?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
a bounds checking code for the ARCH=x86_64
[tinycc.git] / tccelf.c
blob98815e53cd4e9b7eed340e1621ef638ac24286d7
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, addr_t 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 < -2147483648LL || diff > 2147483647LL) {
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 static int tcc_add_support(TCCState *s1, const char *filename)
1544 {
1545 char buf[1024];
1546 snprintf(buf, sizeof(buf), "%s/%s/%s", s1->tcc_lib_path,
1547 /* an cpu specific path inside tcc_lib_path, mainly for keeping libtcc1.a */
1548 #ifdef TCC_TARGET_I386
1549 "i386"
1550 #endif
1551 #ifdef TCC_TARGET_X86_64
1552 "x86-64"
1553 #endif
1554 #ifdef TCC_TARGET_ARM
1555 "arm"
1556 #endif
1557 #ifdef TCC_TARGET_ARM64
1558 "arm64"
1559 #endif
1560 #ifdef TCC_TARGET_C67
1561 "C67"
1562 #endif
1563 ,filename);
1564
1565 return tcc_add_file(s1, buf);
1566 }
1567
1568 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1569 {
1570 #ifdef CONFIG_TCC_BCHECK
1571 addr_t *ptr;
1572
1573 if (0 == s1->do_bounds_check)
1574 return;
1575
1576 /* XXX: add an object file to do that */
1577 ptr = section_ptr_add(bounds_section, sizeof(*ptr));
1578 *ptr = 0;
1579 add_elf_sym(symtab_section, 0, 0,
1580 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1581 bounds_section->sh_num, "__bounds_start");
1582 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1583 /* add 'call __bound_init()' in .init section */
1584
1585 /* XXX not called on MSYS, reason is unknown. For this
1586 case a call to __bound_init is performed in bcheck.c
1587 when __bound_ptr_add, __bound_new_region,
1588 __bound_delete_region called */
1589
1590 int sym_index = find_elf_sym(symtab_section, "__bound_init");
1591 if (sym_index) {
1592 Section *init_section = find_section(s1, ".init");
1593 unsigned char *pinit = section_ptr_add(init_section, 5);
1594 pinit[0] = 0xe8;
1595 put32(pinit + 1, -4);
1596 put_elf_reloc(symtab_section, init_section,
1597 init_section->data_offset - 4, R_386_PC32, sym_index);
1598 }
1599 else
1600 tcc_warning("__bound_init not defined");
1601 }
1602 #endif
1603 }
1604
1605 /* add tcc runtime libraries */
1606 ST_FUNC void tcc_add_runtime(TCCState *s1)
1607 {
1608 /* add libc */
1609 if (!s1->nostdlib) {
1610 tcc_add_library(s1, "c");
1611 #ifdef CONFIG_USE_LIBGCC
1612 if (!s1->static_link) {
1613 tcc_add_file(s1, TCC_LIBGCC);
1614 }
1615 #endif
1616 tcc_add_support(s1, "libtcc1.a");
1617 }
1618
1619 /* tcc_add_bcheck tries to relocate a call to __bound_init in _init so
1620 libtcc1.a must be loaded before for __bound_init to be defined and
1621 crtn.o must be loaded after to not finalize _init too early. */
1622 tcc_add_bcheck(s1);
1623
1624 if (!s1->nostdlib) {
1625 /* add crt end if not memory output */
1626 if (s1->output_type != TCC_OUTPUT_MEMORY)
1627 tcc_add_crt(s1, "crtn.o");
1628 }
1629 }
1630
1631 /* add various standard linker symbols (must be done after the
1632 sections are filled (for example after allocating common
1633 symbols)) */
1634 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1635 {
1636 char buf[1024];
1637 int i;
1638 Section *s;
1639
1640 add_elf_sym(symtab_section,
1641 text_section->data_offset, 0,
1642 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1643 text_section->sh_num, "_etext");
1644 add_elf_sym(symtab_section,
1645 data_section->data_offset, 0,
1646 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1647 data_section->sh_num, "_edata");
1648 add_elf_sym(symtab_section,
1649 bss_section->data_offset, 0,
1650 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1651 bss_section->sh_num, "_end");
1652 /* horrible new standard ldscript defines */
1653 add_init_array_defines(s1, ".preinit_array");
1654 add_init_array_defines(s1, ".init_array");
1655 add_init_array_defines(s1, ".fini_array");
1656
1657 /* add start and stop symbols for sections whose name can be
1658 expressed in C */
1659 for(i = 1; i < s1->nb_sections; i++) {
1660 s = s1->sections[i];
1661 if (s->sh_type == SHT_PROGBITS &&
1662 (s->sh_flags & SHF_ALLOC)) {
1663 const char *p;
1664 int ch;
1665
1666 /* check if section name can be expressed in C */
1667 p = s->name;
1668 for(;;) {
1669 ch = *p;
1670 if (!ch)
1671 break;
1672 if (!isid(ch) && !isnum(ch))
1673 goto next_sec;
1674 p++;
1675 }
1676 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1677 add_elf_sym(symtab_section,
1678 0, 0,
1679 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1680 s->sh_num, buf);
1681 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1682 add_elf_sym(symtab_section,
1683 s->data_offset, 0,
1684 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1685 s->sh_num, buf);
1686 }
1687 next_sec: ;
1688 }
1689 }
1690
1691 static void tcc_output_binary(TCCState *s1, FILE *f,
1692 const int *sec_order)
1693 {
1694 Section *s;
1695 int i, offset, size;
1696
1697 offset = 0;
1698 for(i=1;i<s1->nb_sections;i++) {
1699 s = s1->sections[sec_order[i]];
1700 if (s->sh_type != SHT_NOBITS &&
1701 (s->sh_flags & SHF_ALLOC)) {
1702 while (offset < s->sh_offset) {
1703 fputc(0, f);
1704 offset++;
1705 }
1706 size = s->sh_size;
1707 fwrite(s->data, 1, size, f);
1708 offset += size;
1709 }
1710 }
1711 }
1712
1713 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1714 #define HAVE_PHDR 1
1715 #define EXTRA_RELITEMS 14
1716
1717 /* move the relocation value from .dynsym to .got */
1718 void patch_dynsym_undef(TCCState *s1, Section *s)
1719 {
1720 uint32_t *gotd = (void *)s1->got->data;
1721 ElfW(Sym) *sym;
1722
1723 gotd += 3; /* dummy entries in .got */
1724 /* relocate symbols in .dynsym */
1725 for_each_elem(s, 1, sym, ElfW(Sym)) {
1726 if (sym->st_shndx == SHN_UNDEF) {
1727 *gotd++ = sym->st_value + 6; /* XXX 6 is magic ? */
1728 sym->st_value = 0;
1729 }
1730 }
1731 }
1732 #else
1733 #define HAVE_PHDR 1
1734 #define EXTRA_RELITEMS 9
1735
1736 /* zero plt offsets of weak symbols in .dynsym */
1737 void patch_dynsym_undef(TCCState *s1, Section *s)
1738 {
1739 ElfW(Sym) *sym;
1740
1741 for_each_elem(s, 1, sym, ElfW(Sym))
1742 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1743 sym->st_value = 0;
1744 }
1745 #endif
1746
1747 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1748 {
1749 int sym_index = ELFW(R_SYM) (rel->r_info);
1750 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1751 unsigned long offset;
1752
1753 if (sym_index >= s1->nb_sym_attrs)
1754 return;
1755 offset = s1->sym_attrs[sym_index].got_offset;
1756 section_reserve(s1->got, offset + PTR_SIZE);
1757 #ifdef TCC_TARGET_X86_64
1758 /* only works for x86-64 */
1759 put32(s1->got->data + offset + 4, sym->st_value >> 32);
1760 #endif
1761 put32(s1->got->data + offset, sym->st_value & 0xffffffff);
1762 }
1763
1764 /* Perform relocation to GOT or PLT entries */
1765 ST_FUNC void fill_got(TCCState *s1)
1766 {
1767 Section *s;
1768 ElfW_Rel *rel;
1769 int i;
1770
1771 for(i = 1; i < s1->nb_sections; i++) {
1772 s = s1->sections[i];
1773 if (s->sh_type != SHT_RELX)
1774 continue;
1775 /* no need to handle got relocations */
1776 if (s->link != symtab_section)
1777 continue;
1778 for_each_elem(s, 0, rel, ElfW_Rel) {
1779 switch (ELFW(R_TYPE) (rel->r_info)) {
1780 case R_X86_64_GOT32:
1781 case R_X86_64_GOTPCREL:
1782 case R_X86_64_PLT32:
1783 fill_got_entry(s1, rel);
1784 break;
1785 }
1786 }
1787 }
1788 }
1789
1790 /* Bind symbols of executable: resolve undefined symbols from exported symbols
1791 in shared libraries and export non local defined symbols to shared libraries
1792 if -rdynamic switch was given on command line */
1793 static void bind_exe_dynsyms(TCCState *s1)
1794 {
1795 const char *name;
1796 int sym_index, index;
1797 ElfW(Sym) *sym, *esym;
1798 int type;
1799
1800 /* Resolve undefined symbols from dynamic symbols. When there is a match:
1801 - if STT_FUNC or STT_GNU_IFUNC symbol -> add it in PLT
1802 - if STT_OBJECT symbol -> add it in .bss section with suitable reloc */
1803 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1804 if (sym->st_shndx == SHN_UNDEF) {
1805 name = (char *) symtab_section->link->data + sym->st_name;
1806 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1807 if (sym_index) {
1808 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1809 type = ELFW(ST_TYPE)(esym->st_info);
1810 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1811 /* Indirect functions shall have STT_FUNC type in executable
1812 * dynsym section. Indeed, a dlsym call following a lazy
1813 * resolution would pick the symbol value from the
1814 * executable dynsym entry which would contain the address
1815 * of the function wanted by the caller of dlsym instead of
1816 * the address of the function that would return that
1817 * address */
1818 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1819 ELFW(ST_INFO)(STB_GLOBAL,STT_FUNC),
1820 sym - (ElfW(Sym) *)symtab_section->data);
1821 } else if (type == STT_OBJECT) {
1822 unsigned long offset;
1823 ElfW(Sym) *dynsym;
1824 offset = bss_section->data_offset;
1825 /* XXX: which alignment ? */
1826 offset = (offset + 16 - 1) & -16;
1827 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1828 esym->st_info, 0, bss_section->sh_num,
1829 name);
1830 /* Ensure R_COPY works for weak symbol aliases */
1831 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1832 for_each_elem(s1->dynsymtab_section, 1, dynsym, ElfW(Sym)) {
1833 if ((dynsym->st_value == esym->st_value)
1834 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1835 char *dynname = (char *) s1->dynsymtab_section->link->data
1836 + dynsym->st_name;
1837 put_elf_sym(s1->dynsym, offset, dynsym->st_size,
1838 dynsym->st_info, 0,
1839 bss_section->sh_num, dynname);
1840 break;
1841 }
1842 }
1843 }
1844 put_elf_reloc(s1->dynsym, bss_section,
1845 offset, R_COPY, index);
1846 offset += esym->st_size;
1847 bss_section->data_offset = offset;
1848 }
1849 } else {
1850 /* STB_WEAK undefined symbols are accepted */
1851 /* XXX: _fp_hw seems to be part of the ABI, so we ignore it */
1852 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1853 !strcmp(name, "_fp_hw")) {
1854 } else {
1855 tcc_error_noabort("undefined symbol '%s'", name);
1856 }
1857 }
1858 } else if (s1->rdynamic && ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1859 /* if -rdynamic option, then export all non local symbols */
1860 name = (char *) symtab_section->link->data + sym->st_name;
1861 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size, sym->st_info,
1862 0, sym->st_shndx, name);
1863 }
1864 }
1865 }
1866
1867 /* Bind symbols of libraries: export non local symbols of executable that
1868 resolve undefined symbols of shared libraries */
1869 static void bind_libs_dynsyms(TCCState *s1)
1870 {
1871 const char *name;
1872 int sym_index;
1873 ElfW(Sym) *sym, *esym;
1874
1875 /* now look at unresolved dynamic symbols and export
1876 corresponding symbol */
1877 for_each_elem(s1->dynsymtab_section, 1, esym, ElfW(Sym)) {
1878 if (esym->st_shndx == SHN_UNDEF) {
1879 name = (char *) s1->dynsymtab_section->link->data + esym->st_name;
1880 sym_index = find_elf_sym(symtab_section, name);
1881 if (sym_index) {
1882 /* XXX: avoid adding a symbol if already present because of
1883 -rdynamic ? */
1884 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1885 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1886 sym->st_info, 0, sym->st_shndx, name);
1887 } else {
1888 /* weak symbols can stay undefined */
1889 if (ELFW(ST_BIND)(esym->st_info) != STB_WEAK)
1890 tcc_warning("undefined dynamic symbol '%s'", name);
1891 }
1892 }
1893 }
1894 }
1895
1896 /* Export all non local symbols (for shared libraries) */
1897 static void export_global_syms(TCCState *s1)
1898 {
1899 int nb_syms, dynindex, index;
1900 const char *name;
1901 ElfW(Sym) *sym;
1902
1903 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1904 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1905 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1906 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1907 name = (char *) symtab_section->link->data + sym->st_name;
1908 dynindex = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1909 sym->st_info, 0, sym->st_shndx, name);
1910 index = sym - (ElfW(Sym) *) symtab_section->data;
1911 s1->symtab_to_dynsym[index] = dynindex;
1912 }
1913 }
1914 }
1915
1916 /* relocate the PLT: compute addresses and offsets in the PLT now that final
1917 address for PLT and GOT are known (see fill_program_header) */
1918 ST_FUNC void relocate_plt(TCCState *s1)
1919 {
1920 uint8_t *p, *p_end;
1921
1922 if (!s1->plt)
1923 return;
1924
1925 p = s1->plt->data;
1926 p_end = p + s1->plt->data_offset;
1927 if (p < p_end) {
1928 #if defined(TCC_TARGET_I386)
1929 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1930 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
1931 p += 16;
1932 while (p < p_end) {
1933 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1934 p += 16;
1935 }
1936 #elif defined(TCC_TARGET_X86_64)
1937 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
1938 put32(p + 2, get32(p + 2) + x);
1939 put32(p + 8, get32(p + 8) + x - 6);
1940 p += 16;
1941 while (p < p_end) {
1942 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
1943 p += 16;
1944 }
1945 #elif defined(TCC_TARGET_ARM)
1946 int x;
1947 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
1948 p += 16;
1949 while (p < p_end) {
1950 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
1951 p += 4;
1952 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
1953 p += 16;
1954 }
1955 #elif defined(TCC_TARGET_ARM64)
1956 uint64_t plt = s1->plt->sh_addr;
1957 uint64_t got = s1->got->sh_addr;
1958 uint64_t off = (got >> 12) - (plt >> 12);
1959 if ((off + ((uint32_t)1 << 20)) >> 21)
1960 tcc_error("Failed relocating PLT (off=0x%lx, got=0x%lx, plt=0x%lx)", off, got, plt);
1961 put32(p, 0xa9bf7bf0); // stp x16,x30,[sp,#-16]!
1962 put32(p + 4, (0x90000010 | // adrp x16,...
1963 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1964 put32(p + 8, (0xf9400211 | // ldr x17,[x16,#...]
1965 (got & 0xff8) << 7));
1966 put32(p + 12, (0x91000210 | // add x16,x16,#...
1967 (got & 0xfff) << 10));
1968 put32(p + 16, 0xd61f0220); // br x17
1969 put32(p + 20, 0xd503201f); // nop
1970 put32(p + 24, 0xd503201f); // nop
1971 put32(p + 28, 0xd503201f); // nop
1972 p += 32;
1973 while (p < p_end) {
1974 uint64_t pc = plt + (p - s1->plt->data);
1975 uint64_t addr = got +
1976 (get32(p) | (uint64_t)get32(p + 4) << 32);
1977 uint32_t off = (addr >> 12) - (pc >> 12);
1978 if ((off + ((uint32_t)1 << 20)) >> 21)
1979 tcc_error("Failed relocating PLT (off=0x%lx, addr=0x%lx, pc=0x%lx)", off, addr, pc);
1980 put32(p, (0x90000010 | // adrp x16,...
1981 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1982 put32(p + 4, (0xf9400211 | // ldr x17,[x16,#...]
1983 (addr & 0xff8) << 7));
1984 put32(p + 8, (0x91000210 | // add x16,x16,#...
1985 (addr & 0xfff) << 10));
1986 put32(p + 12, 0xd61f0220); // br x17
1987 p += 16;
1988 }
1989 #elif defined(TCC_TARGET_C67)
1990 /* XXX: TODO */
1991 #else
1992 #error unsupported CPU
1993 #endif
1994 }
1995 }
1996
1997 /* Allocate strings for section names and decide if an unallocated section
1998 should be output.
1999
2000 NOTE: the strsec section comes last, so its size is also correct ! */
2001 static void alloc_sec_names(TCCState *s1, int file_type, Section *strsec)
2002 {
2003 int i;
2004 Section *s;
2005
2006 /* Allocate strings for section names */
2007 for(i = 1; i < s1->nb_sections; i++) {
2008 s = s1->sections[i];
2009 s->sh_name = put_elf_str(strsec, s->name);
2010 /* when generating a DLL, we include relocations but we may
2011 patch them */
2012 if (file_type == TCC_OUTPUT_DLL &&
2013 s->sh_type == SHT_RELX &&
2014 !(s->sh_flags & SHF_ALLOC)) {
2015 /* gr: avoid bogus relocs for empty (debug) sections */
2016 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
2017 prepare_dynamic_rel(s1, s);
2018 else if (s1->do_debug)
2019 s->sh_size = s->data_offset;
2020 } else if (s1->do_debug ||
2021 file_type == TCC_OUTPUT_OBJ ||
2022 file_type == TCC_OUTPUT_EXE ||
2023 (s->sh_flags & SHF_ALLOC) ||
2024 i == (s1->nb_sections - 1)) {
2025 /* we output all sections if debug or object file */
2026 s->sh_size = s->data_offset;
2027 }
2028 }
2029 }
2030
2031 /* Info to be copied in dynamic section */
2032 struct dyn_inf {
2033 Section *dynamic;
2034 Section *dynstr;
2035 unsigned long dyn_rel_off;
2036 addr_t rel_addr;
2037 addr_t rel_size;
2038 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2039 addr_t bss_addr;
2040 addr_t bss_size;
2041 #endif
2042 };
2043
2044 /* Assign sections to segments and decide how are sections laid out when loaded
2045 in memory. This function also fills corresponding program headers. */
2046 static int layout_sections(TCCState *s1, ElfW(Phdr) *phdr, int phnum,
2047 Section *interp, Section* strsec,
2048 struct dyn_inf *dyninf, int *sec_order)
2049 {
2050 int i, j, k, file_type, sh_order_index, file_offset;
2051 unsigned long s_align;
2052 long long tmp;
2053 addr_t addr;
2054 ElfW(Phdr) *ph;
2055 Section *s;
2056
2057 file_type = s1->output_type;
2058 sh_order_index = 1;
2059 file_offset = 0;
2060 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2061 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2062 s_align = ELF_PAGE_SIZE;
2063 if (s1->section_align)
2064 s_align = s1->section_align;
2065
2066 if (phnum > 0) {
2067 if (s1->has_text_addr) {
2068 int a_offset, p_offset;
2069 addr = s1->text_addr;
2070 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
2071 ELF_PAGE_SIZE */
2072 a_offset = (int) (addr & (s_align - 1));
2073 p_offset = file_offset & (s_align - 1);
2074 if (a_offset < p_offset)
2075 a_offset += s_align;
2076 file_offset += (a_offset - p_offset);
2077 } else {
2078 if (file_type == TCC_OUTPUT_DLL)
2079 addr = 0;
2080 else
2081 addr = ELF_START_ADDR;
2082 /* compute address after headers */
2083 addr += (file_offset & (s_align - 1));
2084 }
2085
2086 ph = &phdr[0];
2087 /* Leave one program headers for the program interpreter and one for
2088 the program header table itself if needed. These are done later as
2089 they require section layout to be done first. */
2090 if (interp)
2091 ph += 1 + HAVE_PHDR;
2092
2093 /* dynamic relocation table information, for .dynamic section */
2094 dyninf->rel_addr = dyninf->rel_size = 0;
2095 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2096 dyninf->bss_addr = dyninf->bss_size = 0;
2097 #endif
2098
2099 for(j = 0; j < 2; j++) {
2100 ph->p_type = PT_LOAD;
2101 if (j == 0)
2102 ph->p_flags = PF_R | PF_X;
2103 else
2104 ph->p_flags = PF_R | PF_W;
2105 ph->p_align = s_align;
2106
2107 /* Decide the layout of sections loaded in memory. This must
2108 be done before program headers are filled since they contain
2109 info about the layout. We do the following ordering: interp,
2110 symbol tables, relocations, progbits, nobits */
2111 /* XXX: do faster and simpler sorting */
2112 for(k = 0; k < 5; k++) {
2113 for(i = 1; i < s1->nb_sections; i++) {
2114 s = s1->sections[i];
2115 /* compute if section should be included */
2116 if (j == 0) {
2117 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2118 SHF_ALLOC)
2119 continue;
2120 } else {
2121 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2122 (SHF_ALLOC | SHF_WRITE))
2123 continue;
2124 }
2125 if (s == interp) {
2126 if (k != 0)
2127 continue;
2128 } else if (s->sh_type == SHT_DYNSYM ||
2129 s->sh_type == SHT_STRTAB ||
2130 s->sh_type == SHT_HASH) {
2131 if (k != 1)
2132 continue;
2133 } else if (s->sh_type == SHT_RELX) {
2134 if (k != 2)
2135 continue;
2136 } else if (s->sh_type == SHT_NOBITS) {
2137 if (k != 4)
2138 continue;
2139 } else {
2140 if (k != 3)
2141 continue;
2142 }
2143 sec_order[sh_order_index++] = i;
2144
2145 /* section matches: we align it and add its size */
2146 tmp = addr;
2147 addr = (addr + s->sh_addralign - 1) &
2148 ~(s->sh_addralign - 1);
2149 file_offset += (int) ( addr - tmp );
2150 s->sh_offset = file_offset;
2151 s->sh_addr = addr;
2152
2153 /* update program header infos */
2154 if (ph->p_offset == 0) {
2155 ph->p_offset = file_offset;
2156 ph->p_vaddr = addr;
2157 ph->p_paddr = ph->p_vaddr;
2158 }
2159 /* update dynamic relocation infos */
2160 if (s->sh_type == SHT_RELX) {
2161 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2162 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) {
2163 dyninf->rel_addr = addr;
2164 dyninf->rel_size += s->sh_size; /* XXX only first rel. */
2165 }
2166 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) {
2167 dyninf->bss_addr = addr;
2168 dyninf->bss_size = s->sh_size; /* XXX only first rel. */
2169 }
2170 #else
2171 if (dyninf->rel_size == 0)
2172 dyninf->rel_addr = addr;
2173 dyninf->rel_size += s->sh_size;
2174 #endif
2175 }
2176 addr += s->sh_size;
2177 if (s->sh_type != SHT_NOBITS)
2178 file_offset += s->sh_size;
2179 }
2180 }
2181 if (j == 0) {
2182 /* Make the first PT_LOAD segment include the program
2183 headers itself (and the ELF header as well), it'll
2184 come out with same memory use but will make various
2185 tools like binutils strip work better. */
2186 ph->p_offset &= ~(ph->p_align - 1);
2187 ph->p_vaddr &= ~(ph->p_align - 1);
2188 ph->p_paddr &= ~(ph->p_align - 1);
2189 }
2190 ph->p_filesz = file_offset - ph->p_offset;
2191 ph->p_memsz = addr - ph->p_vaddr;
2192 ph++;
2193 if (j == 0) {
2194 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
2195 /* if in the middle of a page, we duplicate the page in
2196 memory so that one copy is RX and the other is RW */
2197 if ((addr & (s_align - 1)) != 0)
2198 addr += s_align;
2199 } else {
2200 addr = (addr + s_align - 1) & ~(s_align - 1);
2201 file_offset = (file_offset + s_align - 1) & ~(s_align - 1);
2202 }
2203 }
2204 }
2205 }
2206
2207 /* all other sections come after */
2208 for(i = 1; i < s1->nb_sections; i++) {
2209 s = s1->sections[i];
2210 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
2211 continue;
2212 sec_order[sh_order_index++] = i;
2213
2214 file_offset = (file_offset + s->sh_addralign - 1) &
2215 ~(s->sh_addralign - 1);
2216 s->sh_offset = file_offset;
2217 if (s->sh_type != SHT_NOBITS)
2218 file_offset += s->sh_size;
2219 }
2220
2221 return file_offset;
2222 }
2223
2224 static void fill_unloadable_phdr(ElfW(Phdr) *phdr, int phnum, Section *interp,
2225 Section *dynamic)
2226 {
2227 ElfW(Phdr) *ph;
2228
2229 /* if interpreter, then add corresponding program header */
2230 if (interp) {
2231 ph = &phdr[0];
2232
2233 if (HAVE_PHDR)
2234 {
2235 int len = phnum * sizeof(ElfW(Phdr));
2236
2237 ph->p_type = PT_PHDR;
2238 ph->p_offset = sizeof(ElfW(Ehdr));
2239 ph->p_vaddr = interp->sh_addr - len;
2240 ph->p_paddr = ph->p_vaddr;
2241 ph->p_filesz = ph->p_memsz = len;
2242 ph->p_flags = PF_R | PF_X;
2243 ph->p_align = 4; /* interp->sh_addralign; */
2244 ph++;
2245 }
2246
2247 ph->p_type = PT_INTERP;
2248 ph->p_offset = interp->sh_offset;
2249 ph->p_vaddr = interp->sh_addr;
2250 ph->p_paddr = ph->p_vaddr;
2251 ph->p_filesz = interp->sh_size;
2252 ph->p_memsz = interp->sh_size;
2253 ph->p_flags = PF_R;
2254 ph->p_align = interp->sh_addralign;
2255 }
2256
2257 /* if dynamic section, then add corresponding program header */
2258 if (dynamic) {
2259 ph = &phdr[phnum - 1];
2260
2261 ph->p_type = PT_DYNAMIC;
2262 ph->p_offset = dynamic->sh_offset;
2263 ph->p_vaddr = dynamic->sh_addr;
2264 ph->p_paddr = ph->p_vaddr;
2265 ph->p_filesz = dynamic->sh_size;
2266 ph->p_memsz = dynamic->sh_size;
2267 ph->p_flags = PF_R | PF_W;
2268 ph->p_align = dynamic->sh_addralign;
2269 }
2270 }
2271
2272 /* Fill the dynamic section with tags describing the address and size of
2273 sections */
2274 static void fill_dynamic(TCCState *s1, struct dyn_inf *dyninf)
2275 {
2276 Section *dynamic;
2277
2278 dynamic = dyninf->dynamic;
2279
2280 /* put dynamic section entries */
2281 dynamic->data_offset = dyninf->dyn_rel_off;
2282 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2283 put_dt(dynamic, DT_STRTAB, dyninf->dynstr->sh_addr);
2284 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2285 put_dt(dynamic, DT_STRSZ, dyninf->dynstr->data_offset);
2286 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2287 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
2288 put_dt(dynamic, DT_RELA, dyninf->rel_addr);
2289 put_dt(dynamic, DT_RELASZ, dyninf->rel_size);
2290 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2291 #else
2292 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2293 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2294 put_dt(dynamic, DT_PLTRELSZ, dyninf->rel_size);
2295 put_dt(dynamic, DT_JMPREL, dyninf->rel_addr);
2296 put_dt(dynamic, DT_PLTREL, DT_REL);
2297 put_dt(dynamic, DT_REL, dyninf->bss_addr);
2298 put_dt(dynamic, DT_RELSZ, dyninf->bss_size);
2299 #else
2300 put_dt(dynamic, DT_REL, dyninf->rel_addr);
2301 put_dt(dynamic, DT_RELSZ, dyninf->rel_size);
2302 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2303 #endif
2304 #endif
2305 if (s1->do_debug)
2306 put_dt(dynamic, DT_DEBUG, 0);
2307 put_dt(dynamic, DT_NULL, 0);
2308 }
2309
2310 /* Relocate remaining sections and symbols (that is those not related to
2311 dynamic linking) */
2312 static int final_sections_reloc(TCCState *s1)
2313 {
2314 int i;
2315 Section *s;
2316
2317 relocate_syms(s1, 0);
2318
2319 if (s1->nb_errors != 0)
2320 return -1;
2321
2322 /* relocate sections */
2323 /* XXX: ignore sections with allocated relocations ? */
2324 for(i = 1; i < s1->nb_sections; i++) {
2325 s = s1->sections[i];
2326 #ifdef TCC_TARGET_I386
2327 if (s->reloc && s != s1->got && (s->sh_flags & SHF_ALLOC)) //gr
2328 /* On X86 gdb 7.3 works in any case but gdb 6.6 will crash if SHF_ALLOC
2329 checking is removed */
2330 #else
2331 if (s->reloc && s != s1->got)
2332 /* On X86_64 gdb 7.3 will crash if SHF_ALLOC checking is present */
2333 #endif
2334 relocate_section(s1, s);
2335 }
2336
2337 /* relocate relocation entries if the relocation tables are
2338 allocated in the executable */
2339 for(i = 1; i < s1->nb_sections; i++) {
2340 s = s1->sections[i];
2341 if ((s->sh_flags & SHF_ALLOC) &&
2342 s->sh_type == SHT_RELX) {
2343 relocate_rel(s1, s);
2344 }
2345 }
2346 return 0;
2347 }
2348
2349 /* Create an ELF file on disk.
2350 This function handle ELF specific layout requirements */
2351 static void tcc_output_elf(TCCState *s1, FILE *f, int phnum, ElfW(Phdr) *phdr,
2352 int file_offset, int *sec_order)
2353 {
2354 int i, shnum, offset, size, file_type;
2355 Section *s;
2356 ElfW(Ehdr) ehdr;
2357 ElfW(Shdr) shdr, *sh;
2358
2359 file_type = s1->output_type;
2360 shnum = s1->nb_sections;
2361
2362 memset(&ehdr, 0, sizeof(ehdr));
2363
2364 if (phnum > 0) {
2365 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2366 ehdr.e_phnum = phnum;
2367 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2368 }
2369
2370 /* align to 4 */
2371 file_offset = (file_offset + 3) & -4;
2372
2373 /* fill header */
2374 ehdr.e_ident[0] = ELFMAG0;
2375 ehdr.e_ident[1] = ELFMAG1;
2376 ehdr.e_ident[2] = ELFMAG2;
2377 ehdr.e_ident[3] = ELFMAG3;
2378 ehdr.e_ident[4] = ELFCLASSW;
2379 ehdr.e_ident[5] = ELFDATA2LSB;
2380 ehdr.e_ident[6] = EV_CURRENT;
2381 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2382 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2383 #endif
2384 #ifdef TCC_TARGET_ARM
2385 #ifdef TCC_ARM_EABI
2386 ehdr.e_ident[EI_OSABI] = 0;
2387 ehdr.e_flags = EF_ARM_EABI_VER4;
2388 if (file_type == TCC_OUTPUT_EXE || file_type == TCC_OUTPUT_DLL)
2389 ehdr.e_flags |= EF_ARM_HASENTRY;
2390 if (s1->float_abi == ARM_HARD_FLOAT)
2391 ehdr.e_flags |= EF_ARM_VFP_FLOAT;
2392 else
2393 ehdr.e_flags |= EF_ARM_SOFT_FLOAT;
2394 #else
2395 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2396 #endif
2397 #endif
2398 switch(file_type) {
2399 default:
2400 case TCC_OUTPUT_EXE:
2401 ehdr.e_type = ET_EXEC;
2402 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2403 break;
2404 case TCC_OUTPUT_DLL:
2405 ehdr.e_type = ET_DYN;
2406 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2407 break;
2408 case TCC_OUTPUT_OBJ:
2409 ehdr.e_type = ET_REL;
2410 break;
2411 }
2412 ehdr.e_machine = EM_TCC_TARGET;
2413 ehdr.e_version = EV_CURRENT;
2414 ehdr.e_shoff = file_offset;
2415 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2416 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2417 ehdr.e_shnum = shnum;
2418 ehdr.e_shstrndx = shnum - 1;
2419
2420 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2421 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2422 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2423
2424 sort_syms(s1, symtab_section);
2425 for(i = 1; i < s1->nb_sections; i++) {
2426 s = s1->sections[sec_order[i]];
2427 if (s->sh_type != SHT_NOBITS) {
2428 if (s->sh_type == SHT_DYNSYM)
2429 patch_dynsym_undef(s1, s);
2430 while (offset < s->sh_offset) {
2431 fputc(0, f);
2432 offset++;
2433 }
2434 size = s->sh_size;
2435 fwrite(s->data, 1, size, f);
2436 offset += size;
2437 }
2438 }
2439
2440 /* output section headers */
2441 while (offset < ehdr.e_shoff) {
2442 fputc(0, f);
2443 offset++;
2444 }
2445
2446 for(i = 0; i < s1->nb_sections; i++) {
2447 sh = &shdr;
2448 memset(sh, 0, sizeof(ElfW(Shdr)));
2449 s = s1->sections[i];
2450 if (s) {
2451 sh->sh_name = s->sh_name;
2452 sh->sh_type = s->sh_type;
2453 sh->sh_flags = s->sh_flags;
2454 sh->sh_entsize = s->sh_entsize;
2455 sh->sh_info = s->sh_info;
2456 if (s->link)
2457 sh->sh_link = s->link->sh_num;
2458 sh->sh_addralign = s->sh_addralign;
2459 sh->sh_addr = s->sh_addr;
2460 sh->sh_offset = s->sh_offset;
2461 sh->sh_size = s->sh_size;
2462 }
2463 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2464 }
2465 }
2466
2467 /* Write an elf, coff or "binary" file */
2468 static int tcc_write_elf_file(TCCState *s1, const char *filename, int phnum,
2469 ElfW(Phdr) *phdr, int file_offset, int *sec_order)
2470 {
2471 int fd, mode, file_type;
2472 FILE *f;
2473
2474 file_type = s1->output_type;
2475 if (file_type == TCC_OUTPUT_OBJ)
2476 mode = 0666;
2477 else
2478 mode = 0777;
2479 unlink(filename);
2480 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2481 if (fd < 0) {
2482 tcc_error_noabort("could not write '%s'", filename);
2483 return -1;
2484 }
2485 f = fdopen(fd, "wb");
2486 if (s1->verbose)
2487 printf("<- %s\n", filename);
2488
2489 #ifdef TCC_TARGET_COFF
2490 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF)
2491 tcc_output_coff(s1, f);
2492 else
2493 #endif
2494 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2495 tcc_output_elf(s1, f, phnum, phdr, file_offset, sec_order);
2496 else
2497 tcc_output_binary(s1, f, sec_order);
2498 fclose(f);
2499
2500 return 0;
2501 }
2502
2503 /* Output an elf, coff or binary file */
2504 /* XXX: suppress unneeded sections */
2505 static int elf_output_file(TCCState *s1, const char *filename)
2506 {
2507 int i, ret, phnum, shnum, file_type, file_offset, *sec_order;
2508 struct dyn_inf dyninf;
2509 ElfW(Phdr) *phdr;
2510 ElfW(Sym) *sym;
2511 Section *strsec, *interp, *dynamic, *dynstr;
2512
2513 file_type = s1->output_type;
2514 s1->nb_errors = 0;
2515
2516 /* if linking, also link in runtime libraries (libc, libgcc, etc.) */
2517 if (file_type != TCC_OUTPUT_OBJ) {
2518 tcc_add_runtime(s1);
2519 }
2520
2521 phdr = NULL;
2522 sec_order = NULL;
2523 interp = dynamic = dynstr = NULL; /* avoid warning */
2524 dyninf.dyn_rel_off = 0; /* avoid warning */
2525
2526 if (file_type != TCC_OUTPUT_OBJ) {
2527 relocate_common_syms();
2528
2529 tcc_add_linker_symbols(s1);
2530
2531 if (!s1->static_link) {
2532 if (file_type == TCC_OUTPUT_EXE) {
2533 char *ptr;
2534 /* allow override the dynamic loader */
2535 const char *elfint = getenv("LD_SO");
2536 if (elfint == NULL)
2537 elfint = DEFAULT_ELFINTERP(s1);
2538 /* add interpreter section only if executable */
2539 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
2540 interp->sh_addralign = 1;
2541 ptr = section_ptr_add(interp, 1 + strlen(elfint));
2542 strcpy(ptr, elfint);
2543 }
2544
2545 /* add dynamic symbol table */
2546 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
2547 ".dynstr",
2548 ".hash", SHF_ALLOC);
2549 dynstr = s1->dynsym->link;
2550
2551 /* add dynamic section */
2552 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
2553 SHF_ALLOC | SHF_WRITE);
2554 dynamic->link = dynstr;
2555 dynamic->sh_entsize = sizeof(ElfW(Dyn));
2556
2557 build_got(s1);
2558
2559 if (file_type == TCC_OUTPUT_EXE) {
2560 bind_exe_dynsyms(s1);
2561
2562 if (s1->nb_errors) {
2563 ret = -1;
2564 goto the_end;
2565 }
2566
2567 bind_libs_dynsyms(s1);
2568 } else /* shared library case: simply export all global symbols */
2569 export_global_syms(s1);
2570
2571 build_got_entries(s1);
2572
2573 /* add a list of needed dlls */
2574 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2575 DLLReference *dllref = s1->loaded_dlls[i];
2576 if (dllref->level == 0)
2577 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
2578 }
2579
2580 if (s1->rpath)
2581 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
2582
2583 /* XXX: currently, since we do not handle PIC code, we
2584 must relocate the readonly segments */
2585 if (file_type == TCC_OUTPUT_DLL) {
2586 if (s1->soname)
2587 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
2588 put_dt(dynamic, DT_TEXTREL, 0);
2589 }
2590
2591 if (s1->symbolic)
2592 put_dt(dynamic, DT_SYMBOLIC, 0);
2593
2594 /* add necessary space for other entries */
2595 dyninf.dyn_rel_off = dynamic->data_offset;
2596 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
2597 } else {
2598 /* still need to build got entries in case of static link */
2599 build_got_entries(s1);
2600 }
2601 }
2602
2603 /* we add a section for symbols */
2604 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
2605 put_elf_str(strsec, "");
2606
2607 /* compute number of sections */
2608 shnum = s1->nb_sections;
2609
2610 /* this array is used to reorder sections in the output file */
2611 sec_order = tcc_malloc(sizeof(int) * shnum);
2612 sec_order[0] = 0;
2613
2614 /* compute number of program headers */
2615 switch(file_type) {
2616 default:
2617 case TCC_OUTPUT_OBJ:
2618 phnum = 0;
2619 break;
2620 case TCC_OUTPUT_EXE:
2621 if (!s1->static_link)
2622 phnum = 4 + HAVE_PHDR;
2623 else
2624 phnum = 2;
2625 break;
2626 case TCC_OUTPUT_DLL:
2627 phnum = 3;
2628 break;
2629 }
2630
2631 /* Allocate strings for section names */
2632 alloc_sec_names(s1, file_type, strsec);
2633
2634 /* allocate program segment headers */
2635 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
2636
2637 /* compute section to program header mapping */
2638 file_offset = layout_sections(s1, phdr, phnum, interp, strsec, &dyninf,
2639 sec_order);
2640
2641 /* Fill remaining program header and finalize relocation related to dynamic
2642 linking. */
2643 if (phnum > 0) {
2644 fill_unloadable_phdr(phdr, phnum, interp, dynamic);
2645 if (dynamic) {
2646 dyninf.dynamic = dynamic;
2647 dyninf.dynstr = dynstr;
2648
2649 fill_dynamic(s1, &dyninf);
2650
2651 /* put in GOT the dynamic section address and relocate PLT */
2652 put32(s1->got->data, dynamic->sh_addr);
2653 if (file_type == TCC_OUTPUT_EXE
2654 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2655 || file_type == TCC_OUTPUT_DLL
2656 #endif
2657 )
2658 relocate_plt(s1);
2659
2660 /* relocate symbols in .dynsym now that final addresses are known */
2661 for_each_elem(s1->dynsym, 1, sym, ElfW(Sym)) {
2662 if (sym->st_shndx == SHN_UNDEF) {
2663 /* relocate to PLT if symbol corresponds to a PLT entry,
2664 but not if it's a weak symbol */
2665 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
2666 sym->st_value = 0;
2667 else if (sym->st_value)
2668 sym->st_value += s1->plt->sh_addr;
2669 } else if (sym->st_shndx < SHN_LORESERVE) {
2670 /* do symbol relocation */
2671 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2672 }
2673 }
2674 }
2675 }
2676
2677 /* if building executable or DLL, then relocate each section
2678 except the GOT which is already relocated */
2679 if (file_type != TCC_OUTPUT_OBJ) {
2680 ret = final_sections_reloc(s1);
2681 if (ret)
2682 goto the_end;
2683 }
2684
2685 /* Perform relocation to GOT or PLT entries */
2686 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2687 fill_got(s1);
2688
2689 /* Create the ELF file with name 'filename' */
2690 ret = tcc_write_elf_file(s1, filename, phnum, phdr, file_offset, sec_order);
2691 if (s1->do_strip) {
2692 const char *strip = "sstrip "; // super strip utility from ELFkickers
2693 const char *null = " 2> /dev/null";
2694 int len = strlen(strip) + strlen(filename) + strlen(null) + 1;
2695 {
2696 int rc;
2697 char buf[len];
2698 sprintf(buf, "%s%s%s", strip, filename, null);
2699 rc = system(buf);
2700 if (rc) {
2701 system(buf+1); // call a strip utility from binutils
2702 }
2703 }
2704 }
2705 the_end:
2706 tcc_free(s1->symtab_to_dynsym);
2707 tcc_free(sec_order);
2708 tcc_free(phdr);
2709 tcc_free(s1->sym_attrs);
2710 s1->sym_attrs = NULL;
2711 return ret;
2712 }
2713
2714 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2715 {
2716 int ret;
2717 #ifdef TCC_TARGET_PE
2718 if (s->output_type != TCC_OUTPUT_OBJ) {
2719 ret = pe_output_file(s, filename);
2720 } else
2721 #endif
2722 ret = elf_output_file(s, filename);
2723 return ret;
2724 }
2725
2726 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2727 {
2728 void *data;
2729
2730 data = tcc_malloc(size);
2731 lseek(fd, file_offset, SEEK_SET);
2732 read(fd, data, size);
2733 return data;
2734 }
2735
2736 typedef struct SectionMergeInfo {
2737 Section *s; /* corresponding existing section */
2738 unsigned long offset; /* offset of the new section in the existing section */
2739 uint8_t new_section; /* true if section 's' was added */
2740 uint8_t link_once; /* true if link once section */
2741 } SectionMergeInfo;
2742
2743 /* load an object file and merge it with current files */
2744 /* XXX: handle correctly stab (debug) info */
2745 ST_FUNC int tcc_load_object_file(TCCState *s1,
2746 int fd, unsigned long file_offset)
2747 {
2748 ElfW(Ehdr) ehdr;
2749 ElfW(Shdr) *shdr, *sh;
2750 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2751 unsigned char *strsec, *strtab;
2752 int *old_to_new_syms;
2753 char *sh_name, *name;
2754 SectionMergeInfo *sm_table, *sm;
2755 ElfW(Sym) *sym, *symtab;
2756 ElfW_Rel *rel;
2757 Section *s;
2758
2759 int stab_index;
2760 int stabstr_index;
2761
2762 stab_index = stabstr_index = 0;
2763
2764 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2765 goto fail1;
2766 if (ehdr.e_ident[0] != ELFMAG0 ||
2767 ehdr.e_ident[1] != ELFMAG1 ||
2768 ehdr.e_ident[2] != ELFMAG2 ||
2769 ehdr.e_ident[3] != ELFMAG3)
2770 goto fail1;
2771 /* test if object file */
2772 if (ehdr.e_type != ET_REL)
2773 goto fail1;
2774 /* test CPU specific stuff */
2775 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2776 ehdr.e_machine != EM_TCC_TARGET) {
2777 fail1:
2778 tcc_error_noabort("invalid object file");
2779 return -1;
2780 }
2781 /* read sections */
2782 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2783 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2784 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2785
2786 /* load section names */
2787 sh = &shdr[ehdr.e_shstrndx];
2788 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2789
2790 /* load symtab and strtab */
2791 old_to_new_syms = NULL;
2792 symtab = NULL;
2793 strtab = NULL;
2794 nb_syms = 0;
2795 for(i = 1; i < ehdr.e_shnum; i++) {
2796 sh = &shdr[i];
2797 if (sh->sh_type == SHT_SYMTAB) {
2798 if (symtab) {
2799 tcc_error_noabort("object must contain only one symtab");
2800 fail:
2801 ret = -1;
2802 goto the_end;
2803 }
2804 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2805 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2806 sm_table[i].s = symtab_section;
2807
2808 /* now load strtab */
2809 sh = &shdr[sh->sh_link];
2810 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2811 }
2812 }
2813
2814 /* now examine each section and try to merge its content with the
2815 ones in memory */
2816 for(i = 1; i < ehdr.e_shnum; i++) {
2817 /* no need to examine section name strtab */
2818 if (i == ehdr.e_shstrndx)
2819 continue;
2820 sh = &shdr[i];
2821 sh_name = (char *) strsec + sh->sh_name;
2822 /* ignore sections types we do not handle */
2823 if (sh->sh_type != SHT_PROGBITS &&
2824 sh->sh_type != SHT_RELX &&
2825 #ifdef TCC_ARM_EABI
2826 sh->sh_type != SHT_ARM_EXIDX &&
2827 #endif
2828 sh->sh_type != SHT_NOBITS &&
2829 sh->sh_type != SHT_PREINIT_ARRAY &&
2830 sh->sh_type != SHT_INIT_ARRAY &&
2831 sh->sh_type != SHT_FINI_ARRAY &&
2832 strcmp(sh_name, ".stabstr")
2833 )
2834 continue;
2835 if (sh->sh_addralign < 1)
2836 sh->sh_addralign = 1;
2837 /* find corresponding section, if any */
2838 for(j = 1; j < s1->nb_sections;j++) {
2839 s = s1->sections[j];
2840 if (!strcmp(s->name, sh_name)) {
2841 if (!strncmp(sh_name, ".gnu.linkonce",
2842 sizeof(".gnu.linkonce") - 1)) {
2843 /* if a 'linkonce' section is already present, we
2844 do not add it again. It is a little tricky as
2845 symbols can still be defined in
2846 it. */
2847 sm_table[i].link_once = 1;
2848 goto next;
2849 } else {
2850 goto found;
2851 }
2852 }
2853 }
2854 /* not found: create new section */
2855 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2856 /* take as much info as possible from the section. sh_link and
2857 sh_info will be updated later */
2858 s->sh_addralign = sh->sh_addralign;
2859 s->sh_entsize = sh->sh_entsize;
2860 sm_table[i].new_section = 1;
2861 found:
2862 if (sh->sh_type != s->sh_type) {
2863 tcc_error_noabort("invalid section type");
2864 goto fail;
2865 }
2866
2867 /* align start of section */
2868 offset = s->data_offset;
2869
2870 if (0 == strcmp(sh_name, ".stab")) {
2871 stab_index = i;
2872 goto no_align;
2873 }
2874 if (0 == strcmp(sh_name, ".stabstr")) {
2875 stabstr_index = i;
2876 goto no_align;
2877 }
2878
2879 size = sh->sh_addralign - 1;
2880 offset = (offset + size) & ~size;
2881 if (sh->sh_addralign > s->sh_addralign)
2882 s->sh_addralign = sh->sh_addralign;
2883 s->data_offset = offset;
2884 no_align:
2885 sm_table[i].offset = offset;
2886 sm_table[i].s = s;
2887 /* concatenate sections */
2888 size = sh->sh_size;
2889 if (sh->sh_type != SHT_NOBITS) {
2890 unsigned char *ptr;
2891 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2892 ptr = section_ptr_add(s, size);
2893 read(fd, ptr, size);
2894 } else {
2895 s->data_offset += size;
2896 }
2897 next: ;
2898 }
2899
2900 /* gr relocate stab strings */
2901 if (stab_index && stabstr_index) {
2902 Stab_Sym *a, *b;
2903 unsigned o;
2904 s = sm_table[stab_index].s;
2905 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2906 b = (Stab_Sym *)(s->data + s->data_offset);
2907 o = sm_table[stabstr_index].offset;
2908 while (a < b)
2909 a->n_strx += o, a++;
2910 }
2911
2912 /* second short pass to update sh_link and sh_info fields of new
2913 sections */
2914 for(i = 1; i < ehdr.e_shnum; i++) {
2915 s = sm_table[i].s;
2916 if (!s || !sm_table[i].new_section)
2917 continue;
2918 sh = &shdr[i];
2919 if (sh->sh_link > 0)
2920 s->link = sm_table[sh->sh_link].s;
2921 if (sh->sh_type == SHT_RELX) {
2922 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2923 /* update backward link */
2924 s1->sections[s->sh_info]->reloc = s;
2925 }
2926 }
2927 sm = sm_table;
2928
2929 /* resolve symbols */
2930 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2931
2932 sym = symtab + 1;
2933 for(i = 1; i < nb_syms; i++, sym++) {
2934 if (sym->st_shndx != SHN_UNDEF &&
2935 sym->st_shndx < SHN_LORESERVE) {
2936 sm = &sm_table[sym->st_shndx];
2937 if (sm->link_once) {
2938 /* if a symbol is in a link once section, we use the
2939 already defined symbol. It is very important to get
2940 correct relocations */
2941 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2942 name = (char *) strtab + sym->st_name;
2943 sym_index = find_elf_sym(symtab_section, name);
2944 if (sym_index)
2945 old_to_new_syms[i] = sym_index;
2946 }
2947 continue;
2948 }
2949 /* if no corresponding section added, no need to add symbol */
2950 if (!sm->s)
2951 continue;
2952 /* convert section number */
2953 sym->st_shndx = sm->s->sh_num;
2954 /* offset value */
2955 sym->st_value += sm->offset;
2956 }
2957 /* add symbol */
2958 name = (char *) strtab + sym->st_name;
2959 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2960 sym->st_info, sym->st_other,
2961 sym->st_shndx, name);
2962 old_to_new_syms[i] = sym_index;
2963 }
2964
2965 /* third pass to patch relocation entries */
2966 for(i = 1; i < ehdr.e_shnum; i++) {
2967 s = sm_table[i].s;
2968 if (!s)
2969 continue;
2970 sh = &shdr[i];
2971 offset = sm_table[i].offset;
2972 switch(s->sh_type) {
2973 case SHT_RELX:
2974 /* take relocation offset information */
2975 offseti = sm_table[sh->sh_info].offset;
2976 for_each_elem(s, (offset / sizeof(*rel)), rel, ElfW_Rel) {
2977 int type;
2978 unsigned sym_index;
2979 /* convert symbol index */
2980 type = ELFW(R_TYPE)(rel->r_info);
2981 sym_index = ELFW(R_SYM)(rel->r_info);
2982 /* NOTE: only one symtab assumed */
2983 if (sym_index >= nb_syms)
2984 goto invalid_reloc;
2985 sym_index = old_to_new_syms[sym_index];
2986 /* ignore link_once in rel section. */
2987 if (!sym_index && !sm->link_once
2988 #ifdef TCC_TARGET_ARM
2989 && type != R_ARM_V4BX
2990 #endif
2991 ) {
2992 invalid_reloc:
2993 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2994 i, strsec + sh->sh_name, rel->r_offset);
2995 goto fail;
2996 }
2997 rel->r_info = ELFW(R_INFO)(sym_index, type);
2998 /* offset the relocation offset */
2999 rel->r_offset += offseti;
3000 #ifdef TCC_TARGET_ARM
3001 /* Jumps and branches from a Thumb code to a PLT entry need
3002 special handling since PLT entries are ARM code.
3003 Unconditional bl instructions referencing PLT entries are
3004 handled by converting these instructions into blx
3005 instructions. Other case of instructions referencing a PLT
3006 entry require to add a Thumb stub before the PLT entry to
3007 switch to ARM mode. We set bit plt_thumb_stub of the
3008 attribute of a symbol to indicate such a case. */
3009 if (type == R_ARM_THM_JUMP24)
3010 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
3011 #endif
3012 }
3013 break;
3014 default:
3015 break;
3016 }
3017 }
3018
3019 ret = 0;
3020 the_end:
3021 tcc_free(symtab);
3022 tcc_free(strtab);
3023 tcc_free(old_to_new_syms);
3024 tcc_free(sm_table);
3025 tcc_free(strsec);
3026 tcc_free(shdr);
3027 return ret;
3028 }
3029
3030 typedef struct ArchiveHeader {
3031 char ar_name[16]; /* name of this member */
3032 char ar_date[12]; /* file mtime */
3033 char ar_uid[6]; /* owner uid; printed as decimal */
3034 char ar_gid[6]; /* owner gid; printed as decimal */
3035 char ar_mode[8]; /* file mode, printed as octal */
3036 char ar_size[10]; /* file size, printed as decimal */
3037 char ar_fmag[2]; /* should contain ARFMAG */
3038 } ArchiveHeader;
3039
3040 static int get_be32(const uint8_t *b)
3041 {
3042 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
3043 }
3044
3045 /* load only the objects which resolve undefined symbols */
3046 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
3047 {
3048 int i, bound, nsyms, sym_index, off, ret;
3049 uint8_t *data;
3050 const char *ar_names, *p;
3051 const uint8_t *ar_index;
3052 ElfW(Sym) *sym;
3053
3054 data = tcc_malloc(size);
3055 if (read(fd, data, size) != size)
3056 goto fail;
3057 nsyms = get_be32(data);
3058 ar_index = data + 4;
3059 ar_names = (char *) ar_index + nsyms * 4;
3060
3061 do {
3062 bound = 0;
3063 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
3064 sym_index = find_elf_sym(symtab_section, p);
3065 if(sym_index) {
3066 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
3067 if(sym->st_shndx == SHN_UNDEF) {
3068 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
3069 ++bound;
3070 lseek(fd, off, SEEK_SET);
3071 if(tcc_load_object_file(s1, fd, off) < 0) {
3072 fail:
3073 ret = -1;
3074 goto the_end;
3075 }
3076 }
3077 }
3078 }
3079 } while(bound);
3080 ret = 0;
3081 the_end:
3082 tcc_free(data);
3083 return ret;
3084 }
3085
3086 /* load a '.a' file */
3087 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
3088 {
3089 ArchiveHeader hdr;
3090 char ar_size[11];
3091 char ar_name[17];
3092 char magic[8];
3093 int size, len, i;
3094 unsigned long file_offset;
3095
3096 /* skip magic which was already checked */
3097 read(fd, magic, sizeof(magic));
3098
3099 for(;;) {
3100 len = read(fd, &hdr, sizeof(hdr));
3101 if (len == 0)
3102 break;
3103 if (len != sizeof(hdr)) {
3104 tcc_error_noabort("invalid archive");
3105 return -1;
3106 }
3107 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
3108 ar_size[sizeof(hdr.ar_size)] = '\0';
3109 size = strtol(ar_size, NULL, 0);
3110 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
3111 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
3112 if (ar_name[i] != ' ')
3113 break;
3114 }
3115 ar_name[i + 1] = '\0';
3116 file_offset = lseek(fd, 0, SEEK_CUR);
3117 /* align to even */
3118 size = (size + 1) & ~1;
3119 if (!strcmp(ar_name, "/")) {
3120 /* coff symbol table : we handle it */
3121 if(s1->alacarte_link)
3122 return tcc_load_alacarte(s1, fd, size);
3123 } else if (!strcmp(ar_name, "//") ||
3124 !strcmp(ar_name, "__.SYMDEF") ||
3125 !strcmp(ar_name, "__.SYMDEF/") ||
3126 !strcmp(ar_name, "ARFILENAMES/")) {
3127 /* skip symbol table or archive names */
3128 } else {
3129 if (tcc_load_object_file(s1, fd, file_offset) < 0)
3130 return -1;
3131 }
3132 lseek(fd, file_offset + size, SEEK_SET);
3133 }
3134 return 0;
3135 }
3136
3137 #ifndef TCC_TARGET_PE
3138 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
3139 is referenced by the user (so it should be added as DT_NEEDED in
3140 the generated ELF file) */
3141 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
3142 {
3143 ElfW(Ehdr) ehdr;
3144 ElfW(Shdr) *shdr, *sh, *sh1;
3145 int i, j, nb_syms, nb_dts, sym_bind, ret;
3146 ElfW(Sym) *sym, *dynsym;
3147 ElfW(Dyn) *dt, *dynamic;
3148 unsigned char *dynstr;
3149 const char *name, *soname;
3150 DLLReference *dllref;
3151
3152 read(fd, &ehdr, sizeof(ehdr));
3153
3154 /* test CPU specific stuff */
3155 if (ehdr.e_ident[5] != ELFDATA2LSB ||
3156 ehdr.e_machine != EM_TCC_TARGET) {
3157 tcc_error_noabort("bad architecture");
3158 return -1;
3159 }
3160
3161 /* read sections */
3162 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
3163
3164 /* load dynamic section and dynamic symbols */
3165 nb_syms = 0;
3166 nb_dts = 0;
3167 dynamic = NULL;
3168 dynsym = NULL; /* avoid warning */
3169 dynstr = NULL; /* avoid warning */
3170 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
3171 switch(sh->sh_type) {
3172 case SHT_DYNAMIC:
3173 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
3174 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
3175 break;
3176 case SHT_DYNSYM:
3177 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
3178 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
3179 sh1 = &shdr[sh->sh_link];
3180 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
3181 break;
3182 default:
3183 break;
3184 }
3185 }
3186
3187 /* compute the real library name */
3188 soname = tcc_basename(filename);
3189
3190 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3191 if (dt->d_tag == DT_SONAME) {
3192 soname = (char *) dynstr + dt->d_un.d_val;
3193 }
3194 }
3195
3196 /* if the dll is already loaded, do not load it */
3197 for(i = 0; i < s1->nb_loaded_dlls; i++) {
3198 dllref = s1->loaded_dlls[i];
3199 if (!strcmp(soname, dllref->name)) {
3200 /* but update level if needed */
3201 if (level < dllref->level)
3202 dllref->level = level;
3203 ret = 0;
3204 goto the_end;
3205 }
3206 }
3207
3208 /* add the dll and its level */
3209 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
3210 dllref->level = level;
3211 strcpy(dllref->name, soname);
3212 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
3213
3214 /* add dynamic symbols in dynsym_section */
3215 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
3216 sym_bind = ELFW(ST_BIND)(sym->st_info);
3217 if (sym_bind == STB_LOCAL)
3218 continue;
3219 name = (char *) dynstr + sym->st_name;
3220 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
3221 sym->st_info, sym->st_other, sym->st_shndx, name);
3222 }
3223
3224 /* load all referenced DLLs */
3225 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3226 switch(dt->d_tag) {
3227 case DT_NEEDED:
3228 name = (char *) dynstr + dt->d_un.d_val;
3229 for(j = 0; j < s1->nb_loaded_dlls; j++) {
3230 dllref = s1->loaded_dlls[j];
3231 if (!strcmp(name, dllref->name))
3232 goto already_loaded;
3233 }
3234 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
3235 tcc_error_noabort("referenced dll '%s' not found", name);
3236 ret = -1;
3237 goto the_end;
3238 }
3239 already_loaded:
3240 break;
3241 }
3242 }
3243 ret = 0;
3244 the_end:
3245 tcc_free(dynstr);
3246 tcc_free(dynsym);
3247 tcc_free(dynamic);
3248 tcc_free(shdr);
3249 return ret;
3250 }
3251
3252 #define LD_TOK_NAME 256
3253 #define LD_TOK_EOF (-1)
3254
3255 /* return next ld script token */
3256 static int ld_next(TCCState *s1, char *name, int name_size)
3257 {
3258 int c;
3259 char *q;
3260
3261 redo:
3262 switch(ch) {
3263 case ' ':
3264 case '\t':
3265 case '\f':
3266 case '\v':
3267 case '\r':
3268 case '\n':
3269 inp();
3270 goto redo;
3271 case '/':
3272 minp();
3273 if (ch == '*') {
3274 file->buf_ptr = parse_comment(file->buf_ptr);
3275 ch = file->buf_ptr[0];
3276 goto redo;
3277 } else {
3278 q = name;
3279 *q++ = '/';
3280 goto parse_name;
3281 }
3282 break;
3283 /* case 'a' ... 'z': */
3284 case 'a':
3285 case 'b':
3286 case 'c':
3287 case 'd':
3288 case 'e':
3289 case 'f':
3290 case 'g':
3291 case 'h':
3292 case 'i':
3293 case 'j':
3294 case 'k':
3295 case 'l':
3296 case 'm':
3297 case 'n':
3298 case 'o':
3299 case 'p':
3300 case 'q':
3301 case 'r':
3302 case 's':
3303 case 't':
3304 case 'u':
3305 case 'v':
3306 case 'w':
3307 case 'x':
3308 case 'y':
3309 case 'z':
3310 /* case 'A' ... 'z': */
3311 case 'A':
3312 case 'B':
3313 case 'C':
3314 case 'D':
3315 case 'E':
3316 case 'F':
3317 case 'G':
3318 case 'H':
3319 case 'I':
3320 case 'J':
3321 case 'K':
3322 case 'L':
3323 case 'M':
3324 case 'N':
3325 case 'O':
3326 case 'P':
3327 case 'Q':
3328 case 'R':
3329 case 'S':
3330 case 'T':
3331 case 'U':
3332 case 'V':
3333 case 'W':
3334 case 'X':
3335 case 'Y':
3336 case 'Z':
3337 case '_':
3338 case '\\':
3339 case '.':
3340 case '$':
3341 case '~':
3342 q = name;
3343 parse_name:
3344 for(;;) {
3345 if (!((ch >= 'a' && ch <= 'z') ||
3346 (ch >= 'A' && ch <= 'Z') ||
3347 (ch >= '0' && ch <= '9') ||
3348 strchr("/.-_+=$:\\,~", ch)))
3349 break;
3350 if ((q - name) < name_size - 1) {
3351 *q++ = ch;
3352 }
3353 minp();
3354 }
3355 *q = '\0';
3356 c = LD_TOK_NAME;
3357 break;
3358 case CH_EOF:
3359 c = LD_TOK_EOF;
3360 break;
3361 default:
3362 c = ch;
3363 inp();
3364 break;
3365 }
3366 return c;
3367 }
3368
3369 static int ld_add_file(TCCState *s1, const char filename[])
3370 {
3371 int ret;
3372
3373 ret = tcc_add_file_internal(s1, filename, 0);
3374 if (ret)
3375 ret = tcc_add_dll(s1, filename, 0);
3376 return ret;
3377 }
3378
3379 static inline int new_undef_syms(void)
3380 {
3381 int ret = 0;
3382 ret = new_undef_sym;
3383 new_undef_sym = 0;
3384 return ret;
3385 }
3386
3387 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3388 {
3389 char filename[1024], libname[1024];
3390 int t, group, nblibs = 0, ret = 0;
3391 char **libs = NULL;
3392
3393 group = !strcmp(cmd, "GROUP");
3394 if (!as_needed)
3395 new_undef_syms();
3396 t = ld_next(s1, filename, sizeof(filename));
3397 if (t != '(')
3398 expect("(");
3399 t = ld_next(s1, filename, sizeof(filename));
3400 for(;;) {
3401 libname[0] = '\0';
3402 if (t == LD_TOK_EOF) {
3403 tcc_error_noabort("unexpected end of file");
3404 ret = -1;
3405 goto lib_parse_error;
3406 } else if (t == ')') {
3407 break;
3408 } else if (t == '-') {
3409 t = ld_next(s1, filename, sizeof(filename));
3410 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3411 tcc_error_noabort("library name expected");
3412 ret = -1;
3413 goto lib_parse_error;
3414 }
3415 pstrcpy(libname, sizeof libname, &filename[1]);
3416 if (s1->static_link) {
3417 snprintf(filename, sizeof filename, "lib%s.a", libname);
3418 } else {
3419 snprintf(filename, sizeof filename, "lib%s.so", libname);
3420 }
3421 } else if (t != LD_TOK_NAME) {
3422 tcc_error_noabort("filename expected");
3423 ret = -1;
3424 goto lib_parse_error;
3425 }
3426 if (!strcmp(filename, "AS_NEEDED")) {
3427 ret = ld_add_file_list(s1, cmd, 1);
3428 if (ret)
3429 goto lib_parse_error;
3430 } else {
3431 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3432 if (!as_needed) {
3433 ret = ld_add_file(s1, filename);
3434 if (ret)
3435 goto lib_parse_error;
3436 if (group) {
3437 /* Add the filename *and* the libname to avoid future conversions */
3438 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3439 if (libname[0] != '\0')
3440 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3441 }
3442 }
3443 }
3444 t = ld_next(s1, filename, sizeof(filename));
3445 if (t == ',') {
3446 t = ld_next(s1, filename, sizeof(filename));
3447 }
3448 }
3449 if (group && !as_needed) {
3450 while (new_undef_syms()) {
3451 int i;
3452
3453 for (i = 0; i < nblibs; i ++)
3454 ld_add_file(s1, libs[i]);
3455 }
3456 }
3457 lib_parse_error:
3458 dynarray_reset(&libs, &nblibs);
3459 return ret;
3460 }
3461
3462 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3463 files */
3464 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3465 {
3466 char cmd[64];
3467 char filename[1024];
3468 int t, ret;
3469
3470 ch = file->buf_ptr[0];
3471 ch = handle_eob();
3472 for(;;) {
3473 t = ld_next(s1, cmd, sizeof(cmd));
3474 if (t == LD_TOK_EOF)
3475 return 0;
3476 else if (t != LD_TOK_NAME)
3477 return -1;
3478 if (!strcmp(cmd, "INPUT") ||
3479 !strcmp(cmd, "GROUP")) {
3480 ret = ld_add_file_list(s1, cmd, 0);
3481 if (ret)
3482 return ret;
3483 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3484 !strcmp(cmd, "TARGET")) {
3485 /* ignore some commands */
3486 t = ld_next(s1, cmd, sizeof(cmd));
3487 if (t != '(')
3488 expect("(");
3489 for(;;) {
3490 t = ld_next(s1, filename, sizeof(filename));
3491 if (t == LD_TOK_EOF) {
3492 tcc_error_noabort("unexpected end of file");
3493 return -1;
3494 } else if (t == ')') {
3495 break;
3496 }
3497 }
3498 } else {
3499 return -1;
3500 }
3501 }
3502 return 0;
3503 }
3504 #endif /* !TCC_TARGET_PE */
Tiny C Compiler main repository