for(i = 0; i < s1->nb_loaded_dlls; i++) {
3156 dllref = s1->loaded_dlls[i];
3157 if (!strcmp(soname, dllref->name)) {
3158 /* but update level if needed */
3159 if (level < dllref->level)
3160 dllref->level = level;
3161 ret = 0;
3162 goto the_end;
3163 }
3164 }
3165
3166 /* add the dll and its level */
3167 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
3168 dllref->level = level;
3169 strcpy(dllref->name, soname);
3170 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
3171
3172 /* add dynamic symbols in dynsym_section */
3173 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
3174 sym_bind = ELFW(ST_BIND)(sym->st_info);
3175 if (sym_bind == STB_LOCAL)
3176 continue;
3177 name = (char *) dynstr + sym->st_name;
3178 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
3179 sym->st_info, sym->st_other, sym->st_shndx, name);
3180 }
3181
3182 /* load all referenced DLLs */
3183 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3184 switch(dt->d_tag) {
3185 case DT_NEEDED:
3186 name = (char *) dynstr + dt->d_un.d_val;
3187 for(j = 0; j < s1->nb_loaded_dlls; j++) {
3188 dllref = s1->loaded_dlls[j];
3189 if (!strcmp(name, dllref->name))
3190 goto already_loaded;
3191 }
3192 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
3193 tcc_error_noabort("referenced dll '%s' not found", name);
3194 ret = -1;
3195 goto the_end;
3196 }
3197 already_loaded:
3198 break;
3199 }
3200 }
3201 ret = 0;
3202 the_end:
3203 tcc_free(dynstr);
3204 tcc_free(dynsym);
3205 tcc_free(dynamic);
3206 tcc_free(shdr);
3207 return ret;
3208 }
3209
3210 #define LD_TOK_NAME 256
3211 #define LD_TOK_EOF (-1)
3212
3213 /* return next ld script token */
3214 static int ld_next(TCCState *s1, char *name, int name_size)
3215 {
3216 int c;
3217 char *q;
3218
3219 redo:
3220 switch(ch) {
3221 case ' ':
3222 case '\t':
3223 case '\f':
3224 case '\v':
3225 case '\r':
3226 case '\n':
3227 inp();
3228 goto redo;
3229 case '/':
3230 minp();
3231 if (ch == '*') {
3232 file->buf_ptr = parse_comment(file->buf_ptr);
3233 ch = file->buf_ptr[0];
3234 goto redo;
3235 } else {
3236 q = name;
3237 *q++ = '/';
3238 goto parse_name;
3239 }
3240 break;
3241 /* case 'a' ... 'z': */
3242 case 'a':
3243 case 'b':
3244 case 'c':
3245 case 'd':
3246 case 'e':
3247 case 'f':
3248 case 'g':
3249 case 'h':
3250 case 'i':
3251 case 'j':
3252 case 'k':
3253 case 'l':
3254 case 'm':
3255 case 'n':
3256 case 'o':
3257 case 'p':
3258 case 'q':
3259 case 'r':
3260 case 's':
3261 case 't':
3262 case 'u':
3263 case 'v':
3264 case 'w':
3265 case 'x':
3266 case 'y':
3267 case 'z':
3268 /* case 'A' ... 'z': */
3269 case 'A':
3270 case 'B':
3271 case 'C':
3272 case 'D':
3273 case 'E':
3274 case 'F':
3275 case 'G':
3276 case 'H':
3277 case 'I':
3278 case 'J':
3279 case 'K':
3280 case 'L':
3281 case 'M':
3282 case 'N':
3283 case 'O':
3284 case 'P':
3285 case 'Q':
3286 case 'R':
3287 case 'S':
3288 case 'T':
3289 case 'U':
3290 case 'V':
3291 case 'W':
3292 case 'X':
3293 case 'Y':
3294 case 'Z':
3295 case '_':
3296 case '\\':
3297 case '.':
3298 case '$':
3299 case '~':
3300 q = name;
3301 parse_name:
3302 for(;;) {
3303 if (!((ch >= 'a' && ch <= 'z') ||
3304 (ch >= 'A' && ch <= 'Z') ||
3305 (ch >= '0' && ch <= '9') ||
3306 strchr("/.-_+=$:\\,~", ch)))
3307 break;
3308 if ((q - name) < name_size - 1) {
3309 *q++ = ch;
3310 }
3311 minp();
3312 }
3313 *q = '\0';
3314 c = LD_TOK_NAME;
3315 break;
3316 case CH_EOF:
3317 c = LD_TOK_EOF;
3318 break;
3319 default:
3320 c = ch;
3321 inp();
3322 break;
3323 }
3324 return c;
3325 }
3326
3327 static int ld_add_file(TCCState *s1, const char filename[])
3328 {
3329 int ret;
3330
3331 ret = tcc_add_file_internal(s1, filename, 0);
3332 if (ret)
3333 ret = tcc_add_dll(s1, filename, 0);
3334 return ret;
3335 }
3336
3337 static inline int new_undef_syms(void)
3338 {
3339 int ret = 0;
3340 ret = new_undef_sym;
3341 new_undef_sym = 0;
3342 return ret;
3343 }
3344
3345 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3346 {
3347 char filename[1024], libname[1024];
3348 int t, group, nblibs = 0, ret = 0;
3349 char **libs = NULL;
3350
3351 group = !strcmp(cmd, "GROUP");
3352 if (!as_needed)
3353 new_undef_syms();
3354 t = ld_next(s1, filename, sizeof(filename));
3355 if (t != '(')
3356 expect("(");
3357 t = ld_next(s1, filename, sizeof(filename));
3358 for(;;) {
3359 libname[0] = '\0';
3360 if (t == LD_TOK_EOF) {
3361 tcc_error_noabort("unexpected end of file");
3362 ret = -1;
3363 goto lib_parse_error;
3364 } else if (t == ')') {
3365 break;
3366 } else if (t == '-') {
3367 t = ld_next(s1, filename, sizeof(filename));
3368 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3369 tcc_error_noabort("library name expected");
3370 ret = -1;
3371 goto lib_parse_error;
3372 }
3373 pstrcpy(libname, sizeof libname, &filename[1]);
3374 if (s1->static_link) {
3375 snprintf(filename, sizeof filename, "lib%s.a", libname);
3376 } else {
3377 snprintf(filename, sizeof filename, "lib%s.so", libname);
3378 }
3379 } else if (t != LD_TOK_NAME) {
3380 tcc_error_noabort("filename expected");
3381 ret = -1;
3382 goto lib_parse_error;
3383 }
3384 if (!strcmp(filename, "AS_NEEDED")) {
3385 ret = ld_add_file_list(s1, cmd, 1);
3386 if (ret)
3387 goto lib_parse_error;
3388 } else {
3389 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3390 if (!as_needed) {
3391 ret = ld_add_file(s1, filename);
3392 if (ret)
3393 goto lib_parse_error;
3394 if (group) {
3395 /* Add the filename *and* the libname to avoid future conversions */
3396 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3397 if (libname[0] != '\0')
3398 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3399 }
3400 }
3401 }
3402 t = ld_next(s1, filename, sizeof(filename));
3403 if (t == ',') {
3404 t = ld_next(s1, filename, sizeof(filename));
3405 }
3406 }
3407 if (group && !as_needed) {
3408 while (new_undef_syms()) {
3409 int i;
3410
3411 for (i = 0; i < nblibs; i ++)
3412 ld_add_file(s1, libs[i]);
3413 }
3414 }
3415 lib_parse_error:
3416 dynarray_reset(&libs, &nblibs);
3417 return ret;
3418 }
3419
3420 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3421 files */
3422 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3423 {
3424 char cmd[64];
3425 char filename[1024];
3426 int t, ret;
3427
3428 ch = file->buf_ptr[0];
3429 ch = handle_eob();
3430 for(;;) {
3431 t = ld_next(s1, cmd, sizeof(cmd));
3432 if (t == LD_TOK_EOF)
3433 return 0;
3434 else if (t != LD_TOK_NAME)
3435 return -1;
3436 if (!strcmp(cmd, "INPUT") ||
3437 !strcmp(cmd, "GROUP")) {
3438 ret = ld_add_file_list(s1, cmd, 0);
3439 if (ret)
3440 return ret;
3441 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3442 !strcmp(cmd, "TARGET")) {
3443 /* ignore some commands */
3444 t = ld_next(s1, cmd, sizeof(cmd));
3445 if (t != '(')
3446 expect("(");
3447 for(;;) {
3448 t = ld_next(s1, filename, sizeof(filename));
3449 if (t == LD_TOK_EOF) {
3450 tcc_error_noabort("unexpected end of file");
3451 return -1;
3452 } else if (t == ')') {
3453 break;
3454 }
3455 }
3456 } else {
3457 return -1;
3458 }
3459 }
3460 return 0;
3461 }
3462 #endif /* !TCC_TARGET_PE */
Tiny C Compiler main repository