?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
fix a potential end-of-buffer issue in tccelf.c
[tinycc.git] / tccelf.c
blob581ebefa0cd64a9374e1a84f9a45ae728158ec4c
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, TCC_FILETYPE_BINARY);
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, TCC_FILETYPE_BINARY);
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 name = (char *) s1->dynsymtab_section->link->data + esym->st_name;
1879 sym_index = find_elf_sym(symtab_section, name);
1880 if (sym_index) {
1881 /* XXX: avoid adding a symbol if already present because of
1882 -rdynamic ? */
1883 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1884 if (sym->st_shndx != SHN_UNDEF)
1885 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1886 sym->st_info, 0, sym->st_shndx, name);
1887 } else if (esym->st_shndx == SHN_UNDEF) {
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 /* Export all non local symbols (for shared libraries) */
1896 static void export_global_syms(TCCState *s1)
1897 {
1898 int nb_syms, dynindex, index;
1899 const char *name;
1900 ElfW(Sym) *sym;
1901
1902 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1903 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1904 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1905 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1906 name = (char *) symtab_section->link->data + sym->st_name;
1907 dynindex = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1908 sym->st_info, 0, sym->st_shndx, name);
1909 index = sym - (ElfW(Sym) *) symtab_section->data;
1910 s1->symtab_to_dynsym[index] = dynindex;
1911 }
1912 }
1913 }
1914
1915 /* relocate the PLT: compute addresses and offsets in the PLT now that final
1916 address for PLT and GOT are known (see fill_program_header) */
1917 ST_FUNC void relocate_plt(TCCState *s1)
1918 {
1919 uint8_t *p, *p_end;
1920
1921 if (!s1->plt)
1922 return;
1923
1924 p = s1->plt->data;
1925 p_end = p + s1->plt->data_offset;
1926 if (p < p_end) {
1927 #if defined(TCC_TARGET_I386)
1928 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1929 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
1930 p += 16;
1931 while (p < p_end) {
1932 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1933 p += 16;
1934 }
1935 #elif defined(TCC_TARGET_X86_64)
1936 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
1937 put32(p + 2, get32(p + 2) + x);
1938 put32(p + 8, get32(p + 8) + x - 6);
1939 p += 16;
1940 while (p < p_end) {
1941 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
1942 p += 16;
1943 }
1944 #elif defined(TCC_TARGET_ARM)
1945 int x;
1946 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
1947 p += 16;
1948 while (p < p_end) {
1949 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
1950 p += 4;
1951 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
1952 p += 16;
1953 }
1954 #elif defined(TCC_TARGET_ARM64)
1955 uint64_t plt = s1->plt->sh_addr;
1956 uint64_t got = s1->got->sh_addr;
1957 uint64_t off = (got >> 12) - (plt >> 12);
1958 if ((off + ((uint32_t)1 << 20)) >> 21)
1959 tcc_error("Failed relocating PLT (off=0x%lx, got=0x%lx, plt=0x%lx)", off, got, plt);
1960 put32(p, 0xa9bf7bf0); // stp x16,x30,[sp,#-16]!
1961 put32(p + 4, (0x90000010 | // adrp x16,...
1962 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1963 put32(p + 8, (0xf9400211 | // ldr x17,[x16,#...]
1964 (got & 0xff8) << 7));
1965 put32(p + 12, (0x91000210 | // add x16,x16,#...
1966 (got & 0xfff) << 10));
1967 put32(p + 16, 0xd61f0220); // br x17
1968 put32(p + 20, 0xd503201f); // nop
1969 put32(p + 24, 0xd503201f); // nop
1970 put32(p + 28, 0xd503201f); // nop
1971 p += 32;
1972 while (p < p_end) {
1973 uint64_t pc = plt + (p - s1->plt->data);
1974 uint64_t addr = got +
1975 (get32(p) | (uint64_t)get32(p + 4) << 32);
1976 uint32_t off = (addr >> 12) - (pc >> 12);
1977 if ((off + ((uint32_t)1 << 20)) >> 21)
1978 tcc_error("Failed relocating PLT (off=0x%lx, addr=0x%lx, pc=0x%lx)", off, addr, pc);
1979 put32(p, (0x90000010 | // adrp x16,...
1980 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1981 put32(p + 4, (0xf9400211 | // ldr x17,[x16,#...]
1982 (addr & 0xff8) << 7));
1983 put32(p + 8, (0x91000210 | // add x16,x16,#...
1984 (addr & 0xfff) << 10));
1985 put32(p + 12, 0xd61f0220); // br x17
1986 p += 16;
1987 }
1988 #elif defined(TCC_TARGET_C67)
1989 /* XXX: TODO */
1990 #else
1991 #error unsupported CPU
1992 #endif
1993 }
1994 }
1995
1996 /* Allocate strings for section names and decide if an unallocated section
1997 should be output.
1998
1999 NOTE: the strsec section comes last, so its size is also correct ! */
2000 static void alloc_sec_names(TCCState *s1, int file_type, Section *strsec)
2001 {
2002 int i;
2003 Section *s;
2004
2005 /* Allocate strings for section names */
2006 for(i = 1; i < s1->nb_sections; i++) {
2007 s = s1->sections[i];
2008 s->sh_name = put_elf_str(strsec, s->name);
2009 /* when generating a DLL, we include relocations but we may
2010 patch them */
2011 if (file_type == TCC_OUTPUT_DLL &&
2012 s->sh_type == SHT_RELX &&
2013 !(s->sh_flags & SHF_ALLOC)) {
2014 /* gr: avoid bogus relocs for empty (debug) sections */
2015 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
2016 prepare_dynamic_rel(s1, s);
2017 else if (s1->do_debug)
2018 s->sh_size = s->data_offset;
2019 } else if (s1->do_debug ||
2020 file_type == TCC_OUTPUT_OBJ ||
2021 file_type == TCC_OUTPUT_EXE ||
2022 (s->sh_flags & SHF_ALLOC) ||
2023 i == (s1->nb_sections - 1)) {
2024 /* we output all sections if debug or object file */
2025 s->sh_size = s->data_offset;
2026 }
2027 }
2028 }
2029
2030 /* Info to be copied in dynamic section */
2031 struct dyn_inf {
2032 Section *dynamic;
2033 Section *dynstr;
2034 unsigned long dyn_rel_off;
2035 addr_t rel_addr;
2036 addr_t rel_size;
2037 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2038 addr_t bss_addr;
2039 addr_t bss_size;
2040 #endif
2041 };
2042
2043 /* Assign sections to segments and decide how are sections laid out when loaded
2044 in memory. This function also fills corresponding program headers. */
2045 static int layout_sections(TCCState *s1, ElfW(Phdr) *phdr, int phnum,
2046 Section *interp, Section* strsec,
2047 struct dyn_inf *dyninf, int *sec_order)
2048 {
2049 int i, j, k, file_type, sh_order_index, file_offset;
2050 unsigned long s_align;
2051 long long tmp;
2052 addr_t addr;
2053 ElfW(Phdr) *ph;
2054 Section *s;
2055
2056 file_type = s1->output_type;
2057 sh_order_index = 1;
2058 file_offset = 0;
2059 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2060 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2061 s_align = ELF_PAGE_SIZE;
2062 if (s1->section_align)
2063 s_align = s1->section_align;
2064
2065 if (phnum > 0) {
2066 if (s1->has_text_addr) {
2067 int a_offset, p_offset;
2068 addr = s1->text_addr;
2069 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
2070 ELF_PAGE_SIZE */
2071 a_offset = (int) (addr & (s_align - 1));
2072 p_offset = file_offset & (s_align - 1);
2073 if (a_offset < p_offset)
2074 a_offset += s_align;
2075 file_offset += (a_offset - p_offset);
2076 } else {
2077 if (file_type == TCC_OUTPUT_DLL)
2078 addr = 0;
2079 else
2080 addr = ELF_START_ADDR;
2081 /* compute address after headers */
2082 addr += (file_offset & (s_align - 1));
2083 }
2084
2085 ph = &phdr[0];
2086 /* Leave one program headers for the program interpreter and one for
2087 the program header table itself if needed. These are done later as
2088 they require section layout to be done first. */
2089 if (interp)
2090 ph += 1 + HAVE_PHDR;
2091
2092 /* dynamic relocation table information, for .dynamic section */
2093 dyninf->rel_addr = dyninf->rel_size = 0;
2094 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2095 dyninf->bss_addr = dyninf->bss_size = 0;
2096 #endif
2097
2098 for(j = 0; j < 2; j++) {
2099 ph->p_type = PT_LOAD;
2100 if (j == 0)
2101 ph->p_flags = PF_R | PF_X;
2102 else
2103 ph->p_flags = PF_R | PF_W;
2104 ph->p_align = s_align;
2105
2106 /* Decide the layout of sections loaded in memory. This must
2107 be done before program headers are filled since they contain
2108 info about the layout. We do the following ordering: interp,
2109 symbol tables, relocations, progbits, nobits */
2110 /* XXX: do faster and simpler sorting */
2111 for(k = 0; k < 5; k++) {
2112 for(i = 1; i < s1->nb_sections; i++) {
2113 s = s1->sections[i];
2114 /* compute if section should be included */
2115 if (j == 0) {
2116 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2117 SHF_ALLOC)
2118 continue;
2119 } else {
2120 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2121 (SHF_ALLOC | SHF_WRITE))
2122 continue;
2123 }
2124 if (s == interp) {
2125 if (k != 0)
2126 continue;
2127 } else if (s->sh_type == SHT_DYNSYM ||
2128 s->sh_type == SHT_STRTAB ||
2129 s->sh_type == SHT_HASH) {
2130 if (k != 1)
2131 continue;
2132 } else if (s->sh_type == SHT_RELX) {
2133 if (k != 2)
2134 continue;
2135 } else if (s->sh_type == SHT_NOBITS) {
2136 if (k != 4)
2137 continue;
2138 } else {
2139 if (k != 3)
2140 continue;
2141 }
2142 sec_order[sh_order_index++] = i;
2143
2144 /* section matches: we align it and add its size */
2145 tmp = addr;
2146 addr = (addr + s->sh_addralign - 1) &
2147 ~(s->sh_addralign - 1);
2148 file_offset += (int) ( addr - tmp );
2149 s->sh_offset = file_offset;
2150 s->sh_addr = addr;
2151
2152 /* update program header infos */
2153 if (ph->p_offset == 0) {
2154 ph->p_offset = file_offset;
2155 ph->p_vaddr = addr;
2156 ph->p_paddr = ph->p_vaddr;
2157 }
2158 /* update dynamic relocation infos */
2159 if (s->sh_type == SHT_RELX) {
2160 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2161 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) {
2162 dyninf->rel_addr = addr;
2163 dyninf->rel_size += s->sh_size; /* XXX only first rel. */
2164 }
2165 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) {
2166 dyninf->bss_addr = addr;
2167 dyninf->bss_size = s->sh_size; /* XXX only first rel. */
2168 }
2169 #else
2170 if (dyninf->rel_size == 0)
2171 dyninf->rel_addr = addr;
2172 dyninf->rel_size += s->sh_size;
2173 #endif
2174 }
2175 addr += s->sh_size;
2176 if (s->sh_type != SHT_NOBITS)
2177 file_offset += s->sh_size;
2178 }
2179 }
2180 if (j == 0) {
2181 /* Make the first PT_LOAD segment include the program
2182 headers itself (and the ELF header as well), it'll
2183 come out with same memory use but will make various
2184 tools like binutils strip work better. */
2185 ph->p_offset &= ~(ph->p_align - 1);
2186 ph->p_vaddr &= ~(ph->p_align - 1);
2187 ph->p_paddr &= ~(ph->p_align - 1);
2188 }
2189 ph->p_filesz = file_offset - ph->p_offset;
2190 ph->p_memsz = addr - ph->p_vaddr;
2191 ph++;
2192 if (j == 0) {
2193 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
2194 /* if in the middle of a page, we duplicate the page in
2195 memory so that one copy is RX and the other is RW */
2196 if ((addr & (s_align - 1)) != 0)
2197 addr += s_align;
2198 } else {
2199 addr = (addr + s_align - 1) & ~(s_align - 1);
2200 file_offset = (file_offset + s_align - 1) & ~(s_align - 1);
2201 }
2202 }
2203 }
2204 }
2205
2206 /* all other sections come after */
2207 for(i = 1; i < s1->nb_sections; i++) {
2208 s = s1->sections[i];
2209 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
2210 continue;
2211 sec_order[sh_order_index++] = i;
2212
2213 file_offset = (file_offset + s->sh_addralign - 1) &
2214 ~(s->sh_addralign - 1);
2215 s->sh_offset = file_offset;
2216 if (s->sh_type != SHT_NOBITS)
2217 file_offset += s->sh_size;
2218 }
2219
2220 return file_offset;
2221 }
2222
2223 static void fill_unloadable_phdr(ElfW(Phdr) *phdr, int phnum, Section *interp,
2224 Section *dynamic)
2225 {
2226 ElfW(Phdr) *ph;
2227
2228 /* if interpreter, then add corresponding program header */
2229 if (interp) {
2230 ph = &phdr[0];
2231
2232 if (HAVE_PHDR)
2233 {
2234 int len = phnum * sizeof(ElfW(Phdr));
2235
2236 ph->p_type = PT_PHDR;
2237 ph->p_offset = sizeof(ElfW(Ehdr));
2238 ph->p_vaddr = interp->sh_addr - len;
2239 ph->p_paddr = ph->p_vaddr;
2240 ph->p_filesz = ph->p_memsz = len;
2241 ph->p_flags = PF_R | PF_X;
2242 ph->p_align = 4; /* interp->sh_addralign; */
2243 ph++;
2244 }
2245
2246 ph->p_type = PT_INTERP;
2247 ph->p_offset = interp->sh_offset;
2248 ph->p_vaddr = interp->sh_addr;
2249 ph->p_paddr = ph->p_vaddr;
2250 ph->p_filesz = interp->sh_size;
2251 ph->p_memsz = interp->sh_size;
2252 ph->p_flags = PF_R;
2253 ph->p_align = interp->sh_addralign;
2254 }
2255
2256 /* if dynamic section, then add corresponding program header */
2257 if (dynamic) {
2258 ph = &phdr[phnum - 1];
2259
2260 ph->p_type = PT_DYNAMIC;
2261 ph->p_offset = dynamic->sh_offset;
2262 ph->p_vaddr = dynamic->sh_addr;
2263 ph->p_paddr = ph->p_vaddr;
2264 ph->p_filesz = dynamic->sh_size;
2265 ph->p_memsz = dynamic->sh_size;
2266 ph->p_flags = PF_R | PF_W;
2267 ph->p_align = dynamic->sh_addralign;
2268 }
2269 }
2270
2271 /* Fill the dynamic section with tags describing the address and size of
2272 sections */
2273 static void fill_dynamic(TCCState *s1, struct dyn_inf *dyninf)
2274 {
2275 Section *dynamic;
2276
2277 dynamic = dyninf->dynamic;
2278
2279 /* put dynamic section entries */
2280 dynamic->data_offset = dyninf->dyn_rel_off;
2281 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2282 put_dt(dynamic, DT_STRTAB, dyninf->dynstr->sh_addr);
2283 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2284 put_dt(dynamic, DT_STRSZ, dyninf->dynstr->data_offset);
2285 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2286 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
2287 put_dt(dynamic, DT_RELA, dyninf->rel_addr);
2288 put_dt(dynamic, DT_RELASZ, dyninf->rel_size);
2289 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2290 #else
2291 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2292 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2293 put_dt(dynamic, DT_PLTRELSZ, dyninf->rel_size);
2294 put_dt(dynamic, DT_JMPREL, dyninf->rel_addr);
2295 put_dt(dynamic, DT_PLTREL, DT_REL);
2296 put_dt(dynamic, DT_REL, dyninf->bss_addr);
2297 put_dt(dynamic, DT_RELSZ, dyninf->bss_size);
2298 #else
2299 put_dt(dynamic, DT_REL, dyninf->rel_addr);
2300 put_dt(dynamic, DT_RELSZ, dyninf->rel_size);
2301 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2302 #endif
2303 #endif
2304 if (s1->do_debug)
2305 put_dt(dynamic, DT_DEBUG, 0);
2306 put_dt(dynamic, DT_NULL, 0);
2307 }
2308
2309 /* Relocate remaining sections and symbols (that is those not related to
2310 dynamic linking) */
2311 static int final_sections_reloc(TCCState *s1)
2312 {
2313 int i;
2314 Section *s;
2315
2316 relocate_syms(s1, 0);
2317
2318 if (s1->nb_errors != 0)
2319 return -1;
2320
2321 /* relocate sections */
2322 /* XXX: ignore sections with allocated relocations ? */
2323 for(i = 1; i < s1->nb_sections; i++) {
2324 s = s1->sections[i];
2325 #ifdef TCC_TARGET_I386
2326 if (s->reloc && s != s1->got && (s->sh_flags & SHF_ALLOC)) //gr
2327 /* On X86 gdb 7.3 works in any case but gdb 6.6 will crash if SHF_ALLOC
2328 checking is removed */
2329 #else
2330 if (s->reloc && s != s1->got)
2331 /* On X86_64 gdb 7.3 will crash if SHF_ALLOC checking is present */
2332 #endif
2333 relocate_section(s1, s);
2334 }
2335
2336 /* relocate relocation entries if the relocation tables are
2337 allocated in the executable */
2338 for(i = 1; i < s1->nb_sections; i++) {
2339 s = s1->sections[i];
2340 if ((s->sh_flags & SHF_ALLOC) &&
2341 s->sh_type == SHT_RELX) {
2342 relocate_rel(s1, s);
2343 }
2344 }
2345 return 0;
2346 }
2347
2348 /* Create an ELF file on disk.
2349 This function handle ELF specific layout requirements */
2350 static void tcc_output_elf(TCCState *s1, FILE *f, int phnum, ElfW(Phdr) *phdr,
2351 int file_offset, int *sec_order)
2352 {
2353 int i, shnum, offset, size, file_type;
2354 Section *s;
2355 ElfW(Ehdr) ehdr;
2356 ElfW(Shdr) shdr, *sh;
2357
2358 file_type = s1->output_type;
2359 shnum = s1->nb_sections;
2360
2361 memset(&ehdr, 0, sizeof(ehdr));
2362
2363 if (phnum > 0) {
2364 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2365 ehdr.e_phnum = phnum;
2366 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2367 }
2368
2369 /* align to 4 */
2370 file_offset = (file_offset + 3) & -4;
2371
2372 /* fill header */
2373 ehdr.e_ident[0] = ELFMAG0;
2374 ehdr.e_ident[1] = ELFMAG1;
2375 ehdr.e_ident[2] = ELFMAG2;
2376 ehdr.e_ident[3] = ELFMAG3;
2377 ehdr.e_ident[4] = ELFCLASSW;
2378 ehdr.e_ident[5] = ELFDATA2LSB;
2379 ehdr.e_ident[6] = EV_CURRENT;
2380 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2381 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2382 #endif
2383 #ifdef TCC_TARGET_ARM
2384 #ifdef TCC_ARM_EABI
2385 ehdr.e_ident[EI_OSABI] = 0;
2386 ehdr.e_flags = EF_ARM_EABI_VER4;
2387 if (file_type == TCC_OUTPUT_EXE || file_type == TCC_OUTPUT_DLL)
2388 ehdr.e_flags |= EF_ARM_HASENTRY;
2389 if (s1->float_abi == ARM_HARD_FLOAT)
2390 ehdr.e_flags |= EF_ARM_VFP_FLOAT;
2391 else
2392 ehdr.e_flags |= EF_ARM_SOFT_FLOAT;
2393 #else
2394 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2395 #endif
2396 #endif
2397 switch(file_type) {
2398 default:
2399 case TCC_OUTPUT_EXE:
2400 ehdr.e_type = ET_EXEC;
2401 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2402 break;
2403 case TCC_OUTPUT_DLL:
2404 ehdr.e_type = ET_DYN;
2405 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2406 break;
2407 case TCC_OUTPUT_OBJ:
2408 ehdr.e_type = ET_REL;
2409 break;
2410 }
2411 ehdr.e_machine = EM_TCC_TARGET;
2412 ehdr.e_version = EV_CURRENT;
2413 ehdr.e_shoff = file_offset;
2414 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2415 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2416 ehdr.e_shnum = shnum;
2417 ehdr.e_shstrndx = shnum - 1;
2418
2419 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2420 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2421 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2422
2423 sort_syms(s1, symtab_section);
2424 for(i = 1; i < s1->nb_sections; i++) {
2425 s = s1->sections[sec_order[i]];
2426 if (s->sh_type != SHT_NOBITS) {
2427 if (s->sh_type == SHT_DYNSYM)
2428 patch_dynsym_undef(s1, s);
2429 while (offset < s->sh_offset) {
2430 fputc(0, f);
2431 offset++;
2432 }
2433 size = s->sh_size;
2434 fwrite(s->data, 1, size, f);
2435 offset += size;
2436 }
2437 }
2438
2439 /* output section headers */
2440 while (offset < ehdr.e_shoff) {
2441 fputc(0, f);
2442 offset++;
2443 }
2444
2445 for(i = 0; i < s1->nb_sections; i++) {
2446 sh = &shdr;
2447 memset(sh, 0, sizeof(ElfW(Shdr)));
2448 s = s1->sections[i];
2449 if (s) {
2450 sh->sh_name = s->sh_name;
2451 sh->sh_type = s->sh_type;
2452 sh->sh_flags = s->sh_flags;
2453 sh->sh_entsize = s->sh_entsize;
2454 sh->sh_info = s->sh_info;
2455 if (s->link)
2456 sh->sh_link = s->link->sh_num;
2457 sh->sh_addralign = s->sh_addralign;
2458 sh->sh_addr = s->sh_addr;
2459 sh->sh_offset = s->sh_offset;
2460 sh->sh_size = s->sh_size;
2461 }
2462 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2463 }
2464 }
2465
2466 /* Write an elf, coff or "binary" file */
2467 static int tcc_write_elf_file(TCCState *s1, const char *filename, int phnum,
2468 ElfW(Phdr) *phdr, int file_offset, int *sec_order)
2469 {
2470 int fd, mode, file_type;
2471 FILE *f;
2472
2473 file_type = s1->output_type;
2474 if (file_type == TCC_OUTPUT_OBJ)
2475 mode = 0666;
2476 else
2477 mode = 0777;
2478 unlink(filename);
2479 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2480 if (fd < 0) {
2481 tcc_error_noabort("could not write '%s'", filename);
2482 return -1;
2483 }
2484 f = fdopen(fd, "wb");
2485 if (s1->verbose)
2486 printf("<- %s\n", filename);
2487
2488 #ifdef TCC_TARGET_COFF
2489 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF)
2490 tcc_output_coff(s1, f);
2491 else
2492 #endif
2493 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2494 tcc_output_elf(s1, f, phnum, phdr, file_offset, sec_order);
2495 else
2496 tcc_output_binary(s1, f, sec_order);
2497 fclose(f);
2498
2499 return 0;
2500 }
2501
2502 /* Output an elf, coff or binary file */
2503 /* XXX: suppress unneeded sections */
2504 static int elf_output_file(TCCState *s1, const char *filename)
2505 {
2506 int i, ret, phnum, shnum, file_type, file_offset, *sec_order;
2507 struct dyn_inf dyninf;
2508 ElfW(Phdr) *phdr;
2509 ElfW(Sym) *sym;
2510 Section *strsec, *interp, *dynamic, *dynstr;
2511
2512 file_type = s1->output_type;
2513 s1->nb_errors = 0;
2514
2515 /* if linking, also link in runtime libraries (libc, libgcc, etc.) */
2516 if (file_type != TCC_OUTPUT_OBJ) {
2517 tcc_add_runtime(s1);
2518 }
2519
2520 phdr = NULL;
2521 sec_order = NULL;
2522 interp = dynamic = dynstr = NULL; /* avoid warning */
2523 dyninf.dyn_rel_off = 0; /* avoid warning */
2524
2525 if (file_type != TCC_OUTPUT_OBJ) {
2526 relocate_common_syms();
2527
2528 tcc_add_linker_symbols(s1);
2529
2530 if (!s1->static_link) {
2531 if (file_type == TCC_OUTPUT_EXE) {
2532 char *ptr;
2533 /* allow override the dynamic loader */
2534 const char *elfint = getenv("LD_SO");
2535 if (elfint == NULL)
2536 elfint = DEFAULT_ELFINTERP(s1);
2537 /* add interpreter section only if executable */
2538 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
2539 interp->sh_addralign = 1;
2540 ptr = section_ptr_add(interp, 1 + strlen(elfint));
2541 strcpy(ptr, elfint);
2542 }
2543
2544 /* add dynamic symbol table */
2545 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
2546 ".dynstr",
2547 ".hash", SHF_ALLOC);
2548 dynstr = s1->dynsym->link;
2549
2550 /* add dynamic section */
2551 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
2552 SHF_ALLOC | SHF_WRITE);
2553 dynamic->link = dynstr;
2554 dynamic->sh_entsize = sizeof(ElfW(Dyn));
2555
2556 build_got(s1);
2557
2558 if (file_type == TCC_OUTPUT_EXE) {
2559 bind_exe_dynsyms(s1);
2560
2561 if (s1->nb_errors) {
2562 ret = -1;
2563 goto the_end;
2564 }
2565
2566 bind_libs_dynsyms(s1);
2567 } else /* shared library case: simply export all global symbols */
2568 export_global_syms(s1);
2569
2570 build_got_entries(s1);
2571
2572 /* add a list of needed dlls */
2573 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2574 DLLReference *dllref = s1->loaded_dlls[i];
2575 if (dllref->level == 0)
2576 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
2577 }
2578
2579 if (s1->rpath)
2580 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
2581
2582 /* XXX: currently, since we do not handle PIC code, we
2583 must relocate the readonly segments */
2584 if (file_type == TCC_OUTPUT_DLL) {
2585 if (s1->soname)
2586 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
2587 put_dt(dynamic, DT_TEXTREL, 0);
2588 }
2589
2590 if (s1->symbolic)
2591 put_dt(dynamic, DT_SYMBOLIC, 0);
2592
2593 /* add necessary space for other entries */
2594 dyninf.dyn_rel_off = dynamic->data_offset;
2595 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
2596 } else {
2597 /* still need to build got entries in case of static link */
2598 build_got_entries(s1);
2599 }
2600 }
2601
2602 /* we add a section for symbols */
2603 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
2604 put_elf_str(strsec, "");
2605
2606 /* compute number of sections */
2607 shnum = s1->nb_sections;
2608
2609 /* this array is used to reorder sections in the output file */
2610 sec_order = tcc_malloc(sizeof(int) * shnum);
2611 sec_order[0] = 0;
2612
2613 /* compute number of program headers */
2614 switch(file_type) {
2615 default:
2616 case TCC_OUTPUT_OBJ:
2617 phnum = 0;
2618 break;
2619 case TCC_OUTPUT_EXE:
2620 if (!s1->static_link)
2621 phnum = 4 + HAVE_PHDR;
2622 else
2623 phnum = 2;
2624 break;
2625 case TCC_OUTPUT_DLL:
2626 phnum = 3;
2627 break;
2628 }
2629
2630 /* Allocate strings for section names */
2631 alloc_sec_names(s1, file_type, strsec);
2632
2633 /* allocate program segment headers */
2634 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
2635
2636 /* compute section to program header mapping */
2637 file_offset = layout_sections(s1, phdr, phnum, interp, strsec, &dyninf,
2638 sec_order);
2639
2640 /* Fill remaining program header and finalize relocation related to dynamic
2641 linking. */
2642 if (phnum > 0) {
2643 fill_unloadable_phdr(phdr, phnum, interp, dynamic);
2644 if (dynamic) {
2645 dyninf.dynamic = dynamic;
2646 dyninf.dynstr = dynstr;
2647
2648 fill_dynamic(s1, &dyninf);
2649
2650 /* put in GOT the dynamic section address and relocate PLT */
2651 put32(s1->got->data, dynamic->sh_addr);
2652 if (file_type == TCC_OUTPUT_EXE
2653 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2654 || file_type == TCC_OUTPUT_DLL
2655 #endif
2656 )
2657 relocate_plt(s1);
2658
2659 /* relocate symbols in .dynsym now that final addresses are known */
2660 for_each_elem(s1->dynsym, 1, sym, ElfW(Sym)) {
2661 if (sym->st_shndx == SHN_UNDEF) {
2662 /* relocate to PLT if symbol corresponds to a PLT entry,
2663 but not if it's a weak symbol */
2664 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
2665 sym->st_value = 0;
2666 else if (sym->st_value)
2667 sym->st_value += s1->plt->sh_addr;
2668 } else if (sym->st_shndx < SHN_LORESERVE) {
2669 /* do symbol relocation */
2670 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2671 }
2672 }
2673 }
2674 }
2675
2676 /* if building executable or DLL, then relocate each section
2677 except the GOT which is already relocated */
2678 if (file_type != TCC_OUTPUT_OBJ) {
2679 ret = final_sections_reloc(s1);
2680 if (ret)
2681 goto the_end;
2682 }
2683
2684 /* Perform relocation to GOT or PLT entries */
2685 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2686 fill_got(s1);
2687
2688 /* Create the ELF file with name 'filename' */
2689 ret = tcc_write_elf_file(s1, filename, phnum, phdr, file_offset, sec_order);
2690 if (s1->do_strip) {
2691 int rc;
2692 const char *strip_cmd = "sstrip "; // super strip utility from ELFkickers
2693 const char *null_dev = " 2> /dev/null";
2694 char buf[1050];
2695 snprintf(buf, sizeof(buf), "%s%s%s", strip_cmd, filename, null_dev);
2696 rc = system(buf);
2697 if (rc)
2698 system(buf+1); // call a strip utility from binutils
2699 }
2700 the_end:
2701 tcc_free(s1->symtab_to_dynsym);
2702 tcc_free(sec_order);
2703 tcc_free(phdr);
2704 tcc_free(s1->sym_attrs);
2705 s1->sym_attrs = NULL;
2706 return ret;
2707 }
2708
2709 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2710 {
2711 int ret;
2712 #ifdef TCC_TARGET_PE
2713 if (s->output_type != TCC_OUTPUT_OBJ) {
2714 ret = pe_output_file(s, filename);
2715 } else
2716 #endif
2717 ret = elf_output_file(s, filename);
2718 return ret;
2719 }
2720
2721 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2722 {
2723 void *data;
2724
2725 data = tcc_malloc(size);
2726 lseek(fd, file_offset, SEEK_SET);
2727 read(fd, data, size);
2728 return data;
2729 }
2730
2731 typedef struct SectionMergeInfo {
2732 Section *s; /* corresponding existing section */
2733 unsigned long offset; /* offset of the new section in the existing section */
2734 uint8_t new_section; /* true if section 's' was added */
2735 uint8_t link_once; /* true if link once section */
2736 } SectionMergeInfo;
2737
2738 /* load an object file and merge it with current files */
2739 /* XXX: handle correctly stab (debug) info */
2740 ST_FUNC int tcc_load_object_file(TCCState *s1,
2741 int fd, unsigned long file_offset)
2742 {
2743 ElfW(Ehdr) ehdr;
2744 ElfW(Shdr) *shdr, *sh;
2745 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2746 unsigned char *strsec, *strtab;
2747 int *old_to_new_syms;
2748 char *sh_name, *name;
2749 SectionMergeInfo *sm_table, *sm;
2750 ElfW(Sym) *sym, *symtab;
2751 ElfW_Rel *rel;
2752 Section *s;
2753
2754 int stab_index;
2755 int stabstr_index;
2756
2757 stab_index = stabstr_index = 0;
2758
2759 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2760 goto fail1;
2761 if (ehdr.e_ident[0] != ELFMAG0 ||
2762 ehdr.e_ident[1] != ELFMAG1 ||
2763 ehdr.e_ident[2] != ELFMAG2 ||
2764 ehdr.e_ident[3] != ELFMAG3)
2765 goto fail1;
2766 /* test if object file */
2767 if (ehdr.e_type != ET_REL)
2768 goto fail1;
2769 /* test CPU specific stuff */
2770 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2771 ehdr.e_machine != EM_TCC_TARGET) {
2772 fail1:
2773 tcc_error_noabort("invalid object file");
2774 return -1;
2775 }
2776 /* read sections */
2777 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2778 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2779 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2780
2781 /* load section names */
2782 sh = &shdr[ehdr.e_shstrndx];
2783 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2784
2785 /* load symtab and strtab */
2786 old_to_new_syms = NULL;
2787 symtab = NULL;
2788 strtab = NULL;
2789 nb_syms = 0;
2790 for(i = 1; i < ehdr.e_shnum; i++) {
2791 sh = &shdr[i];
2792 if (sh->sh_type == SHT_SYMTAB) {
2793 if (symtab) {
2794 tcc_error_noabort("object must contain only one symtab");
2795 fail:
2796 ret = -1;
2797 goto the_end;
2798 }
2799 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2800 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2801 sm_table[i].s = symtab_section;
2802
2803 /* now load strtab */
2804 sh = &shdr[sh->sh_link];
2805 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2806 }
2807 }
2808
2809 /* now examine each section and try to merge its content with the
2810 ones in memory */
2811 for(i = 1; i < ehdr.e_shnum; i++) {
2812 /* no need to examine section name strtab */
2813 if (i == ehdr.e_shstrndx)
2814 continue;
2815 sh = &shdr[i];
2816 sh_name = (char *) strsec + sh->sh_name;
2817 /* ignore sections types we do not handle */
2818 if (sh->sh_type != SHT_PROGBITS &&
2819 sh->sh_type != SHT_RELX &&
2820 #ifdef TCC_ARM_EABI
2821 sh->sh_type != SHT_ARM_EXIDX &&
2822 #endif
2823 sh->sh_type != SHT_NOBITS &&
2824 sh->sh_type != SHT_PREINIT_ARRAY &&
2825 sh->sh_type != SHT_INIT_ARRAY &&
2826 sh->sh_type != SHT_FINI_ARRAY &&
2827 strcmp(sh_name, ".stabstr")
2828 )
2829 continue;
2830 if (sh->sh_addralign < 1)
2831 sh->sh_addralign = 1;
2832 /* find corresponding section, if any */
2833 for(j = 1; j < s1->nb_sections;j++) {
2834 s = s1->sections[j];
2835 if (!strcmp(s->name, sh_name)) {
2836 if (!strncmp(sh_name, ".gnu.linkonce",
2837 sizeof(".gnu.linkonce") - 1)) {
2838 /* if a 'linkonce' section is already present, we
2839 do not add it again. It is a little tricky as
2840 symbols can still be defined in
2841 it. */
2842 sm_table[i].link_once = 1;
2843 goto next;
2844 } else {
2845 goto found;
2846 }
2847 }
2848 }
2849 /* not found: create new section */
2850 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2851 /* take as much info as possible from the section. sh_link and
2852 sh_info will be updated later */
2853 s->sh_addralign = sh->sh_addralign;
2854 s->sh_entsize = sh->sh_entsize;
2855 sm_table[i].new_section = 1;
2856 found:
2857 if (sh->sh_type != s->sh_type) {
2858 tcc_error_noabort("invalid section type");
2859 goto fail;
2860 }
2861
2862 /* align start of section */
2863 offset = s->data_offset;
2864
2865 if (0 == strcmp(sh_name, ".stab")) {
2866 stab_index = i;
2867 goto no_align;
2868 }
2869 if (0 == strcmp(sh_name, ".stabstr")) {
2870 stabstr_index = i;
2871 goto no_align;
2872 }
2873
2874 size = sh->sh_addralign - 1;
2875 offset = (offset + size) & ~size;
2876 if (sh->sh_addralign > s->sh_addralign)
2877 s->sh_addralign = sh->sh_addralign;
2878 s->data_offset = offset;
2879 no_align:
2880 sm_table[i].offset = offset;
2881 sm_table[i].s = s;
2882 /* concatenate sections */
2883 size = sh->sh_size;
2884 if (sh->sh_type != SHT_NOBITS) {
2885 unsigned char *ptr;
2886 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2887 ptr = section_ptr_add(s, size);
2888 read(fd, ptr, size);
2889 } else {
2890 s->data_offset += size;
2891 }
2892 next: ;
2893 }
2894
2895 /* gr relocate stab strings */
2896 if (stab_index && stabstr_index) {
2897 Stab_Sym *a, *b;
2898 unsigned o;
2899 s = sm_table[stab_index].s;
2900 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2901 b = (Stab_Sym *)(s->data + s->data_offset);
2902 o = sm_table[stabstr_index].offset;
2903 while (a < b)
2904 a->n_strx += o, a++;
2905 }
2906
2907 /* second short pass to update sh_link and sh_info fields of new
2908 sections */
2909 for(i = 1; i < ehdr.e_shnum; i++) {
2910 s = sm_table[i].s;
2911 if (!s || !sm_table[i].new_section)
2912 continue;
2913 sh = &shdr[i];
2914 if (sh->sh_link > 0)
2915 s->link = sm_table[sh->sh_link].s;
2916 if (sh->sh_type == SHT_RELX) {
2917 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2918 /* update backward link */
2919 s1->sections[s->sh_info]->reloc = s;
2920 }
2921 }
2922 sm = sm_table;
2923
2924 /* resolve symbols */
2925 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2926
2927 sym = symtab + 1;
2928 for(i = 1; i < nb_syms; i++, sym++) {
2929 if (sym->st_shndx != SHN_UNDEF &&
2930 sym->st_shndx < SHN_LORESERVE) {
2931 sm = &sm_table[sym->st_shndx];
2932 if (sm->link_once) {
2933 /* if a symbol is in a link once section, we use the
2934 already defined symbol. It is very important to get
2935 correct relocations */
2936 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2937 name = (char *) strtab + sym->st_name;
2938 sym_index = find_elf_sym(symtab_section, name);
2939 if (sym_index)
2940 old_to_new_syms[i] = sym_index;
2941 }
2942 continue;
2943 }
2944 /* if no corresponding section added, no need to add symbol */
2945 if (!sm->s)
2946 continue;
2947 /* convert section number */
2948 sym->st_shndx = sm->s->sh_num;
2949 /* offset value */
2950 sym->st_value += sm->offset;
2951 }
2952 /* add symbol */
2953 name = (char *) strtab + sym->st_name;
2954 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2955 sym->st_info, sym->st_other,
2956 sym->st_shndx, name);
2957 old_to_new_syms[i] = sym_index;
2958 }
2959
2960 /* third pass to patch relocation entries */
2961 for(i = 1; i < ehdr.e_shnum; i++) {
2962 s = sm_table[i].s;
2963 if (!s)
2964 continue;
2965 sh = &shdr[i];
2966 offset = sm_table[i].offset;
2967 switch(s->sh_type) {
2968 case SHT_RELX:
2969 /* take relocation offset information */
2970 offseti = sm_table[sh->sh_info].offset;
2971 for_each_elem(s, (offset / sizeof(*rel)), rel, ElfW_Rel) {
2972 int type;
2973 unsigned sym_index;
2974 /* convert symbol index */
2975 type = ELFW(R_TYPE)(rel->r_info);
2976 sym_index = ELFW(R_SYM)(rel->r_info);
2977 /* NOTE: only one symtab assumed */
2978 if (sym_index >= nb_syms)
2979 goto invalid_reloc;
2980 sym_index = old_to_new_syms[sym_index];
2981 /* ignore link_once in rel section. */
2982 if (!sym_index && !sm->link_once
2983 #ifdef TCC_TARGET_ARM
2984 && type != R_ARM_V4BX
2985 #endif
2986 ) {
2987 invalid_reloc:
2988 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2989 i, strsec + sh->sh_name, rel->r_offset);
2990 goto fail;
2991 }
2992 rel->r_info = ELFW(R_INFO)(sym_index, type);
2993 /* offset the relocation offset */
2994 rel->r_offset += offseti;
2995 #ifdef TCC_TARGET_ARM
2996 /* Jumps and branches from a Thumb code to a PLT entry need
2997 special handling since PLT entries are ARM code.
2998 Unconditional bl instructions referencing PLT entries are
2999 handled by converting these instructions into blx
3000 instructions. Other case of instructions referencing a PLT
3001 entry require to add a Thumb stub before the PLT entry to
3002 switch to ARM mode. We set bit plt_thumb_stub of the
3003 attribute of a symbol to indicate such a case. */
3004 if (type == R_ARM_THM_JUMP24)
3005 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
3006 #endif
3007 }
3008 break;
3009 default:
3010 break;
3011 }
3012 }
3013
3014 ret = 0;
3015 the_end:
3016 tcc_free(symtab);
3017 tcc_free(strtab);
3018 tcc_free(old_to_new_syms);
3019 tcc_free(sm_table);
3020 tcc_free(strsec);
3021 tcc_free(shdr);
3022 return ret;
3023 }
3024
3025 typedef struct ArchiveHeader {
3026 char ar_name[16]; /* name of this member */
3027 char ar_date[12]; /* file mtime */
3028 char ar_uid[6]; /* owner uid; printed as decimal */
3029 char ar_gid[6]; /* owner gid; printed as decimal */
3030 char ar_mode[8]; /* file mode, printed as octal */
3031 char ar_size[10]; /* file size, printed as decimal */
3032 char ar_fmag[2]; /* should contain ARFMAG */
3033 } ArchiveHeader;
3034
3035 static int get_be32(const uint8_t *b)
3036 {
3037 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
3038 }
3039
3040 /* load only the objects which resolve undefined symbols */
3041 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
3042 {
3043 int i, bound, nsyms, sym_index, off, ret;
3044 uint8_t *data;
3045 const char *ar_names, *p;
3046 const uint8_t *ar_index;
3047 ElfW(Sym) *sym;
3048
3049 data = tcc_malloc(size);
3050 if (read(fd, data, size) != size)
3051 goto fail;
3052 nsyms = get_be32(data);
3053 ar_index = data + 4;
3054 ar_names = (char *) ar_index + nsyms * 4;
3055
3056 do {
3057 bound = 0;
3058 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
3059 sym_index = find_elf_sym(symtab_section, p);
3060 if(sym_index) {
3061 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
3062 if(sym->st_shndx == SHN_UNDEF) {
3063 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
3064 ++bound;
3065 lseek(fd, off, SEEK_SET);
3066 if(tcc_load_object_file(s1, fd, off) < 0) {
3067 fail:
3068 ret = -1;
3069 goto the_end;
3070 }
3071 }
3072 }
3073 }
3074 } while(bound);
3075 ret = 0;
3076 the_end:
3077 tcc_free(data);
3078 return ret;
3079 }
3080
3081 /* load a '.a' file */
3082 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
3083 {
3084 ArchiveHeader hdr;
3085 char ar_size[11];
3086 char ar_name[17];
3087 char magic[8];
3088 int size, len, i;
3089 unsigned long file_offset;
3090
3091 /* skip magic which was already checked */
3092 read(fd, magic, sizeof(magic));
3093
3094 for(;;) {
3095 len = read(fd, &hdr, sizeof(hdr));
3096 if (len == 0)
3097 break;
3098 if (len != sizeof(hdr)) {
3099 tcc_error_noabort("invalid archive");
3100 return -1;
3101 }
3102 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
3103 ar_size[sizeof(hdr.ar_size)] = '\0';
3104 size = strtol(ar_size, NULL, 0);
3105 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
3106 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
3107 if (ar_name[i] != ' ')
3108 break;
3109 }
3110 ar_name[i + 1] = '\0';
3111 file_offset = lseek(fd, 0, SEEK_CUR);
3112 /* align to even */
3113 size = (size + 1) & ~1;
3114 if (!strcmp(ar_name, "/")) {
3115 /* coff symbol table : we handle it */
3116 if(s1->alacarte_link)
3117 return tcc_load_alacarte(s1, fd, size);
3118 } else if (!strcmp(ar_name, "//") ||
3119 !strcmp(ar_name, "__.SYMDEF") ||
3120 !strcmp(ar_name, "__.SYMDEF/") ||
3121 !strcmp(ar_name, "ARFILENAMES/")) {
3122 /* skip symbol table or archive names */
3123 } else {
3124 if (tcc_load_object_file(s1, fd, file_offset) < 0)
3125 return -1;
3126 }
3127 lseek(fd, file_offset + size, SEEK_SET);
3128 }
3129 return 0;
3130 }
3131
3132 #ifndef TCC_TARGET_PE
3133 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
3134 is referenced by the user (so it should be added as DT_NEEDED in
3135 the generated ELF file) */
3136 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
3137 {
3138 ElfW(Ehdr) ehdr;
3139 ElfW(Shdr) *shdr, *sh, *sh1;
3140 int i, j, nb_syms, nb_dts, sym_bind, ret;
3141 ElfW(Sym) *sym, *dynsym;
3142 ElfW(Dyn) *dt, *dynamic;
3143 unsigned char *dynstr;
3144 const char *name, *soname;
3145 DLLReference *dllref;
3146
3147 read(fd, &ehdr, sizeof(ehdr));
3148
3149 /* test CPU specific stuff */
3150 if (ehdr.e_ident[5] != ELFDATA2LSB ||
3151 ehdr.e_machine != EM_TCC_TARGET) {
3152 tcc_error_noabort("bad architecture");
3153 return -1;
3154 }
3155
3156 /* read sections */
3157 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
3158
3159 /* load dynamic section and dynamic symbols */
3160 nb_syms = 0;
3161 nb_dts = 0;
3162 dynamic = NULL;
3163 dynsym = NULL; /* avoid warning */
3164 dynstr = NULL; /* avoid warning */
3165 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
3166 switch(sh->sh_type) {
3167 case SHT_DYNAMIC:
3168 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
3169 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
3170 break;
3171 case SHT_DYNSYM:
3172 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
3173 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
3174 sh1 = &shdr[sh->sh_link];
3175 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
3176 break;
3177 default:
3178 break;
3179 }
3180 }
3181
3182 /* compute the real library name */
3183 soname = tcc_basename(filename);
3184
3185 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3186 if (dt->d_tag == DT_SONAME) {
3187 soname = (char *) dynstr + dt->d_un.d_val;
3188 }
3189 }
3190
3191 /* if the dll is already loaded, do not load it */
3192 for(i = 0; i < s1->nb_loaded_dlls; i++) {
3193 dllref = s1->loaded_dlls[i];
3194 if (!strcmp(soname, dllref->name)) {
3195 /* but update level if needed */
3196 if (level < dllref->level)
3197 dllref->level = level;
3198 ret = 0;
3199 goto the_end;
3200 }
3201 }
3202
3203 /* add the dll and its level */
3204 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
3205 dllref->level = level;
3206 strcpy(dllref->name, soname);
3207 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
3208
3209 /* add dynamic symbols in dynsym_section */
3210 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
3211 sym_bind = ELFW(ST_BIND)(sym->st_info);
3212 if (sym_bind == STB_LOCAL)
3213 continue;
3214 name = (char *) dynstr + sym->st_name;
3215 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
3216 sym->st_info, sym->st_other, sym->st_shndx, name);
3217 }
3218
3219 /* load all referenced DLLs */
3220 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3221 switch(dt->d_tag) {
3222 case DT_NEEDED:
3223 name = (char *) dynstr + dt->d_un.d_val;
3224 for(j = 0; j < s1->nb_loaded_dlls; j++) {
3225 dllref = s1->loaded_dlls[j];
3226 if (!strcmp(name, dllref->name))
3227 goto already_loaded;
3228 }
3229 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
3230 tcc_error_noabort("referenced dll '%s' not found", name);
3231 ret = -1;
3232 goto the_end;
3233 }
3234 already_loaded:
3235 break;
3236 }
3237 }
3238 ret = 0;
3239 the_end:
3240 tcc_free(dynstr);
3241 tcc_free(dynsym);
3242 tcc_free(dynamic);
3243 tcc_free(shdr);
3244 return ret;
3245 }
3246
3247 #define LD_TOK_NAME 256
3248 #define LD_TOK_EOF (-1)
3249
3250 /* return next ld script token */
3251 static int ld_next(TCCState *s1, char *name, int name_size)
3252 {
3253 int c;
3254 char *q;
3255
3256 redo:
3257 switch(ch) {
3258 case ' ':
3259 case '\t':
3260 case '\f':
3261 case '\v':
3262 case '\r':
3263 case '\n':
3264 inp();
3265 goto redo;
3266 case '/':
3267 minp();
3268 if (ch == '*') {
3269 file->buf_ptr = parse_comment(file->buf_ptr);
3270 ch = file->buf_ptr[0];
3271 goto redo;
3272 } else {
3273 q = name;
3274 *q++ = '/';
3275 goto parse_name;
3276 }
3277 break;
3278 case '\\':
3279 ch = handle_eob();
3280 if (ch != '\\')
3281 goto redo;
3282 /* fall through */
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 q = name;
3342 parse_name:
3343 for(;;) {
3344 if (!((ch >= 'a' && ch <= 'z') ||
3345 (ch >= 'A' && ch <= 'Z') ||
3346 (ch >= '0' && ch <= '9') ||
3347 strchr("/.-_+=$:\\,~", ch)))
3348 break;
3349 if ((q - name) < name_size - 1) {
3350 *q++ = ch;
3351 }
3352 minp();
3353 }
3354 *q = '\0';
3355 c = LD_TOK_NAME;
3356 break;
3357 case CH_EOF:
3358 c = LD_TOK_EOF;
3359 break;
3360 default:
3361 c = ch;
3362 inp();
3363 break;
3364 }
3365 return c;
3366 }
3367
3368 static int ld_add_file(TCCState *s1, const char filename[])
3369 {
3370 int ret;
3371
3372 ret = tcc_add_file_internal(s1, filename, 0, TCC_FILETYPE_BINARY);
3373 if (ret)
3374 ret = tcc_add_dll(s1, filename, 0);
3375 return ret;
3376 }
3377
3378 static inline int new_undef_syms(void)
3379 {
3380 int ret = 0;
3381 ret = new_undef_sym;
3382 new_undef_sym = 0;
3383 return ret;
3384 }
3385
3386 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3387 {
3388 char filename[1024], libname[1024];
3389 int t, group, nblibs = 0, ret = 0;
3390 char **libs = NULL;
3391
3392 group = !strcmp(cmd, "GROUP");
3393 if (!as_needed)
3394 new_undef_syms();
3395 t = ld_next(s1, filename, sizeof(filename));
3396 if (t != '(')
3397 expect("(");
3398 t = ld_next(s1, filename, sizeof(filename));
3399 for(;;) {
3400 libname[0] = '\0';
3401 if (t == LD_TOK_EOF) {
3402 tcc_error_noabort("unexpected end of file");
3403 ret = -1;
3404 goto lib_parse_error;
3405 } else if (t == ')') {
3406 break;
3407 } else if (t == '-') {
3408 t = ld_next(s1, filename, sizeof(filename));
3409 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3410 tcc_error_noabort("library name expected");
3411 ret = -1;
3412 goto lib_parse_error;
3413 }
3414 pstrcpy(libname, sizeof libname, &filename[1]);
3415 if (s1->static_link) {
3416 snprintf(filename, sizeof filename, "lib%s.a", libname);
3417 } else {
3418 snprintf(filename, sizeof filename, "lib%s.so", libname);
3419 }
3420 } else if (t != LD_TOK_NAME) {
3421 tcc_error_noabort("filename expected");
3422 ret = -1;
3423 goto lib_parse_error;
3424 }
3425 if (!strcmp(filename, "AS_NEEDED")) {
3426 ret = ld_add_file_list(s1, cmd, 1);
3427 if (ret)
3428 goto lib_parse_error;
3429 } else {
3430 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3431 if (!as_needed) {
3432 ret = ld_add_file(s1, filename);
3433 if (ret)
3434 goto lib_parse_error;
3435 if (group) {
3436 /* Add the filename *and* the libname to avoid future conversions */
3437 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3438 if (libname[0] != '\0')
3439 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3440 }
3441 }
3442 }
3443 t = ld_next(s1, filename, sizeof(filename));
3444 if (t == ',') {
3445 t = ld_next(s1, filename, sizeof(filename));
3446 }
3447 }
3448 if (group && !as_needed) {
3449 while (new_undef_syms()) {
3450 int i;
3451
3452 for (i = 0; i < nblibs; i ++)
3453 ld_add_file(s1, libs[i]);
3454 }
3455 }
3456 lib_parse_error:
3457 dynarray_reset(&libs, &nblibs);
3458 return ret;
3459 }
3460
3461 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3462 files */
3463 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3464 {
3465 char cmd[64];
3466 char filename[1024];
3467 int t, ret;
3468
3469 ch = handle_eob();
3470 for(;;) {
3471 t = ld_next(s1, cmd, sizeof(cmd));
3472 if (t == LD_TOK_EOF)
3473 return 0;
3474 else if (t != LD_TOK_NAME)
3475 return -1;
3476 if (!strcmp(cmd, "INPUT") ||
3477 !strcmp(cmd, "GROUP")) {
3478 ret = ld_add_file_list(s1, cmd, 0);
3479 if (ret)
3480 return ret;
3481 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3482 !strcmp(cmd, "TARGET")) {
3483 /* ignore some commands */
3484 t = ld_next(s1, cmd, sizeof(cmd));
3485 if (t != '(')
3486 expect("(");
3487 for(;;) {
3488 t = ld_next(s1, filename, sizeof(filename));
3489 if (t == LD_TOK_EOF) {
3490 tcc_error_noabort("unexpected end of file");
3491 return -1;
3492 } else if (t == ')') {
3493 break;
3494 }
3495 }
3496 } else {
3497 return -1;
3498 }
3499 }
3500 return 0;
3501 }
3502 #endif /* !TCC_TARGET_PE */
Tiny C Compiler main repository