Update Changelog from git changelog entries
[tinycc.git] / tccelf.c
blob932cf031e0b498656437504023c526ad81828081
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 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
806 qrel->r_addend = *(long long *)ptr + val;
807 qrel++;
809 *(long long *)ptr += val;
810 break;
811 case R_X86_64_32:
812 case R_X86_64_32S:
813 if (s1->output_type == TCC_OUTPUT_DLL) {
814 /* XXX: this logic may depend on TCC's codegen
815 now TCC uses R_X86_64_32 even for a 64bit pointer */
816 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
817 qrel->r_addend = *(int *)ptr + val;
818 qrel++;
820 *(int *)ptr += val;
821 break;
823 case R_X86_64_PC32:
824 if (s1->output_type == TCC_OUTPUT_DLL) {
825 /* DLL relocation */
826 esym_index = s1->symtab_to_dynsym[sym_index];
827 if (esym_index) {
828 qrel->r_offset = rel->r_offset;
829 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_PC32);
830 qrel->r_addend = *(int *)ptr;
831 qrel++;
832 break;
835 /* fall through */
836 case R_X86_64_PLT32: {
837 long long diff;
838 diff = (long long)val - addr;
839 if (diff <= -2147483647 || diff > 2147483647) {
840 #ifdef TCC_HAS_RUNTIME_PLTGOT
841 /* XXX: naive support for over 32bit jump */
842 if (s1->output_type == TCC_OUTPUT_MEMORY) {
843 val = (add_jmp_table(s1, val - rel->r_addend) +
844 rel->r_addend);
845 diff = val - addr;
847 #endif
848 if (diff <= -2147483647 || diff > 2147483647) {
849 tcc_error("internal error: relocation failed");
852 *(int *)ptr += diff;
854 break;
855 case R_X86_64_GLOB_DAT:
856 case R_X86_64_JUMP_SLOT:
857 /* They don't need addend */
858 *(int *)ptr = val - rel->r_addend;
859 break;
860 case R_X86_64_GOTPCREL:
861 #ifdef TCC_HAS_RUNTIME_PLTGOT
862 if (s1->output_type == TCC_OUTPUT_MEMORY) {
863 val = add_got_table(s1, val - rel->r_addend) + rel->r_addend;
864 *(int *)ptr += val - addr;
865 break;
867 #endif
868 *(int *)ptr += (s1->got->sh_addr - addr +
869 s1->sym_attrs[sym_index].got_offset - 4);
870 break;
871 case R_X86_64_GOTTPOFF:
872 *(int *)ptr += val - s1->got->sh_addr;
873 break;
874 case R_X86_64_GOT32:
875 /* we load the got offset */
876 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
877 break;
878 #else
879 #error unsupported processor
880 #endif
883 /* if the relocation is allocated, we change its symbol table */
884 if (sr->sh_flags & SHF_ALLOC)
885 sr->link = s1->dynsym;
888 /* relocate relocation table in 'sr' */
889 static void relocate_rel(TCCState *s1, Section *sr)
891 Section *s;
892 ElfW_Rel *rel;
894 s = s1->sections[sr->sh_info];
895 for_each_elem(sr, 0, rel, ElfW_Rel)
896 rel->r_offset += s->sh_addr;
899 /* count the number of dynamic relocations so that we can reserve
900 their space */
901 static int prepare_dynamic_rel(TCCState *s1, Section *sr)
903 ElfW_Rel *rel;
904 int sym_index, esym_index, type, count;
906 count = 0;
907 for_each_elem(sr, 0, rel, ElfW_Rel) {
908 sym_index = ELFW(R_SYM)(rel->r_info);
909 type = ELFW(R_TYPE)(rel->r_info);
910 switch(type) {
911 #if defined(TCC_TARGET_I386)
912 case R_386_32:
913 #elif defined(TCC_TARGET_X86_64)
914 case R_X86_64_32:
915 case R_X86_64_32S:
916 case R_X86_64_64:
917 #endif
918 count++;
919 break;
920 #if defined(TCC_TARGET_I386)
921 case R_386_PC32:
922 #elif defined(TCC_TARGET_X86_64)
923 case R_X86_64_PC32:
924 #endif
925 esym_index = s1->symtab_to_dynsym[sym_index];
926 if (esym_index)
927 count++;
928 break;
929 default:
930 break;
933 if (count) {
934 /* allocate the section */
935 sr->sh_flags |= SHF_ALLOC;
936 sr->sh_size = count * sizeof(ElfW_Rel);
938 return count;
941 static struct sym_attr *alloc_sym_attr(TCCState *s1, int index)
943 int n;
944 struct sym_attr *tab;
946 if (index >= s1->nb_sym_attrs) {
947 /* find immediately bigger power of 2 and reallocate array */
948 n = 1;
949 while (index >= n)
950 n *= 2;
951 tab = tcc_realloc(s1->sym_attrs, n * sizeof(*s1->sym_attrs));
952 s1->sym_attrs = tab;
953 memset(s1->sym_attrs + s1->nb_sym_attrs, 0,
954 (n - s1->nb_sym_attrs) * sizeof(*s1->sym_attrs));
955 s1->nb_sym_attrs = n;
957 return &s1->sym_attrs[index];
960 /* XXX: suppress that */
961 static void put32(unsigned char *p, uint32_t val)
963 p[0] = val;
964 p[1] = val >> 8;
965 p[2] = val >> 16;
966 p[3] = val >> 24;
969 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_ARM) || \
970 defined(TCC_TARGET_X86_64)
971 static uint32_t get32(unsigned char *p)
973 return p[0] | (p[1] << 8) | (p[2] << 16) | (p[3] << 24);
975 #endif
977 static void build_got(TCCState *s1)
979 unsigned char *ptr;
981 /* if no got, then create it */
982 s1->got = new_section(s1, ".got", SHT_PROGBITS, SHF_ALLOC | SHF_WRITE);
983 s1->got->sh_entsize = 4;
984 add_elf_sym(symtab_section, 0, 4, ELFW(ST_INFO)(STB_GLOBAL, STT_OBJECT),
985 0, s1->got->sh_num, "_GLOBAL_OFFSET_TABLE_");
986 ptr = section_ptr_add(s1->got, 3 * PTR_SIZE);
987 #if PTR_SIZE == 4
988 /* keep space for _DYNAMIC pointer, if present */
989 put32(ptr, 0);
990 /* two dummy got entries */
991 put32(ptr + 4, 0);
992 put32(ptr + 8, 0);
993 #else
994 /* keep space for _DYNAMIC pointer, if present */
995 put32(ptr, 0);
996 put32(ptr + 4, 0);
997 /* two dummy got entries */
998 put32(ptr + 8, 0);
999 put32(ptr + 12, 0);
1000 put32(ptr + 16, 0);
1001 put32(ptr + 20, 0);
1002 #endif
1005 /* put a got entry corresponding to a symbol in symtab_section. 'size'
1006 and 'info' can be modifed if more precise info comes from the DLL */
1007 static void put_got_entry(TCCState *s1,
1008 int reloc_type, unsigned long size, int info,
1009 int sym_index)
1011 int index, need_plt_entry, got_entry_present = 0;
1012 const char *name;
1013 ElfW(Sym) *sym;
1014 unsigned long offset;
1015 int *ptr;
1017 if (!s1->got)
1018 build_got(s1);
1020 need_plt_entry = s1->dynsym &&
1021 #ifdef TCC_TARGET_X86_64
1022 (reloc_type == R_X86_64_JUMP_SLOT);
1023 #elif defined(TCC_TARGET_I386)
1024 (reloc_type == R_386_JMP_SLOT);
1025 #elif defined(TCC_TARGET_ARM)
1026 (reloc_type == R_ARM_JUMP_SLOT);
1027 #else
1029 #endif
1031 /* if a got entry already exists for that symbol, no need to add one */
1032 if (sym_index < s1->nb_sym_attrs &&
1033 s1->sym_attrs[sym_index].got_offset) {
1034 if (!need_plt_entry || s1->sym_attrs[sym_index].has_plt_entry)
1035 return;
1036 else
1037 got_entry_present = 1;
1040 alloc_sym_attr(s1, sym_index)->got_offset = s1->got->data_offset;
1042 if (s1->dynsym) {
1043 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1044 name = (char *) symtab_section->link->data + sym->st_name;
1045 offset = sym->st_value;
1046 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_X86_64)
1047 if (need_plt_entry) {
1048 Section *plt;
1049 uint8_t *p;
1050 int modrm;
1052 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1053 modrm = 0x25;
1054 #else
1055 /* if we build a DLL, we add a %ebx offset */
1056 if (s1->output_type == TCC_OUTPUT_DLL)
1057 modrm = 0xa3;
1058 else
1059 modrm = 0x25;
1060 #endif
1062 /* add a PLT entry */
1063 plt = s1->plt;
1064 if (plt->data_offset == 0) {
1065 /* first plt entry */
1066 p = section_ptr_add(plt, 16);
1067 p[0] = 0xff; /* pushl got + PTR_SIZE */
1068 p[1] = modrm + 0x10;
1069 put32(p + 2, PTR_SIZE);
1070 p[6] = 0xff; /* jmp *(got + PTR_SIZE * 2) */
1071 p[7] = modrm;
1072 put32(p + 8, PTR_SIZE * 2);
1075 p = section_ptr_add(plt, 16);
1076 p[0] = 0xff; /* jmp *(got + x) */
1077 p[1] = modrm;
1078 put32(p + 2, s1->got->data_offset);
1079 p[6] = 0x68; /* push $xxx */
1080 put32(p + 7, (plt->data_offset - 32) >> 1);
1081 p[11] = 0xe9; /* jmp plt_start */
1082 put32(p + 12, -(plt->data_offset));
1084 /* the symbol is modified so that it will be relocated to
1085 the PLT */
1086 #if !defined(TCC_OUTPUT_DLL_WITH_PLT)
1087 if (s1->output_type == TCC_OUTPUT_EXE)
1088 #endif
1089 offset = plt->data_offset - 16;
1090 s1->sym_attrs[sym_index].has_plt_entry = 1;
1092 #elif defined(TCC_TARGET_ARM)
1093 if (need_plt_entry) {
1094 Section *plt;
1095 uint8_t *p;
1097 /* if we build a DLL, we add a %ebx offset */
1098 if (s1->output_type == TCC_OUTPUT_DLL)
1099 tcc_error("DLLs unimplemented!");
1101 /* add a PLT entry */
1102 plt = s1->plt;
1103 if (plt->data_offset == 0) {
1104 /* first plt entry */
1105 p = section_ptr_add(plt, 16);
1106 put32(p, 0xe52de004); /* push {lr} */
1107 put32(p+4, 0xe59fe010); /* ldr lr, [pc, #16] */
1108 put32(p+8, 0xe08fe00e); /* add lr, pc, lr */
1109 put32(p+12, 0xe5bef008); /* ldr pc, [lr, #8]! */
1112 if (s1->sym_attrs[sym_index].plt_thumb_stub) {
1113 p = section_ptr_add(plt, 20);
1114 put32(p, 0x4778); /* bx pc */
1115 put32(p+2, 0x46c0); /* nop */
1116 p += 4;
1117 } else
1118 p = section_ptr_add(plt, 16);
1119 put32(p, 0xe59fc004); /* ldr ip, [pc, #4] ; GOT entry offset */
1120 put32(p+4, 0xe08fc00c); /* add ip, pc, ip ; addr of GOT entry */
1121 put32(p+8, 0xe59cf000); /* ldr pc, [ip] ; jump to GOT entry */
1122 put32(p+12, s1->got->data_offset); /* GOT entry off once patched */
1124 /* the symbol is modified so that it will be relocated to
1125 the PLT */
1126 if (s1->output_type == TCC_OUTPUT_EXE)
1127 offset = plt->data_offset - 16;
1128 s1->sym_attrs[sym_index].has_plt_entry = 1;
1130 #elif defined(TCC_TARGET_C67)
1131 tcc_error("C67 got not implemented");
1132 #else
1133 #error unsupported CPU
1134 #endif
1135 index = put_elf_sym(s1->dynsym, offset,
1136 size, info, 0, sym->st_shndx, name);
1137 if (got_entry_present) {
1138 put_elf_reloc(s1->dynsym, s1->got,
1139 s1->sym_attrs[sym_index].got_offset,
1140 reloc_type, index);
1141 return;
1144 /* put a got entry */
1145 put_elf_reloc(s1->dynsym, s1->got,
1146 s1->got->data_offset,
1147 reloc_type, index);
1149 ptr = section_ptr_add(s1->got, PTR_SIZE);
1150 *ptr = 0;
1153 /* build GOT and PLT entries */
1154 ST_FUNC void build_got_entries(TCCState *s1)
1156 Section *s;
1157 ElfW_Rel *rel;
1158 ElfW(Sym) *sym;
1159 int i, type, reloc_type, sym_index;
1161 for(i = 1; i < s1->nb_sections; i++) {
1162 s = s1->sections[i];
1163 if (s->sh_type != SHT_RELX)
1164 continue;
1165 /* no need to handle got relocations */
1166 if (s->link != symtab_section)
1167 continue;
1168 for_each_elem(s, 0, rel, ElfW_Rel) {
1169 type = ELFW(R_TYPE)(rel->r_info);
1170 switch(type) {
1171 #if defined(TCC_TARGET_I386)
1172 case R_386_GOT32:
1173 case R_386_GOTOFF:
1174 case R_386_GOTPC:
1175 case R_386_PLT32:
1176 if (!s1->got)
1177 build_got(s1);
1178 if (type == R_386_GOT32 || type == R_386_PLT32) {
1179 sym_index = ELFW(R_SYM)(rel->r_info);
1180 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1181 /* look at the symbol got offset. If none, then add one */
1182 if (type == R_386_GOT32)
1183 reloc_type = R_386_GLOB_DAT;
1184 else
1185 reloc_type = R_386_JMP_SLOT;
1186 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1187 sym_index);
1189 break;
1190 #elif defined(TCC_TARGET_ARM)
1191 case R_ARM_GOT32:
1192 case R_ARM_GOTOFF:
1193 case R_ARM_GOTPC:
1194 case R_ARM_PLT32:
1195 if (!s1->got)
1196 build_got(s1);
1197 if (type == R_ARM_GOT32 || type == R_ARM_PLT32) {
1198 sym_index = ELFW(R_SYM)(rel->r_info);
1199 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1200 /* look at the symbol got offset. If none, then add one */
1201 if (type == R_ARM_GOT32)
1202 reloc_type = R_ARM_GLOB_DAT;
1203 else
1204 reloc_type = R_ARM_JUMP_SLOT;
1205 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1206 sym_index);
1208 break;
1209 case R_ARM_THM_JUMP24:
1210 sym_index = ELFW(R_SYM)(rel->r_info);
1211 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1212 /* We are relocating a jump from thumb code to arm code */
1213 if (sym->st_shndx != SHN_UNDEF && !(sym->st_value & 1)) {
1214 int index;
1215 uint8_t *p;
1216 char *name, buf[1024];
1217 Section *text_section;
1219 name = (char *) symtab_section->link->data + sym->st_name;
1220 text_section = s1->sections[sym->st_shndx];
1221 /* Modify reloc to target a thumb stub to switch to ARM */
1222 snprintf(buf, sizeof(buf), "%s_from_thumb", name);
1223 index = put_elf_sym(symtab_section,
1224 text_section->data_offset + 1,
1225 sym->st_size, sym->st_info, 0,
1226 sym->st_shndx, buf);
1227 rel->r_info = ELFW(R_INFO)(index, type);
1228 /* Create a thumb stub fonction to switch to ARM mode */
1229 put_elf_reloc(symtab_section, text_section,
1230 text_section->data_offset + 4, R_ARM_JUMP24,
1231 sym_index);
1232 p = section_ptr_add(text_section, 8);
1233 put32(p, 0x4778); /* bx pc */
1234 put32(p+2, 0x46c0); /* nop */
1235 put32(p+4, 0xeafffffe); /* b $sym */
1237 #elif defined(TCC_TARGET_C67)
1238 case R_C60_GOT32:
1239 case R_C60_GOTOFF:
1240 case R_C60_GOTPC:
1241 case R_C60_PLT32:
1242 if (!s1->got)
1243 build_got(s1);
1244 if (type == R_C60_GOT32 || type == R_C60_PLT32) {
1245 sym_index = ELFW(R_SYM)(rel->r_info);
1246 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1247 /* look at the symbol got offset. If none, then add one */
1248 if (type == R_C60_GOT32)
1249 reloc_type = R_C60_GLOB_DAT;
1250 else
1251 reloc_type = R_C60_JMP_SLOT;
1252 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1253 sym_index);
1255 break;
1256 #elif defined(TCC_TARGET_X86_64)
1257 case R_X86_64_GOT32:
1258 case R_X86_64_GOTTPOFF:
1259 case R_X86_64_GOTPCREL:
1260 case R_X86_64_PLT32:
1261 if (!s1->got)
1262 build_got(s1);
1263 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL ||
1264 type == R_X86_64_PLT32) {
1265 sym_index = ELFW(R_SYM)(rel->r_info);
1266 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1267 /* look at the symbol got offset. If none, then add one */
1268 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL)
1269 reloc_type = R_X86_64_GLOB_DAT;
1270 else
1271 reloc_type = R_X86_64_JUMP_SLOT;
1272 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1273 sym_index);
1275 break;
1276 #else
1277 #error unsupported CPU
1278 #endif
1279 default:
1280 break;
1286 ST_FUNC Section *new_symtab(TCCState *s1,
1287 const char *symtab_name, int sh_type, int sh_flags,
1288 const char *strtab_name,
1289 const char *hash_name, int hash_sh_flags)
1291 Section *symtab, *strtab, *hash;
1292 int *ptr, nb_buckets;
1294 symtab = new_section(s1, symtab_name, sh_type, sh_flags);
1295 symtab->sh_entsize = sizeof(ElfW(Sym));
1296 strtab = new_section(s1, strtab_name, SHT_STRTAB, sh_flags);
1297 put_elf_str(strtab, "");
1298 symtab->link = strtab;
1299 put_elf_sym(symtab, 0, 0, 0, 0, 0, NULL);
1301 nb_buckets = 1;
1303 hash = new_section(s1, hash_name, SHT_HASH, hash_sh_flags);
1304 hash->sh_entsize = sizeof(int);
1305 symtab->hash = hash;
1306 hash->link = symtab;
1308 ptr = section_ptr_add(hash, (2 + nb_buckets + 1) * sizeof(int));
1309 ptr[0] = nb_buckets;
1310 ptr[1] = 1;
1311 memset(ptr + 2, 0, (nb_buckets + 1) * sizeof(int));
1312 return symtab;
1315 /* put dynamic tag */
1316 static void put_dt(Section *dynamic, int dt, addr_t val)
1318 ElfW(Dyn) *dyn;
1319 dyn = section_ptr_add(dynamic, sizeof(ElfW(Dyn)));
1320 dyn->d_tag = dt;
1321 dyn->d_un.d_val = val;
1324 static void add_init_array_defines(TCCState *s1, const char *section_name)
1326 Section *s;
1327 long end_offset;
1328 char sym_start[1024];
1329 char sym_end[1024];
1331 snprintf(sym_start, sizeof(sym_start), "__%s_start", section_name + 1);
1332 snprintf(sym_end, sizeof(sym_end), "__%s_end", section_name + 1);
1334 s = find_section(s1, section_name);
1335 if (!s) {
1336 end_offset = 0;
1337 s = data_section;
1338 } else {
1339 end_offset = s->data_offset;
1342 add_elf_sym(symtab_section,
1343 0, 0,
1344 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1345 s->sh_num, sym_start);
1346 add_elf_sym(symtab_section,
1347 end_offset, 0,
1348 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1349 s->sh_num, sym_end);
1352 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1354 #ifdef CONFIG_TCC_BCHECK
1355 unsigned long *ptr;
1356 Section *init_section;
1357 unsigned char *pinit;
1358 int sym_index;
1360 if (0 == s1->do_bounds_check)
1361 return;
1363 /* XXX: add an object file to do that */
1364 ptr = section_ptr_add(bounds_section, sizeof(unsigned long));
1365 *ptr = 0;
1366 add_elf_sym(symtab_section, 0, 0,
1367 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1368 bounds_section->sh_num, "__bounds_start");
1369 #ifdef TCC_TARGET_I386
1370 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1371 /* add 'call __bound_init()' in .init section */
1372 init_section = find_section(s1, ".init");
1373 pinit = section_ptr_add(init_section, 5);
1374 pinit[0] = 0xe8;
1375 put32(pinit + 1, -4);
1376 sym_index = find_elf_sym(symtab_section, "__bound_init");
1377 put_elf_reloc(symtab_section, init_section,
1378 init_section->data_offset - 4, R_386_PC32, sym_index);
1380 #endif
1381 #endif
1384 static inline int tcc_add_support(TCCState *s1, const char *filename)
1386 char buf[1024];
1387 snprintf(buf, sizeof(buf), "%s/%s", s1->tcc_lib_path, filename);
1388 return tcc_add_file(s1, buf);
1391 /* add tcc runtime libraries */
1392 ST_FUNC void tcc_add_runtime(TCCState *s1)
1394 /* add libc */
1395 if (!s1->nostdlib) {
1396 tcc_add_library(s1, "c");
1397 #ifdef CONFIG_USE_LIBGCC
1398 if (!s1->static_link) {
1399 tcc_add_file(s1, TCC_LIBGCC);
1400 tcc_add_support(s1, "libtcc1.a");
1401 } else
1402 tcc_add_support(s1, "libtcc1.a");
1403 #else
1404 tcc_add_support(s1, "libtcc1.a");
1405 #endif
1408 /* tcc_add_bcheck tries to relocate a call to __bound_init in _init so
1409 libtcc1.a must be loaded before for __bound_init to be defined and
1410 crtn.o must be loaded after to not finalize _init too early. */
1411 tcc_add_bcheck(s1);
1413 if (!s1->nostdlib) {
1414 /* add crt end if not memory output */
1415 if (s1->output_type != TCC_OUTPUT_MEMORY)
1416 tcc_add_crt(s1, "crtn.o");
1420 /* add various standard linker symbols (must be done after the
1421 sections are filled (for example after allocating common
1422 symbols)) */
1423 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1425 char buf[1024];
1426 int i;
1427 Section *s;
1429 add_elf_sym(symtab_section,
1430 text_section->data_offset, 0,
1431 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1432 text_section->sh_num, "_etext");
1433 add_elf_sym(symtab_section,
1434 data_section->data_offset, 0,
1435 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1436 data_section->sh_num, "_edata");
1437 add_elf_sym(symtab_section,
1438 bss_section->data_offset, 0,
1439 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1440 bss_section->sh_num, "_end");
1441 /* horrible new standard ldscript defines */
1442 add_init_array_defines(s1, ".preinit_array");
1443 add_init_array_defines(s1, ".init_array");
1444 add_init_array_defines(s1, ".fini_array");
1446 /* add start and stop symbols for sections whose name can be
1447 expressed in C */
1448 for(i = 1; i < s1->nb_sections; i++) {
1449 s = s1->sections[i];
1450 if (s->sh_type == SHT_PROGBITS &&
1451 (s->sh_flags & SHF_ALLOC)) {
1452 const char *p;
1453 int ch;
1455 /* check if section name can be expressed in C */
1456 p = s->name;
1457 for(;;) {
1458 ch = *p;
1459 if (!ch)
1460 break;
1461 if (!isid(ch) && !isnum(ch))
1462 goto next_sec;
1463 p++;
1465 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1466 add_elf_sym(symtab_section,
1467 0, 0,
1468 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1469 s->sh_num, buf);
1470 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1471 add_elf_sym(symtab_section,
1472 s->data_offset, 0,
1473 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1474 s->sh_num, buf);
1476 next_sec: ;
1480 static void tcc_output_binary(TCCState *s1, FILE *f,
1481 const int *sec_order)
1483 Section *s;
1484 int i, offset, size;
1486 offset = 0;
1487 for(i=1;i<s1->nb_sections;i++) {
1488 s = s1->sections[sec_order[i]];
1489 if (s->sh_type != SHT_NOBITS &&
1490 (s->sh_flags & SHF_ALLOC)) {
1491 while (offset < s->sh_offset) {
1492 fputc(0, f);
1493 offset++;
1495 size = s->sh_size;
1496 fwrite(s->data, 1, size, f);
1497 offset += size;
1502 // making this evaluate to true allow valgrind to work on linux
1503 // but when compiled with debug info and then striped
1504 // the compiled programs segfault
1505 // more tought must be applyed here
1506 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1507 #define HAVE_PHDR 1
1508 #define EXTRA_RELITEMS 14
1510 /* move the relocation value from .dynsym to .got */
1511 void patch_dynsym_undef(TCCState *s1, Section *s)
1513 uint32_t *gotd = (void *)s1->got->data;
1514 ElfW(Sym) *sym;
1516 gotd += 3; /* dummy entries in .got */
1517 /* relocate symbols in .dynsym */
1518 for_each_elem(s, 1, sym, ElfW(Sym)) {
1519 if (sym->st_shndx == SHN_UNDEF) {
1520 *gotd++ = sym->st_value + 6; /* XXX 6 is magic ? */
1521 sym->st_value = 0;
1525 #else
1526 #define HAVE_PHDR 0
1527 #define EXTRA_RELITEMS 9
1529 /* zero plt offsets of weak symbols in .dynsym */
1530 void patch_dynsym_undef(TCCState *s1, Section *s)
1532 ElfW(Sym) *sym;
1534 for_each_elem(s, 1, sym, ElfW(Sym))
1535 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1536 sym->st_value = 0;
1538 #endif
1540 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1542 int sym_index = ELFW(R_SYM) (rel->r_info);
1543 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1544 unsigned long offset;
1546 if (sym_index >= s1->nb_sym_attrs)
1547 return;
1548 offset = s1->sym_attrs[sym_index].got_offset;
1549 section_reserve(s1->got, offset + PTR_SIZE);
1550 #ifdef TCC_TARGET_X86_64
1551 /* only works for x86-64 */
1552 put32(s1->got->data + offset + 4, sym->st_value >> 32);
1553 #endif
1554 put32(s1->got->data + offset, sym->st_value & 0xffffffff);
1557 /* Perform relocation to GOT or PLT entries */
1558 ST_FUNC void fill_got(TCCState *s1)
1560 Section *s;
1561 ElfW_Rel *rel;
1562 int i;
1564 for(i = 1; i < s1->nb_sections; i++) {
1565 s = s1->sections[i];
1566 if (s->sh_type != SHT_RELX)
1567 continue;
1568 /* no need to handle got relocations */
1569 if (s->link != symtab_section)
1570 continue;
1571 for_each_elem(s, 0, rel, ElfW_Rel) {
1572 switch (ELFW(R_TYPE) (rel->r_info)) {
1573 case R_X86_64_GOT32:
1574 case R_X86_64_GOTPCREL:
1575 case R_X86_64_PLT32:
1576 fill_got_entry(s1, rel);
1577 break;
1583 /* Bind symbols of executable: resolve undefined symbols from exported symbols
1584 in shared libraries and export non local defined symbols to shared libraries
1585 if -rdynamic switch was given on command line */
1586 static void bind_exe_dynsyms(TCCState *s1)
1588 const char *name;
1589 int sym_index, index;
1590 ElfW(Sym) *sym, *esym;
1591 int type;
1593 /* Resolve undefined symbols from dynamic symbols. When there is a match:
1594 - if STT_FUNC or STT_GNU_IFUNC symbol -> add it in PLT
1595 - if STT_OBJECT symbol -> add it in .bss section with suitable reloc */
1596 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1597 if (sym->st_shndx == SHN_UNDEF) {
1598 name = (char *) symtab_section->link->data + sym->st_name;
1599 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1600 if (sym_index) {
1601 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1602 type = ELFW(ST_TYPE)(esym->st_info);
1603 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1604 /* Indirect functions shall have STT_FUNC type in executable
1605 * dynsym section. Indeed, a dlsym call following a lazy
1606 * resolution would pick the symbol value from the
1607 * executable dynsym entry which would contain the address
1608 * of the function wanted by the caller of dlsym instead of
1609 * the address of the function that would return that
1610 * address */
1611 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1612 ELFW(ST_INFO)(STB_GLOBAL,STT_FUNC),
1613 sym - (ElfW(Sym) *)symtab_section->data);
1614 } else if (type == STT_OBJECT) {
1615 unsigned long offset;
1616 ElfW(Sym) *dynsym;
1617 offset = bss_section->data_offset;
1618 /* XXX: which alignment ? */
1619 offset = (offset + 16 - 1) & -16;
1620 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1621 esym->st_info, 0, bss_section->sh_num,
1622 name);
1623 /* Ensure R_COPY works for weak symbol aliases */
1624 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1625 for_each_elem(s1->dynsymtab_section, 1, dynsym, ElfW(Sym)) {
1626 if ((dynsym->st_value == esym->st_value)
1627 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1628 char *dynname = (char *) s1->dynsymtab_section->link->data
1629 + dynsym->st_name;
1630 put_elf_sym(s1->dynsym, offset, dynsym->st_size,
1631 dynsym->st_info, 0,
1632 bss_section->sh_num, dynname);
1633 break;
1637 put_elf_reloc(s1->dynsym, bss_section,
1638 offset, R_COPY, index);
1639 offset += esym->st_size;
1640 bss_section->data_offset = offset;
1642 } else {
1643 /* STB_WEAK undefined symbols are accepted */
1644 /* XXX: _fp_hw seems to be part of the ABI, so we ignore it */
1645 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1646 !strcmp(name, "_fp_hw")) {
1647 } else {
1648 tcc_error_noabort("undefined symbol '%s'", name);
1651 } else if (s1->rdynamic && ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1652 /* if -rdynamic option, then export all non local symbols */
1653 name = (char *) symtab_section->link->data + sym->st_name;
1654 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size, sym->st_info,
1655 0, sym->st_shndx, name);
1660 /* Bind symbols of libraries: export non local symbols of executable that
1661 resolve undefined symbols of shared libraries */
1662 static void bind_libs_dynsyms(TCCState *s1)
1664 const char *name;
1665 int sym_index;
1666 ElfW(Sym) *sym, *esym;
1668 /* now look at unresolved dynamic symbols and export
1669 corresponding symbol */
1670 for_each_elem(s1->dynsymtab_section, 1, esym, ElfW(Sym)) {
1671 if (esym->st_shndx == SHN_UNDEF) {
1672 name = (char *) s1->dynsymtab_section->link->data + esym->st_name;
1673 sym_index = find_elf_sym(symtab_section, name);
1674 if (sym_index) {
1675 /* XXX: avoid adding a symbol if already present because of
1676 -rdynamic ? */
1677 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1678 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1679 sym->st_info, 0, sym->st_shndx, name);
1680 } else {
1681 /* weak symbols can stay undefined */
1682 if (ELFW(ST_BIND)(esym->st_info) != STB_WEAK)
1683 tcc_warning("undefined dynamic symbol '%s'", name);
1689 /* Export all non local symbols (for shared libraries) */
1690 static void export_global_syms(TCCState *s1)
1692 int nb_syms, dynindex, index;
1693 const char *name;
1694 ElfW(Sym) *sym;
1696 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1697 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1698 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1699 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1700 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1701 int type = ELFW(ST_TYPE)(sym->st_info);
1702 if ((type == STT_FUNC || type == STT_GNU_IFUNC)
1703 && sym->st_shndx == SHN_UNDEF) {
1704 int visibility = ELFW(ST_BIND)(sym->st_info);
1705 put_got_entry(s1, R_JMP_SLOT, sym->st_size,
1706 ELFW(ST_INFO)(visibility, STT_FUNC),
1707 sym - (ElfW(Sym) *) symtab_section->data);
1708 } else if (type == STT_OBJECT) {
1709 put_got_entry(s1, R_X86_64_GLOB_DAT, sym->st_size, sym->st_info,
1710 sym - (ElfW(Sym) *) symtab_section->data);
1711 } else
1712 #endif
1714 name = (char *) symtab_section->link->data + sym->st_name;
1715 dynindex = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1716 sym->st_info, 0, sym->st_shndx, name);
1717 index = sym - (ElfW(Sym) *) symtab_section->data;
1718 s1->symtab_to_dynsym[index] = dynindex;
1724 /* relocate the PLT: compute addresses and offsets in the PLT now that final
1725 address for PLT and GOT are known (see fill_program_header) */
1726 static void relocate_plt(TCCState *s1)
1728 uint8_t *p, *p_end;
1730 p = s1->plt->data;
1731 p_end = p + s1->plt->data_offset;
1732 if (p < p_end) {
1733 #if defined(TCC_TARGET_I386)
1734 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1735 put32(p + 8, get32(p + 8) + s1->got->sh_addr);
1736 p += 16;
1737 while (p < p_end) {
1738 put32(p + 2, get32(p + 2) + s1->got->sh_addr);
1739 p += 16;
1741 #elif defined(TCC_TARGET_X86_64)
1742 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
1743 put32(p + 2, get32(p + 2) + x);
1744 put32(p + 8, get32(p + 8) + x - 6);
1745 p += 16;
1746 while (p < p_end) {
1747 put32(p + 2, get32(p + 2) + x + s1->plt->data - p);
1748 p += 16;
1750 #elif defined(TCC_TARGET_ARM)
1751 int x;
1752 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
1753 p += 16;
1754 while (p < p_end) {
1755 if (get32(p) == 0x46c04778) /* PLT Thumb stub present */
1756 p += 4;
1757 put32(p + 12, x + get32(p + 12) + s1->plt->data - p);
1758 p += 16;
1760 #elif defined(TCC_TARGET_C67)
1761 /* XXX: TODO */
1762 #else
1763 #error unsupported CPU
1764 #endif
1768 /* Allocate strings for section names and decide if an unallocated section
1769 should be output.
1771 NOTE: the strsec section comes last, so its size is also correct ! */
1772 static void alloc_sec_names(TCCState *s1, int file_type, Section *strsec)
1774 int i;
1775 Section *s;
1777 /* Allocate strings for section names */
1778 for(i = 1; i < s1->nb_sections; i++) {
1779 s = s1->sections[i];
1780 s->sh_name = put_elf_str(strsec, s->name);
1781 /* when generating a DLL, we include relocations but we may
1782 patch them */
1783 if (file_type == TCC_OUTPUT_DLL &&
1784 s->sh_type == SHT_RELX &&
1785 !(s->sh_flags & SHF_ALLOC)) {
1786 /* gr: avoid bogus relocs for empty (debug) sections */
1787 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
1788 prepare_dynamic_rel(s1, s);
1789 else if (s1->do_debug)
1790 s->sh_size = s->data_offset;
1791 } else if (s1->do_debug ||
1792 file_type == TCC_OUTPUT_OBJ ||
1793 (s->sh_flags & SHF_ALLOC) ||
1794 i == (s1->nb_sections - 1)) {
1795 /* we output all sections if debug or object file */
1796 s->sh_size = s->data_offset;
1801 /* Info to be copied in dynamic section */
1802 struct dyn_inf {
1803 Section *dynamic;
1804 Section *dynstr;
1805 unsigned long dyn_rel_off;
1806 addr_t rel_addr;
1807 addr_t rel_size;
1808 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1809 addr_t bss_addr;
1810 addr_t bss_size;
1811 #endif
1814 /* Assign sections to segments and decide how are sections laid out when loaded
1815 in memory. This function also fills corresponding program headers. */
1816 static int layout_sections(TCCState *s1, ElfW(Phdr) *phdr, int phnum,
1817 Section *interp, struct dyn_inf *dyninf,
1818 int *sec_order)
1820 int i, j, k, file_type, sh_order_index, file_offset;
1821 long long tmp;
1822 addr_t addr;
1823 ElfW(Phdr) *ph;
1824 Section *s;
1826 file_type = s1->output_type;
1827 sh_order_index = 1;
1828 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
1829 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
1830 else
1831 file_offset = 0;
1833 if (phnum > 0) {
1834 if (s1->has_text_addr) {
1835 int a_offset, p_offset;
1836 addr = s1->text_addr;
1837 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
1838 ELF_PAGE_SIZE */
1839 a_offset = (int) (addr & (s1->section_align - 1));
1840 p_offset = file_offset & (s1->section_align - 1);
1841 if (a_offset < p_offset)
1842 a_offset += s1->section_align;
1843 file_offset += (a_offset - p_offset);
1844 } else {
1845 if (file_type == TCC_OUTPUT_DLL)
1846 addr = 0;
1847 else
1848 addr = ELF_START_ADDR;
1849 /* compute address after headers */
1850 addr += (file_offset & (s1->section_align - 1));
1853 ph = &phdr[0];
1854 /* Leave one program headers for the program interpreter and one for
1855 the program header table itself if needed. These are done later as
1856 they require section layout to be done first. */
1857 if (interp)
1858 ph += 1 + HAVE_PHDR;
1860 /* dynamic relocation table information, for .dynamic section */
1861 dyninf->rel_addr = dyninf->rel_size = 0;
1862 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1863 dyninf->bss_addr = dyninf->bss_size = 0;
1864 #endif
1866 for(j = 0; j < 2; j++) {
1867 ph->p_type = PT_LOAD;
1868 if (j == 0)
1869 ph->p_flags = PF_R | PF_X;
1870 else
1871 ph->p_flags = PF_R | PF_W;
1872 ph->p_align = s1->section_align;
1874 /* Decide the layout of sections loaded in memory. This must
1875 be done before program headers are filled since they contain
1876 info about the layout. We do the following ordering: interp,
1877 symbol tables, relocations, progbits, nobits */
1878 /* XXX: do faster and simpler sorting */
1879 for(k = 0; k < 5; k++) {
1880 for(i = 1; i < s1->nb_sections; i++) {
1881 s = s1->sections[i];
1882 /* compute if section should be included */
1883 if (j == 0) {
1884 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1885 SHF_ALLOC)
1886 continue;
1887 } else {
1888 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
1889 (SHF_ALLOC | SHF_WRITE))
1890 continue;
1892 if (s == interp) {
1893 if (k != 0)
1894 continue;
1895 } else if (s->sh_type == SHT_DYNSYM ||
1896 s->sh_type == SHT_STRTAB ||
1897 s->sh_type == SHT_HASH) {
1898 if (k != 1)
1899 continue;
1900 } else if (s->sh_type == SHT_RELX) {
1901 if (k != 2)
1902 continue;
1903 } else if (s->sh_type == SHT_NOBITS) {
1904 if (k != 4)
1905 continue;
1906 } else {
1907 if (k != 3)
1908 continue;
1910 sec_order[sh_order_index++] = i;
1912 /* section matches: we align it and add its size */
1913 tmp = addr;
1914 addr = (addr + s->sh_addralign - 1) &
1915 ~(s->sh_addralign - 1);
1916 file_offset += (int) ( addr - tmp );
1917 s->sh_offset = file_offset;
1918 s->sh_addr = addr;
1920 /* update program header infos */
1921 if (ph->p_offset == 0) {
1922 ph->p_offset = file_offset;
1923 ph->p_vaddr = addr;
1924 ph->p_paddr = ph->p_vaddr;
1926 /* update dynamic relocation infos */
1927 if (s->sh_type == SHT_RELX) {
1928 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1929 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) {
1930 dyninf->rel_addr = addr;
1931 dyninf->rel_size += s->sh_size; /* XXX only first rel. */
1933 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) {
1934 dyninf->bss_addr = addr;
1935 dyninf->bss_size = s->sh_size; /* XXX only first rel. */
1937 #else
1938 if (dyninf->rel_size == 0)
1939 dyninf->rel_addr = addr;
1940 dyninf->rel_size += s->sh_size;
1941 #endif
1943 addr += s->sh_size;
1944 if (s->sh_type != SHT_NOBITS)
1945 file_offset += s->sh_size;
1948 ph->p_filesz = file_offset - ph->p_offset;
1949 ph->p_memsz = addr - ph->p_vaddr;
1950 ph++;
1951 if (j == 0) {
1952 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
1953 /* if in the middle of a page, we duplicate the page in
1954 memory so that one copy is RX and the other is RW */
1955 if ((addr & (s1->section_align - 1)) != 0)
1956 addr += s1->section_align;
1957 } else {
1958 addr = (addr + s1->section_align - 1) & ~(s1->section_align - 1);
1959 file_offset = (file_offset + s1->section_align - 1) &
1960 ~(s1->section_align - 1);
1966 /* all other sections come after */
1967 for(i = 1; i < s1->nb_sections; i++) {
1968 s = s1->sections[i];
1969 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
1970 continue;
1971 sec_order[sh_order_index++] = i;
1973 file_offset = (file_offset + s->sh_addralign - 1) &
1974 ~(s->sh_addralign - 1);
1975 s->sh_offset = file_offset;
1976 if (s->sh_type != SHT_NOBITS)
1977 file_offset += s->sh_size;
1980 return file_offset;
1983 static void fill_unloadable_phdr(ElfW(Phdr) *phdr, int phnum, Section *interp,
1984 Section *dynamic)
1986 ElfW(Phdr) *ph;
1988 /* if interpreter, then add corresponding program header */
1989 if (interp) {
1990 ph = &phdr[0];
1992 if (HAVE_PHDR)
1994 int len = phnum * sizeof(ElfW(Phdr));
1996 ph->p_type = PT_PHDR;
1997 ph->p_offset = sizeof(ElfW(Ehdr));
1998 ph->p_vaddr = interp->sh_addr - len;
1999 ph->p_paddr = ph->p_vaddr;
2000 ph->p_filesz = ph->p_memsz = len;
2001 ph->p_flags = PF_R | PF_X;
2002 ph->p_align = 4; /* interp->sh_addralign; */
2003 ph++;
2006 ph->p_type = PT_INTERP;
2007 ph->p_offset = interp->sh_offset;
2008 ph->p_vaddr = interp->sh_addr;
2009 ph->p_paddr = ph->p_vaddr;
2010 ph->p_filesz = interp->sh_size;
2011 ph->p_memsz = interp->sh_size;
2012 ph->p_flags = PF_R;
2013 ph->p_align = interp->sh_addralign;
2016 /* if dynamic section, then add corresponding program header */
2017 if (dynamic) {
2018 ph = &phdr[phnum - 1];
2020 ph->p_type = PT_DYNAMIC;
2021 ph->p_offset = dynamic->sh_offset;
2022 ph->p_vaddr = dynamic->sh_addr;
2023 ph->p_paddr = ph->p_vaddr;
2024 ph->p_filesz = dynamic->sh_size;
2025 ph->p_memsz = dynamic->sh_size;
2026 ph->p_flags = PF_R | PF_W;
2027 ph->p_align = dynamic->sh_addralign;
2031 /* Fill the dynamic section with tags describing the address and size of
2032 sections */
2033 static void fill_dynamic(TCCState *s1, struct dyn_inf *dyninf)
2035 Section *dynamic;
2037 dynamic = dyninf->dynamic;
2039 /* put dynamic section entries */
2040 dynamic->data_offset = dyninf->dyn_rel_off;
2041 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2042 put_dt(dynamic, DT_STRTAB, dyninf->dynstr->sh_addr);
2043 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2044 put_dt(dynamic, DT_STRSZ, dyninf->dynstr->data_offset);
2045 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2046 #ifdef TCC_TARGET_X86_64
2047 put_dt(dynamic, DT_RELA, dyninf->rel_addr);
2048 put_dt(dynamic, DT_RELASZ, dyninf->rel_size);
2049 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2050 #else
2051 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2052 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2053 put_dt(dynamic, DT_PLTRELSZ, dyninf->rel_size);
2054 put_dt(dynamic, DT_JMPREL, dyninf->rel_addr);
2055 put_dt(dynamic, DT_PLTREL, DT_REL);
2056 put_dt(dynamic, DT_REL, dyninf->bss_addr);
2057 put_dt(dynamic, DT_RELSZ, dyninf->bss_size);
2058 #else
2059 put_dt(dynamic, DT_REL, dyninf->rel_addr);
2060 put_dt(dynamic, DT_RELSZ, dyninf->rel_size);
2061 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2062 #endif
2063 #endif
2064 if (s1->do_debug)
2065 put_dt(dynamic, DT_DEBUG, 0);
2066 put_dt(dynamic, DT_NULL, 0);
2069 /* Relocate remaining sections and symbols (that is those not related to
2070 dynamic linking) */
2071 static int final_sections_reloc(TCCState *s1)
2073 int i;
2074 Section *s;
2076 relocate_syms(s1, 0);
2078 if (s1->nb_errors != 0)
2079 return -1;
2081 /* relocate sections */
2082 /* XXX: ignore sections with allocated relocations ? */
2083 for(i = 1; i < s1->nb_sections; i++) {
2084 s = s1->sections[i];
2085 if (s->reloc && s != s1->got)
2086 relocate_section(s1, s);
2089 /* relocate relocation entries if the relocation tables are
2090 allocated in the executable */
2091 for(i = 1; i < s1->nb_sections; i++) {
2092 s = s1->sections[i];
2093 if ((s->sh_flags & SHF_ALLOC) &&
2094 s->sh_type == SHT_RELX) {
2095 relocate_rel(s1, s);
2098 return 0;
2101 /* Create an ELF file on disk.
2102 This function handle ELF specific layout requirements */
2103 static void tcc_output_elf(TCCState *s1, FILE *f, int phnum, ElfW(Phdr) *phdr,
2104 int file_offset, int *sec_order)
2106 int i, shnum, offset, size, file_type;
2107 Section *s;
2108 ElfW(Ehdr) ehdr;
2109 ElfW(Shdr) shdr, *sh;
2111 file_type = s1->output_type;
2112 shnum = s1->nb_sections;
2114 memset(&ehdr, 0, sizeof(ehdr));
2116 if (phnum > 0) {
2117 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2118 ehdr.e_phnum = phnum;
2119 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2122 /* align to 4 */
2123 file_offset = (file_offset + 3) & -4;
2125 /* fill header */
2126 ehdr.e_ident[0] = ELFMAG0;
2127 ehdr.e_ident[1] = ELFMAG1;
2128 ehdr.e_ident[2] = ELFMAG2;
2129 ehdr.e_ident[3] = ELFMAG3;
2130 ehdr.e_ident[4] = ELFCLASSW;
2131 ehdr.e_ident[5] = ELFDATA2LSB;
2132 ehdr.e_ident[6] = EV_CURRENT;
2133 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2134 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2135 #endif
2136 #ifdef TCC_TARGET_ARM
2137 #ifdef TCC_ARM_EABI
2138 ehdr.e_ident[EI_OSABI] = 0;
2139 ehdr.e_flags = EF_ARM_EABI_VER4;
2140 if (file_type == TCC_OUTPUT_EXE || file_type == TCC_OUTPUT_DLL)
2141 ehdr.e_flags |= EF_ARM_HASENTRY;
2142 if (s1->float_abi == ARM_HARD_FLOAT)
2143 ehdr.e_flags |= EF_ARM_VFP_FLOAT;
2144 else
2145 ehdr.e_flags |= EF_ARM_SOFT_FLOAT;
2146 #else
2147 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2148 #endif
2149 #endif
2150 switch(file_type) {
2151 default:
2152 case TCC_OUTPUT_EXE:
2153 ehdr.e_type = ET_EXEC;
2154 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2155 break;
2156 case TCC_OUTPUT_DLL:
2157 ehdr.e_type = ET_DYN;
2158 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2159 break;
2160 case TCC_OUTPUT_OBJ:
2161 ehdr.e_type = ET_REL;
2162 break;
2164 ehdr.e_machine = EM_TCC_TARGET;
2165 ehdr.e_version = EV_CURRENT;
2166 ehdr.e_shoff = file_offset;
2167 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2168 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2169 ehdr.e_shnum = shnum;
2170 ehdr.e_shstrndx = shnum - 1;
2172 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2173 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2174 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2176 sort_syms(s1, symtab_section);
2177 for(i = 1; i < s1->nb_sections; i++) {
2178 s = s1->sections[sec_order[i]];
2179 if (s->sh_type != SHT_NOBITS) {
2180 if (s->sh_type == SHT_DYNSYM)
2181 patch_dynsym_undef(s1, s);
2182 while (offset < s->sh_offset) {
2183 fputc(0, f);
2184 offset++;
2186 size = s->sh_size;
2187 fwrite(s->data, 1, size, f);
2188 offset += size;
2192 /* output section headers */
2193 while (offset < ehdr.e_shoff) {
2194 fputc(0, f);
2195 offset++;
2198 for(i = 0; i < s1->nb_sections; i++) {
2199 sh = &shdr;
2200 memset(sh, 0, sizeof(ElfW(Shdr)));
2201 s = s1->sections[i];
2202 if (s) {
2203 sh->sh_name = s->sh_name;
2204 sh->sh_type = s->sh_type;
2205 sh->sh_flags = s->sh_flags;
2206 sh->sh_entsize = s->sh_entsize;
2207 sh->sh_info = s->sh_info;
2208 if (s->link)
2209 sh->sh_link = s->link->sh_num;
2210 sh->sh_addralign = s->sh_addralign;
2211 sh->sh_addr = s->sh_addr;
2212 sh->sh_offset = s->sh_offset;
2213 sh->sh_size = s->sh_size;
2215 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2219 /* Write an elf, coff or "binary" file */
2220 static int tcc_write_elf_file(TCCState *s1, const char *filename, int phnum,
2221 ElfW(Phdr) *phdr, int file_offset, int *sec_order)
2223 int fd, mode, file_type;
2224 FILE *f;
2226 file_type = s1->output_type;
2227 if (file_type == TCC_OUTPUT_OBJ)
2228 mode = 0666;
2229 else
2230 mode = 0777;
2231 unlink(filename);
2232 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2233 if (fd < 0) {
2234 tcc_error_noabort("could not write '%s'", filename);
2235 return -1;
2237 f = fdopen(fd, "wb");
2238 if (s1->verbose)
2239 printf("<- %s\n", filename);
2241 #ifdef TCC_TARGET_COFF
2242 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF)
2243 tcc_output_coff(s1, f);
2244 else
2245 #endif
2246 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2247 tcc_output_elf(s1, f, phnum, phdr, file_offset, sec_order);
2248 else
2249 tcc_output_binary(s1, f, sec_order);
2250 fclose(f);
2252 return 0;
2255 /* Output an elf, coff or binary file */
2256 /* XXX: suppress unneeded sections */
2257 static int elf_output_file(TCCState *s1, const char *filename)
2259 int i, ret, phnum, shnum, file_type, file_offset, *sec_order;
2260 struct dyn_inf dyninf;
2261 ElfW(Phdr) *phdr;
2262 ElfW(Sym) *sym;
2263 Section *strsec, *interp, *dynamic, *dynstr;
2265 file_type = s1->output_type;
2266 s1->nb_errors = 0;
2268 /* if linking, also link in runtime libraries (libc, libgcc, etc.) */
2269 if (file_type != TCC_OUTPUT_OBJ) {
2270 tcc_add_runtime(s1);
2273 phdr = NULL;
2274 sec_order = NULL;
2275 interp = dynamic = dynstr = NULL; /* avoid warning */
2276 dyninf.dyn_rel_off = 0; /* avoid warning */
2278 if (file_type != TCC_OUTPUT_OBJ) {
2279 relocate_common_syms();
2281 tcc_add_linker_symbols(s1);
2283 if (!s1->static_link) {
2284 if (file_type == TCC_OUTPUT_EXE) {
2285 char *ptr;
2286 /* allow override the dynamic loader */
2287 const char *elfint = getenv("LD_SO");
2288 if (elfint == NULL)
2289 elfint = DEFAULT_ELFINTERP(s1);
2290 /* add interpreter section only if executable */
2291 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
2292 interp->sh_addralign = 1;
2293 ptr = section_ptr_add(interp, 1 + strlen(elfint));
2294 strcpy(ptr, elfint);
2297 /* add dynamic symbol table */
2298 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
2299 ".dynstr",
2300 ".hash", SHF_ALLOC);
2301 dynstr = s1->dynsym->link;
2303 /* add dynamic section */
2304 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
2305 SHF_ALLOC | SHF_WRITE);
2306 dynamic->link = dynstr;
2307 dynamic->sh_entsize = sizeof(ElfW(Dyn));
2309 /* add PLT */
2310 s1->plt = new_section(s1, ".plt", SHT_PROGBITS,
2311 SHF_ALLOC | SHF_EXECINSTR);
2312 s1->plt->sh_entsize = 4;
2314 build_got(s1);
2316 if (file_type == TCC_OUTPUT_EXE) {
2317 bind_exe_dynsyms(s1);
2319 if (s1->nb_errors) {
2320 ret = -1;
2321 goto the_end;
2324 bind_libs_dynsyms(s1);
2325 } else /* shared library case: simply export all global symbols */
2326 export_global_syms(s1);
2328 build_got_entries(s1);
2330 /* add a list of needed dlls */
2331 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2332 DLLReference *dllref = s1->loaded_dlls[i];
2333 if (dllref->level == 0)
2334 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
2337 if (s1->rpath)
2338 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
2340 /* XXX: currently, since we do not handle PIC code, we
2341 must relocate the readonly segments */
2342 if (file_type == TCC_OUTPUT_DLL) {
2343 if (s1->soname)
2344 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
2345 put_dt(dynamic, DT_TEXTREL, 0);
2348 if (s1->symbolic)
2349 put_dt(dynamic, DT_SYMBOLIC, 0);
2351 /* add necessary space for other entries */
2352 dyninf.dyn_rel_off = dynamic->data_offset;
2353 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
2354 } else {
2355 /* still need to build got entries in case of static link */
2356 build_got_entries(s1);
2360 /* we add a section for symbols */
2361 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
2362 put_elf_str(strsec, "");
2364 /* compute number of sections */
2365 shnum = s1->nb_sections;
2367 /* this array is used to reorder sections in the output file */
2368 sec_order = tcc_malloc(sizeof(int) * shnum);
2369 sec_order[0] = 0;
2371 /* compute number of program headers */
2372 switch(file_type) {
2373 default:
2374 case TCC_OUTPUT_OBJ:
2375 phnum = 0;
2376 break;
2377 case TCC_OUTPUT_EXE:
2378 if (!s1->static_link)
2379 phnum = 4 + HAVE_PHDR;
2380 else
2381 phnum = 2;
2382 break;
2383 case TCC_OUTPUT_DLL:
2384 phnum = 3;
2385 break;
2388 /* Allocate strings for section names */
2389 alloc_sec_names(s1, file_type, strsec);
2391 /* allocate program segment headers */
2392 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
2394 /* compute section to program header mapping */
2395 file_offset = layout_sections(s1, phdr, phnum, interp, &dyninf, sec_order);
2397 /* Fill remaining program header and finalize relocation related to dynamic
2398 linking. */
2399 if (phnum > 0) {
2400 fill_unloadable_phdr(phdr, phnum, interp, dynamic);
2401 if (dynamic) {
2402 dyninf.dynamic = dynamic;
2403 dyninf.dynstr = dynstr;
2405 fill_dynamic(s1, &dyninf);
2407 /* put in GOT the dynamic section address and relocate PLT */
2408 put32(s1->got->data, dynamic->sh_addr);
2409 if (file_type == TCC_OUTPUT_EXE
2410 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2411 || file_type == TCC_OUTPUT_DLL
2412 #endif
2414 relocate_plt(s1);
2416 /* relocate symbols in .dynsym now that final addresses are known */
2417 for_each_elem(s1->dynsym, 1, sym, ElfW(Sym)) {
2418 /* relocate to PLT if symbol corresponds to a PLT entry */
2419 if (sym->st_shndx == SHN_UNDEF) {
2420 if (sym->st_value)
2421 sym->st_value += s1->plt->sh_addr;
2422 } else if (sym->st_shndx < SHN_LORESERVE) {
2423 /* do symbol relocation */
2424 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2430 /* if building executable or DLL, then relocate each section
2431 except the GOT which is already relocated */
2432 if (file_type != TCC_OUTPUT_OBJ) {
2433 ret = final_sections_reloc(s1);
2434 if (ret)
2435 goto the_end;
2438 /* Perform relocation to GOT or PLT entries */
2439 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2440 fill_got(s1);
2442 /* Create the ELF file with name 'filename' */
2443 ret = tcc_write_elf_file(s1, filename, phnum, phdr, file_offset, sec_order);
2444 the_end:
2445 tcc_free(s1->symtab_to_dynsym);
2446 tcc_free(sec_order);
2447 tcc_free(phdr);
2448 tcc_free(s1->sym_attrs);
2449 s1->sym_attrs = NULL;
2450 return ret;
2453 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2455 int ret;
2456 #ifdef TCC_TARGET_PE
2457 if (s->output_type != TCC_OUTPUT_OBJ) {
2458 ret = pe_output_file(s, filename);
2459 } else
2460 #endif
2461 ret = elf_output_file(s, filename);
2462 return ret;
2465 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2467 void *data;
2469 data = tcc_malloc(size);
2470 lseek(fd, file_offset, SEEK_SET);
2471 read(fd, data, size);
2472 return data;
2475 typedef struct SectionMergeInfo {
2476 Section *s; /* corresponding existing section */
2477 unsigned long offset; /* offset of the new section in the existing section */
2478 uint8_t new_section; /* true if section 's' was added */
2479 uint8_t link_once; /* true if link once section */
2480 } SectionMergeInfo;
2482 /* load an object file and merge it with current files */
2483 /* XXX: handle correctly stab (debug) info */
2484 ST_FUNC int tcc_load_object_file(TCCState *s1,
2485 int fd, unsigned long file_offset)
2487 ElfW(Ehdr) ehdr;
2488 ElfW(Shdr) *shdr, *sh;
2489 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2490 unsigned char *strsec, *strtab;
2491 int *old_to_new_syms;
2492 char *sh_name, *name;
2493 SectionMergeInfo *sm_table, *sm;
2494 ElfW(Sym) *sym, *symtab;
2495 ElfW_Rel *rel;
2496 Section *s;
2498 int stab_index;
2499 int stabstr_index;
2501 stab_index = stabstr_index = 0;
2503 if (read(fd, &ehdr, sizeof(ehdr)) != sizeof(ehdr))
2504 goto fail1;
2505 if (ehdr.e_ident[0] != ELFMAG0 ||
2506 ehdr.e_ident[1] != ELFMAG1 ||
2507 ehdr.e_ident[2] != ELFMAG2 ||
2508 ehdr.e_ident[3] != ELFMAG3)
2509 goto fail1;
2510 /* test if object file */
2511 if (ehdr.e_type != ET_REL)
2512 goto fail1;
2513 /* test CPU specific stuff */
2514 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2515 ehdr.e_machine != EM_TCC_TARGET) {
2516 fail1:
2517 tcc_error_noabort("invalid object file");
2518 return -1;
2520 /* read sections */
2521 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2522 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2523 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2525 /* load section names */
2526 sh = &shdr[ehdr.e_shstrndx];
2527 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2529 /* load symtab and strtab */
2530 old_to_new_syms = NULL;
2531 symtab = NULL;
2532 strtab = NULL;
2533 nb_syms = 0;
2534 for(i = 1; i < ehdr.e_shnum; i++) {
2535 sh = &shdr[i];
2536 if (sh->sh_type == SHT_SYMTAB) {
2537 if (symtab) {
2538 tcc_error_noabort("object must contain only one symtab");
2539 fail:
2540 ret = -1;
2541 goto the_end;
2543 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2544 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2545 sm_table[i].s = symtab_section;
2547 /* now load strtab */
2548 sh = &shdr[sh->sh_link];
2549 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2553 /* now examine each section and try to merge its content with the
2554 ones in memory */
2555 for(i = 1; i < ehdr.e_shnum; i++) {
2556 /* no need to examine section name strtab */
2557 if (i == ehdr.e_shstrndx)
2558 continue;
2559 sh = &shdr[i];
2560 sh_name = (char *) strsec + sh->sh_name;
2561 /* ignore sections types we do not handle */
2562 if (sh->sh_type != SHT_PROGBITS &&
2563 sh->sh_type != SHT_RELX &&
2564 #ifdef TCC_ARM_EABI
2565 sh->sh_type != SHT_ARM_EXIDX &&
2566 #endif
2567 sh->sh_type != SHT_NOBITS &&
2568 sh->sh_type != SHT_PREINIT_ARRAY &&
2569 sh->sh_type != SHT_INIT_ARRAY &&
2570 sh->sh_type != SHT_FINI_ARRAY &&
2571 strcmp(sh_name, ".stabstr")
2573 continue;
2574 if (sh->sh_addralign < 1)
2575 sh->sh_addralign = 1;
2576 /* find corresponding section, if any */
2577 for(j = 1; j < s1->nb_sections;j++) {
2578 s = s1->sections[j];
2579 if (!strcmp(s->name, sh_name)) {
2580 if (!strncmp(sh_name, ".gnu.linkonce",
2581 sizeof(".gnu.linkonce") - 1)) {
2582 /* if a 'linkonce' section is already present, we
2583 do not add it again. It is a little tricky as
2584 symbols can still be defined in
2585 it. */
2586 sm_table[i].link_once = 1;
2587 goto next;
2588 } else {
2589 goto found;
2593 /* not found: create new section */
2594 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2595 /* take as much info as possible from the section. sh_link and
2596 sh_info will be updated later */
2597 s->sh_addralign = sh->sh_addralign;
2598 s->sh_entsize = sh->sh_entsize;
2599 sm_table[i].new_section = 1;
2600 found:
2601 if (sh->sh_type != s->sh_type) {
2602 tcc_error_noabort("invalid section type");
2603 goto fail;
2606 /* align start of section */
2607 offset = s->data_offset;
2609 if (0 == strcmp(sh_name, ".stab")) {
2610 stab_index = i;
2611 goto no_align;
2613 if (0 == strcmp(sh_name, ".stabstr")) {
2614 stabstr_index = i;
2615 goto no_align;
2618 size = sh->sh_addralign - 1;
2619 offset = (offset + size) & ~size;
2620 if (sh->sh_addralign > s->sh_addralign)
2621 s->sh_addralign = sh->sh_addralign;
2622 s->data_offset = offset;
2623 no_align:
2624 sm_table[i].offset = offset;
2625 sm_table[i].s = s;
2626 /* concatenate sections */
2627 size = sh->sh_size;
2628 if (sh->sh_type != SHT_NOBITS) {
2629 unsigned char *ptr;
2630 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2631 ptr = section_ptr_add(s, size);
2632 read(fd, ptr, size);
2633 } else {
2634 s->data_offset += size;
2636 next: ;
2639 /* gr relocate stab strings */
2640 if (stab_index && stabstr_index) {
2641 Stab_Sym *a, *b;
2642 unsigned o;
2643 s = sm_table[stab_index].s;
2644 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2645 b = (Stab_Sym *)(s->data + s->data_offset);
2646 o = sm_table[stabstr_index].offset;
2647 while (a < b)
2648 a->n_strx += o, a++;
2651 /* second short pass to update sh_link and sh_info fields of new
2652 sections */
2653 for(i = 1; i < ehdr.e_shnum; i++) {
2654 s = sm_table[i].s;
2655 if (!s || !sm_table[i].new_section)
2656 continue;
2657 sh = &shdr[i];
2658 if (sh->sh_link > 0)
2659 s->link = sm_table[sh->sh_link].s;
2660 if (sh->sh_type == SHT_RELX) {
2661 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2662 /* update backward link */
2663 s1->sections[s->sh_info]->reloc = s;
2666 sm = sm_table;
2668 /* resolve symbols */
2669 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2671 sym = symtab + 1;
2672 for(i = 1; i < nb_syms; i++, sym++) {
2673 if (sym->st_shndx != SHN_UNDEF &&
2674 sym->st_shndx < SHN_LORESERVE) {
2675 sm = &sm_table[sym->st_shndx];
2676 if (sm->link_once) {
2677 /* if a symbol is in a link once section, we use the
2678 already defined symbol. It is very important to get
2679 correct relocations */
2680 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2681 name = (char *) strtab + sym->st_name;
2682 sym_index = find_elf_sym(symtab_section, name);
2683 if (sym_index)
2684 old_to_new_syms[i] = sym_index;
2686 continue;
2688 /* if no corresponding section added, no need to add symbol */
2689 if (!sm->s)
2690 continue;
2691 /* convert section number */
2692 sym->st_shndx = sm->s->sh_num;
2693 /* offset value */
2694 sym->st_value += sm->offset;
2696 /* add symbol */
2697 name = (char *) strtab + sym->st_name;
2698 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2699 sym->st_info, sym->st_other,
2700 sym->st_shndx, name);
2701 old_to_new_syms[i] = sym_index;
2704 /* third pass to patch relocation entries */
2705 for(i = 1; i < ehdr.e_shnum; i++) {
2706 s = sm_table[i].s;
2707 if (!s)
2708 continue;
2709 sh = &shdr[i];
2710 offset = sm_table[i].offset;
2711 switch(s->sh_type) {
2712 case SHT_RELX:
2713 /* take relocation offset information */
2714 offseti = sm_table[sh->sh_info].offset;
2715 for_each_elem(s, (offset / sizeof(*rel)), rel, ElfW_Rel) {
2716 int type;
2717 unsigned sym_index;
2718 /* convert symbol index */
2719 type = ELFW(R_TYPE)(rel->r_info);
2720 sym_index = ELFW(R_SYM)(rel->r_info);
2721 /* NOTE: only one symtab assumed */
2722 if (sym_index >= nb_syms)
2723 goto invalid_reloc;
2724 sym_index = old_to_new_syms[sym_index];
2725 /* ignore link_once in rel section. */
2726 if (!sym_index && !sm->link_once
2727 #ifdef TCC_TARGET_ARM
2728 && type != R_ARM_V4BX
2729 #endif
2731 invalid_reloc:
2732 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2733 i, strsec + sh->sh_name, rel->r_offset);
2734 goto fail;
2736 rel->r_info = ELFW(R_INFO)(sym_index, type);
2737 /* offset the relocation offset */
2738 rel->r_offset += offseti;
2739 #ifdef TCC_TARGET_ARM
2740 /* Jumps and branches from a Thumb code to a PLT entry need
2741 special handling since PLT entries are ARM code.
2742 Unconditional bl instructions referencing PLT entries are
2743 handled by converting these instructions into blx
2744 instructions. Other case of instructions referencing a PLT
2745 entry require to add a Thumb stub before the PLT entry to
2746 switch to ARM mode. We set bit plt_thumb_stub of the
2747 attribute of a symbol to indicate such a case. */
2748 if (type == R_ARM_THM_JUMP24)
2749 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
2750 #endif
2752 break;
2753 default:
2754 break;
2758 ret = 0;
2759 the_end:
2760 tcc_free(symtab);
2761 tcc_free(strtab);
2762 tcc_free(old_to_new_syms);
2763 tcc_free(sm_table);
2764 tcc_free(strsec);
2765 tcc_free(shdr);
2766 return ret;
2769 typedef struct ArchiveHeader {
2770 char ar_name[16]; /* name of this member */
2771 char ar_date[12]; /* file mtime */
2772 char ar_uid[6]; /* owner uid; printed as decimal */
2773 char ar_gid[6]; /* owner gid; printed as decimal */
2774 char ar_mode[8]; /* file mode, printed as octal */
2775 char ar_size[10]; /* file size, printed as decimal */
2776 char ar_fmag[2]; /* should contain ARFMAG */
2777 } ArchiveHeader;
2779 static int get_be32(const uint8_t *b)
2781 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
2784 /* load only the objects which resolve undefined symbols */
2785 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
2787 int i, bound, nsyms, sym_index, off, ret;
2788 uint8_t *data;
2789 const char *ar_names, *p;
2790 const uint8_t *ar_index;
2791 ElfW(Sym) *sym;
2793 data = tcc_malloc(size);
2794 if (read(fd, data, size) != size)
2795 goto fail;
2796 nsyms = get_be32(data);
2797 ar_index = data + 4;
2798 ar_names = (char *) ar_index + nsyms * 4;
2800 do {
2801 bound = 0;
2802 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
2803 sym_index = find_elf_sym(symtab_section, p);
2804 if(sym_index) {
2805 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
2806 if(sym->st_shndx == SHN_UNDEF) {
2807 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
2808 ++bound;
2809 lseek(fd, off, SEEK_SET);
2810 if(tcc_load_object_file(s1, fd, off) < 0) {
2811 fail:
2812 ret = -1;
2813 goto the_end;
2818 } while(bound);
2819 ret = 0;
2820 the_end:
2821 tcc_free(data);
2822 return ret;
2825 /* load a '.a' file */
2826 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
2828 ArchiveHeader hdr;
2829 char ar_size[11];
2830 char ar_name[17];
2831 char magic[8];
2832 int size, len, i;
2833 unsigned long file_offset;
2835 /* skip magic which was already checked */
2836 read(fd, magic, sizeof(magic));
2838 for(;;) {
2839 len = read(fd, &hdr, sizeof(hdr));
2840 if (len == 0)
2841 break;
2842 if (len != sizeof(hdr)) {
2843 tcc_error_noabort("invalid archive");
2844 return -1;
2846 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
2847 ar_size[sizeof(hdr.ar_size)] = '\0';
2848 size = strtol(ar_size, NULL, 0);
2849 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
2850 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
2851 if (ar_name[i] != ' ')
2852 break;
2854 ar_name[i + 1] = '\0';
2855 file_offset = lseek(fd, 0, SEEK_CUR);
2856 /* align to even */
2857 size = (size + 1) & ~1;
2858 if (!strcmp(ar_name, "/")) {
2859 /* coff symbol table : we handle it */
2860 if(s1->alacarte_link)
2861 return tcc_load_alacarte(s1, fd, size);
2862 } else if (!strcmp(ar_name, "//") ||
2863 !strcmp(ar_name, "__.SYMDEF") ||
2864 !strcmp(ar_name, "__.SYMDEF/") ||
2865 !strcmp(ar_name, "ARFILENAMES/")) {
2866 /* skip symbol table or archive names */
2867 } else {
2868 if (tcc_load_object_file(s1, fd, file_offset) < 0)
2869 return -1;
2871 lseek(fd, file_offset + size, SEEK_SET);
2873 return 0;
2876 #ifndef TCC_TARGET_PE
2877 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
2878 is referenced by the user (so it should be added as DT_NEEDED in
2879 the generated ELF file) */
2880 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
2882 ElfW(Ehdr) ehdr;
2883 ElfW(Shdr) *shdr, *sh, *sh1;
2884 int i, j, nb_syms, nb_dts, sym_bind, ret;
2885 ElfW(Sym) *sym, *dynsym;
2886 ElfW(Dyn) *dt, *dynamic;
2887 unsigned char *dynstr;
2888 const char *name, *soname;
2889 DLLReference *dllref;
2891 read(fd, &ehdr, sizeof(ehdr));
2893 /* test CPU specific stuff */
2894 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2895 ehdr.e_machine != EM_TCC_TARGET) {
2896 tcc_error_noabort("bad architecture");
2897 return -1;
2900 /* read sections */
2901 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2903 /* load dynamic section and dynamic symbols */
2904 nb_syms = 0;
2905 nb_dts = 0;
2906 dynamic = NULL;
2907 dynsym = NULL; /* avoid warning */
2908 dynstr = NULL; /* avoid warning */
2909 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
2910 switch(sh->sh_type) {
2911 case SHT_DYNAMIC:
2912 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
2913 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
2914 break;
2915 case SHT_DYNSYM:
2916 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2917 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
2918 sh1 = &shdr[sh->sh_link];
2919 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
2920 break;
2921 default:
2922 break;
2926 /* compute the real library name */
2927 soname = tcc_basename(filename);
2929 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2930 if (dt->d_tag == DT_SONAME) {
2931 soname = (char *) dynstr + dt->d_un.d_val;
2935 /* if the dll is already loaded, do not load it */
2936 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2937 dllref = s1->loaded_dlls[i];
2938 if (!strcmp(soname, dllref->name)) {
2939 /* but update level if needed */
2940 if (level < dllref->level)
2941 dllref->level = level;
2942 ret = 0;
2943 goto the_end;
2947 /* add the dll and its level */
2948 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
2949 dllref->level = level;
2950 strcpy(dllref->name, soname);
2951 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
2953 /* add dynamic symbols in dynsym_section */
2954 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
2955 sym_bind = ELFW(ST_BIND)(sym->st_info);
2956 if (sym_bind == STB_LOCAL)
2957 continue;
2958 name = (char *) dynstr + sym->st_name;
2959 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
2960 sym->st_info, sym->st_other, sym->st_shndx, name);
2963 /* load all referenced DLLs */
2964 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
2965 switch(dt->d_tag) {
2966 case DT_NEEDED:
2967 name = (char *) dynstr + dt->d_un.d_val;
2968 for(j = 0; j < s1->nb_loaded_dlls; j++) {
2969 dllref = s1->loaded_dlls[j];
2970 if (!strcmp(name, dllref->name))
2971 goto already_loaded;
2973 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
2974 tcc_error_noabort("referenced dll '%s' not found", name);
2975 ret = -1;
2976 goto the_end;
2978 already_loaded:
2979 break;
2982 ret = 0;
2983 the_end:
2984 tcc_free(dynstr);
2985 tcc_free(dynsym);
2986 tcc_free(dynamic);
2987 tcc_free(shdr);
2988 return ret;
2991 #define LD_TOK_NAME 256
2992 #define LD_TOK_EOF (-1)
2994 /* return next ld script token */
2995 static int ld_next(TCCState *s1, char *name, int name_size)
2997 int c;
2998 char *q;
3000 redo:
3001 switch(ch) {
3002 case ' ':
3003 case '\t':
3004 case '\f':
3005 case '\v':
3006 case '\r':
3007 case '\n':
3008 inp();
3009 goto redo;
3010 case '/':
3011 minp();
3012 if (ch == '*') {
3013 file->buf_ptr = parse_comment(file->buf_ptr);
3014 ch = file->buf_ptr[0];
3015 goto redo;
3016 } else {
3017 q = name;
3018 *q++ = '/';
3019 goto parse_name;
3021 break;
3022 /* case 'a' ... 'z': */
3023 case 'a':
3024 case 'b':
3025 case 'c':
3026 case 'd':
3027 case 'e':
3028 case 'f':
3029 case 'g':
3030 case 'h':
3031 case 'i':
3032 case 'j':
3033 case 'k':
3034 case 'l':
3035 case 'm':
3036 case 'n':
3037 case 'o':
3038 case 'p':
3039 case 'q':
3040 case 'r':
3041 case 's':
3042 case 't':
3043 case 'u':
3044 case 'v':
3045 case 'w':
3046 case 'x':
3047 case 'y':
3048 case 'z':
3049 /* case 'A' ... 'z': */
3050 case 'A':
3051 case 'B':
3052 case 'C':
3053 case 'D':
3054 case 'E':
3055 case 'F':
3056 case 'G':
3057 case 'H':
3058 case 'I':
3059 case 'J':
3060 case 'K':
3061 case 'L':
3062 case 'M':
3063 case 'N':
3064 case 'O':
3065 case 'P':
3066 case 'Q':
3067 case 'R':
3068 case 'S':
3069 case 'T':
3070 case 'U':
3071 case 'V':
3072 case 'W':
3073 case 'X':
3074 case 'Y':
3075 case 'Z':
3076 case '_':
3077 case '\\':
3078 case '.':
3079 case '$':
3080 case '~':
3081 q = name;
3082 parse_name:
3083 for(;;) {
3084 if (!((ch >= 'a' && ch <= 'z') ||
3085 (ch >= 'A' && ch <= 'Z') ||
3086 (ch >= '0' && ch <= '9') ||
3087 strchr("/.-_+=$:\\,~", ch)))
3088 break;
3089 if ((q - name) < name_size - 1) {
3090 *q++ = ch;
3092 minp();
3094 *q = '\0';
3095 c = LD_TOK_NAME;
3096 break;
3097 case CH_EOF:
3098 c = LD_TOK_EOF;
3099 break;
3100 default:
3101 c = ch;
3102 inp();
3103 break;
3105 return c;
3108 static int ld_add_file(TCCState *s1, const char filename[])
3110 int ret;
3112 ret = tcc_add_file_internal(s1, filename, 0);
3113 if (ret)
3114 ret = tcc_add_dll(s1, filename, 0);
3115 return ret;
3118 static inline int new_undef_syms(void)
3120 int ret = 0;
3121 ret = new_undef_sym;
3122 new_undef_sym = 0;
3123 return ret;
3126 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3128 char filename[1024], libname[1024];
3129 int t, group, nblibs = 0, ret = 0;
3130 char **libs = NULL;
3132 group = !strcmp(cmd, "GROUP");
3133 if (!as_needed)
3134 new_undef_syms();
3135 t = ld_next(s1, filename, sizeof(filename));
3136 if (t != '(')
3137 expect("(");
3138 t = ld_next(s1, filename, sizeof(filename));
3139 for(;;) {
3140 libname[0] = '\0';
3141 if (t == LD_TOK_EOF) {
3142 tcc_error_noabort("unexpected end of file");
3143 ret = -1;
3144 goto lib_parse_error;
3145 } else if (t == ')') {
3146 break;
3147 } else if (t == '-') {
3148 t = ld_next(s1, filename, sizeof(filename));
3149 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3150 tcc_error_noabort("library name expected");
3151 ret = -1;
3152 goto lib_parse_error;
3154 pstrcpy(libname, sizeof libname, &filename[1]);
3155 if (s1->static_link) {
3156 snprintf(filename, sizeof filename, "lib%s.a", libname);
3157 } else {
3158 snprintf(filename, sizeof filename, "lib%s.so", libname);
3160 } else if (t != LD_TOK_NAME) {
3161 tcc_error_noabort("filename expected");
3162 ret = -1;
3163 goto lib_parse_error;
3165 if (!strcmp(filename, "AS_NEEDED")) {
3166 ret = ld_add_file_list(s1, cmd, 1);
3167 if (ret)
3168 goto lib_parse_error;
3169 } else {
3170 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3171 if (!as_needed) {
3172 ret = ld_add_file(s1, filename);
3173 if (ret)
3174 goto lib_parse_error;
3175 if (group) {
3176 /* Add the filename *and* the libname to avoid future conversions */
3177 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3178 if (libname[0] != '\0')
3179 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3183 t = ld_next(s1, filename, sizeof(filename));
3184 if (t == ',') {
3185 t = ld_next(s1, filename, sizeof(filename));
3188 if (group && !as_needed) {
3189 while (new_undef_syms()) {
3190 int i;
3192 for (i = 0; i < nblibs; i ++)
3193 ld_add_file(s1, libs[i]);
3196 lib_parse_error:
3197 dynarray_reset(&libs, &nblibs);
3198 return ret;
3201 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3202 files */
3203 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3205 char cmd[64];
3206 char filename[1024];
3207 int t, ret;
3209 ch = file->buf_ptr[0];
3210 ch = handle_eob();
3211 for(;;) {
3212 t = ld_next(s1, cmd, sizeof(cmd));
3213 if (t == LD_TOK_EOF)
3214 return 0;
3215 else if (t != LD_TOK_NAME)
3216 return -1;
3217 if (!strcmp(cmd, "INPUT") ||
3218 !strcmp(cmd, "GROUP")) {
3219 ret = ld_add_file_list(s1, cmd, 0);
3220 if (ret)
3221 return ret;
3222 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3223 !strcmp(cmd, "TARGET")) {
3224 /* ignore some commands */
3225 t = ld_next(s1, cmd, sizeof(cmd));
3226 if (t != '(')
3227 expect("(");
3228 for(;;) {
3229 t = ld_next(s1, filename, sizeof(filename));
3230 if (t == LD_TOK_EOF) {
3231 tcc_error_noabort("unexpected end of file");
3232 return -1;
3233 } else if (t == ')') {
3234 break;
3237 } else {
3238 return -1;
3241 return 0;
3243 #endif /* !TCC_TARGET_PE */