shared libs: Build libtcc1.a with -fPIC
[tinycc.git] / tccelf.c
blob7544d2e7f7c8f3562dbe50f3e9f42f08cf61b7b9
1 /*
2 * ELF file handling for TCC
4 * Copyright (c) 2001-2004 Fabrice Bellard
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.
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.
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
21 #include "tcc.h"
23 /* XXX: avoid static variable */
24 static int new_undef_sym = 0; /* Is there a new undefined sym since last new_undef_sym() */
26 ST_FUNC int put_elf_str(Section *s, const char *sym)
28 int offset, len;
29 char *ptr;
31 len = strlen(sym) + 1;
32 offset = s->data_offset;
33 ptr = section_ptr_add(s, len);
34 memcpy(ptr, sym, len);
35 return offset;
38 /* elf symbol hashing function */
39 static unsigned long elf_hash(const unsigned char *name)
41 unsigned long h = 0, g;
43 while (*name) {
44 h = (h << 4) + *name++;
45 g = h & 0xf0000000;
46 if (g)
47 h ^= g >> 24;
48 h &= ~g;
50 return h;
53 /* rebuild hash table of section s */
54 /* NOTE: we do factorize the hash table code to go faster */
55 static void rebuild_hash(Section *s, unsigned int nb_buckets)
57 ElfW(Sym) *sym;
58 int *ptr, *hash, nb_syms, sym_index, h;
59 unsigned char *strtab;
61 strtab = s->link->data;
62 nb_syms = s->data_offset / sizeof(ElfW(Sym));
64 s->hash->data_offset = 0;
65 ptr = section_ptr_add(s->hash, (2 + nb_buckets + nb_syms) * sizeof(int));
66 ptr[0] = nb_buckets;
67 ptr[1] = nb_syms;
68 ptr += 2;
69 hash = ptr;
70 memset(hash, 0, (nb_buckets + 1) * sizeof(int));
71 ptr += nb_buckets + 1;
73 sym = (ElfW(Sym) *)s->data + 1;
74 for(sym_index = 1; sym_index < nb_syms; sym_index++) {
75 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
76 h = elf_hash(strtab + sym->st_name) % nb_buckets;
77 *ptr = hash[h];
78 hash[h] = sym_index;
79 } else {
80 *ptr = 0;
82 ptr++;
83 sym++;
87 /* return the symbol number */
88 ST_FUNC int put_elf_sym(Section *s, addr_t value, unsigned long size,
89 int info, int other, int shndx, const char *name)
91 int name_offset, sym_index;
92 int nbuckets, h;
93 ElfW(Sym) *sym;
94 Section *hs;
96 sym = section_ptr_add(s, sizeof(ElfW(Sym)));
97 if (name)
98 name_offset = put_elf_str(s->link, name);
99 else
100 name_offset = 0;
101 /* XXX: endianness */
102 sym->st_name = name_offset;
103 sym->st_value = value;
104 sym->st_size = size;
105 sym->st_info = info;
106 sym->st_other = other;
107 sym->st_shndx = shndx;
108 sym_index = sym - (ElfW(Sym) *)s->data;
109 hs = s->hash;
110 if (hs) {
111 int *ptr, *base;
112 ptr = section_ptr_add(hs, sizeof(int));
113 base = (int *)hs->data;
114 /* only add global or weak symbols */
115 if (ELFW(ST_BIND)(info) != STB_LOCAL) {
116 /* add another hashing entry */
117 nbuckets = base[0];
118 h = elf_hash((unsigned char *) name) % nbuckets;
119 *ptr = base[2 + h];
120 base[2 + h] = sym_index;
121 base[1]++;
122 /* we resize the hash table */
123 hs->nb_hashed_syms++;
124 if (hs->nb_hashed_syms > 2 * nbuckets) {
125 rebuild_hash(s, 2 * nbuckets);
127 } else {
128 *ptr = 0;
129 base[1]++;
132 return sym_index;
135 /* find global ELF symbol 'name' and return its index. Return 0 if not
136 found. */
137 ST_FUNC int find_elf_sym(Section *s, const char *name)
139 ElfW(Sym) *sym;
140 Section *hs;
141 int nbuckets, sym_index, h;
142 const char *name1;
144 hs = s->hash;
145 if (!hs)
146 return 0;
147 nbuckets = ((int *)hs->data)[0];
148 h = elf_hash((unsigned char *) name) % nbuckets;
149 sym_index = ((int *)hs->data)[2 + h];
150 while (sym_index != 0) {
151 sym = &((ElfW(Sym) *)s->data)[sym_index];
152 name1 = (char *) s->link->data + sym->st_name;
153 if (!strcmp(name, name1))
154 return sym_index;
155 sym_index = ((int *)hs->data)[2 + nbuckets + sym_index];
157 return 0;
160 /* return elf symbol value, signal error if 'err' is nonzero */
161 ST_FUNC addr_t get_elf_sym_addr(TCCState *s, const char *name, int err)
163 int sym_index;
164 ElfW(Sym) *sym;
166 sym_index = find_elf_sym(s->symtab, name);
167 sym = &((ElfW(Sym) *)s->symtab->data)[sym_index];
168 if (!sym_index || sym->st_shndx == SHN_UNDEF) {
169 if (err)
170 tcc_error("%s not defined", name);
171 return 0;
173 return sym->st_value;
176 /* return elf symbol value */
177 LIBTCCAPI void *tcc_get_symbol(TCCState *s, const char *name)
179 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 0);
182 #if defined TCC_IS_NATIVE || defined TCC_TARGET_PE
183 /* return elf symbol value or error */
184 ST_FUNC void* tcc_get_symbol_err(TCCState *s, const char *name)
186 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 1);
188 #endif
190 /* add an elf symbol : check if it is already defined and patch
191 it. Return symbol index. NOTE that sh_num can be SHN_UNDEF. */
192 ST_FUNC int add_elf_sym(Section *s, addr_t value, unsigned long size,
193 int info, int other, int sh_num, const char *name)
195 ElfW(Sym) *esym;
196 int sym_bind, sym_index, sym_type, esym_bind;
197 unsigned char sym_vis, esym_vis, new_vis;
199 sym_bind = ELFW(ST_BIND)(info);
200 sym_type = ELFW(ST_TYPE)(info);
201 sym_vis = ELFW(ST_VISIBILITY)(other);
203 if (sym_bind != STB_LOCAL) {
204 /* we search global or weak symbols */
205 sym_index = find_elf_sym(s, name);
206 if (!sym_index)
207 goto do_def;
208 esym = &((ElfW(Sym) *)s->data)[sym_index];
209 if (esym->st_shndx != SHN_UNDEF) {
210 esym_bind = ELFW(ST_BIND)(esym->st_info);
211 /* propagate the most constraining visibility */
212 /* STV_DEFAULT(0)<STV_PROTECTED(3)<STV_HIDDEN(2)<STV_INTERNAL(1) */
213 esym_vis = ELFW(ST_VISIBILITY)(esym->st_other);
214 if (esym_vis == STV_DEFAULT) {
215 new_vis = sym_vis;
216 } else if (sym_vis == STV_DEFAULT) {
217 new_vis = esym_vis;
218 } else {
219 new_vis = (esym_vis < sym_vis) ? esym_vis : sym_vis;
221 esym->st_other = (esym->st_other & ~ELFW(ST_VISIBILITY)(-1))
222 | new_vis;
223 other = esym->st_other; /* in case we have to patch esym */
224 if (sh_num == SHN_UNDEF) {
225 /* ignore adding of undefined symbol if the
226 corresponding symbol is already defined */
227 } else if (sym_bind == STB_GLOBAL && esym_bind == STB_WEAK) {
228 /* global overrides weak, so patch */
229 goto do_patch;
230 } else if (sym_bind == STB_WEAK && esym_bind == STB_GLOBAL) {
231 /* weak is ignored if already global */
232 } else if (sym_bind == STB_WEAK && esym_bind == STB_WEAK) {
233 /* keep first-found weak definition, ignore subsequents */
234 } else if (sym_vis == STV_HIDDEN || sym_vis == STV_INTERNAL) {
235 /* ignore hidden symbols after */
236 } else if (esym->st_shndx == SHN_COMMON
237 && (sh_num < SHN_LORESERVE || sh_num == SHN_COMMON)) {
238 /* gr: Happens with 'tcc ... -static tcctest.c' on e.g. Ubuntu 6.01
239 No idea if this is the correct solution ... */
240 goto do_patch;
241 } else if (s == tcc_state->dynsymtab_section) {
242 /* we accept that two DLL define the same symbol */
243 } else {
244 #if 0
245 printf("new_bind=%x new_shndx=%x new_vis=%x old_bind=%x old_shndx=%x old_vis=%x\n",
246 sym_bind, sh_num, new_vis, esym_bind, esym->st_shndx, esym_vis);
247 #endif
248 tcc_error_noabort("'%s' defined twice", name);
250 } else {
251 do_patch:
252 esym->st_info = ELFW(ST_INFO)(sym_bind, sym_type);
253 esym->st_shndx = sh_num;
254 new_undef_sym = 1;
255 esym->st_value = value;
256 esym->st_size = size;
257 esym->st_other = other;
259 } else {
260 do_def:
261 sym_index = put_elf_sym(s, value, size,
262 ELFW(ST_INFO)(sym_bind, sym_type), other,
263 sh_num, name);
265 return sym_index;
268 /* put relocation */
269 ST_FUNC void put_elf_reloc(Section *symtab, Section *s, unsigned long offset,
270 int type, int symbol)
272 char buf[256];
273 Section *sr;
274 ElfW_Rel *rel;
276 sr = s->reloc;
277 if (!sr) {
278 /* if no relocation section, create it */
279 snprintf(buf, sizeof(buf), REL_SECTION_FMT, s->name);
280 /* if the symtab is allocated, then we consider the relocation
281 are also */
282 sr = new_section(tcc_state, buf, SHT_RELX, symtab->sh_flags);
283 sr->sh_entsize = sizeof(ElfW_Rel);
284 sr->link = symtab;
285 sr->sh_info = s->sh_num;
286 s->reloc = sr;
288 rel = section_ptr_add(sr, sizeof(ElfW_Rel));
289 rel->r_offset = offset;
290 rel->r_info = ELFW(R_INFO)(symbol, type);
291 #ifdef TCC_TARGET_X86_64
292 rel->r_addend = 0;
293 #endif
296 /* put stab debug information */
298 ST_FUNC void put_stabs(const char *str, int type, int other, int desc,
299 unsigned long value)
301 Stab_Sym *sym;
303 sym = section_ptr_add(stab_section, sizeof(Stab_Sym));
304 if (str) {
305 sym->n_strx = put_elf_str(stabstr_section, str);
306 } else {
307 sym->n_strx = 0;
309 sym->n_type = type;
310 sym->n_other = other;
311 sym->n_desc = desc;
312 sym->n_value = value;
315 ST_FUNC void put_stabs_r(const char *str, int type, int other, int desc,
316 unsigned long value, Section *sec, int sym_index)
318 put_stabs(str, type, other, desc, value);
319 put_elf_reloc(symtab_section, stab_section,
320 stab_section->data_offset - sizeof(unsigned int),
321 R_DATA_32, sym_index);
324 ST_FUNC void put_stabn(int type, int other, int desc, int value)
326 put_stabs(NULL, type, other, desc, value);
329 ST_FUNC void put_stabd(int type, int other, int desc)
331 put_stabs(NULL, type, other, desc, 0);
334 /* Browse each elem of type <type> in section <sec> starting at elem <startoff>
335 using variable <elem> */
336 #define for_each_elem(sec, startoff, elem, type) \
337 for (elem = (type *) sec->data + startoff; \
338 elem < (type *) (sec->data + sec->data_offset); elem++)
340 /* In an ELF file symbol table, the local symbols must appear below
341 the global and weak ones. Since TCC cannot sort it while generating
342 the code, we must do it after. All the relocation tables are also
343 modified to take into account the symbol table sorting */
344 static void sort_syms(TCCState *s1, Section *s)
346 int *old_to_new_syms;
347 ElfW(Sym) *new_syms;
348 int nb_syms, i;
349 ElfW(Sym) *p, *q;
350 ElfW_Rel *rel;
351 Section *sr;
352 int type, sym_index;
354 nb_syms = s->data_offset / sizeof(ElfW(Sym));
355 new_syms = tcc_malloc(nb_syms * sizeof(ElfW(Sym)));
356 old_to_new_syms = tcc_malloc(nb_syms * sizeof(int));
358 /* first pass for local symbols */
359 p = (ElfW(Sym) *)s->data;
360 q = new_syms;
361 for(i = 0; i < nb_syms; i++) {
362 if (ELFW(ST_BIND)(p->st_info) == STB_LOCAL) {
363 old_to_new_syms[i] = q - new_syms;
364 *q++ = *p;
366 p++;
368 /* save the number of local symbols in section header */
369 s->sh_info = q - new_syms;
371 /* then second pass for non local symbols */
372 p = (ElfW(Sym) *)s->data;
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;
378 p++;
381 /* we copy the new symbols to the old */
382 memcpy(s->data, new_syms, nb_syms * sizeof(ElfW(Sym)));
383 tcc_free(new_syms);
385 /* now we modify all the relocations */
386 for(i = 1; i < s1->nb_sections; i++) {
387 sr = s1->sections[i];
388 if (sr->sh_type == SHT_RELX && sr->link == s) {
389 for_each_elem(sr, 0, rel, ElfW_Rel) {
390 sym_index = ELFW(R_SYM)(rel->r_info);
391 type = ELFW(R_TYPE)(rel->r_info);
392 sym_index = old_to_new_syms[sym_index];
393 rel->r_info = ELFW(R_INFO)(sym_index, type);
398 tcc_free(old_to_new_syms);
401 /* relocate common symbols in the .bss section */
402 ST_FUNC void relocate_common_syms(void)
404 ElfW(Sym) *sym;
405 unsigned long offset, align;
407 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
408 if (sym->st_shndx == SHN_COMMON) {
409 /* align symbol */
410 align = sym->st_value;
411 offset = bss_section->data_offset;
412 offset = (offset + align - 1) & -align;
413 sym->st_value = offset;
414 sym->st_shndx = bss_section->sh_num;
415 offset += sym->st_size;
416 bss_section->data_offset = offset;
421 /* relocate symbol table, resolve undefined symbols if do_resolve is
422 true and output error if undefined symbol. */
423 ST_FUNC void relocate_syms(TCCState *s1, int do_resolve)
425 ElfW(Sym) *sym, *esym;
426 int sym_bind, sh_num, sym_index;
427 const char *name;
429 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
430 sh_num = sym->st_shndx;
431 if (sh_num == SHN_UNDEF) {
432 name = (char *) strtab_section->data + sym->st_name;
433 /* Use ld.so to resolve symbol for us (for tcc -run) */
434 if (do_resolve) {
435 #if defined TCC_IS_NATIVE && !defined _WIN32
436 void *addr;
437 name = (char *) symtab_section->link->data + sym->st_name;
438 addr = resolve_sym(s1, name);
439 if (addr) {
440 sym->st_value = (addr_t)addr;
441 goto found;
443 #endif
444 } else if (s1->dynsym) {
445 /* if dynamic symbol exist, then use it */
446 sym_index = find_elf_sym(s1->dynsym, name);
447 if (sym_index) {
448 esym = &((ElfW(Sym) *)s1->dynsym->data)[sym_index];
449 sym->st_value = esym->st_value;
450 goto found;
453 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
454 it */
455 if (!strcmp(name, "_fp_hw"))
456 goto found;
457 /* only weak symbols are accepted to be undefined. Their
458 value is zero */
459 sym_bind = ELFW(ST_BIND)(sym->st_info);
460 if (sym_bind == STB_WEAK) {
461 sym->st_value = 0;
462 } else {
463 tcc_error_noabort("undefined symbol '%s'", name);
465 } else if (sh_num < SHN_LORESERVE) {
466 /* add section base */
467 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
469 found: ;
473 #ifdef TCC_HAS_RUNTIME_PLTGOT
474 #ifdef TCC_TARGET_X86_64
475 #define JMP_TABLE_ENTRY_SIZE 14
476 static addr_t add_jmp_table(TCCState *s1, addr_t val)
478 char *p = s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset;
479 s1->runtime_plt_and_got_offset += JMP_TABLE_ENTRY_SIZE;
480 /* jmp *0x0(%rip) */
481 p[0] = 0xff;
482 p[1] = 0x25;
483 *(int *)(p + 2) = 0;
484 *(addr_t *)(p + 6) = val;
485 return (addr_t)p;
488 static addr_t add_got_table(TCCState *s1, addr_t val)
490 addr_t *p = (addr_t *)(s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset);
491 s1->runtime_plt_and_got_offset += sizeof(addr_t);
492 *p = val;
493 return (addr_t)p;
495 #elif defined TCC_TARGET_ARM
496 #define JMP_TABLE_ENTRY_SIZE 8
497 static addr_t add_jmp_table(TCCState *s1, int val)
499 uint32_t *p = (uint32_t *)(s1->runtime_plt_and_got + s1->runtime_plt_and_got_offset);
500 s1->runtime_plt_and_got_offset += JMP_TABLE_ENTRY_SIZE;
501 /* ldr pc, [pc, #-4] */
502 p[0] = 0xE51FF004;
503 p[1] = val;
504 return (addr_t)p;
506 #endif
507 #endif /* def TCC_HAS_RUNTIME_PLTGOT */
509 /* relocate a given section (CPU dependent) by applying the relocations
510 in the associated relocation section */
511 ST_FUNC void relocate_section(TCCState *s1, Section *s)
513 Section *sr = s->reloc;
514 ElfW_Rel *rel;
515 ElfW(Sym) *sym;
516 int type, sym_index;
517 unsigned char *ptr;
518 addr_t val, addr;
519 #if defined TCC_TARGET_I386 || defined TCC_TARGET_X86_64
520 ElfW_Rel *qrel = (ElfW_Rel *) sr->data; /* ptr to next reloc entry reused */
521 int esym_index;
522 #endif
524 for_each_elem(sr, 0, rel, ElfW_Rel) {
525 ptr = s->data + rel->r_offset;
527 sym_index = ELFW(R_SYM)(rel->r_info);
528 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
529 val = sym->st_value;
530 #ifdef TCC_TARGET_X86_64
531 val += rel->r_addend;
532 #endif
533 type = ELFW(R_TYPE)(rel->r_info);
534 addr = s->sh_addr + rel->r_offset;
536 /* CPU specific */
537 switch(type) {
538 #if defined(TCC_TARGET_I386)
539 case R_386_32:
540 if (s1->output_type == TCC_OUTPUT_DLL) {
541 esym_index = s1->symtab_to_dynsym[sym_index];
542 qrel->r_offset = rel->r_offset;
543 if (esym_index) {
544 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_32);
545 qrel++;
546 break;
547 } else {
548 qrel->r_info = ELFW(R_INFO)(0, R_386_RELATIVE);
549 qrel++;
552 *(int *)ptr += val;
553 break;
554 case R_386_PC32:
555 if (s1->output_type == TCC_OUTPUT_DLL) {
556 /* DLL relocation */
557 esym_index = s1->symtab_to_dynsym[sym_index];
558 if (esym_index) {
559 qrel->r_offset = rel->r_offset;
560 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_PC32);
561 qrel++;
562 break;
565 *(int *)ptr += val - addr;
566 break;
567 case R_386_PLT32:
568 *(int *)ptr += val - addr;
569 break;
570 case R_386_GLOB_DAT:
571 case R_386_JMP_SLOT:
572 *(int *)ptr = val;
573 break;
574 case R_386_GOTPC:
575 *(int *)ptr += s1->got->sh_addr - addr;
576 break;
577 case R_386_GOTOFF:
578 *(int *)ptr += val - s1->got->sh_addr;
579 break;
580 case R_386_GOT32:
581 /* we load the got offset */
582 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
583 break;
584 case R_386_16:
585 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY) {
586 output_file:
587 tcc_error("can only produce 16-bit binary files");
589 *(short *)ptr += val;
590 break;
591 case R_386_PC16:
592 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY)
593 goto output_file;
594 *(short *)ptr += val - addr;
595 break;
596 #elif defined(TCC_TARGET_ARM)
597 case R_ARM_PC24:
598 case R_ARM_CALL:
599 case R_ARM_JUMP24:
600 case R_ARM_PLT32:
602 int x, is_thumb, is_call, h, blx_avail, is_bl, th_ko;
603 x = (*(int *) ptr) & 0xffffff;
604 (*(int *)ptr) &= 0xff000000;
605 if (x & 0x800000)
606 x -= 0x1000000;
607 x <<= 2;
608 blx_avail = (TCC_ARM_VERSION >= 5);
609 is_thumb = val & 1;
610 is_bl = (*(unsigned *) ptr) >> 24 == 0xeb;
611 is_call = (type == R_ARM_CALL || (type == R_ARM_PC24 && is_bl));
612 x += val - addr;
613 h = x & 2;
614 th_ko = (x & 3) && (!blx_avail || !is_call);
615 #ifdef TCC_HAS_RUNTIME_PLTGOT
616 if (s1->output_type == TCC_OUTPUT_MEMORY) {
617 if (th_ko || x >= 0x2000000 || x < -0x2000000) {
618 x += add_jmp_table(s1, val) - val; /* add veneer */
619 th_ko = (x & 3) && (!blx_avail || !is_call);
620 is_thumb = 0; /* Veneer uses ARM instructions */
623 #endif
624 if (th_ko || x >= 0x2000000 || x < -0x2000000)
625 tcc_error("can't relocate value at %x",addr);
626 x >>= 2;
627 x &= 0xffffff;
628 /* Only reached if blx is avail and it is a call */
629 if (is_thumb) {
630 x |= h << 24;
631 (*(int *)ptr) = 0xfa << 24; /* bl -> blx */
633 (*(int *) ptr) |= x;
635 break;
636 /* Since these relocations only concern Thumb-2 and blx instruction was
637 introduced before Thumb-2, we can assume blx is available and not
638 guard its use */
639 case R_ARM_THM_PC22:
640 case R_ARM_THM_JUMP24:
642 int x, hi, lo, s, j1, j2, i1, i2, imm10, imm11;
643 int to_thumb, is_call, to_plt, blx_bit = 1 << 12;
644 Section *plt;
646 /* weak reference */
647 if (sym->st_shndx == SHN_UNDEF &&
648 ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
649 break;
651 /* Get initial offset */
652 hi = (*(uint16_t *)ptr);
653 lo = (*(uint16_t *)(ptr+2));
654 s = (hi >> 10) & 1;
655 j1 = (lo >> 13) & 1;
656 j2 = (lo >> 11) & 1;
657 i1 = (j1 ^ s) ^ 1;
658 i2 = (j2 ^ s) ^ 1;
659 imm10 = hi & 0x3ff;
660 imm11 = lo & 0x7ff;
661 x = (s << 24) | (i1 << 23) | (i2 << 22) |
662 (imm10 << 12) | (imm11 << 1);
663 if (x & 0x01000000)
664 x -= 0x02000000;
666 /* Relocation infos */
667 to_thumb = val & 1;
668 plt = s1->plt;
669 to_plt = (val >= plt->sh_addr) &&
670 (val < plt->sh_addr + plt->data_offset);
671 is_call = (type == R_ARM_THM_PC22);
673 /* Compute final offset */
674 if (to_plt && !is_call) /* Point to 1st instr of Thumb stub */
675 x -= 4;
676 x += val - addr;
677 if (!to_thumb && is_call) {
678 blx_bit = 0; /* bl -> blx */
679 x = (x + 3) & -4; /* Compute offset from aligned PC */
682 /* Check that relocation is possible
683 * offset must not be out of range
684 * if target is to be entered in arm mode:
685 - bit 1 must not set
686 - instruction must be a call (bl) or a jump to PLT */
687 if (!to_thumb || x >= 0x1000000 || x < -0x1000000)
688 if (to_thumb || (val & 2) || (!is_call && !to_plt))
689 tcc_error("can't relocate value at %x",addr);
691 /* Compute and store final offset */
692 s = (x >> 24) & 1;
693 i1 = (x >> 23) & 1;
694 i2 = (x >> 22) & 1;
695 j1 = s ^ (i1 ^ 1);
696 j2 = s ^ (i2 ^ 1);
697 imm10 = (x >> 12) & 0x3ff;
698 imm11 = (x >> 1) & 0x7ff;
699 (*(uint16_t *)ptr) = (uint16_t) ((hi & 0xf800) |
700 (s << 10) | imm10);
701 (*(uint16_t *)(ptr+2)) = (uint16_t) ((lo & 0xc000) |
702 (j1 << 13) | blx_bit | (j2 << 11) |
703 imm11);
705 break;
706 case R_ARM_MOVT_ABS:
707 case R_ARM_MOVW_ABS_NC:
709 int x, imm4, imm12;
710 if (type == R_ARM_MOVT_ABS)
711 val >>= 16;
712 imm12 = val & 0xfff;
713 imm4 = (val >> 12) & 0xf;
714 x = (imm4 << 16) | imm12;
715 if (type == R_ARM_THM_MOVT_ABS)
716 *(int *)ptr |= x;
717 else
718 *(int *)ptr += x;
720 break;
721 case R_ARM_THM_MOVT_ABS:
722 case R_ARM_THM_MOVW_ABS_NC:
724 int x, i, imm4, imm3, imm8;
725 if (type == R_ARM_THM_MOVT_ABS)
726 val >>= 16;
727 imm8 = val & 0xff;
728 imm3 = (val >> 8) & 0x7;
729 i = (val >> 11) & 1;
730 imm4 = (val >> 12) & 0xf;
731 x = (imm3 << 28) | (imm8 << 16) | (i << 10) | imm4;
732 if (type == R_ARM_THM_MOVT_ABS)
733 *(int *)ptr |= x;
734 else
735 *(int *)ptr += x;
737 break;
738 case R_ARM_PREL31:
740 int x;
741 x = (*(int *)ptr) & 0x7fffffff;
742 (*(int *)ptr) &= 0x80000000;
743 x = (x * 2) / 2;
744 x += val - addr;
745 if((x^(x>>1))&0x40000000)
746 tcc_error("can't relocate value at %x",addr);
747 (*(int *)ptr) |= x & 0x7fffffff;
749 case R_ARM_ABS32:
750 *(int *)ptr += val;
751 break;
752 case R_ARM_REL32:
753 *(int *)ptr += val - addr;
754 break;
755 case R_ARM_GOTPC:
756 *(int *)ptr += s1->got->sh_addr - addr;
757 break;
758 case R_ARM_GOTOFF:
759 *(int *)ptr += val - s1->got->sh_addr;
760 break;
761 case R_ARM_GOT32:
762 /* we load the got offset */
763 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
764 break;
765 case R_ARM_COPY:
766 break;
767 case R_ARM_V4BX:
768 /* trade Thumb support for ARMv4 support */
769 if ((0x0ffffff0 & *(int*)ptr) == 0x012FFF10)
770 *(int*)ptr ^= 0xE12FFF10 ^ 0xE1A0F000; /* BX Rm -> MOV PC, Rm */
771 break;
772 default:
773 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
774 type, (unsigned)addr, ptr, (unsigned)val);
775 break;
776 #elif defined(TCC_TARGET_C67)
777 case R_C60_32:
778 *(int *)ptr += val;
779 break;
780 case R_C60LO16:
782 uint32_t orig;
784 /* put the low 16 bits of the absolute address
785 add to what is already there */
787 orig = ((*(int *)(ptr )) >> 7) & 0xffff;
788 orig |= (((*(int *)(ptr+4)) >> 7) & 0xffff) << 16;
790 /* patch both at once - assumes always in pairs Low - High */
792 *(int *) ptr = (*(int *) ptr & (~(0xffff << 7)) ) | (((val+orig) & 0xffff) << 7);
793 *(int *)(ptr+4) = (*(int *)(ptr+4) & (~(0xffff << 7)) ) | ((((val+orig)>>16) & 0xffff) << 7);
795 break;
796 case R_C60HI16:
797 break;
798 default:
799 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
800 type, (unsigned)addr, ptr, (unsigned)val);
801 break;
802 #elif defined(TCC_TARGET_X86_64)
803 case R_X86_64_64:
804 if (s1->output_type == TCC_OUTPUT_DLL) {
805 esym_index = s1->symtab_to_dynsym[sym_index];
806 qrel->r_offset = rel->r_offset;
807 if (esym_index) {
808 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_64);
809 qrel->r_addend = rel->r_addend;
810 qrel++;
811 break;
812 } else {
813 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
814 qrel->r_addend = *(long long *)ptr + val;
815 qrel++;
818 *(long long *)ptr += val;
819 break;
820 case R_X86_64_32:
821 case R_X86_64_32S:
822 if (s1->output_type == TCC_OUTPUT_DLL) {
823 /* XXX: this logic may depend on TCC's codegen
824 now TCC uses R_X86_64_32 even for a 64bit pointer */
825 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
826 qrel->r_addend = *(int *)ptr + val;
827 qrel++;
829 *(int *)ptr += val;
830 break;
832 case R_X86_64_PC32:
833 if (s1->output_type == TCC_OUTPUT_DLL) {
834 /* DLL relocation */
835 esym_index = s1->symtab_to_dynsym[sym_index];
836 if (esym_index) {
837 qrel->r_offset = rel->r_offset;
838 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_PC32);
839 qrel->r_addend = *(int *)ptr;
840 qrel++;
841 break;
844 goto plt32pc32;
846 case R_X86_64_PLT32:
847 /* We've put the PLT slot offset into r_addend when generating
848 it, and that's what we must use as relocation value (adjusted
849 by section offset of course). */
850 if (s1->output_type != TCC_OUTPUT_MEMORY)
851 val = s1->plt->sh_addr + rel->r_addend;
852 /* fallthrough. */
854 plt32pc32:
856 long long diff;
857 diff = (long long)val - addr;
858 if (diff <= -2147483647 || diff > 2147483647) {
859 #ifdef TCC_HAS_RUNTIME_PLTGOT
860 /* XXX: naive support for over 32bit jump */
861 if (s1->output_type == TCC_OUTPUT_MEMORY) {
862 val = (add_jmp_table(s1, val - rel->r_addend) +
863 rel->r_addend);
864 diff = val - addr;
866 #endif
867 if (diff <= -2147483647 || diff > 2147483647) {
868 tcc_error("internal error: relocation failed");
871 *(int *)ptr += diff;
873 break;
874 case R_X86_64_GLOB_DAT:
875 case R_X86_64_JUMP_SLOT:
876 /* They don't need addend */
877 *(int *)ptr = val - rel->r_addend;
878 break;
879 case R_X86_64_GOTPCREL:
880 #ifdef TCC_HAS_RUNTIME_PLTGOT
881 if (s1->output_type == TCC_OUTPUT_MEMORY) {
882 val = add_got_table(s1, val - rel->r_addend) + rel->r_addend;
883 *(int *)ptr += val - addr;
884 break;
886 #endif
887 *(int *)ptr += (s1->got->sh_addr - addr +
888 s1->sym_attrs[sym_index].got_offset - 4);
889 break;
890 case R_X86_64_GOTTPOFF:
891 *(int *)ptr += val - s1->got->sh_addr;
892 break;
893 case R_X86_64_GOT32:
894 /* we load the got offset */
895 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
896 break;
897 #else
898 #error unsupported processor
899 #endif
902 /* if the relocation is allocated, we change its symbol table */
903 if (sr->sh_flags & SHF_ALLOC)
904 sr->link = s1->dynsym;
907 /* relocate relocation table in 'sr' */
908 static void relocate_rel(TCCState *s1, Section *sr)
910 Section *s;
911 ElfW_Rel *rel;
913 s = s1->sections[sr->sh_info];
914 for_each_elem(sr, 0, rel, ElfW_Rel)
915 rel->r_offset += s->sh_addr;
918 /* count the number of dynamic relocations so that we can reserve
919 their space */
920 static int prepare_dynamic_rel(TCCState *s1, Section *sr)
922 ElfW_Rel *rel;
923 int sym_index, esym_index, type, count;
925 count = 0;
926 for_each_elem(sr, 0, rel, ElfW_Rel) {
927 sym_index = ELFW(R_SYM)(rel->r_info);
928 type = ELFW(R_TYPE)(rel->r_info);
929 switch(type) {
930 #if defined(TCC_TARGET_I386)
931 case R_386_32:
932 #elif defined(TCC_TARGET_X86_64)
933 case R_X86_64_32:
934 case R_X86_64_32S:
935 case R_X86_64_64:
936 #endif
937 count++;
938 break;
939 #if defined(TCC_TARGET_I386)
940 case R_386_PC32:
941 #elif defined(TCC_TARGET_X86_64)
942 case R_X86_64_PC32:
943 #endif
944 esym_index = s1->symtab_to_dynsym[sym_index];
945 if (esym_index)
946 count++;
947 break;
948 default:
949 break;
952 if (count) {
953 /* allocate the section */
954 sr->sh_flags |= SHF_ALLOC;
955 sr->sh_size = count * sizeof(ElfW_Rel);
957 return count;
960 static struct sym_attr *alloc_sym_attr(TCCState *s1, int index)
962 int n;
963 struct sym_attr *tab;
965 if (index >= s1->nb_sym_attrs) {
966 /* find immediately bigger power of 2 and reallocate array */
967 n = 1;
968 while (index >= n)
969 n *= 2;
970 tab = tcc_realloc(s1->sym_attrs, n * sizeof(*s1->sym_attrs));
971 s1->sym_attrs = tab;
972 memset(s1->sym_attrs + s1->nb_sym_attrs, 0,
973 (n - s1->nb_sym_attrs) * sizeof(*s1->sym_attrs));
974 s1->nb_sym_attrs = n;
976 return &s1->sym_attrs[index];
979 /* XXX: suppress that */
980 static void put32(unsigned char *p, uint32_t val)
982 p[0] = val;
983 p[1] = val >> 8;
984 p[2] = val >> 16;
985 p[3] = val >> 24;
988 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_ARM) || \
989 defined(TCC_TARGET_X86_64)
990 static uint32_t get32(unsigned char *p)
992 return p[0] | (p[1] << 8) | (p[2] << 16) | (p[3] << 24);
994 #endif
996 static void build_got(TCCState *s1)
998 unsigned char *ptr;
1000 /* if no got, then create it */
1001 s1->got = new_section(s1, ".got", SHT_PROGBITS, SHF_ALLOC | SHF_WRITE);
1002 s1->got->sh_entsize = 4;
1003 add_elf_sym(symtab_section, 0, 4, ELFW(ST_INFO)(STB_GLOBAL, STT_OBJECT),
1004 0, s1->got->sh_num, "_GLOBAL_OFFSET_TABLE_");
1005 ptr = section_ptr_add(s1->got, 3 * PTR_SIZE);
1006 #if PTR_SIZE == 4
1007 /* keep space for _DYNAMIC pointer, if present */
1008 put32(ptr, 0);
1009 /* two dummy got entries */
1010 put32(ptr + 4, 0);
1011 put32(ptr + 8, 0);
1012 #else
1013 /* keep space for _DYNAMIC pointer, if present */
1014 put32(ptr, 0);
1015 put32(ptr + 4, 0);
1016 /* two dummy got entries */
1017 put32(ptr + 8, 0);
1018 put32(ptr + 12, 0);
1019 put32(ptr + 16, 0);
1020 put32(ptr + 20, 0);
1021 #endif
1024 /* put a got or plt entry corresponding to a symbol in symtab_section. 'size'
1025 and 'info' can be modifed if more precise info comes from the DLL.
1026 Returns offset of GOT or PLT slot. */
1027 static unsigned long put_got_entry(TCCState *s1,
1028 int reloc_type, unsigned long size, int info,
1029 int sym_index)
1031 int index, need_plt_entry;
1032 const char *name;
1033 ElfW(Sym) *sym;
1034 unsigned long offset;
1035 int *ptr;
1036 struct sym_attr *symattr;
1038 if (!s1->got)
1039 build_got(s1);
1041 need_plt_entry = s1->dynsym &&
1042 #ifdef TCC_TARGET_X86_64
1043 (reloc_type == R_X86_64_JUMP_SLOT);
1044 #elif defined(TCC_TARGET_I386)
1045 (reloc_type == R_386_JMP_SLOT);
1046 #elif defined(TCC_TARGET_ARM)
1047 (reloc_type == R_ARM_JUMP_SLOT);
1048 #else
1050 #endif
1052 /* If a got/plt entry already exists for that symbol, no need to add one */
1053 if (sym_index < s1->nb_sym_attrs) {
1054 if (need_plt_entry && s1->sym_attrs[sym_index].plt_offset)
1055 return s1->sym_attrs[sym_index].plt_offset;
1056 else if (!need_plt_entry && s1->sym_attrs[sym_index].got_offset)
1057 return s1->sym_attrs[sym_index].got_offset;
1060 symattr = alloc_sym_attr(s1, sym_index);
1062 /* Only store the GOT offset if it's not generated for the PLT entry. */
1063 if (!need_plt_entry)
1064 symattr->got_offset = s1->got->data_offset;
1066 if (s1->dynsym) {
1067 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1068 name = (char *) symtab_section->link->data + sym->st_name;
1069 offset = sym->st_value;
1070 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_X86_64)
1071 if (need_plt_entry) {
1072 Section *plt;
1073 uint8_t *p;
1074 int modrm;
1075 unsigned long relofs;
1077 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1078 modrm = 0x25;
1079 #else
1080 /* if we build a DLL, we add a %ebx offset */
1081 if (s1->output_type == TCC_OUTPUT_DLL)
1082 modrm = 0xa3;
1083 else
1084 modrm = 0x25;
1085 #endif
1087 /* add a PLT entry */
1088 plt = s1->plt;
1089 if (plt->data_offset == 0) {
1090 /* first plt entry */
1091 p = section_ptr_add(plt, 16);
1092 p[0] = 0xff; /* pushl got + PTR_SIZE */
1093 p[1] = modrm + 0x10;
1094 put32(p + 2, PTR_SIZE);
1095 p[6] = 0xff; /* jmp *(got + PTR_SIZE * 2) */
1096 p[7] = modrm;
1097 put32(p + 8, PTR_SIZE * 2);
1100 /* The PLT slot refers to the relocation entry it needs
1101 via offset. The reloc entry is created below, so its
1102 offset is the current data_offset. */
1103 relofs = s1->got->reloc ? s1->got->reloc->data_offset : 0;
1104 symattr->plt_offset = plt->data_offset;
1105 p = section_ptr_add(plt, 16);
1106 p[0] = 0xff; /* jmp *(got + x) */
1107 p[1] = modrm;
1108 put32(p + 2, s1->got->data_offset);
1109 p[6] = 0x68; /* push $xxx */
1110 #ifdef TCC_TARGET_X86_64
1111 /* On x86-64, the relocation is referred to by _index_. */
1112 put32(p + 7, relofs / sizeof (ElfW_Rel));
1113 #else
1114 put32(p + 7, relofs);
1115 #endif
1116 p[11] = 0xe9; /* jmp plt_start */
1117 put32(p + 12, -(plt->data_offset));
1119 /* If this was an UNDEF symbol set the offset in the
1120 dynsymtab to the PLT slot, so that PC32 relocs to it
1121 can be resolved. */
1122 if (sym->st_shndx == SHN_UNDEF)
1123 offset = plt->data_offset - 16;
1125 #elif defined(TCC_TARGET_ARM)
1126 if (need_plt_entry) {
1127 Section *plt;
1128 uint8_t *p;
1130 /* if we build a DLL, we add a %ebx offset */
1131 if (s1->output_type == TCC_OUTPUT_DLL)
1132 tcc_error("DLLs unimplemented!");
1134 /* add a PLT entry */
1135 plt = s1->plt;
1136 if (plt->data_offset == 0) {
1137 /* first plt entry */
1138 p = section_ptr_add(plt, 16);
1139 put32(p, 0xe52de004); /* push {lr} */
1140 put32(p+4, 0xe59fe010); /* ldr lr, [pc, #16] */
1141 put32(p+8, 0xe08fe00e); /* add lr, pc, lr */
1142 put32(p+12, 0xe5bef008); /* ldr pc, [lr, #8]! */
1145 symattr->plt_offset = plt->data_offset;
1146 if (symattr->plt_thumb_stub) {
1147 p = section_ptr_add(plt, 20);
1148 put32(p, 0x4778); /* bx pc */
1149 put32(p+2, 0x46c0); /* nop */
1150 p += 4;
1151 } else
1152 p = section_ptr_add(plt, 16);
1153 put32(p, 0xe59fc004); /* ldr ip, [pc, #4] ; GOT entry offset */
1154 put32(p+4, 0xe08fc00c); /* add ip, pc, ip ; addr of GOT entry */
1155 put32(p+8, 0xe59cf000); /* ldr pc, [ip] ; jump to GOT entry */
1156 put32(p+12, s1->got->data_offset); /* GOT entry off once patched */
1158 /* the symbol is modified so that it will be relocated to
1159 the PLT */
1160 if (s1->output_type == TCC_OUTPUT_EXE)
1161 offset = plt->data_offset - 16;
1163 #elif defined(TCC_TARGET_C67)
1164 tcc_error("C67 got not implemented");
1165 #else
1166 #error unsupported CPU
1167 #endif
1168 /* XXX This might generate multiple syms for name. */
1169 index = put_elf_sym(s1->dynsym, offset,
1170 size, info, 0, sym->st_shndx, name);
1171 /* Create the relocation (it's against the GOT for PLT
1172 and GOT relocs). */
1173 put_elf_reloc(s1->dynsym, s1->got,
1174 s1->got->data_offset,
1175 reloc_type, index);
1177 /* And now create the GOT slot itself. */
1178 ptr = section_ptr_add(s1->got, PTR_SIZE);
1179 *ptr = 0;
1180 if (need_plt_entry)
1181 return symattr->plt_offset;
1182 else
1183 return symattr->got_offset;
1186 /* build GOT and PLT entries */
1187 ST_FUNC void build_got_entries(TCCState *s1)
1189 Section *s;
1190 ElfW_Rel *rel;
1191 ElfW(Sym) *sym;
1192 int i, type, reloc_type, sym_index;
1194 for(i = 1; i < s1->nb_sections; i++) {
1195 s = s1->sections[i];
1196 if (s->sh_type != SHT_RELX)
1197 continue;
1198 /* no need to handle got relocations */
1199 if (s->link != symtab_section)
1200 continue;
1201 for_each_elem(s, 0, rel, ElfW_Rel) {
1202 type = ELFW(R_TYPE)(rel->r_info);
1203 switch(type) {
1204 #if defined(TCC_TARGET_I386)
1205 case R_386_GOT32:
1206 case R_386_GOTOFF:
1207 case R_386_GOTPC:
1208 case R_386_PLT32:
1209 if (!s1->got)
1210 build_got(s1);
1211 if (type == R_386_GOT32 || type == R_386_PLT32) {
1212 sym_index = ELFW(R_SYM)(rel->r_info);
1213 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1214 /* look at the symbol got offset. If none, then add one */
1215 if (type == R_386_GOT32)
1216 reloc_type = R_386_GLOB_DAT;
1217 else
1218 reloc_type = R_386_JMP_SLOT;
1219 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1220 sym_index);
1222 break;
1223 #elif defined(TCC_TARGET_ARM)
1224 case R_ARM_GOT32:
1225 case R_ARM_GOTOFF:
1226 case R_ARM_GOTPC:
1227 case R_ARM_PLT32:
1228 if (!s1->got)
1229 build_got(s1);
1230 if (type == R_ARM_GOT32 || type == R_ARM_PLT32) {
1231 sym_index = ELFW(R_SYM)(rel->r_info);
1232 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1233 /* look at the symbol got offset. If none, then add one */
1234 if (type == R_ARM_GOT32)
1235 reloc_type = R_ARM_GLOB_DAT;
1236 else
1237 reloc_type = R_ARM_JUMP_SLOT;
1238 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1239 sym_index);
1241 break;
1242 case R_ARM_THM_JUMP24:
1243 sym_index = ELFW(R_SYM)(rel->r_info);
1244 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1245 /* We are relocating a jump from thumb code to arm code */
1246 if (sym->st_shndx != SHN_UNDEF && !(sym->st_value & 1)) {
1247 int index;
1248 uint8_t *p;
1249 char *name, buf[1024];
1250 Section *text_section;
1252 name = (char *) symtab_section->link->data + sym->st_name;
1253 text_section = s1->sections[sym->st_shndx];
1254 /* Modify reloc to target a thumb stub to switch to ARM */
1255 snprintf(buf, sizeof(buf), "%s_from_thumb", name);
1256 index = put_elf_sym(symtab_section,
1257 text_section->data_offset + 1,
1258 sym->st_size, sym->st_info, 0,
1259 sym->st_shndx, buf);
1260 rel->r_info = ELFW(R_INFO)(index, type);
1261 /* Create a thumb stub fonction to switch to ARM mode */
1262 put_elf_reloc(symtab_section, text_section,
1263 text_section->data_offset + 4, R_ARM_JUMP24,
1264 sym_index);
1265 p = section_ptr_add(text_section, 8);
1266 put32(p, 0x4778); /* bx pc */
1267 put32(p+2, 0x46c0); /* nop */
1268 put32(p+4, 0xeafffffe); /* b $sym */
1270 #elif defined(TCC_TARGET_C67)
1271 case R_C60_GOT32:
1272 case R_C60_GOTOFF:
1273 case R_C60_GOTPC:
1274 case R_C60_PLT32:
1275 if (!s1->got)
1276 build_got(s1);
1277 if (type == R_C60_GOT32 || type == R_C60_PLT32) {
1278 sym_index = ELFW(R_SYM)(rel->r_info);
1279 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1280 /* look at the symbol got offset. If none, then add one */
1281 if (type == R_C60_GOT32)
1282 reloc_type = R_C60_GLOB_DAT;
1283 else
1284 reloc_type = R_C60_JMP_SLOT;
1285 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1286 sym_index);
1288 break;
1289 #elif defined(TCC_TARGET_X86_64)
1290 case R_X86_64_GOT32:
1291 case R_X86_64_GOTTPOFF:
1292 case R_X86_64_GOTPCREL:
1293 case R_X86_64_PLT32:
1294 if (!s1->got)
1295 build_got(s1);
1296 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL ||
1297 type == R_X86_64_PLT32) {
1298 unsigned long ofs;
1299 sym_index = ELFW(R_SYM)(rel->r_info);
1300 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1301 /* look at the symbol got offset. If none, then add one */
1302 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL)
1303 reloc_type = R_X86_64_GLOB_DAT;
1304 else
1305 reloc_type = R_X86_64_JUMP_SLOT;
1306 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1307 sym->st_info, sym_index);
1308 if (type == R_X86_64_PLT32
1309 && s1->output_type != TCC_OUTPUT_MEMORY)
1310 /* We store the place of the generated PLT slot
1311 in our addend. */
1312 rel->r_addend += ofs;
1314 break;
1315 #else
1316 #error unsupported CPU
1317 #endif
1318 default:
1319 break;
1325 ST_FUNC Section *new_symtab(TCCState *s1,
1326 const char *symtab_name, int sh_type, int sh_flags,
1327 const char *strtab_name,
1328 const char *hash_name, int hash_sh_flags)
1330 Section *symtab, *strtab, *hash;
1331 int *ptr, nb_buckets;
1333 symtab = new_section(s1, symtab_name, sh_type, sh_flags);
1334 symtab->sh_entsize = sizeof(ElfW(Sym));
1335 strtab = new_section(s1, strtab_name, SHT_STRTAB, sh_flags);
1336 put_elf_str(strtab, "");
1337 symtab->link = strtab;
1338 put_elf_sym(symtab, 0, 0, 0, 0, 0, NULL);
1340 nb_buckets = 1;
1342 hash = new_section(s1, hash_name, SHT_HASH, hash_sh_flags);
1343 hash->sh_entsize = sizeof(int);
1344 symtab->hash = hash;
1345 hash->link = symtab;
1347 ptr = section_ptr_add(hash, (2 + nb_buckets + 1) * sizeof(int));
1348 ptr[0] = nb_buckets;
1349 ptr[1] = 1;
1350 memset(ptr + 2, 0, (nb_buckets + 1) * sizeof(int));
1351 return symtab;
1354 /* put dynamic tag */
1355 static void put_dt(Section *dynamic, int dt, addr_t val)
1357 ElfW(Dyn) *dyn;
1358 dyn = section_ptr_add(dynamic, sizeof(ElfW(Dyn)));
1359 dyn->d_tag = dt;
1360 dyn->d_un.d_val = val;
1363 static void add_init_array_defines(TCCState *s1, const char *section_name)
1365 Section *s;
1366 long end_offset;
1367 char sym_start[1024];
1368 char sym_end[1024];
1370 snprintf(sym_start, sizeof(sym_start), "__%s_start", section_name + 1);
1371 snprintf(sym_end, sizeof(sym_end), "__%s_end", section_name + 1);
1373 s = find_section(s1, section_name);
1374 if (!s) {
1375 end_offset = 0;
1376 s = data_section;
1377 } else {
1378 end_offset = s->data_offset;
1381 add_elf_sym(symtab_section,
1382 0, 0,
1383 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1384 s->sh_num, sym_start);
1385 add_elf_sym(symtab_section,
1386 end_offset, 0,
1387 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1388 s->sh_num, sym_end);
1391 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1393 #ifdef CONFIG_TCC_BCHECK
1394 unsigned long *ptr;
1395 Section *init_section;
1396 unsigned char *pinit;
1397 int sym_index;
1399 if (0 == s1->do_bounds_check)
1400 return;
1402 /* XXX: add an object file to do that */
1403 ptr = section_ptr_add(bounds_section, sizeof(unsigned long));
1404 *ptr = 0;
1405 add_elf_sym(symtab_section, 0, 0,
1406 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1407 bounds_section->sh_num, "__bounds_start");
1408 #ifdef TCC_TARGET_I386
1409 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1410 /* add 'call __bound_init()' in .init section */
1411 init_section = find_section(s1, ".init");
1412 pinit = section_ptr_add(init_section, 5);
1413 pinit[0] = 0xe8;
1414 put32(pinit + 1, -4);
1415 sym_index = find_elf_sym(symtab_section, "__bound_init");
1416 put_elf_reloc(symtab_section, init_section,
1417 init_section->data_offset - 4, R_386_PC32, sym_index);
1419 #endif
1420 #endif
1423 static inline int tcc_add_support(TCCState *s1, const char *filename)
1425 char buf[1024];
1426 snprintf(buf, sizeof(buf), "%s/%s", s1->tcc_lib_path, filename);
1427 return tcc_add_file(s1, buf);
1430 /* add tcc runtime libraries */
1431 ST_FUNC void tcc_add_runtime(TCCState *s1)
1433 /* add libc */
1434 if (!s1->nostdlib) {
1435 tcc_add_library(s1, "c");
1436 #ifdef CONFIG_USE_LIBGCC
1437 if (!s1->static_link) {
1438 tcc_add_file(s1, TCC_LIBGCC);
1439 tcc_add_support(s1, "libtcc1.a");
1440 } else
1441 tcc_add_support(s1, "libtcc1.a");
1442 #else
1443 tcc_add_support(s1, "libtcc1.a");
1444 #endif
1447 /* tcc_add_bcheck tries to relocate a call to __bound_init in _init so
1448 libtcc1.a must be loaded before for __bound_init to be defined and
1449 crtn.o must be loaded after to not finalize _init too early. */
1450 tcc_add_bcheck(s1);
1452 if (!s1->nostdlib) {
1453 /* add crt end if not memory output */
1454 if (s1->output_type != TCC_OUTPUT_MEMORY)
1455 tcc_add_crt(s1, "crtn.o");
1459 /* add various standard linker symbols (must be done after the
1460 sections are filled (for example after allocating common
1461 symbols)) */
1462 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1464 char buf[1024];
1465 int i;
1466 Section *s;
1468 add_elf_sym(symtab_section,
1469 text_section->data_offset, 0,
1470 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1471 text_section->sh_num, "_etext");
1472 add_elf_sym(symtab_section,
1473 data_section->data_offset, 0,
1474 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1475 data_section->sh_num, "_edata");
1476 add_elf_sym(symtab_section,
1477 bss_section->data_offset, 0,
1478 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1479 bss_section->sh_num, "_end");
1480 /* horrible new standard ldscript defines */
1481 add_init_array_defines(s1, ".preinit_array");
1482 add_init_array_defines(s1, ".init_array");
1483 add_init_array_defines(s1, ".fini_array");
1485 /* add start and stop symbols for sections whose name can be
1486 expressed in C */
1487 for(i = 1; i < s1->nb_sections; i++) {
1488 s = s1->sections[i];
1489 if (s->sh_type == SHT_PROGBITS &&
1490 (s->sh_flags & SHF_ALLOC)) {
1491 const char *p;
1492 int ch;
1494 /* check if section name can be expressed in C */
1495 p = s->name;
1496 for(;;) {
1497 ch = *p;
1498 if (!ch)
1499 break;
1500 if (!isid(ch) && !isnum(ch))
1501 goto next_sec;
1502 p++;
1504 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1505 add_elf_sym(symtab_section,
1506 0, 0,
1507 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1508 s->sh_num, buf);
1509 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1510 add_elf_sym(symtab_section,
1511 s->data_offset, 0,
1512 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1513 s->sh_num, buf);
1515 next_sec: ;
1519 static void tcc_output_binary(TCCState *s1, FILE *f,
1520 const int *sec_order)
1522 Section *s;
1523 int i, offset, size;
1525 offset = 0;
1526 for(i=1;i<s1->nb_sections;i++) {
1527 s = s1->sections[sec_order[i]];
1528 if (s->sh_type != SHT_NOBITS &&
1529 (s->sh_flags & SHF_ALLOC)) {
1530 while (offset < s->sh_offset) {
1531 fputc(0, f);
1532 offset++;
1534 size = s->sh_size;
1535 fwrite(s->data, 1, size, f);
1536 offset += size;
1541 // making this evaluate to true allow valgrind to work on linux
1542 // but when compiled with debug info and then striped
1543 // the compiled programs segfault
1544 // more tought must be applyed here
1545 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1546 #define HAVE_PHDR 1
1547 #define EXTRA_RELITEMS 14
1549 /* move the relocation value from .dynsym to .got */
1550 void patch_dynsym_undef(TCCState *s1, Section *s)
1552 uint32_t *gotd = (void *)s1->got->data;
1553 ElfW(Sym) *sym;
1555 gotd += 3; /* dummy entries in .got */
1556 /* relocate symbols in .dynsym */
1557 for_each_elem(s, 1, sym, ElfW(Sym)) {
1558 if (sym->st_shndx == SHN_UNDEF) {
1559 *gotd++ = sym->st_value + 6; /* XXX 6 is magic ? */
1560 sym->st_value = 0;
1564 #else
1565 #define HAVE_PHDR 0
1566 #define EXTRA_RELITEMS 9
1568 /* zero plt offsets of weak symbols in .dynsym */
1569 void patch_dynsym_undef(TCCState *s1, Section *s)
1571 ElfW(Sym) *sym;
1573 for_each_elem(s, 1, sym, ElfW(Sym))
1574 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1575 sym->st_value = 0;
1577 #endif
1579 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1581 int sym_index = ELFW(R_SYM) (rel->r_info);
1582 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1583 unsigned long offset;
1585 if (sym_index >= s1->nb_sym_attrs)
1586 return;
1587 offset = s1->sym_attrs[sym_index].got_offset;
1588 section_reserve(s1->got, offset + PTR_SIZE);
1589 #ifdef TCC_TARGET_X86_64
1590 /* only works for x86-64 */
1591 put32(s1->got->data + offset + 4, sym->st_value >> 32);
1592 #endif
1593 put32(s1->got->data + offset, sym->st_value & 0xffffffff);
1596 /* Perform relocation to GOT or PLT entries */
1597 ST_FUNC void fill_got(TCCState *s1)
1599 Section *s;
1600 ElfW_Rel *rel;
1601 int i;
1603 for(i = 1; i < s1->nb_sections; i++) {
1604 s = s1->sections[i];
1605 if (s->sh_type != SHT_RELX)
1606 continue;
1607 /* no need to handle got relocations */
1608 if (s->link != symtab_section)
1609 continue;
1610 for_each_elem(s, 0, rel, ElfW_Rel) {
1611 switch (ELFW(R_TYPE) (rel->r_info)) {
1612 case R_X86_64_GOT32:
1613 case R_X86_64_GOTPCREL:
1614 case R_X86_64_PLT32:
1615 fill_got_entry(s1, rel);
1616 break;
1622 /* Bind symbols of executable: resolve undefined symbols from exported symbols
1623 in shared libraries and export non local defined symbols to shared libraries
1624 if -rdynamic switch was given on command line */
1625 static void bind_exe_dynsyms(TCCState *s1)
1627 const char *name;
1628 int sym_index, index;
1629 ElfW(Sym) *sym, *esym;
1630 int type;
1632 /* Resolve undefined symbols from dynamic symbols. When there is a match:
1633 - if STT_FUNC or STT_GNU_IFUNC symbol -> add it in PLT
1634 - if STT_OBJECT symbol -> add it in .bss section with suitable reloc */
1635 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1636 if (sym->st_shndx == SHN_UNDEF) {
1637 name = (char *) symtab_section->link->data + sym->st_name;
1638 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1639 if (sym_index) {
1640 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1641 type = ELFW(ST_TYPE)(esym->st_info);
1642 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1643 /* Indirect functions shall have STT_FUNC type in executable
1644 * dynsym section. Indeed, a dlsym call following a lazy
1645 * resolution would pick the symbol value from the
1646 * executable dynsym entry which would contain the address
1647 * of the function wanted by the caller of dlsym instead of
1648 * the address of the function that would return that
1649 * address */
1650 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1651 ELFW(ST_INFO)(STB_GLOBAL,STT_FUNC),
1652 sym - (ElfW(Sym) *)symtab_section->data);
1653 } else if (type == STT_OBJECT) {
1654 unsigned long offset;
1655 ElfW(Sym) *dynsym;
1656 offset = bss_section->data_offset;
1657 /* XXX: which alignment ? */
1658 offset = (offset + 16 - 1) & -16;
1659 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1660 esym->st_info, 0, bss_section->sh_num,
1661 name);
1662 /* Ensure R_COPY works for weak symbol aliases */
1663 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1664 for_each_elem(s1->dynsymtab_section, 1, dynsym, ElfW(Sym)) {
1665 if ((dynsym->st_value == esym->st_value)
1666 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1667 char *dynname = (char *) s1->dynsymtab_section->link->data
1668 + dynsym->st_name;
1669 put_elf_sym(s1->dynsym, offset, dynsym->st_size,
1670 dynsym->st_info, 0,
1671 bss_section->sh_num, dynname);
1672 break;
1676 put_elf_reloc(s1->dynsym, bss_section,
1677 offset, R_COPY, index);
1678 offset += esym->st_size;
1679 bss_section->data_offset = offset;
1681 } else {
1682 /* STB_WEAK undefined symbols are accepted */
1683 /* XXX: _fp_hw seems to be part of the ABI, so we ignore it */
1684 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1685 !strcmp(name, "_fp_hw")) {
1686 } else {
1687 tcc_error_noabort("undefined symbol '%s'", name);
1690 } else if (s1->rdynamic && ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1691 /* if -rdynamic option, then export all non local symbols */
1692 name = (char *) symtab_section->link->data + sym->st_name;
1693 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size, sym->st_info,
1694 0, sym->st_shndx, name);
1699 /* Bind symbols of libraries: export non local symbols of executable that
1700 resolve undefined symbols of shared libraries */
1701 static void bind_libs_dynsyms(TCCState *s1)
1703 const char *name;
1704 int sym_index;
1705 ElfW(Sym) *sym, *esym;
1707 /* now look at unresolved dynamic symbols and export
1708 corresponding symbol */
1709 for_each_elem(s1->dynsymtab_section, 1, esym, ElfW(Sym)) {
1710 if (esym->st_shndx == SHN_UNDEF) {
1711 name = (char *) s1->dynsymtab_section->link->data + esym->st_name;
1712 sym_index = find_elf_sym(symtab_section, name);
1713 if (sym_index) {
1714 /* XXX: avoid adding a symbol if already present because of
1715 -rdynamic ? */
1716 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1717 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1718 sym->st_info, 0, sym->st_shndx, name);
1719 } else {
1720 /* weak symbols can stay undefined */
1721 if (ELFW(ST_BIND)(esym->st_info) != STB_WEAK)
1722 tcc_warning("undefined dynamic symbol '%s'", name);
1728 /* Export all non local symbols (for shared libraries) */
1729 static void export_global_syms(TCCState *s1)
1731 int nb_syms, dynindex, index;
1732 const char *name;
1733 ElfW(Sym) *sym;
1735 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1736 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1737 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1738 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1739 name = (char *) symtab_section->link->data + sym->st_name;
1740 dynindex = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1741 sym->st_info, 0, sym->st_shndx, name);
1742 index = sym - (ElfW(Sym) *) symtab_section->data;
1743 s1->symtab_to_dynsym[index] = dynindex;
1748 /* relocate the PLT: compute addresses and offsets in the PLT now that final
1749 address for PLT and GOT are known (see fill_program_header) */
1750 static void relocate_plt(TCCState *s1)
1752 uint8_t *p, *p_end;
1754 p = s1->plt->data;
1755 p_end = p + s1->plt->data_offset;
1756 if (p < p_end) {
1757 #if defined(TCC_TARGET_I386)
1758 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1759 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
1760 p += 16;
1761 while (p < p_end) {
1762 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1763 p += 16;
1765 #elif defined(TCC_TARGET_X86_64)
1766 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
1767 put32(p + 2, get32(p + 2) + x);
1768 put32(p + 8, get32(p + 8) + x - 6);
1769 p += 16;
1770 while (p < p_end) {
1771 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
1772 p += 16;
1774 #elif defined(TCC_TARGET_ARM)
1775 int x;
1776 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
1777 p += 16;
1778 while (p < p_end) {
1779 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
1780 p += 4;
1781 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
1782 p += 16;
1784 #elif defined(TCC_TARGET_C67)
1785 /* XXX: TODO */
1786 #else
1787 #error unsupported CPU
1788 #endif
1792 /* Allocate strings for section names and decide if an unallocated section
1793 should be output.
1795 NOTE: the strsec section comes last, so its size is also correct ! */
1796 static void alloc_sec_names(TCCState *s1, int file_type, Section *strsec)
1798 int i;
1799 Section *s;
1801 /* Allocate strings for section names */
1802 for(i = 1; i < s1->nb_sections; i++) {
1803 s = s1->sections[i];
1804 s->sh_name = put_elf_str(strsec, s->name);
1805 /* when generating a DLL, we include relocations but we may
1806 patch them */
1807 if (file_type == TCC_OUTPUT_DLL &&
1808 s->sh_type == SHT_RELX &&
1809 !(s->sh_flags & SHF_ALLOC)) {
1810 /* gr: avoid bogus relocs for empty (debug) sections */
1811 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
1812 prepare_dynamic_rel(s1, s);
1813 else if (s1->do_debug)
1814 s->sh_size = s->data_offset;
1815 } else if (s1->do_debug ||
1816 file_type == TCC_OUTPUT_OBJ ||
1817 (s->sh_flags & SHF_ALLOC) ||
1818 i == (s1->nb_sections - 1)) {
1819 /* we output all sections if debug or object file */
1820 s->sh_size = s->data_offset;
1825 /* Info to be copied in dynamic section */
1826 struct dyn_inf {
1827 Section *dynamic;
1828 Section *dynstr;
1829 unsigned long dyn_rel_off;
1830 addr_t rel_addr;
1831 addr_t rel_size;
1832 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1833 addr_t bss_addr;
1834 addr_t bss_size;
1835 #endif
1838 /* Assign sections to segments and decide how are sections laid out when loaded
1839 in memory. This function also fills corresponding program headers. */
1840 static int layout_sections(TCCState *s1, ElfW(Phdr) *phdr, int phnum,
1841 Section *interp, struct dyn_inf *dyninf,
1842 int *sec_order)
1844 int i, j, k, file_type, sh_order_index, file_offset;
1845 long long tmp;
1846 addr_t addr;
1847 ElfW(Phdr) *ph;
1848 Section *s;
1850 file_type = s1->output_type;
1851 sh_order_index = 1;
1852 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
1853 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
1854 else
1855 file_offset = 0;
1857 if (phnum > 0) {
1858 if (s1->has_text_addr) {
1859 int a_offset, p_offset;
1860 addr = s1->text_addr;
1861 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
1862 ELF_PAGE_SIZE */
1863 a_offset = (int) (addr & (s1->section_align - 1));
1864 p_offset = file_offset & (s1->section_align - 1);
1865 if (a_offset < p_offset)
1866 a_offset += s1->section_align;
1867 file_offset += (a_offset - p_offset);
1868 } else {
1869 if (file_type == TCC_OUTPUT_DLL)
1870 addr = 0;
1871 else
1872 addr = ELF_START_ADDR;
1873 /* compute address after headers */
1874 addr += (file_offset & (s1->section_align - 1));
1877 ph = &phdr[0];
1878 /* Leave one program headers for the program interpreter and one for
1879 the program header table itself if needed. These are done later as
1880 they require section layout to be done first. */
1881 if (interp)
1882 ph += 1 + HAVE_PHDR;
1884 /* dynamic relocation table information, for .dynamic section */
1885 dyninf->rel_addr = dyninf->rel_size = 0;
1886 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1887 dyninf->bss_addr = dyninf->bss_size = 0;
1888 #endif
1890 for(j = 0; j < 2; j++) {
1891 ph->p_type = PT_LOAD;
1892 if (j == 0)
1893 ph->p_flags = PF_R | PF_X;
1894 else
1895 ph->p_flags = PF_R | PF_W;
1896 ph->p_align = s1->section_align;
1898 /* Decide the layout of sections loaded in memory. This must
1899 be done before program headers are filled since they contain
1900 info about the layout. We do the following ordering: interp,
1901 symbol tables, relocations, progbits, nobits */
1902 /* XXX: do faster and simpler sorting */
1903 for(k = 0; k < 5; k++) {
1904 for(i = 1; i < s1->nb_sections; i++) {
1905 s = s1->sections[i];
1906 /* compute if section should be included */
1907 if (j == 0) {
1908 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1909 SHF_ALLOC)
1910 continue;
1911 } else {
1912 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1913 (SHF_ALLOC | SHF_WRITE))
1914 continue;
1916 if (s == interp) {
1917 if (k != 0)
1918 continue;
1919 } else if (s->sh_type == SHT_DYNSYM ||
1920 s->sh_type == SHT_STRTAB ||
1921 s->sh_type == SHT_HASH) {
1922 if (k != 1)
1923 continue;
1924 } else if (s->sh_type == SHT_RELX) {
1925 if (k != 2)
1926 continue;
1927 } else if (s->sh_type == SHT_NOBITS) {
1928 if (k != 4)
1929 continue;
1930 } else {
1931 if (k != 3)
1932 continue;
1934 sec_order[sh_order_index++] = i;
1936 /* section matches: we align it and add its size */
1937 tmp = addr;
1938 addr = (addr + s->sh_addralign - 1) &
1939 ~(s->sh_addralign - 1);
1940 file_offset += (int) ( addr - tmp );
1941 s->sh_offset = file_offset;
1942 s->sh_addr = addr;
1944 /* update program header infos */
1945 if (ph->p_offset == 0) {
1946 ph->p_offset = file_offset;
1947 ph->p_vaddr = addr;
1948 ph->p_paddr = ph->p_vaddr;
1950 /* update dynamic relocation infos */
1951 if (s->sh_type == SHT_RELX) {
1952 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1953 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) {
1954 dyninf->rel_addr = addr;
1955 dyninf->rel_size += s->sh_size; /* XXX only first rel. */
1957 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) {
1958 dyninf->bss_addr = addr;
1959 dyninf->bss_size = s->sh_size; /* XXX only first rel. */
1961 #else
1962 if (dyninf->rel_size == 0)
1963 dyninf->rel_addr = addr;
1964 dyninf->rel_size += s->sh_size;
1965 #endif
1967 addr += s->sh_size;
1968 if (s->sh_type != SHT_NOBITS)
1969 file_offset += s->sh_size;
1972 ph->p_filesz = file_offset - ph->p_offset;
1973 ph->p_memsz = addr - ph->p_vaddr;
1974 ph++;
1975 if (j == 0) {
1976 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
1977 /* if in the middle of a page, we duplicate the page in
1978 memory so that one copy is RX and the other is RW */
1979 if ((addr & (s1->section_align - 1)) != 0)
1980 addr += s1->section_align;
1981 } else {
1982 addr = (addr + s1->section_align - 1) & ~(s1->section_align - 1);
1983 file_offset = (file_offset + s1->section_align - 1) &
1984 ~(s1->section_align - 1);
1990 /* all other sections come after */
1991 for(i = 1; i < s1->nb_sections; i++) {
1992 s = s1->sections[i];
1993 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
1994 continue;
1995 sec_order[sh_order_index++] = i;
1997 file_offset = (file_offset + s->sh_addralign - 1) &
1998 ~(s->sh_addralign - 1);
1999 s->sh_offset = file_offset;
2000 if (s->sh_type != SHT_NOBITS)
2001 file_offset += s->sh_size;
2004 return file_offset;
2007 static void fill_unloadable_phdr(ElfW(Phdr) *phdr, int phnum, Section *interp,
2008 Section *dynamic)
2010 ElfW(Phdr) *ph;
2012 /* if interpreter, then add corresponding program header */
2013 if (interp) {
2014 ph = &phdr[0];
2016 if (HAVE_PHDR)
2018 int len = phnum * sizeof(ElfW(Phdr));
2020 ph->p_type = PT_PHDR;
2021 ph->p_offset = sizeof(ElfW(Ehdr));
2022 ph->p_vaddr = interp->sh_addr - len;
2023 ph->p_paddr = ph->p_vaddr;
2024 ph->p_filesz = ph->p_memsz = len;
2025 ph->p_flags = PF_R | PF_X;
2026 ph->p_align = 4; /* interp->sh_addralign; */
2027 ph++;
2030 ph->p_type = PT_INTERP;
2031 ph->p_offset = interp->sh_offset;
2032 ph->p_vaddr = interp->sh_addr;
2033 ph->p_paddr = ph->p_vaddr;
2034 ph->p_filesz = interp->sh_size;
2035 ph->p_memsz = interp->sh_size;
2036 ph->p_flags = PF_R;
2037 ph->p_align = interp->sh_addralign;
2040 /* if dynamic section, then add corresponding program header */
2041 if (dynamic) {
2042 ph = &phdr[phnum - 1];
2044 ph->p_type = PT_DYNAMIC;
2045 ph->p_offset = dynamic->sh_offset;
2046 ph->p_vaddr = dynamic->sh_addr;
2047 ph->p_paddr = ph->p_vaddr;
2048 ph->p_filesz = dynamic->sh_size;
2049 ph->p_memsz = dynamic->sh_size;
2050 ph->p_flags = PF_R | PF_W;
2051 ph->p_align = dynamic->sh_addralign;
2055 /* Fill the dynamic section with tags describing the address and size of
2056 sections */
2057 static void fill_dynamic(TCCState *s1, struct dyn_inf *dyninf)
2059 Section *dynamic;
2061 dynamic = dyninf->dynamic;
2063 /* put dynamic section entries */
2064 dynamic->data_offset = dyninf->dyn_rel_off;
2065 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2066 put_dt(dynamic, DT_STRTAB, dyninf->dynstr->sh_addr);
2067 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2068 put_dt(dynamic, DT_STRSZ, dyninf->dynstr->data_offset);
2069 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2070 #ifdef TCC_TARGET_X86_64
2071 put_dt(dynamic, DT_RELA, dyninf->rel_addr);
2072 put_dt(dynamic, DT_RELASZ, dyninf->rel_size);
2073 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2074 #else
2075 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2076 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2077 put_dt(dynamic, DT_PLTRELSZ, dyninf->rel_size);
2078 put_dt(dynamic, DT_JMPREL, dyninf->rel_addr);
2079 put_dt(dynamic, DT_PLTREL, DT_REL);
2080 put_dt(dynamic, DT_REL, dyninf->bss_addr);
2081 put_dt(dynamic, DT_RELSZ, dyninf->bss_size);
2082 #else
2083 put_dt(dynamic, DT_REL, dyninf->rel_addr);
2084 put_dt(dynamic, DT_RELSZ, dyninf->rel_size);
2085 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2086 #endif
2087 #endif
2088 if (s1->do_debug)
2089 put_dt(dynamic, DT_DEBUG, 0);
2090 put_dt(dynamic, DT_NULL, 0);
2093 /* Relocate remaining sections and symbols (that is those not related to
2094 dynamic linking) */
2095 static int final_sections_reloc(TCCState *s1)
2097 int i;
2098 Section *s;
2100 relocate_syms(s1, 0);
2102 if (s1->nb_errors != 0)
2103 return -1;
2105 /* relocate sections */
2106 /* XXX: ignore sections with allocated relocations ? */
2107 for(i = 1; i < s1->nb_sections; i++) {
2108 s = s1->sections[i];
2109 if (s->reloc && s != s1->got)
2110 relocate_section(s1, s);
2113 /* relocate relocation entries if the relocation tables are
2114 allocated in the executable */
2115 for(i = 1; i < s1->nb_sections; i++) {
2116 s = s1->sections[i];
2117 if ((s->sh_flags & SHF_ALLOC) &&
2118 s->sh_type == SHT_RELX) {
2119 relocate_rel(s1, s);
2122 return 0;
2125 /* Create an ELF file on disk.
2126 This function handle ELF specific layout requirements */
2127 static void tcc_output_elf(TCCState *s1, FILE *f, int phnum, ElfW(Phdr) *phdr,
2128 int file_offset, int *sec_order)
2130 int i, shnum, offset, size, file_type;
2131 Section *s;
2132 ElfW(Ehdr) ehdr;
2133 ElfW(Shdr) shdr, *sh;
2135 file_type = s1->output_type;
2136 shnum = s1->nb_sections;
2138 memset(&ehdr, 0, sizeof(ehdr));
2140 if (phnum > 0) {
2141 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2142 ehdr.e_phnum = phnum;
2143 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2146 /* align to 4 */
2147 file_offset = (file_offset + 3) & -4;
2149 /* fill header */
2150 ehdr.e_ident[0] = ELFMAG0;
2151 ehdr.e_ident[1] = ELFMAG1;
2152 ehdr.e_ident[2] = ELFMAG2;
2153 ehdr.e_ident[3] = ELFMAG3;
2154 ehdr.e_ident[4] = ELFCLASSW;
2155 ehdr.e_ident[5] = ELFDATA2LSB;
2156 ehdr.e_ident[6] = EV_CURRENT;
2157 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2158 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2159 #endif
2160 #ifdef TCC_TARGET_ARM
2161 #ifdef TCC_ARM_EABI
2162 ehdr.e_ident[EI_OSABI] = 0;
2163 ehdr.e_flags = EF_ARM_EABI_VER4;
2164 if (file_type == TCC_OUTPUT_EXE || file_type == TCC_OUTPUT_DLL)
2165 ehdr.e_flags |= EF_ARM_HASENTRY;
2166 if (s1->float_abi == ARM_HARD_FLOAT)
2167 ehdr.e_flags |= EF_ARM_VFP_FLOAT;
2168 else
2169 ehdr.e_flags |= EF_ARM_SOFT_FLOAT;
2170 #else
2171 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2172 #endif
2173 #endif
2174 switch(file_type) {
2175 default:
2176 case TCC_OUTPUT_EXE:
2177 ehdr.e_type = ET_EXEC;
2178 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2179 break;
2180 case TCC_OUTPUT_DLL:
2181 ehdr.e_type = ET_DYN;
2182 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2183 break;
2184 case TCC_OUTPUT_OBJ:
2185 ehdr.e_type = ET_REL;
2186 break;
2188 ehdr.e_machine = EM_TCC_TARGET;
2189 ehdr.e_version = EV_CURRENT;
2190 ehdr.e_shoff = file_offset;
2191 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2192 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2193 ehdr.e_shnum = shnum;
2194 ehdr.e_shstrndx = shnum - 1;
2196 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2197 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2198 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2200 sort_syms(s1, symtab_section);
2201 for(i = 1; i < s1->nb_sections; i++) {
2202 s = s1->sections[sec_order[i]];
2203 if (s->sh_type != SHT_NOBITS) {
2204 if (s->sh_type == SHT_DYNSYM)
2205 patch_dynsym_undef(s1, s);
2206 while (offset < s->sh_offset) {
2207 fputc(0, f);
2208 offset++;
2210 size = s->sh_size;
2211 fwrite(s->data, 1, size, f);
2212 offset += size;
2216 /* output section headers */
2217 while (offset < ehdr.e_shoff) {
2218 fputc(0, f);
2219 offset++;
2222 for(i = 0; i < s1->nb_sections; i++) {
2223 sh = &shdr;
2224 memset(sh, 0, sizeof(ElfW(Shdr)));
2225 s = s1->sections[i];
2226 if (s) {
2227 sh->sh_name = s->sh_name;
2228 sh->sh_type = s->sh_type;
2229 sh->sh_flags = s->sh_flags;
2230 sh->sh_entsize = s->sh_entsize;
2231 sh->sh_info = s->sh_info;
2232 if (s->link)
2233 sh->sh_link = s->link->sh_num;
2234 sh->sh_addralign = s->sh_addralign;
2235 sh->sh_addr = s->sh_addr;
2236 sh->sh_offset = s->sh_offset;
2237 sh->sh_size = s->sh_size;
2239 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2243 /* Write an elf, coff or "binary" file */
2244 static int tcc_write_elf_file(TCCState *s1, const char *filename, int phnum,
2245 ElfW(Phdr) *phdr, int file_offset, int *sec_order)
2247 int fd, mode, file_type;
2248 FILE *f;
2250 file_type = s1->output_type;
2251 if (file_type == TCC_OUTPUT_OBJ)
2252 mode = 0666;
2253 else
2254 mode = 0777;
2255 unlink(filename);
2256 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2257 if (fd < 0) {
2258 tcc_error_noabort("could not write '%s'", filename);
2259 return -1;
2261 f = fdopen(fd, "wb");
2262 if (s1->verbose)
2263 printf("<- %s\n", filename);
2265 #ifdef TCC_TARGET_COFF
2266 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF)
2267 tcc_output_coff(s1, f);
2268 else
2269 #endif
2270 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2271 tcc_output_elf(s1, f, phnum, phdr, file_offset, sec_order);
2272 else
2273 tcc_output_binary(s1, f, sec_order);
2274 fclose(f);
2276 return 0;
2279 /* Output an elf, coff or binary file */
2280 /* XXX: suppress unneeded sections */
2281 static int elf_output_file(TCCState *s1, const char *filename)
2283 int i, ret, phnum, shnum, file_type, file_offset, *sec_order;
2284 struct dyn_inf dyninf;
2285 ElfW(Phdr) *phdr;
2286 ElfW(Sym) *sym;
2287 Section *strsec, *interp, *dynamic, *dynstr;
2289 file_type = s1->output_type;
2290 s1->nb_errors = 0;
2292 /* if linking, also link in runtime libraries (libc, libgcc, etc.) */
2293 if (file_type != TCC_OUTPUT_OBJ) {
2294 tcc_add_runtime(s1);
2297 phdr = NULL;
2298 sec_order = NULL;
2299 interp = dynamic = dynstr = NULL; /* avoid warning */
2300 dyninf.dyn_rel_off = 0; /* avoid warning */
2302 if (file_type != TCC_OUTPUT_OBJ) {
2303 relocate_common_syms();
2305 tcc_add_linker_symbols(s1);
2307 if (!s1->static_link) {
2308 if (file_type == TCC_OUTPUT_EXE) {
2309 char *ptr;
2310 /* allow override the dynamic loader */
2311 const char *elfint = getenv("LD_SO");
2312 if (elfint == NULL)
2313 elfint = DEFAULT_ELFINTERP(s1);
2314 /* add interpreter section only if executable */
2315 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
2316 interp->sh_addralign = 1;
2317 ptr = section_ptr_add(interp, 1 + strlen(elfint));
2318 strcpy(ptr, elfint);
2321 /* add dynamic symbol table */
2322 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
2323 ".dynstr",
2324 ".hash", SHF_ALLOC);
2325 dynstr = s1->dynsym->link;
2327 /* add dynamic section */
2328 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
2329 SHF_ALLOC | SHF_WRITE);
2330 dynamic->link = dynstr;
2331 dynamic->sh_entsize = sizeof(ElfW(Dyn));
2333 /* add PLT */
2334 s1->plt = new_section(s1, ".plt", SHT_PROGBITS,
2335 SHF_ALLOC | SHF_EXECINSTR);
2336 s1->plt->sh_entsize = 4;
2338 build_got(s1);
2340 if (file_type == TCC_OUTPUT_EXE) {
2341 bind_exe_dynsyms(s1);
2343 if (s1->nb_errors) {
2344 ret = -1;
2345 goto the_end;
2348 bind_libs_dynsyms(s1);
2349 } else /* shared library case: simply export all global symbols */
2350 export_global_syms(s1);
2352 build_got_entries(s1);
2354 /* add a list of needed dlls */
2355 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2356 DLLReference *dllref = s1->loaded_dlls[i];
2357 if (dllref->level == 0)
2358 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
2361 if (s1->rpath)
2362 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
2364 /* XXX: currently, since we do not handle PIC code, we
2365 must relocate the readonly segments */
2366 if (file_type == TCC_OUTPUT_DLL) {
2367 if (s1->soname)
2368 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
2369 put_dt(dynamic, DT_TEXTREL, 0);
2372 if (s1->symbolic)
2373 put_dt(dynamic, DT_SYMBOLIC, 0);
2375 /* add necessary space for other entries */
2376 dyninf.dyn_rel_off = dynamic->data_offset;
2377 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
2378 } else {
2379 /* still need to build got entries in case of static link */
2380 build_got_entries(s1);
2384 /* we add a section for symbols */
2385 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
2386 put_elf_str(strsec, "");
2388 /* compute number of sections */
2389 shnum = s1->nb_sections;
2391 /* this array is used to reorder sections in the output file */
2392 sec_order = tcc_malloc(sizeof(int) * shnum);
2393 sec_order[0] = 0;
2395 /* compute number of program headers */
2396 switch(file_type) {
2397 default:
2398 case TCC_OUTPUT_OBJ:
2399 phnum = 0;
2400 break;
2401 case TCC_OUTPUT_EXE:
2402 if (!s1->static_link)
2403 phnum = 4 + HAVE_PHDR;
2404 else
2405 phnum = 2;
2406 break;
2407 case TCC_OUTPUT_DLL:
2408 phnum = 3;
2409 break;
2412 /* Allocate strings for section names */
2413 alloc_sec_names(s1, file_type, strsec);
2415 /* allocate program segment headers */
2416 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
2418 /* compute section to program header mapping */
2419 file_offset = layout_sections(s1, phdr, phnum, interp, &dyninf, sec_order);
2421 /* Fill remaining program header and finalize relocation related to dynamic
2422 linking. */
2423 if (phnum > 0) {
2424 fill_unloadable_phdr(phdr, phnum, interp, dynamic);
2425 if (dynamic) {
2426 dyninf.dynamic = dynamic;
2427 dyninf.dynstr = dynstr;
2429 fill_dynamic(s1, &dyninf);
2431 /* put in GOT the dynamic section address and relocate PLT */
2432 put32(s1->got->data, dynamic->sh_addr);
2433 if (file_type == TCC_OUTPUT_EXE
2434 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2435 || file_type == TCC_OUTPUT_DLL
2436 #endif
2438 relocate_plt(s1);
2440 /* relocate symbols in .dynsym now that final addresses are known */
2441 for_each_elem(s1->dynsym, 1, sym, ElfW(Sym)) {
2442 /* relocate to PLT if symbol corresponds to a PLT entry */
2443 if (sym->st_shndx == SHN_UNDEF) {
2444 if (sym->st_value)
2445 sym->st_value += s1->plt->sh_addr;
2446 } else if (sym->st_shndx < SHN_LORESERVE) {
2447 /* do symbol relocation */
2448 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2454 /* if building executable or DLL, then relocate each section
2455 except the GOT which is already relocated */
2456 if (file_type != TCC_OUTPUT_OBJ) {
2457 ret = final_sections_reloc(s1);
2458 if (ret)
2459 goto the_end;
2462 /* Perform relocation to GOT or PLT entries */
2463 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2464 fill_got(s1);
2466 /* Create the ELF file with name 'filename' */
2467 ret = tcc_write_elf_file(s1, filename, phnum, phdr, file_offset, sec_order);
2468 the_end:
2469 tcc_free(s1->symtab_to_dynsym);
2470 tcc_free(sec_order);
2471 tcc_free(phdr);
2472 tcc_free(s1->sym_attrs);
2473 s1->sym_attrs = NULL;
2474 return ret;
2477 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2479 int ret;
2480 #ifdef TCC_TARGET_PE
2481 if (s->output_type != TCC_OUTPUT_OBJ) {
2482 ret = pe_output_file(s, filename);
2483 } else
2484 #endif
2485 ret = elf_output_file(s, filename);
2486 return ret;
2489 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2491 void *data;
2493 data = tcc_malloc(size);
2494 lseek(fd, file_offset, SEEK_SET);
2495 read(fd, data, size);
2496 return data;
2499 typedef struct SectionMergeInfo {
2500 Section *s; /* corresponding existing section */
2501 unsigned long offset; /* offset of the new section in the existing section */
2502 uint8_t new_section; /* true if section 's' was added */
2503 uint8_t link_once; /* true if link once section */
2504 } SectionMergeInfo;
2506 /* load an object file and merge it with current files */
2507 /* XXX: handle correctly stab (debug) info */
2508 ST_FUNC int tcc_load_object_file(TCCState *s1,
2509 int fd, unsigned long file_offset)
2511 ElfW(Ehdr) ehdr;
2512 ElfW(Shdr) *shdr, *sh;
2513 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2514 unsigned char *strsec, *strtab;
2515 int *old_to_new_syms;
2516 char *sh_name, *name;
2517 SectionMergeInfo *sm_table, *sm;
2518 ElfW(Sym) *sym, *symtab;
2519 ElfW_Rel *rel;
2520 Section *s;
2522 int stab_index;
2523 int stabstr_index;
2525 stab_index = stabstr_index = 0;
2527 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2528 goto fail1;
2529 if (ehdr.e_ident[0] != ELFMAG0 ||
2530 ehdr.e_ident[1] != ELFMAG1 ||
2531 ehdr.e_ident[2] != ELFMAG2 ||
2532 ehdr.e_ident[3] != ELFMAG3)
2533 goto fail1;
2534 /* test if object file */
2535 if (ehdr.e_type != ET_REL)
2536 goto fail1;
2537 /* test CPU specific stuff */
2538 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2539 ehdr.e_machine != EM_TCC_TARGET) {
2540 fail1:
2541 tcc_error_noabort("invalid object file");
2542 return -1;
2544 /* read sections */
2545 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2546 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2547 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2549 /* load section names */
2550 sh = &shdr[ehdr.e_shstrndx];
2551 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2553 /* load symtab and strtab */
2554 old_to_new_syms = NULL;
2555 symtab = NULL;
2556 strtab = NULL;
2557 nb_syms = 0;
2558 for(i = 1; i < ehdr.e_shnum; i++) {
2559 sh = &shdr[i];
2560 if (sh->sh_type == SHT_SYMTAB) {
2561 if (symtab) {
2562 tcc_error_noabort("object must contain only one symtab");
2563 fail:
2564 ret = -1;
2565 goto the_end;
2567 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2568 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2569 sm_table[i].s = symtab_section;
2571 /* now load strtab */
2572 sh = &shdr[sh->sh_link];
2573 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2577 /* now examine each section and try to merge its content with the
2578 ones in memory */
2579 for(i = 1; i < ehdr.e_shnum; i++) {
2580 /* no need to examine section name strtab */
2581 if (i == ehdr.e_shstrndx)
2582 continue;
2583 sh = &shdr[i];
2584 sh_name = (char *) strsec + sh->sh_name;
2585 /* ignore sections types we do not handle */
2586 if (sh->sh_type != SHT_PROGBITS &&
2587 sh->sh_type != SHT_RELX &&
2588 #ifdef TCC_ARM_EABI
2589 sh->sh_type != SHT_ARM_EXIDX &&
2590 #endif
2591 sh->sh_type != SHT_NOBITS &&
2592 sh->sh_type != SHT_PREINIT_ARRAY &&
2593 sh->sh_type != SHT_INIT_ARRAY &&
2594 sh->sh_type != SHT_FINI_ARRAY &&
2595 strcmp(sh_name, ".stabstr")
2597 continue;
2598 if (sh->sh_addralign < 1)
2599 sh->sh_addralign = 1;
2600 /* find corresponding section, if any */
2601 for(j = 1; j < s1->nb_sections;j++) {
2602 s = s1->sections[j];
2603 if (!strcmp(s->name, sh_name)) {
2604 if (!strncmp(sh_name, ".gnu.linkonce",
2605 sizeof(".gnu.linkonce") - 1)) {
2606 /* if a 'linkonce' section is already present, we
2607 do not add it again. It is a little tricky as
2608 symbols can still be defined in
2609 it. */
2610 sm_table[i].link_once = 1;
2611 goto next;
2612 } else {
2613 goto found;
2617 /* not found: create new section */
2618 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2619 /* take as much info as possible from the section. sh_link and
2620 sh_info will be updated later */
2621 s->sh_addralign = sh->sh_addralign;
2622 s->sh_entsize = sh->sh_entsize;
2623 sm_table[i].new_section = 1;
2624 found:
2625 if (sh->sh_type != s->sh_type) {
2626 tcc_error_noabort("invalid section type");
2627 goto fail;
2630 /* align start of section */
2631 offset = s->data_offset;
2633 if (0 == strcmp(sh_name, ".stab")) {
2634 stab_index = i;
2635 goto no_align;
2637 if (0 == strcmp(sh_name, ".stabstr")) {
2638 stabstr_index = i;
2639 goto no_align;
2642 size = sh->sh_addralign - 1;
2643 offset = (offset + size) & ~size;
2644 if (sh->sh_addralign > s->sh_addralign)
2645 s->sh_addralign = sh->sh_addralign;
2646 s->data_offset = offset;
2647 no_align:
2648 sm_table[i].offset = offset;
2649 sm_table[i].s = s;
2650 /* concatenate sections */
2651 size = sh->sh_size;
2652 if (sh->sh_type != SHT_NOBITS) {
2653 unsigned char *ptr;
2654 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2655 ptr = section_ptr_add(s, size);
2656 read(fd, ptr, size);
2657 } else {
2658 s->data_offset += size;
2660 next: ;
2663 /* gr relocate stab strings */
2664 if (stab_index && stabstr_index) {
2665 Stab_Sym *a, *b;
2666 unsigned o;
2667 s = sm_table[stab_index].s;
2668 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2669 b = (Stab_Sym *)(s->data + s->data_offset);
2670 o = sm_table[stabstr_index].offset;
2671 while (a < b)
2672 a->n_strx += o, a++;
2675 /* second short pass to update sh_link and sh_info fields of new
2676 sections */
2677 for(i = 1; i < ehdr.e_shnum; i++) {
2678 s = sm_table[i].s;
2679 if (!s || !sm_table[i].new_section)
2680 continue;
2681 sh = &shdr[i];
2682 if (sh->sh_link > 0)
2683 s->link = sm_table[sh->sh_link].s;
2684 if (sh->sh_type == SHT_RELX) {
2685 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2686 /* update backward link */
2687 s1->sections[s->sh_info]->reloc = s;
2690 sm = sm_table;
2692 /* resolve symbols */
2693 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2695 sym = symtab + 1;
2696 for(i = 1; i < nb_syms; i++, sym++) {
2697 if (sym->st_shndx != SHN_UNDEF &&
2698 sym->st_shndx < SHN_LORESERVE) {
2699 sm = &sm_table[sym->st_shndx];
2700 if (sm->link_once) {
2701 /* if a symbol is in a link once section, we use the
2702 already defined symbol. It is very important to get
2703 correct relocations */
2704 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2705 name = (char *) strtab + sym->st_name;
2706 sym_index = find_elf_sym(symtab_section, name);
2707 if (sym_index)
2708 old_to_new_syms[i] = sym_index;
2710 continue;
2712 /* if no corresponding section added, no need to add symbol */
2713 if (!sm->s)
2714 continue;
2715 /* convert section number */
2716 sym->st_shndx = sm->s->sh_num;
2717 /* offset value */
2718 sym->st_value += sm->offset;
2720 /* add symbol */
2721 name = (char *) strtab + sym->st_name;
2722 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2723 sym->st_info, sym->st_other,
2724 sym->st_shndx, name);
2725 old_to_new_syms[i] = sym_index;
2728 /* third pass to patch relocation entries */
2729 for(i = 1; i < ehdr.e_shnum; i++) {
2730 s = sm_table[i].s;
2731 if (!s)
2732 continue;
2733 sh = &shdr[i];
2734 offset = sm_table[i].offset;
2735 switch(s->sh_type) {
2736 case SHT_RELX:
2737 /* take relocation offset information */
2738 offseti = sm_table[sh->sh_info].offset;
2739 for_each_elem(s, (offset / sizeof(*rel)), rel, ElfW_Rel) {
2740 int type;
2741 unsigned sym_index;
2742 /* convert symbol index */
2743 type = ELFW(R_TYPE)(rel->r_info);
2744 sym_index = ELFW(R_SYM)(rel->r_info);
2745 /* NOTE: only one symtab assumed */
2746 if (sym_index >= nb_syms)
2747 goto invalid_reloc;
2748 sym_index = old_to_new_syms[sym_index];
2749 /* ignore link_once in rel section. */
2750 if (!sym_index && !sm->link_once
2751 #ifdef TCC_TARGET_ARM
2752 && type != R_ARM_V4BX
2753 #endif
2755 invalid_reloc:
2756 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2757 i, strsec + sh->sh_name, rel->r_offset);
2758 goto fail;
2760 rel->r_info = ELFW(R_INFO)(sym_index, type);
2761 /* offset the relocation offset */
2762 rel->r_offset += offseti;
2763 #ifdef TCC_TARGET_ARM
2764 /* Jumps and branches from a Thumb code to a PLT entry need
2765 special handling since PLT entries are ARM code.
2766 Unconditional bl instructions referencing PLT entries are
2767 handled by converting these instructions into blx
2768 instructions. Other case of instructions referencing a PLT
2769 entry require to add a Thumb stub before the PLT entry to
2770 switch to ARM mode. We set bit plt_thumb_stub of the
2771 attribute of a symbol to indicate such a case. */
2772 if (type == R_ARM_THM_JUMP24)
2773 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
2774 #endif
2776 break;
2777 default:
2778 break;
2782 ret = 0;
2783 the_end:
2784 tcc_free(symtab);
2785 tcc_free(strtab);
2786 tcc_free(old_to_new_syms);
2787 tcc_free(sm_table);
2788 tcc_free(strsec);
2789 tcc_free(shdr);
2790 return ret;
2793 typedef struct ArchiveHeader {
2794 char ar_name[16]; /* name of this member */
2795 char ar_date[12]; /* file mtime */
2796 char ar_uid[6]; /* owner uid; printed as decimal */
2797 char ar_gid[6]; /* owner gid; printed as decimal */
2798 char ar_mode[8]; /* file mode, printed as octal */
2799 char ar_size[10]; /* file size, printed as decimal */
2800 char ar_fmag[2]; /* should contain ARFMAG */
2801 } ArchiveHeader;
2803 static int get_be32(const uint8_t *b)
2805 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
2808 /* load only the objects which resolve undefined symbols */
2809 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
2811 int i, bound, nsyms, sym_index, off, ret;
2812 uint8_t *data;
2813 const char *ar_names, *p;
2814 const uint8_t *ar_index;
2815 ElfW(Sym) *sym;
2817 data = tcc_malloc(size);
2818 if (read(fd, data, size) != size)
2819 goto fail;
2820 nsyms = get_be32(data);
2821 ar_index = data + 4;
2822 ar_names = (char *) ar_index + nsyms * 4;
2824 do {
2825 bound = 0;
2826 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
2827 sym_index = find_elf_sym(symtab_section, p);
2828 if(sym_index) {
2829 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
2830 if(sym->st_shndx == SHN_UNDEF) {
2831 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
2832 ++bound;
2833 lseek(fd, off, SEEK_SET);
2834 if(tcc_load_object_file(s1, fd, off) < 0) {
2835 fail:
2836 ret = -1;
2837 goto the_end;
2842 } while(bound);
2843 ret = 0;
2844 the_end:
2845 tcc_free(data);
2846 return ret;
2849 /* load a '.a' file */
2850 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
2852 ArchiveHeader hdr;
2853 char ar_size[11];
2854 char ar_name[17];
2855 char magic[8];
2856 int size, len, i;
2857 unsigned long file_offset;
2859 /* skip magic which was already checked */
2860 read(fd, magic, sizeof(magic));
2862 for(;;) {
2863 len = read(fd, &hdr, sizeof(hdr));
2864 if (len == 0)
2865 break;
2866 if (len != sizeof(hdr)) {
2867 tcc_error_noabort("invalid archive");
2868 return -1;
2870 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
2871 ar_size[sizeof(hdr.ar_size)] = '\0';
2872 size = strtol(ar_size, NULL, 0);
2873 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
2874 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
2875 if (ar_name[i] != ' ')
2876 break;
2878 ar_name[i + 1] = '\0';
2879 file_offset = lseek(fd, 0, SEEK_CUR);
2880 /* align to even */
2881 size = (size + 1) & ~1;
2882 if (!strcmp(ar_name, "/")) {
2883 /* coff symbol table : we handle it */
2884 if(s1->alacarte_link)
2885 return tcc_load_alacarte(s1, fd, size);
2886 } else if (!strcmp(ar_name, "//") ||
2887 !strcmp(ar_name, "__.SYMDEF") ||
2888 !strcmp(ar_name, "__.SYMDEF/") ||
2889 !strcmp(ar_name, "ARFILENAMES/")) {
2890 /* skip symbol table or archive names */
2891 } else {
2892 if (tcc_load_object_file(s1, fd, file_offset) < 0)
2893 return -1;
2895 lseek(fd, file_offset + size, SEEK_SET);
2897 return 0;
2900 #ifndef TCC_TARGET_PE
2901 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
2902 is referenced by the user (so it should be added as DT_NEEDED in
2903 the generated ELF file) */
2904 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
2906 ElfW(Ehdr) ehdr;
2907 ElfW(Shdr) *shdr, *sh, *sh1;
2908 int i, j, nb_syms, nb_dts, sym_bind, ret;
2909 ElfW(Sym) *sym, *dynsym;
2910 ElfW(Dyn) *dt, *dynamic;
2911 unsigned char *dynstr;
2912 const char *name, *soname;
2913 DLLReference *dllref;
2915 read(fd, &ehdr, sizeof(ehdr));
2917 /* test CPU specific stuff */
2918 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2919 ehdr.e_machine != EM_TCC_TARGET) {
2920 tcc_error_noabort("bad architecture");
2921 return -1;
2924 /* read sections */
2925 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2927 /* load dynamic section and dynamic symbols */
2928 nb_syms = 0;
2929 nb_dts = 0;
2930 dynamic = NULL;
2931 dynsym = NULL; /* avoid warning */
2932 dynstr = NULL; /* avoid warning */
2933 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
2934 switch(sh->sh_type) {
2935 case SHT_DYNAMIC:
2936 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
2937 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
2938 break;
2939 case SHT_DYNSYM:
2940 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2941 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
2942 sh1 = &shdr[sh->sh_link];
2943 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
2944 break;
2945 default:
2946 break;
2950 /* compute the real library name */
2951 soname = tcc_basename(filename);
2953 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2954 if (dt->d_tag == DT_SONAME) {
2955 soname = (char *) dynstr + dt->d_un.d_val;
2959 /* if the dll is already loaded, do not load it */
2960 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2961 dllref = s1->loaded_dlls[i];
2962 if (!strcmp(soname, dllref->name)) {
2963 /* but update level if needed */
2964 if (level < dllref->level)
2965 dllref->level = level;
2966 ret = 0;
2967 goto the_end;
2971 /* add the dll and its level */
2972 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
2973 dllref->level = level;
2974 strcpy(dllref->name, soname);
2975 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
2977 /* add dynamic symbols in dynsym_section */
2978 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
2979 sym_bind = ELFW(ST_BIND)(sym->st_info);
2980 if (sym_bind == STB_LOCAL)
2981 continue;
2982 name = (char *) dynstr + sym->st_name;
2983 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
2984 sym->st_info, sym->st_other, sym->st_shndx, name);
2987 /* load all referenced DLLs */
2988 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2989 switch(dt->d_tag) {
2990 case DT_NEEDED:
2991 name = (char *) dynstr + dt->d_un.d_val;
2992 for(j = 0; j < s1->nb_loaded_dlls; j++) {
2993 dllref = s1->loaded_dlls[j];
2994 if (!strcmp(name, dllref->name))
2995 goto already_loaded;
2997 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
2998 tcc_error_noabort("referenced dll '%s' not found", name);
2999 ret = -1;
3000 goto the_end;
3002 already_loaded:
3003 break;
3006 ret = 0;
3007 the_end:
3008 tcc_free(dynstr);
3009 tcc_free(dynsym);
3010 tcc_free(dynamic);
3011 tcc_free(shdr);
3012 return ret;
3015 #define LD_TOK_NAME 256
3016 #define LD_TOK_EOF (-1)
3018 /* return next ld script token */
3019 static int ld_next(TCCState *s1, char *name, int name_size)
3021 int c;
3022 char *q;
3024 redo:
3025 switch(ch) {
3026 case ' ':
3027 case '\t':
3028 case '\f':
3029 case '\v':
3030 case '\r':
3031 case '\n':
3032 inp();
3033 goto redo;
3034 case '/':
3035 minp();
3036 if (ch == '*') {
3037 file->buf_ptr = parse_comment(file->buf_ptr);
3038 ch = file->buf_ptr[0];
3039 goto redo;
3040 } else {
3041 q = name;
3042 *q++ = '/';
3043 goto parse_name;
3045 break;
3046 /* case 'a' ... 'z': */
3047 case 'a':
3048 case 'b':
3049 case 'c':
3050 case 'd':
3051 case 'e':
3052 case 'f':
3053 case 'g':
3054 case 'h':
3055 case 'i':
3056 case 'j':
3057 case 'k':
3058 case 'l':
3059 case 'm':
3060 case 'n':
3061 case 'o':
3062 case 'p':
3063 case 'q':
3064 case 'r':
3065 case 's':
3066 case 't':
3067 case 'u':
3068 case 'v':
3069 case 'w':
3070 case 'x':
3071 case 'y':
3072 case 'z':
3073 /* case 'A' ... 'z': */
3074 case 'A':
3075 case 'B':
3076 case 'C':
3077 case 'D':
3078 case 'E':
3079 case 'F':
3080 case 'G':
3081 case 'H':
3082 case 'I':
3083 case 'J':
3084 case 'K':
3085 case 'L':
3086 case 'M':
3087 case 'N':
3088 case 'O':
3089 case 'P':
3090 case 'Q':
3091 case 'R':
3092 case 'S':
3093 case 'T':
3094 case 'U':
3095 case 'V':
3096 case 'W':
3097 case 'X':
3098 case 'Y':
3099 case 'Z':
3100 case '_':
3101 case '\\':
3102 case '.':
3103 case '$':
3104 case '~':
3105 q = name;
3106 parse_name:
3107 for(;;) {
3108 if (!((ch >= 'a' && ch <= 'z') ||
3109 (ch >= 'A' && ch <= 'Z') ||
3110 (ch >= '0' && ch <= '9') ||
3111 strchr("/.-_+=$:\\,~", ch)))
3112 break;
3113 if ((q - name) < name_size - 1) {
3114 *q++ = ch;
3116 minp();
3118 *q = '\0';
3119 c = LD_TOK_NAME;
3120 break;
3121 case CH_EOF:
3122 c = LD_TOK_EOF;
3123 break;
3124 default:
3125 c = ch;
3126 inp();
3127 break;
3129 return c;
3132 static int ld_add_file(TCCState *s1, const char filename[])
3134 int ret;
3136 ret = tcc_add_file_internal(s1, filename, 0);
3137 if (ret)
3138 ret = tcc_add_dll(s1, filename, 0);
3139 return ret;
3142 static inline int new_undef_syms(void)
3144 int ret = 0;
3145 ret = new_undef_sym;
3146 new_undef_sym = 0;
3147 return ret;
3150 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3152 char filename[1024], libname[1024];
3153 int t, group, nblibs = 0, ret = 0;
3154 char **libs = NULL;
3156 group = !strcmp(cmd, "GROUP");
3157 if (!as_needed)
3158 new_undef_syms();
3159 t = ld_next(s1, filename, sizeof(filename));
3160 if (t != '(')
3161 expect("(");
3162 t = ld_next(s1, filename, sizeof(filename));
3163 for(;;) {
3164 libname[0] = '\0';
3165 if (t == LD_TOK_EOF) {
3166 tcc_error_noabort("unexpected end of file");
3167 ret = -1;
3168 goto lib_parse_error;
3169 } else if (t == ')') {
3170 break;
3171 } else if (t == '-') {
3172 t = ld_next(s1, filename, sizeof(filename));
3173 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3174 tcc_error_noabort("library name expected");
3175 ret = -1;
3176 goto lib_parse_error;
3178 pstrcpy(libname, sizeof libname, &filename[1]);
3179 if (s1->static_link) {
3180 snprintf(filename, sizeof filename, "lib%s.a", libname);
3181 } else {
3182 snprintf(filename, sizeof filename, "lib%s.so", libname);
3184 } else if (t != LD_TOK_NAME) {
3185 tcc_error_noabort("filename expected");
3186 ret = -1;
3187 goto lib_parse_error;
3189 if (!strcmp(filename, "AS_NEEDED")) {
3190 ret = ld_add_file_list(s1, cmd, 1);
3191 if (ret)
3192 goto lib_parse_error;
3193 } else {
3194 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3195 if (!as_needed) {
3196 ret = ld_add_file(s1, filename);
3197 if (ret)
3198 goto lib_parse_error;
3199 if (group) {
3200 /* Add the filename *and* the libname to avoid future conversions */
3201 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3202 if (libname[0] != '\0')
3203 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3207 t = ld_next(s1, filename, sizeof(filename));
3208 if (t == ',') {
3209 t = ld_next(s1, filename, sizeof(filename));
3212 if (group && !as_needed) {
3213 while (new_undef_syms()) {
3214 int i;
3216 for (i = 0; i < nblibs; i ++)
3217 ld_add_file(s1, libs[i]);
3220 lib_parse_error:
3221 dynarray_reset(&libs, &nblibs);
3222 return ret;
3225 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3226 files */
3227 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3229 char cmd[64];
3230 char filename[1024];
3231 int t, ret;
3233 ch = file->buf_ptr[0];
3234 ch = handle_eob();
3235 for(;;) {
3236 t = ld_next(s1, cmd, sizeof(cmd));
3237 if (t == LD_TOK_EOF)
3238 return 0;
3239 else if (t != LD_TOK_NAME)
3240 return -1;
3241 if (!strcmp(cmd, "INPUT") ||
3242 !strcmp(cmd, "GROUP")) {
3243 ret = ld_add_file_list(s1, cmd, 0);
3244 if (ret)
3245 return ret;
3246 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3247 !strcmp(cmd, "TARGET")) {
3248 /* ignore some commands */
3249 t = ld_next(s1, cmd, sizeof(cmd));
3250 if (t != '(')
3251 expect("(");
3252 for(;;) {
3253 t = ld_next(s1, filename, sizeof(filename));
3254 if (t == LD_TOK_EOF) {
3255 tcc_error_noabort("unexpected end of file");
3256 return -1;
3257 } else if (t == ')') {
3258 break;
3261 } else {
3262 return -1;
3265 return 0;
3267 #endif /* !TCC_TARGET_PE */