tccasm.c

   1 /*
   2  *  GAS like assembler for TCC
   3  *
   4  *  Copyright (c) 2001-2004 Fabrice Bellard
   5  *
   6  * This library is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2 of the License, or (at your option) any later version.
  10  *
  11  * This library is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with this library; if not, write to the Free Software
  18  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19  */
  20
  21 #include "tcc.h"
  22 #ifdef CONFIG_TCC_ASM
  23
  24 ST_FUNC int asm_get_local_label_name(TCCState *s1, unsigned int n)
  25 {
  26     char buf[64];
  27     TokenSym *ts;
  28
  29     snprintf(buf, sizeof(buf), "L..%u", n);
  30     ts = tok_alloc(buf, strlen(buf));
  31     return ts->tok;
  32 }
  33
  34 ST_FUNC void asm_expr(TCCState *s1, ExprValue *pe);
  35 static int tcc_assemble_internal(TCCState *s1, int do_preprocess);
  36 static Sym sym_dot;
  37
  38 /* Return a symbol we can use inside the assembler, having name NAME.
  39    The assembler symbol table is different from the C symbol table
  40    (and the Sym members are used differently).  But we must be able
  41    to look up file-global C symbols from inside the assembler, e.g.
  42    for global asm blocks to be able to refer to defined C symbols.
  43
  44    This routine gives back either an existing asm-internal
  45    symbol, or a new one.  In the latter case the new asm-internal
  46    symbol is initialized with info from the C symbol table.
  47
  48    If CSYM is non-null we take symbol info from it, otherwise
  49    we look up NAME in the C symbol table and use that.  */
  50 ST_FUNC Sym* get_asm_sym(int name, Sym *csym)
  51 {
  52     Sym *sym = label_find(name);
  53     if (!sym) {
  54         sym = label_push(&tcc_state->asm_labels, name, 0);
  55         sym->type.t = VT_VOID | VT_EXTERN;
  56         if (!csym) {
  57             csym = sym_find(name);
  58             /* We might be called for an asm block from inside a C routine
  59                and so might have local decls on the identifier stack.  Search
  60                for the first global one.  */
  61             while (csym && csym->scope)
  62                 csym = csym->prev_tok;
  63         }
  64         /* Now, if we have a defined global symbol copy over
  65            section and offset.  */
  66         if (csym &&
  67             ((csym->r & (VT_SYM|VT_CONST)) == (VT_SYM|VT_CONST)) &&
  68             csym->c) {
  69             ElfW(Sym) *esym;
  70             esym = &((ElfW(Sym) *)symtab_section->data)[csym->c];
  71             sym->c = csym->c;
  72             sym->r = esym->st_shndx;
  73             sym->jnext = esym->st_value;
  74             /* XXX can't yet store st_size anywhere.  */
  75             sym->type.t &= ~VT_EXTERN;
  76             /* Mark that this asm symbol doesn't need to be fed back.  */
  77             sym->type.t |= VT_IMPORT;
  78         }
  79     }
  80     return sym;
  81 }
  82
  83 /* We do not use the C expression parser to handle symbols. Maybe the
  84    C expression parser could be tweaked to do so. */
  85
  86 static void asm_expr_unary(TCCState *s1, ExprValue *pe)
  87 {
  88     Sym *sym;
  89     int op, label;
  90     unsigned long n;
  91     const char *p;
  92
  93     switch(tok) {
  94     case TOK_PPNUM:
  95         p = tokc.str.data;
  96         n = strtoul(p, (char **)&p, 0);
  97         if (*p == 'b' || *p == 'f') {
  98             /* backward or forward label */
  99             label = asm_get_local_label_name(s1, n);
 100             sym = label_find(label);
 101             if (*p == 'b') {
 102                 /* backward : find the last corresponding defined label */
 103                 if (sym && sym->r == 0)
 104                     sym = sym->prev_tok;
 105                 if (!sym)
 106                     tcc_error("local label '%d' not found backward", n);
 107             } else {
 108                 /* forward */
 109                 if (!sym || sym->r) {
 110                     /* if the last label is defined, then define a new one */
 111                     sym = label_push(&s1->asm_labels, label, 0);
 112                     sym->type.t = VT_STATIC | VT_VOID | VT_EXTERN;
 113                 }
 114             }
 115             pe->v = 0;
 116             pe->sym = sym;
 117             pe->pcrel = 0;
 118         } else if (*p == '\0') {
 119             pe->v = n;
 120             pe->sym = NULL;
 121             pe->pcrel = 0;
 122         } else {
 123             tcc_error("invalid number syntax");
 124         }
 125         next();
 126         break;
 127     case '+':
 128         next();
 129         asm_expr_unary(s1, pe);
 130         break;
 131     case '-':
 132     case '~':
 133         op = tok;
 134         next();
 135         asm_expr_unary(s1, pe);
 136         if (pe->sym)
 137             tcc_error("invalid operation with label");
 138         if (op == '-')
 139             pe->v = -pe->v;
 140         else
 141             pe->v = ~pe->v;
 142         break;
 143     case TOK_CCHAR:
 144     case TOK_LCHAR:
 145         pe->v = tokc.i;
 146         pe->sym = NULL;
 147         pe->pcrel = 0;
 148         next();
 149         break;
 150     case '(':
 151         next();
 152         asm_expr(s1, pe);
 153         skip(')');
 154         break;
 155     case '.':
 156         pe->v = 0;
 157         pe->sym = &sym_dot;
 158         pe->pcrel = 0;
 159         sym_dot.type.t = VT_VOID | VT_STATIC;
 160         sym_dot.r = cur_text_section->sh_num;
 161         sym_dot.jnext = ind;
 162         next();
 163         break;
 164     default:
 165         if (tok >= TOK_IDENT) {
 166             /* label case : if the label was not found, add one */
 167             sym = get_asm_sym(tok, NULL);
 168             if (sym->r == SHN_ABS) {
 169                 /* if absolute symbol, no need to put a symbol value */
 170                 pe->v = sym->jnext;
 171                 pe->sym = NULL;
 172                 pe->pcrel = 0;
 173             } else {
 174                 pe->v = 0;
 175                 pe->sym = sym;
 176                 pe->pcrel = 0;
 177             }
 178             next();
 179         } else {
 180             tcc_error("bad expression syntax [%s]", get_tok_str(tok, &tokc));
 181         }
 182         break;
 183     }
 184 }
 185
 186 static void asm_expr_prod(TCCState *s1, ExprValue *pe)
 187 {
 188     int op;
 189     ExprValue e2;
 190
 191     asm_expr_unary(s1, pe);
 192     for(;;) {
 193         op = tok;
 194         if (op != '*' && op != '/' && op != '%' &&
 195             op != TOK_SHL && op != TOK_SAR)
 196             break;
 197         next();
 198         asm_expr_unary(s1, &e2);
 199         if (pe->sym || e2.sym)
 200             tcc_error("invalid operation with label");
 201         switch(op) {
 202         case '*':
 203             pe->v *= e2.v;
 204             break;
 205         case '/':
 206             if (e2.v == 0) {
 207             div_error:
 208                 tcc_error("division by zero");
 209             }
 210             pe->v /= e2.v;
 211             break;
 212         case '%':
 213             if (e2.v == 0)
 214                 goto div_error;
 215             pe->v %= e2.v;
 216             break;
 217         case TOK_SHL:
 218             pe->v <<= e2.v;
 219             break;
 220         default:
 221         case TOK_SAR:
 222             pe->v >>= e2.v;
 223             break;
 224         }
 225     }
 226 }
 227
 228 static void asm_expr_logic(TCCState *s1, ExprValue *pe)
 229 {
 230     int op;
 231     ExprValue e2;
 232
 233     asm_expr_prod(s1, pe);
 234     for(;;) {
 235         op = tok;
 236         if (op != '&' && op != '|' && op != '^')
 237             break;
 238         next();
 239         asm_expr_prod(s1, &e2);
 240         if (pe->sym || e2.sym)
 241             tcc_error("invalid operation with label");
 242         switch(op) {
 243         case '&':
 244             pe->v &= e2.v;
 245             break;
 246         case '|':
 247             pe->v |= e2.v;
 248             break;
 249         default:
 250         case '^':
 251             pe->v ^= e2.v;
 252             break;
 253         }
 254     }
 255 }
 256
 257 static inline void asm_expr_sum(TCCState *s1, ExprValue *pe)
 258 {
 259     int op;
 260     ExprValue e2;
 261
 262     asm_expr_logic(s1, pe);
 263     for(;;) {
 264         op = tok;
 265         if (op != '+' && op != '-')
 266             break;
 267         next();
 268         asm_expr_logic(s1, &e2);
 269         if (op == '+') {
 270             if (pe->sym != NULL && e2.sym != NULL)
 271                 goto cannot_relocate;
 272             pe->v += e2.v;
 273             if (pe->sym == NULL && e2.sym != NULL)
 274                 pe->sym = e2.sym;
 275         } else {
 276             pe->v -= e2.v;
 277             /* NOTE: we are less powerful than gas in that case
 278                because we store only one symbol in the expression */
 279             if (!e2.sym) {
 280                 /* OK */
 281             } else if (pe->sym == e2.sym) {
 282                 /* OK */
 283                 pe->sym = NULL; /* same symbols can be subtracted to NULL */
 284             } else if (pe->sym && pe->sym->r == e2.sym->r && pe->sym->r != 0) {
 285                 /* we also accept defined symbols in the same section */
 286                 pe->v += pe->sym->jnext - e2.sym->jnext;
 287                 pe->sym = NULL;
 288             } else if (e2.sym->r == cur_text_section->sh_num) {
 289                 /* When subtracting a defined symbol in current section
 290                    this actually makes the value PC-relative.  */
 291                 pe->v -= e2.sym->jnext - ind - 4;
 292                 pe->pcrel = 1;
 293                 e2.sym = NULL;
 294             } else {
 295             cannot_relocate:
 296                 tcc_error("invalid operation with label");
 297             }
 298         }
 299     }
 300 }
 301
 302 static inline void asm_expr_cmp(TCCState *s1, ExprValue *pe)
 303 {
 304     int op;
 305     ExprValue e2;
 306
 307     asm_expr_sum(s1, pe);
 308     for(;;) {
 309         op = tok;
 310         if (op != TOK_EQ && op != TOK_NE
 311             && (op > TOK_GT || op < TOK_ULE))
 312             break;
 313         next();
 314         asm_expr_sum(s1, &e2);
 315         if (pe->sym || e2.sym)
 316             tcc_error("invalid operation with label");
 317         switch(op) {
 318         case TOK_EQ:
 319             pe->v = pe->v == e2.v;
 320             break;
 321         case TOK_NE:
 322             pe->v = pe->v != e2.v;
 323             break;
 324         case TOK_LT:
 325             pe->v = (int64_t)pe->v < (int64_t)e2.v;
 326             break;
 327         case TOK_GE:
 328             pe->v = (int64_t)pe->v >= (int64_t)e2.v;
 329             break;
 330         case TOK_LE:
 331             pe->v = (int64_t)pe->v <= (int64_t)e2.v;
 332             break;
 333         case TOK_GT:
 334             pe->v = (int64_t)pe->v > (int64_t)e2.v;
 335             break;
 336         default:
 337             break;
 338         }
 339         /* GAS compare results are -1/0 not 1/0.  */
 340         pe->v = -(int64_t)pe->v;
 341     }
 342 }
 343
 344 ST_FUNC void asm_expr(TCCState *s1, ExprValue *pe)
 345 {
 346     asm_expr_cmp(s1, pe);
 347 }
 348
 349 ST_FUNC int asm_int_expr(TCCState *s1)
 350 {
 351     ExprValue e;
 352     asm_expr(s1, &e);
 353     if (e.sym)
 354         expect("constant");
 355     return e.v;
 356 }
 357
 358 /* NOTE: the same name space as C labels is used to avoid using too
 359    much memory when storing labels in TokenStrings */
 360 static Sym* asm_new_label1(TCCState *s1, int label, int is_local,
 361                            int sh_num, int value)
 362 {
 363     Sym *sym;
 364
 365     sym = label_find(label);
 366     if (sym) {
 367         /* A VT_EXTERN symbol, even if it has a section is considered
 368            overridable.  This is how we "define" .set targets.  Real
 369            definitions won't have VT_EXTERN set.  */
 370         if (sym->r && !(sym->type.t & VT_EXTERN)) {
 371             /* the label is already defined */
 372             if (!is_local) {
 373                 tcc_error("assembler label '%s' already defined",
 374                       get_tok_str(label, NULL));
 375             } else {
 376                 /* redefinition of local labels is possible */
 377                 goto new_label;
 378             }
 379         }
 380     } else {
 381     new_label:
 382         sym = label_push(&s1->asm_labels, label, 0);
 383         /* If we need a symbol to hold a value, mark it as
 384            tentative only (for .set).  If this is for a real label
 385            we'll remove VT_EXTERN.  */
 386         sym->type.t = VT_STATIC | VT_VOID | VT_EXTERN;
 387     }
 388     sym->r = sh_num;
 389     sym->jnext = value;
 390     return sym;
 391 }
 392
 393 static Sym* asm_new_label(TCCState *s1, int label, int is_local)
 394 {
 395     return asm_new_label1(s1, label, is_local, cur_text_section->sh_num, ind);
 396 }
 397
 398 /* Set the value of LABEL to that of some expression (possibly
 399    involving other symbols).  LABEL can be overwritten later still.  */
 400 static Sym* set_symbol(TCCState *s1, int label)
 401 {
 402     long n;
 403     ExprValue e;
 404     next();
 405     asm_expr(s1, &e);
 406     n = e.v;
 407     if (e.sym)
 408         n += e.sym->jnext;
 409     return asm_new_label1(s1, label, 0, e.sym ? e.sym->r : SHN_ABS, n);
 410 }
 411
 412 static void asm_free_labels(TCCState *st)
 413 {
 414     Sym *s, *s1;
 415     Section *sec;
 416
 417     for(s = st->asm_labels; s != NULL; s = s1) {
 418         s1 = s->prev;
 419         /* define symbol value in object file */
 420         s->type.t &= ~VT_EXTERN;
 421         if (s->r && !(s->type.t & VT_IMPORT)) {
 422             if (s->r == SHN_ABS)
 423                 sec = SECTION_ABS;
 424             else
 425                 sec = st->sections[s->r];
 426             put_extern_sym2(s, sec, s->jnext, 0, 0);
 427         }
 428         /* remove label */
 429         table_ident[s->v - TOK_IDENT]->sym_label = NULL;
 430         sym_free(s);
 431     }
 432     st->asm_labels = NULL;
 433 }
 434
 435 static void use_section1(TCCState *s1, Section *sec)
 436 {
 437     cur_text_section->data_offset = ind;
 438     cur_text_section = sec;
 439     ind = cur_text_section->data_offset;
 440 }
 441
 442 static void use_section(TCCState *s1, const char *name)
 443 {
 444     Section *sec;
 445     sec = find_section(s1, name);
 446     use_section1(s1, sec);
 447 }
 448
 449 static void push_section(TCCState *s1, const char *name)
 450 {
 451     Section *sec = find_section(s1, name);
 452     sec->prev = cur_text_section;
 453     use_section1(s1, sec);
 454 }
 455
 456 static void pop_section(TCCState *s1)
 457 {
 458     Section *prev = cur_text_section->prev;
 459     if (!prev)
 460         tcc_error(".popsection without .pushsection");
 461     cur_text_section->prev = NULL;
 462     use_section1(s1, prev);
 463 }
 464
 465 static void asm_parse_directive(TCCState *s1)
 466 {
 467     int n, offset, v, size, tok1;
 468     Section *sec;
 469     uint8_t *ptr;
 470
 471     /* assembler directive */
 472     sec = cur_text_section;
 473     switch(tok) {
 474     case TOK_ASMDIR_align:
 475     case TOK_ASMDIR_balign:
 476     case TOK_ASMDIR_p2align:
 477     case TOK_ASMDIR_skip:
 478     case TOK_ASMDIR_space:
 479         tok1 = tok;
 480         next();
 481         n = asm_int_expr(s1);
 482         if (tok1 == TOK_ASMDIR_p2align)
 483         {
 484             if (n < 0 || n > 30)
 485                 tcc_error("invalid p2align, must be between 0 and 30");
 486             n = 1 << n;
 487             tok1 = TOK_ASMDIR_align;
 488         }
 489         if (tok1 == TOK_ASMDIR_align || tok1 == TOK_ASMDIR_balign) {
 490             if (n < 0 || (n & (n-1)) != 0)
 491                 tcc_error("alignment must be a positive power of two");
 492             offset = (ind + n - 1) & -n;
 493             size = offset - ind;
 494             /* the section must have a compatible alignment */
 495             if (sec->sh_addralign < n)
 496                 sec->sh_addralign = n;
 497         } else {
 498             if (n < 0)
 499                 n = 0;
 500             size = n;
 501         }
 502         v = 0;
 503         if (tok == ',') {
 504             next();
 505             v = asm_int_expr(s1);
 506         }
 507     zero_pad:
 508         if (sec->sh_type != SHT_NOBITS) {
 509             sec->data_offset = ind;
 510             ptr = section_ptr_add(sec, size);
 511             memset(ptr, v, size);
 512         }
 513         ind += size;
 514         break;
 515     case TOK_ASMDIR_quad:
 516 #ifdef TCC_TARGET_X86_64
 517         size = 8;
 518         goto asm_data;
 519 #else
 520         next();
 521         for(;;) {
 522             uint64_t vl;
 523             const char *p;
 524
 525             p = tokc.str.data;
 526             if (tok != TOK_PPNUM) {
 527             error_constant:
 528                 tcc_error("64 bit constant");
 529             }
 530             vl = strtoll(p, (char **)&p, 0);
 531             if (*p != '\0')
 532                 goto error_constant;
 533             next();
 534             if (sec->sh_type != SHT_NOBITS) {
 535                 /* XXX: endianness */
 536                 gen_le32(vl);
 537                 gen_le32(vl >> 32);
 538             } else {
 539                 ind += 8;
 540             }
 541             if (tok != ',')
 542                 break;
 543             next();
 544         }
 545         break;
 546 #endif
 547     case TOK_ASMDIR_byte:
 548         size = 1;
 549         goto asm_data;
 550     case TOK_ASMDIR_word:
 551     case TOK_ASMDIR_short:
 552         size = 2;
 553         goto asm_data;
 554     case TOK_ASMDIR_long:
 555     case TOK_ASMDIR_int:
 556         size = 4;
 557     asm_data:
 558         next();
 559         for(;;) {
 560             ExprValue e;
 561             asm_expr(s1, &e);
 562             if (sec->sh_type != SHT_NOBITS) {
 563                 if (size == 4) {
 564                     gen_expr32(&e);
 565 #ifdef TCC_TARGET_X86_64
 566                 } else if (size == 8) {
 567                     gen_expr64(&e);
 568 #endif
 569                 } else {
 570                     if (e.sym)
 571                         expect("constant");
 572                     if (size == 1)
 573                         g(e.v);
 574                     else
 575                         gen_le16(e.v);
 576                 }
 577             } else {
 578                 ind += size;
 579             }
 580             if (tok != ',')
 581                 break;
 582             next();
 583         }
 584         break;
 585     case TOK_ASMDIR_fill:
 586         {
 587             int repeat, size, val, i, j;
 588             uint8_t repeat_buf[8];
 589             next();
 590             repeat = asm_int_expr(s1);
 591             if (repeat < 0) {
 592                 tcc_error("repeat < 0; .fill ignored");
 593                 break;
 594             }
 595             size = 1;
 596             val = 0;
 597             if (tok == ',') {
 598                 next();
 599                 size = asm_int_expr(s1);
 600                 if (size < 0) {
 601                     tcc_error("size < 0; .fill ignored");
 602                     break;
 603                 }
 604                 if (size > 8)
 605                     size = 8;
 606                 if (tok == ',') {
 607                     next();
 608                     val = asm_int_expr(s1);
 609                 }
 610             }
 611             /* XXX: endianness */
 612             repeat_buf[0] = val;
 613             repeat_buf[1] = val >> 8;
 614             repeat_buf[2] = val >> 16;
 615             repeat_buf[3] = val >> 24;
 616             repeat_buf[4] = 0;
 617             repeat_buf[5] = 0;
 618             repeat_buf[6] = 0;
 619             repeat_buf[7] = 0;
 620             for(i = 0; i < repeat; i++) {
 621                 for(j = 0; j < size; j++) {
 622                     g(repeat_buf[j]);
 623                 }
 624             }
 625         }
 626         break;
 627     case TOK_ASMDIR_rept:
 628         {
 629             int repeat;
 630             TokenString *init_str;
 631             ParseState saved_parse_state = {0};
 632             next();
 633             repeat = asm_int_expr(s1);
 634             init_str = tok_str_alloc();
 635             next();
 636             while ((tok != TOK_ASMDIR_endr) && (tok != CH_EOF)) {
 637                 tok_str_add_tok(init_str);
 638                 next();
 639             }
 640             if (tok == CH_EOF) tcc_error("we at end of file, .endr not found");
 641             next();
 642             tok_str_add(init_str, -1);
 643             tok_str_add(init_str, 0);
 644             save_parse_state(&saved_parse_state);
 645             begin_macro(init_str, 1);
 646             while (repeat-- > 0) {
 647                 tcc_assemble_internal(s1, (parse_flags & PARSE_FLAG_PREPROCESS));
 648                 macro_ptr = init_str->str;
 649             }
 650             end_macro();
 651             restore_parse_state(&saved_parse_state);
 652             break;
 653         }
 654     case TOK_ASMDIR_org:
 655         {
 656             unsigned long n;
 657             ExprValue e;
 658             next();
 659             asm_expr(s1, &e);
 660             n = e.v;
 661             if (e.sym) {
 662                 if (e.sym->r != cur_text_section->sh_num)
 663                   expect("constant or same-section symbol");
 664                 n += e.sym->jnext;
 665             }
 666             if (n < ind)
 667                 tcc_error("attempt to .org backwards");
 668             v = 0;
 669             size = n - ind;
 670             goto zero_pad;
 671         }
 672         break;
 673     case TOK_ASMDIR_set:
 674         next();
 675         tok1 = tok;
 676         next();
 677         /* Also accept '.set stuff', but don't do anything with this.
 678            It's used in GAS to set various features like '.set mips16'.  */
 679         if (tok == ',')
 680             set_symbol(s1, tok1);
 681         break;
 682     case TOK_ASMDIR_globl:
 683     case TOK_ASMDIR_global:
 684     case TOK_ASMDIR_weak:
 685     case TOK_ASMDIR_hidden:
 686         tok1 = tok;
 687         do {
 688             Sym *sym;
 689
 690             next();
 691             sym = get_asm_sym(tok, NULL);
 692             if (tok1 != TOK_ASMDIR_hidden)
 693                 sym->type.t &= ~VT_STATIC;
 694             if (tok1 == TOK_ASMDIR_weak)
 695                 sym->type.t |= VT_WEAK;
 696             else if (tok1 == TOK_ASMDIR_hidden)
 697                 sym->type.t |= STV_HIDDEN << VT_VIS_SHIFT;
 698             next();
 699         } while (tok == ',');
 700         break;
 701     case TOK_ASMDIR_string:
 702     case TOK_ASMDIR_ascii:
 703     case TOK_ASMDIR_asciz:
 704         {
 705             const uint8_t *p;
 706             int i, size, t;
 707
 708             t = tok;
 709             next();
 710             for(;;) {
 711                 if (tok != TOK_STR)
 712                     expect("string constant");
 713                 p = tokc.str.data;
 714                 size = tokc.str.size;
 715                 if (t == TOK_ASMDIR_ascii && size > 0)
 716                     size--;
 717                 for(i = 0; i < size; i++)
 718                     g(p[i]);
 719                 next();
 720                 if (tok == ',') {
 721                     next();
 722                 } else if (tok != TOK_STR) {
 723                     break;
 724                 }
 725             }
 726         }
 727         break;
 728     case TOK_ASMDIR_text:
 729     case TOK_ASMDIR_data:
 730     case TOK_ASMDIR_bss:
 731         {
 732             char sname[64];
 733             tok1 = tok;
 734             n = 0;
 735             next();
 736             if (tok != ';' && tok != TOK_LINEFEED) {
 737                 n = asm_int_expr(s1);
 738                 next();
 739             }
 740             if (n)
 741                 sprintf(sname, "%s%d", get_tok_str(tok1, NULL), n);
 742             else
 743                 sprintf(sname, "%s", get_tok_str(tok1, NULL));
 744             use_section(s1, sname);
 745         }
 746         break;
 747     case TOK_ASMDIR_file:
 748         {
 749             char filename[512];
 750
 751             filename[0] = '\0';
 752             next();
 753
 754             if (tok == TOK_STR)
 755                 pstrcat(filename, sizeof(filename), tokc.str.data);
 756             else
 757                 pstrcat(filename, sizeof(filename), get_tok_str(tok, NULL));
 758
 759             if (s1->warn_unsupported)
 760                 tcc_warning("ignoring .file %s", filename);
 761
 762             next();
 763         }
 764         break;
 765     case TOK_ASMDIR_ident:
 766         {
 767             char ident[256];
 768
 769             ident[0] = '\0';
 770             next();
 771
 772             if (tok == TOK_STR)
 773                 pstrcat(ident, sizeof(ident), tokc.str.data);
 774             else
 775                 pstrcat(ident, sizeof(ident), get_tok_str(tok, NULL));
 776
 777             if (s1->warn_unsupported)
 778                 tcc_warning("ignoring .ident %s", ident);
 779
 780             next();
 781         }
 782         break;
 783     case TOK_ASMDIR_size:
 784         {
 785             Sym *sym;
 786
 787             next();
 788             sym = label_find(tok);
 789             if (!sym) {
 790                 tcc_error("label not found: %s", get_tok_str(tok, NULL));
 791             }
 792
 793             /* XXX .size name,label2-label1 */
 794             if (s1->warn_unsupported)
 795                 tcc_warning("ignoring .size %s,*", get_tok_str(tok, NULL));
 796
 797             next();
 798             skip(',');
 799             while (tok != TOK_LINEFEED && tok != ';' && tok != CH_EOF) {
 800                 next();
 801             }
 802         }
 803         break;
 804     case TOK_ASMDIR_type:
 805         {
 806             Sym *sym;
 807             const char *newtype;
 808
 809             next();
 810             sym = get_asm_sym(tok, NULL);
 811             next();
 812             skip(',');
 813             if (tok == TOK_STR) {
 814                 newtype = tokc.str.data;
 815             } else {
 816                 if (tok == '@' || tok == '%')
 817                     next();
 818                 newtype = get_tok_str(tok, NULL);
 819             }
 820
 821             if (!strcmp(newtype, "function") || !strcmp(newtype, "STT_FUNC")) {
 822                 sym->type.t = (sym->type.t & ~VT_BTYPE) | VT_FUNC;
 823             }
 824             else if (s1->warn_unsupported)
 825                 tcc_warning("change type of '%s' from 0x%x to '%s' ignored",
 826                     get_tok_str(sym->v, NULL), sym->type.t, newtype);
 827
 828             next();
 829         }
 830         break;
 831     case TOK_ASMDIR_pushsection:
 832     case TOK_ASMDIR_section:
 833         {
 834             char sname[256];
 835             int old_nb_section = s1->nb_sections;
 836
 837             tok1 = tok;
 838             /* XXX: support more options */
 839             next();
 840             sname[0] = '\0';
 841             while (tok != ';' && tok != TOK_LINEFEED && tok != ',') {
 842                 if (tok == TOK_STR)
 843                     pstrcat(sname, sizeof(sname), tokc.str.data);
 844                 else
 845                     pstrcat(sname, sizeof(sname), get_tok_str(tok, NULL));
 846                 next();
 847             }
 848             if (tok == ',') {
 849                 /* skip section options */
 850                 next();
 851                 if (tok != TOK_STR)
 852                     expect("string constant");
 853                 next();
 854                 if (tok == ',') {
 855                     next();
 856                     if (tok == '@' || tok == '%')
 857                         next();
 858                     next();
 859                 }
 860             }
 861             last_text_section = cur_text_section;
 862             if (tok1 == TOK_ASMDIR_section)
 863                 use_section(s1, sname);
 864             else
 865                 push_section(s1, sname);
 866             /* If we just allocated a new section reset its alignment to
 867                1.  new_section normally acts for GCC compatibility and
 868                sets alignment to PTR_SIZE.  The assembler behaves different. */
 869             if (old_nb_section != s1->nb_sections)
 870                 cur_text_section->sh_addralign = 1;
 871         }
 872         break;
 873     case TOK_ASMDIR_previous:
 874         {
 875             Section *sec;
 876             next();
 877             if (!last_text_section)
 878                 tcc_error("no previous section referenced");
 879             sec = cur_text_section;
 880             use_section1(s1, last_text_section);
 881             last_text_section = sec;
 882         }
 883         break;
 884     case TOK_ASMDIR_popsection:
 885         next();
 886         pop_section(s1);
 887         break;
 888 #ifdef TCC_TARGET_I386
 889     case TOK_ASMDIR_code16:
 890         {
 891             next();
 892             s1->seg_size = 16;
 893         }
 894         break;
 895     case TOK_ASMDIR_code32:
 896         {
 897             next();
 898             s1->seg_size = 32;
 899         }
 900         break;
 901 #endif
 902 #ifdef TCC_TARGET_X86_64
 903     /* added for compatibility with GAS */
 904     case TOK_ASMDIR_code64:
 905         next();
 906         break;
 907 #endif
 908     default:
 909         tcc_error("unknown assembler directive '.%s'", get_tok_str(tok, NULL));
 910         break;
 911     }
 912 }
 913
 914
 915 /* assemble a file */
 916 static int tcc_assemble_internal(TCCState *s1, int do_preprocess)
 917 {
 918     int opcode;
 919
 920     /* XXX: undefine C labels */
 921
 922     ch = file->buf_ptr[0];
 923     tok_flags = TOK_FLAG_BOL | TOK_FLAG_BOF;
 924     parse_flags = PARSE_FLAG_ASM_FILE | PARSE_FLAG_TOK_STR;
 925     set_idnum('.', IS_ID);
 926     if (do_preprocess)
 927         parse_flags |= PARSE_FLAG_PREPROCESS;
 928     next();
 929     for(;;) {
 930         if (tok == TOK_EOF)
 931             break;
 932         parse_flags |= PARSE_FLAG_LINEFEED; /* XXX: suppress that hack */
 933     redo:
 934         if (tok == '#') {
 935             /* horrible gas comment */
 936             while (tok != TOK_LINEFEED)
 937                 next();
 938         } else if (tok >= TOK_ASMDIR_FIRST && tok <= TOK_ASMDIR_LAST) {
 939             asm_parse_directive(s1);
 940         } else if (tok == TOK_PPNUM) {
 941             Sym *sym;
 942             const char *p;
 943             int n;
 944             p = tokc.str.data;
 945             n = strtoul(p, (char **)&p, 10);
 946             if (*p != '\0')
 947                 expect("':'");
 948             /* new local label */
 949             sym = asm_new_label(s1, asm_get_local_label_name(s1, n), 1);
 950             /* Remove the marker for tentative definitions.  */
 951             sym->type.t &= ~VT_EXTERN;
 952             next();
 953             skip(':');
 954             goto redo;
 955         } else if (tok >= TOK_IDENT) {
 956             /* instruction or label */
 957             opcode = tok;
 958             next();
 959             if (tok == ':') {
 960                 /* handle "extern void vide(void); __asm__("vide: ret");" as
 961                 "__asm__("globl vide\nvide: ret");" */
 962                 Sym *sym = sym_find(opcode);
 963                 if (sym && (sym->type.t & VT_EXTERN) && nocode_wanted) {
 964                     sym = label_find(opcode);
 965                     if (!sym) {
 966                         sym = label_push(&s1->asm_labels, opcode, 0);
 967                         sym->type.t = VT_VOID | VT_EXTERN;
 968                     }
 969                 }
 970                 /* new label */
 971                 sym = asm_new_label(s1, opcode, 0);
 972                 sym->type.t &= ~VT_EXTERN;
 973                 next();
 974                 goto redo;
 975             } else if (tok == '=') {
 976                 set_symbol(s1, opcode);
 977                 goto redo;
 978             } else {
 979                 asm_opcode(s1, opcode);
 980             }
 981         }
 982         /* end of line */
 983         if (tok != ';' && tok != TOK_LINEFEED){
 984             expect("end of line");
 985         }
 986         parse_flags &= ~PARSE_FLAG_LINEFEED; /* XXX: suppress that hack */
 987         next();
 988     }
 989
 990     asm_free_labels(s1);
 991
 992     return 0;
 993 }
 994
 995 /* Assemble the current file */
 996 ST_FUNC int tcc_assemble(TCCState *s1, int do_preprocess)
 997 {
 998     Sym *define_start;
 999     int ret;
1000
1001     preprocess_start(s1);
1002
1003     /* default section is text */
1004     cur_text_section = text_section;
1005     ind = cur_text_section->data_offset;
1006
1007     define_start = define_stack;
1008
1009     /* an elf symbol of type STT_FILE must be put so that STB_LOCAL
1010        symbols can be safely used */
1011     put_elf_sym(symtab_section, 0, 0,
1012                 ELFW(ST_INFO)(STB_LOCAL, STT_FILE), 0,
1013                 SHN_ABS, file->filename);
1014
1015     ret = tcc_assemble_internal(s1, do_preprocess);
1016
1017     cur_text_section->data_offset = ind;
1018
1019     free_defines(define_start);
1020
1021     return ret;
1022 }
1023
1024 /********************************************************************/
1025 /* GCC inline asm support */
1026
1027 /* assemble the string 'str' in the current C compilation unit without
1028    C preprocessing. NOTE: str is modified by modifying the '\0' at the
1029    end */
1030 static void tcc_assemble_inline(TCCState *s1, char *str, int len)
1031 {
1032     int saved_parse_flags;
1033     const int *saved_macro_ptr;
1034
1035     saved_parse_flags = parse_flags;
1036     saved_macro_ptr = macro_ptr;
1037
1038     tcc_open_bf(s1, ":asm:", len);
1039     memcpy(file->buffer, str, len);
1040
1041     macro_ptr = NULL;
1042     tcc_assemble_internal(s1, 0);
1043     tcc_close();
1044
1045     parse_flags = saved_parse_flags;
1046     set_idnum('.', (parse_flags & PARSE_FLAG_ASM_FILE) ? IS_ID : 0);
1047     macro_ptr = saved_macro_ptr;
1048 }
1049
1050 /* find a constraint by its number or id (gcc 3 extended
1051    syntax). return -1 if not found. Return in *pp in char after the
1052    constraint */
1053 ST_FUNC int find_constraint(ASMOperand *operands, int nb_operands,
1054                            const char *name, const char **pp)
1055 {
1056     int index;
1057     TokenSym *ts;
1058     const char *p;
1059
1060     if (isnum(*name)) {
1061         index = 0;
1062         while (isnum(*name)) {
1063             index = (index * 10) + (*name) - '0';
1064             name++;
1065         }
1066         if ((unsigned)index >= nb_operands)
1067             index = -1;
1068     } else if (*name == '[') {
1069         name++;
1070         p = strchr(name, ']');
1071         if (p) {
1072             ts = tok_alloc(name, p - name);
1073             for(index = 0; index < nb_operands; index++) {
1074                 if (operands[index].id == ts->tok)
1075                     goto found;
1076             }
1077             index = -1;
1078         found:
1079             name = p + 1;
1080         } else {
1081             index = -1;
1082         }
1083     } else {
1084         index = -1;
1085     }
1086     if (pp)
1087         *pp = name;
1088     return index;
1089 }
1090
1091 static void subst_asm_operands(ASMOperand *operands, int nb_operands,
1092                                int nb_outputs,
1093                                CString *out_str, CString *in_str)
1094 {
1095     int c, index, modifier;
1096     const char *str;
1097     ASMOperand *op;
1098     SValue sv;
1099
1100     cstr_new(out_str);
1101     str = in_str->data;
1102     for(;;) {
1103         c = *str++;
1104         if (c == '%') {
1105             if (*str == '%') {
1106                 str++;
1107                 goto add_char;
1108             }
1109             modifier = 0;
1110             if (*str == 'c' || *str == 'n' ||
1111                 *str == 'b' || *str == 'w' || *str == 'h' || *str == 'k' ||
1112                 *str == 'q' ||
1113                 /* P in GCC would add "@PLT" to symbol refs in PIC mode,
1114                    and make literal operands not be decorated with '$'.  */
1115                 *str == 'P')
1116                 modifier = *str++;
1117             index = find_constraint(operands, nb_operands, str, &str);
1118             if (index < 0)
1119                 tcc_error("invalid operand reference after %%");
1120             op = &operands[index];
1121             sv = *op->vt;
1122             if (op->reg >= 0) {
1123                 sv.r = op->reg;
1124                 if ((op->vt->r & VT_VALMASK) == VT_LLOCAL && op->is_memory)
1125                     sv.r |= VT_LVAL;
1126             }
1127             subst_asm_operand(out_str, &sv, modifier);
1128         } else {
1129         add_char:
1130             cstr_ccat(out_str, c);
1131             if (c == '\0')
1132                 break;
1133         }
1134     }
1135 }
1136
1137
1138 static void parse_asm_operands(ASMOperand *operands, int *nb_operands_ptr,
1139                                int is_output)
1140 {
1141     ASMOperand *op;
1142     int nb_operands;
1143
1144     if (tok != ':') {
1145         nb_operands = *nb_operands_ptr;
1146         for(;;) {
1147             CString astr;
1148             if (nb_operands >= MAX_ASM_OPERANDS)
1149                 tcc_error("too many asm operands");
1150             op = &operands[nb_operands++];
1151             op->id = 0;
1152             if (tok == '[') {
1153                 next();
1154                 if (tok < TOK_IDENT)
1155                     expect("identifier");
1156                 op->id = tok;
1157                 next();
1158                 skip(']');
1159             }
1160             parse_mult_str(&astr, "string constant");
1161             op->constraint = tcc_malloc(astr.size);
1162             strcpy(op->constraint, astr.data);
1163             cstr_free(&astr);
1164             skip('(');
1165             gexpr();
1166             if (is_output) {
1167                 if (!(vtop->type.t & VT_ARRAY))
1168                     test_lvalue();
1169             } else {
1170                 /* we want to avoid LLOCAL case, except when the 'm'
1171                    constraint is used. Note that it may come from
1172                    register storage, so we need to convert (reg)
1173                    case */
1174                 if ((vtop->r & VT_LVAL) &&
1175                     ((vtop->r & VT_VALMASK) == VT_LLOCAL ||
1176                      (vtop->r & VT_VALMASK) < VT_CONST) &&
1177                     !strchr(op->constraint, 'm')) {
1178                     gv(RC_INT);
1179                 }
1180             }
1181             op->vt = vtop;
1182             skip(')');
1183             if (tok == ',') {
1184                 next();
1185             } else {
1186                 break;
1187             }
1188         }
1189         *nb_operands_ptr = nb_operands;
1190     }
1191 }
1192
1193 /* parse the GCC asm() instruction */
1194 ST_FUNC void asm_instr(void)
1195 {
1196     CString astr, astr1;
1197     ASMOperand operands[MAX_ASM_OPERANDS];
1198     int nb_outputs, nb_operands, i, must_subst, out_reg;
1199     uint8_t clobber_regs[NB_ASM_REGS];
1200
1201     next();
1202     /* since we always generate the asm() instruction, we can ignore
1203        volatile */
1204     if (tok == TOK_VOLATILE1 || tok == TOK_VOLATILE2 || tok == TOK_VOLATILE3) {
1205         next();
1206     }
1207     parse_asm_str(&astr);
1208     nb_operands = 0;
1209     nb_outputs = 0;
1210     must_subst = 0;
1211     memset(clobber_regs, 0, sizeof(clobber_regs));
1212     if (tok == ':') {
1213         next();
1214         must_subst = 1;
1215         /* output args */
1216         parse_asm_operands(operands, &nb_operands, 1);
1217         nb_outputs = nb_operands;
1218         if (tok == ':') {
1219             next();
1220             if (tok != ')') {
1221                 /* input args */
1222                 parse_asm_operands(operands, &nb_operands, 0);
1223                 if (tok == ':') {
1224                     /* clobber list */
1225                     /* XXX: handle registers */
1226                     next();
1227                     for(;;) {
1228                         if (tok != TOK_STR)
1229                             expect("string constant");
1230                         asm_clobber(clobber_regs, tokc.str.data);
1231                         next();
1232                         if (tok == ',') {
1233                             next();
1234                         } else {
1235                             break;
1236                         }
1237                     }
1238                 }
1239             }
1240         }
1241     }
1242     skip(')');
1243     /* NOTE: we do not eat the ';' so that we can restore the current
1244        token after the assembler parsing */
1245     if (tok != ';')
1246         expect("';'");
1247
1248     /* save all values in the memory */
1249     save_regs(0);
1250
1251     /* compute constraints */
1252     asm_compute_constraints(operands, nb_operands, nb_outputs,
1253                             clobber_regs, &out_reg);
1254
1255     /* substitute the operands in the asm string. No substitution is
1256        done if no operands (GCC behaviour) */
1257 #ifdef ASM_DEBUG
1258     printf("asm: \"%s\"\n", (char *)astr.data);
1259 #endif
1260     if (must_subst) {
1261         subst_asm_operands(operands, nb_operands, nb_outputs, &astr1, &astr);
1262         cstr_free(&astr);
1263     } else {
1264         astr1 = astr;
1265     }
1266 #ifdef ASM_DEBUG
1267     printf("subst_asm: \"%s\"\n", (char *)astr1.data);
1268 #endif
1269
1270     /* generate loads */
1271     asm_gen_code(operands, nb_operands, nb_outputs, 0,
1272                  clobber_regs, out_reg);
1273
1274     /* assemble the string with tcc internal assembler */
1275     tcc_assemble_inline(tcc_state, astr1.data, astr1.size - 1);
1276
1277     /* restore the current C token */
1278     next();
1279
1280     /* store the output values if needed */
1281     asm_gen_code(operands, nb_operands, nb_outputs, 1,
1282                  clobber_regs, out_reg);
1283
1284     /* free everything */
1285     for(i=0;i<nb_operands;i++) {
1286         ASMOperand *op;
1287         op = &operands[i];
1288         tcc_free(op->constraint);
1289         vpop();
1290     }
1291     cstr_free(&astr1);
1292 }
1293
1294 ST_FUNC void asm_global_instr(void)
1295 {
1296     CString astr;
1297
1298     next();
1299     parse_asm_str(&astr);
1300     skip(')');
1301     /* NOTE: we do not eat the ';' so that we can restore the current
1302        token after the assembler parsing */
1303     if (tok != ';')
1304         expect("';'");
1305
1306 #ifdef ASM_DEBUG
1307     printf("asm_global: \"%s\"\n", (char *)astr.data);
1308 #endif
1309     cur_text_section = text_section;
1310     ind = cur_text_section->data_offset;
1311
1312     /* assemble the string with tcc internal assembler */
1313     tcc_assemble_inline(tcc_state, astr.data, astr.size - 1);
1314
1315     cur_text_section->data_offset = ind;
1316
1317     /* restore the current C token */
1318     next();
1319
1320     cstr_free(&astr);
1321 }
1322 #endif /* CONFIG_TCC_ASM */