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, int global);
  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     uint64_t n;
  91     const char *p;
  92
  93     switch(tok) {
  94     case TOK_PPNUM:
  95         p = tokc.str.data;
  96         n = strtoull(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, int global)
 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                                       global);
 649                 macro_ptr = init_str->str;
 650             }
 651             end_macro();
 652             restore_parse_state(&saved_parse_state);
 653             break;
 654         }
 655     case TOK_ASMDIR_org:
 656         {
 657             unsigned long n;
 658             ExprValue e;
 659             next();
 660             asm_expr(s1, &e);
 661             n = e.v;
 662             if (e.sym) {
 663                 if (e.sym->r != cur_text_section->sh_num)
 664                   expect("constant or same-section symbol");
 665                 n += e.sym->jnext;
 666             }
 667             if (n < ind)
 668                 tcc_error("attempt to .org backwards");
 669             v = 0;
 670             size = n - ind;
 671             goto zero_pad;
 672         }
 673         break;
 674     case TOK_ASMDIR_set:
 675         next();
 676         tok1 = tok;
 677         next();
 678         /* Also accept '.set stuff', but don't do anything with this.
 679            It's used in GAS to set various features like '.set mips16'.  */
 680         if (tok == ',')
 681             set_symbol(s1, tok1);
 682         break;
 683     case TOK_ASMDIR_globl:
 684     case TOK_ASMDIR_global:
 685     case TOK_ASMDIR_weak:
 686     case TOK_ASMDIR_hidden:
 687         tok1 = tok;
 688         do {
 689             Sym *sym;
 690
 691             next();
 692             sym = get_asm_sym(tok, NULL);
 693             if (tok1 != TOK_ASMDIR_hidden)
 694                 sym->type.t &= ~VT_STATIC;
 695             if (tok1 == TOK_ASMDIR_weak)
 696                 sym->type.t |= VT_WEAK;
 697             else if (tok1 == TOK_ASMDIR_hidden)
 698                 sym->type.t |= STV_HIDDEN << VT_VIS_SHIFT;
 699             next();
 700         } while (tok == ',');
 701         break;
 702     case TOK_ASMDIR_string:
 703     case TOK_ASMDIR_ascii:
 704     case TOK_ASMDIR_asciz:
 705         {
 706             const uint8_t *p;
 707             int i, size, t;
 708
 709             t = tok;
 710             next();
 711             for(;;) {
 712                 if (tok != TOK_STR)
 713                     expect("string constant");
 714                 p = tokc.str.data;
 715                 size = tokc.str.size;
 716                 if (t == TOK_ASMDIR_ascii && size > 0)
 717                     size--;
 718                 for(i = 0; i < size; i++)
 719                     g(p[i]);
 720                 next();
 721                 if (tok == ',') {
 722                     next();
 723                 } else if (tok != TOK_STR) {
 724                     break;
 725                 }
 726             }
 727         }
 728         break;
 729     case TOK_ASMDIR_text:
 730     case TOK_ASMDIR_data:
 731     case TOK_ASMDIR_bss:
 732         {
 733             char sname[64];
 734             tok1 = tok;
 735             n = 0;
 736             next();
 737             if (tok != ';' && tok != TOK_LINEFEED) {
 738                 n = asm_int_expr(s1);
 739                 next();
 740             }
 741             if (n)
 742                 sprintf(sname, "%s%d", get_tok_str(tok1, NULL), n);
 743             else
 744                 sprintf(sname, "%s", get_tok_str(tok1, NULL));
 745             use_section(s1, sname);
 746         }
 747         break;
 748     case TOK_ASMDIR_file:
 749         {
 750             char filename[512];
 751
 752             filename[0] = '\0';
 753             next();
 754
 755             if (tok == TOK_STR)
 756                 pstrcat(filename, sizeof(filename), tokc.str.data);
 757             else
 758                 pstrcat(filename, sizeof(filename), get_tok_str(tok, NULL));
 759
 760             if (s1->warn_unsupported)
 761                 tcc_warning("ignoring .file %s", filename);
 762
 763             next();
 764         }
 765         break;
 766     case TOK_ASMDIR_ident:
 767         {
 768             char ident[256];
 769
 770             ident[0] = '\0';
 771             next();
 772
 773             if (tok == TOK_STR)
 774                 pstrcat(ident, sizeof(ident), tokc.str.data);
 775             else
 776                 pstrcat(ident, sizeof(ident), get_tok_str(tok, NULL));
 777
 778             if (s1->warn_unsupported)
 779                 tcc_warning("ignoring .ident %s", ident);
 780
 781             next();
 782         }
 783         break;
 784     case TOK_ASMDIR_size:
 785         {
 786             Sym *sym;
 787
 788             next();
 789             sym = label_find(tok);
 790             if (!sym) {
 791                 tcc_error("label not found: %s", get_tok_str(tok, NULL));
 792             }
 793
 794             /* XXX .size name,label2-label1 */
 795             if (s1->warn_unsupported)
 796                 tcc_warning("ignoring .size %s,*", get_tok_str(tok, NULL));
 797
 798             next();
 799             skip(',');
 800             while (tok != TOK_LINEFEED && tok != ';' && tok != CH_EOF) {
 801                 next();
 802             }
 803         }
 804         break;
 805     case TOK_ASMDIR_type:
 806         {
 807             Sym *sym;
 808             const char *newtype;
 809
 810             next();
 811             sym = get_asm_sym(tok, NULL);
 812             next();
 813             skip(',');
 814             if (tok == TOK_STR) {
 815                 newtype = tokc.str.data;
 816             } else {
 817                 if (tok == '@' || tok == '%')
 818                     next();
 819                 newtype = get_tok_str(tok, NULL);
 820             }
 821
 822             if (!strcmp(newtype, "function") || !strcmp(newtype, "STT_FUNC")) {
 823                 sym->type.t = (sym->type.t & ~VT_BTYPE) | VT_FUNC;
 824             }
 825             else if (s1->warn_unsupported)
 826                 tcc_warning("change type of '%s' from 0x%x to '%s' ignored",
 827                     get_tok_str(sym->v, NULL), sym->type.t, newtype);
 828
 829             next();
 830         }
 831         break;
 832     case TOK_ASMDIR_pushsection:
 833     case TOK_ASMDIR_section:
 834         {
 835             char sname[256];
 836             int old_nb_section = s1->nb_sections;
 837
 838             tok1 = tok;
 839             /* XXX: support more options */
 840             next();
 841             sname[0] = '\0';
 842             while (tok != ';' && tok != TOK_LINEFEED && tok != ',') {
 843                 if (tok == TOK_STR)
 844                     pstrcat(sname, sizeof(sname), tokc.str.data);
 845                 else
 846                     pstrcat(sname, sizeof(sname), get_tok_str(tok, NULL));
 847                 next();
 848             }
 849             if (tok == ',') {
 850                 /* skip section options */
 851                 next();
 852                 if (tok != TOK_STR)
 853                     expect("string constant");
 854                 next();
 855                 if (tok == ',') {
 856                     next();
 857                     if (tok == '@' || tok == '%')
 858                         next();
 859                     next();
 860                 }
 861             }
 862             last_text_section = cur_text_section;
 863             if (tok1 == TOK_ASMDIR_section)
 864                 use_section(s1, sname);
 865             else
 866                 push_section(s1, sname);
 867             /* If we just allocated a new section reset its alignment to
 868                1.  new_section normally acts for GCC compatibility and
 869                sets alignment to PTR_SIZE.  The assembler behaves different. */
 870             if (old_nb_section != s1->nb_sections)
 871                 cur_text_section->sh_addralign = 1;
 872         }
 873         break;
 874     case TOK_ASMDIR_previous:
 875         {
 876             Section *sec;
 877             next();
 878             if (!last_text_section)
 879                 tcc_error("no previous section referenced");
 880             sec = cur_text_section;
 881             use_section1(s1, last_text_section);
 882             last_text_section = sec;
 883         }
 884         break;
 885     case TOK_ASMDIR_popsection:
 886         next();
 887         pop_section(s1);
 888         break;
 889 #ifdef TCC_TARGET_I386
 890     case TOK_ASMDIR_code16:
 891         {
 892             next();
 893             s1->seg_size = 16;
 894         }
 895         break;
 896     case TOK_ASMDIR_code32:
 897         {
 898             next();
 899             s1->seg_size = 32;
 900         }
 901         break;
 902 #endif
 903 #ifdef TCC_TARGET_X86_64
 904     /* added for compatibility with GAS */
 905     case TOK_ASMDIR_code64:
 906         next();
 907         break;
 908 #endif
 909     default:
 910         tcc_error("unknown assembler directive '.%s'", get_tok_str(tok, NULL));
 911         break;
 912     }
 913 }
 914
 915
 916 /* assemble a file */
 917 static int tcc_assemble_internal(TCCState *s1, int do_preprocess, int global)
 918 {
 919     int opcode;
 920
 921     /* XXX: undefine C labels */
 922
 923     ch = file->buf_ptr[0];
 924     tok_flags = TOK_FLAG_BOL | TOK_FLAG_BOF;
 925     parse_flags = PARSE_FLAG_ASM_FILE | PARSE_FLAG_TOK_STR;
 926     set_idnum('.', IS_ID);
 927     if (do_preprocess)
 928         parse_flags |= PARSE_FLAG_PREPROCESS;
 929     for(;;) {
 930         next();
 931         if (tok == TOK_EOF)
 932             break;
 933         /* generate line number info */
 934         if (global && s1->do_debug)
 935             tcc_debug_line(s1);
 936         parse_flags |= PARSE_FLAG_LINEFEED; /* XXX: suppress that hack */
 937     redo:
 938         if (tok == '#') {
 939             /* horrible gas comment */
 940             while (tok != TOK_LINEFEED)
 941                 next();
 942         } else if (tok >= TOK_ASMDIR_FIRST && tok <= TOK_ASMDIR_LAST) {
 943             asm_parse_directive(s1, global);
 944         } else if (tok == TOK_PPNUM) {
 945             Sym *sym;
 946             const char *p;
 947             int n;
 948             p = tokc.str.data;
 949             n = strtoul(p, (char **)&p, 10);
 950             if (*p != '\0')
 951                 expect("':'");
 952             /* new local label */
 953             sym = asm_new_label(s1, asm_get_local_label_name(s1, n), 1);
 954             /* Remove the marker for tentative definitions.  */
 955             sym->type.t &= ~VT_EXTERN;
 956             next();
 957             skip(':');
 958             goto redo;
 959         } else if (tok >= TOK_IDENT) {
 960             /* instruction or label */
 961             opcode = tok;
 962             next();
 963             if (tok == ':') {
 964                 /* handle "extern void vide(void); __asm__("vide: ret");" as
 965                 "__asm__("globl vide\nvide: ret");" */
 966                 Sym *sym = sym_find(opcode);
 967                 if (sym && (sym->type.t & VT_EXTERN) && global) {
 968                     sym = label_find(opcode);
 969                     if (!sym) {
 970                         sym = label_push(&s1->asm_labels, opcode, 0);
 971                         sym->type.t = VT_VOID | VT_EXTERN;
 972                     }
 973                 }
 974                 /* new label */
 975                 sym = asm_new_label(s1, opcode, 0);
 976                 sym->type.t &= ~VT_EXTERN;
 977                 next();
 978                 goto redo;
 979             } else if (tok == '=') {
 980                 set_symbol(s1, opcode);
 981                 goto redo;
 982             } else {
 983                 asm_opcode(s1, opcode);
 984             }
 985         }
 986         /* end of line */
 987         if (tok != ';' && tok != TOK_LINEFEED)
 988             expect("end of line");
 989         parse_flags &= ~PARSE_FLAG_LINEFEED; /* XXX: suppress that hack */
 990     }
 991
 992     asm_free_labels(s1);
 993     return 0;
 994 }
 995
 996 /* Assemble the current file */
 997 ST_FUNC int tcc_assemble(TCCState *s1, int do_preprocess)
 998 {
 999     Sym *define_start;
1000     int ret;
1001
1002     define_start = define_stack;
1003     preprocess_start(s1);
1004     tcc_debug_start(s1);
1005
1006     /* default section is text */
1007     cur_text_section = text_section;
1008     ind = cur_text_section->data_offset;
1009     nocode_wanted = 0;
1010
1011     ret = tcc_assemble_internal(s1, do_preprocess, 1);
1012
1013     cur_text_section->data_offset = ind;
1014
1015     tcc_debug_end(s1);
1016     free_defines(define_start);
1017     return ret;
1018 }
1019
1020 /********************************************************************/
1021 /* GCC inline asm support */
1022
1023 /* assemble the string 'str' in the current C compilation unit without
1024    C preprocessing. NOTE: str is modified by modifying the '\0' at the
1025    end */
1026 static void tcc_assemble_inline(TCCState *s1, char *str, int len, int global)
1027 {
1028     int saved_parse_flags;
1029     const int *saved_macro_ptr;
1030
1031     saved_parse_flags = parse_flags;
1032     saved_macro_ptr = macro_ptr;
1033
1034     tcc_open_bf(s1, ":asm:", len);
1035     memcpy(file->buffer, str, len);
1036
1037     macro_ptr = NULL;
1038     tcc_assemble_internal(s1, 0, global);
1039     tcc_close();
1040
1041     parse_flags = saved_parse_flags;
1042     set_idnum('.', (parse_flags & PARSE_FLAG_ASM_FILE) ? IS_ID : 0);
1043     macro_ptr = saved_macro_ptr;
1044 }
1045
1046 /* find a constraint by its number or id (gcc 3 extended
1047    syntax). return -1 if not found. Return in *pp in char after the
1048    constraint */
1049 ST_FUNC int find_constraint(ASMOperand *operands, int nb_operands,
1050                            const char *name, const char **pp)
1051 {
1052     int index;
1053     TokenSym *ts;
1054     const char *p;
1055
1056     if (isnum(*name)) {
1057         index = 0;
1058         while (isnum(*name)) {
1059             index = (index * 10) + (*name) - '0';
1060             name++;
1061         }
1062         if ((unsigned)index >= nb_operands)
1063             index = -1;
1064     } else if (*name == '[') {
1065         name++;
1066         p = strchr(name, ']');
1067         if (p) {
1068             ts = tok_alloc(name, p - name);
1069             for(index = 0; index < nb_operands; index++) {
1070                 if (operands[index].id == ts->tok)
1071                     goto found;
1072             }
1073             index = -1;
1074         found:
1075             name = p + 1;
1076         } else {
1077             index = -1;
1078         }
1079     } else {
1080         index = -1;
1081     }
1082     if (pp)
1083         *pp = name;
1084     return index;
1085 }
1086
1087 static void subst_asm_operands(ASMOperand *operands, int nb_operands,
1088                                CString *out_str, CString *in_str)
1089 {
1090     int c, index, modifier;
1091     const char *str;
1092     ASMOperand *op;
1093     SValue sv;
1094
1095     cstr_new(out_str);
1096     str = in_str->data;
1097     for(;;) {
1098         c = *str++;
1099         if (c == '%') {
1100             if (*str == '%') {
1101                 str++;
1102                 goto add_char;
1103             }
1104             modifier = 0;
1105             if (*str == 'c' || *str == 'n' ||
1106                 *str == 'b' || *str == 'w' || *str == 'h' || *str == 'k' ||
1107                 *str == 'q' ||
1108                 /* P in GCC would add "@PLT" to symbol refs in PIC mode,
1109                    and make literal operands not be decorated with '$'.  */
1110                 *str == 'P')
1111                 modifier = *str++;
1112             index = find_constraint(operands, nb_operands, str, &str);
1113             if (index < 0)
1114                 tcc_error("invalid operand reference after %%");
1115             op = &operands[index];
1116             sv = *op->vt;
1117             if (op->reg >= 0) {
1118                 sv.r = op->reg;
1119                 if ((op->vt->r & VT_VALMASK) == VT_LLOCAL && op->is_memory)
1120                     sv.r |= VT_LVAL;
1121             }
1122             subst_asm_operand(out_str, &sv, modifier);
1123         } else {
1124         add_char:
1125             cstr_ccat(out_str, c);
1126             if (c == '\0')
1127                 break;
1128         }
1129     }
1130 }
1131
1132
1133 static void parse_asm_operands(ASMOperand *operands, int *nb_operands_ptr,
1134                                int is_output)
1135 {
1136     ASMOperand *op;
1137     int nb_operands;
1138
1139     if (tok != ':') {
1140         nb_operands = *nb_operands_ptr;
1141         for(;;) {
1142             CString astr;
1143             if (nb_operands >= MAX_ASM_OPERANDS)
1144                 tcc_error("too many asm operands");
1145             op = &operands[nb_operands++];
1146             op->id = 0;
1147             if (tok == '[') {
1148                 next();
1149                 if (tok < TOK_IDENT)
1150                     expect("identifier");
1151                 op->id = tok;
1152                 next();
1153                 skip(']');
1154             }
1155             parse_mult_str(&astr, "string constant");
1156             op->constraint = tcc_malloc(astr.size);
1157             strcpy(op->constraint, astr.data);
1158             cstr_free(&astr);
1159             skip('(');
1160             gexpr();
1161             if (is_output) {
1162                 if (!(vtop->type.t & VT_ARRAY))
1163                     test_lvalue();
1164             } else {
1165                 /* we want to avoid LLOCAL case, except when the 'm'
1166                    constraint is used. Note that it may come from
1167                    register storage, so we need to convert (reg)
1168                    case */
1169                 if ((vtop->r & VT_LVAL) &&
1170                     ((vtop->r & VT_VALMASK) == VT_LLOCAL ||
1171                      (vtop->r & VT_VALMASK) < VT_CONST) &&
1172                     !strchr(op->constraint, 'm')) {
1173                     gv(RC_INT);
1174                 }
1175             }
1176             op->vt = vtop;
1177             skip(')');
1178             if (tok == ',') {
1179                 next();
1180             } else {
1181                 break;
1182             }
1183         }
1184         *nb_operands_ptr = nb_operands;
1185     }
1186 }
1187
1188 /* parse the GCC asm() instruction */
1189 ST_FUNC void asm_instr(void)
1190 {
1191     CString astr, astr1;
1192     ASMOperand operands[MAX_ASM_OPERANDS];
1193     int nb_outputs, nb_operands, i, must_subst, out_reg;
1194     uint8_t clobber_regs[NB_ASM_REGS];
1195
1196     next();
1197     /* since we always generate the asm() instruction, we can ignore
1198        volatile */
1199     if (tok == TOK_VOLATILE1 || tok == TOK_VOLATILE2 || tok == TOK_VOLATILE3) {
1200         next();
1201     }
1202     parse_asm_str(&astr);
1203     nb_operands = 0;
1204     nb_outputs = 0;
1205     must_subst = 0;
1206     memset(clobber_regs, 0, sizeof(clobber_regs));
1207     if (tok == ':') {
1208         next();
1209         must_subst = 1;
1210         /* output args */
1211         parse_asm_operands(operands, &nb_operands, 1);
1212         nb_outputs = nb_operands;
1213         if (tok == ':') {
1214             next();
1215             if (tok != ')') {
1216                 /* input args */
1217                 parse_asm_operands(operands, &nb_operands, 0);
1218                 if (tok == ':') {
1219                     /* clobber list */
1220                     /* XXX: handle registers */
1221                     next();
1222                     for(;;) {
1223                         if (tok != TOK_STR)
1224                             expect("string constant");
1225                         asm_clobber(clobber_regs, tokc.str.data);
1226                         next();
1227                         if (tok == ',') {
1228                             next();
1229                         } else {
1230                             break;
1231                         }
1232                     }
1233                 }
1234             }
1235         }
1236     }
1237     skip(')');
1238     /* NOTE: we do not eat the ';' so that we can restore the current
1239        token after the assembler parsing */
1240     if (tok != ';')
1241         expect("';'");
1242
1243     /* save all values in the memory */
1244     save_regs(0);
1245
1246     /* compute constraints */
1247     asm_compute_constraints(operands, nb_operands, nb_outputs,
1248                             clobber_regs, &out_reg);
1249
1250     /* substitute the operands in the asm string. No substitution is
1251        done if no operands (GCC behaviour) */
1252 #ifdef ASM_DEBUG
1253     printf("asm: \"%s\"\n", (char *)astr.data);
1254 #endif
1255     if (must_subst) {
1256         subst_asm_operands(operands, nb_operands, &astr1, &astr);
1257         cstr_free(&astr);
1258     } else {
1259         astr1 = astr;
1260     }
1261 #ifdef ASM_DEBUG
1262     printf("subst_asm: \"%s\"\n", (char *)astr1.data);
1263 #endif
1264
1265     /* generate loads */
1266     asm_gen_code(operands, nb_operands, nb_outputs, 0,
1267                  clobber_regs, out_reg);
1268
1269     /* assemble the string with tcc internal assembler */
1270     tcc_assemble_inline(tcc_state, astr1.data, astr1.size - 1, 0);
1271
1272     /* restore the current C token */
1273     next();
1274
1275     /* store the output values if needed */
1276     asm_gen_code(operands, nb_operands, nb_outputs, 1,
1277                  clobber_regs, out_reg);
1278
1279     /* free everything */
1280     for(i=0;i<nb_operands;i++) {
1281         ASMOperand *op;
1282         op = &operands[i];
1283         tcc_free(op->constraint);
1284         vpop();
1285     }
1286     cstr_free(&astr1);
1287 }
1288
1289 ST_FUNC void asm_global_instr(void)
1290 {
1291     CString astr;
1292     int saved_nocode_wanted = nocode_wanted;
1293
1294     /* Global asm blocks are always emitted.  */
1295     nocode_wanted = 0;
1296     next();
1297     parse_asm_str(&astr);
1298     skip(')');
1299     /* NOTE: we do not eat the ';' so that we can restore the current
1300        token after the assembler parsing */
1301     if (tok != ';')
1302         expect("';'");
1303
1304 #ifdef ASM_DEBUG
1305     printf("asm_global: \"%s\"\n", (char *)astr.data);
1306 #endif
1307     cur_text_section = text_section;
1308     ind = cur_text_section->data_offset;
1309
1310     /* assemble the string with tcc internal assembler */
1311     tcc_assemble_inline(tcc_state, astr.data, astr.size - 1, 1);
1312
1313     cur_text_section->data_offset = ind;
1314
1315     /* restore the current C token */
1316     next();
1317
1318     cstr_free(&astr);
1319     nocode_wanted = saved_nocode_wanted;
1320 }
1321 #endif /* CONFIG_TCC_ASM */