1 /* Assembler interface for targets using CGEN. -*- C -*- 2 CGEN: Cpu tools GENerator 3 4 THIS FILE IS MACHINE GENERATED WITH CGEN. 5 - the resultant file is machine generated, cgen-asm.in isn't 6 7 Copyright 1996, 1997, 1998, 1999, 2000, 2001, 2005 8 Free Software Foundation, Inc. 9 10 This file is part of the GNU Binutils and GDB, the GNU debugger. 11 12 This program is free software; you can redistribute it and/or modify 13 it under the terms of the GNU General Public License as published by 14 the Free Software Foundation; either version 2, or (at your option) 15 any later version. 16 17 This program is distributed in the hope that it will be useful, 18 but WITHOUT ANY WARRANTY; without even the implied warranty of 19 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 20 GNU General Public License for more details. 21 22 You should have received a copy of the GNU General Public License 23 along with this program; if not, write to the Free Software Foundation, Inc., 24 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA. */ 25 26 /* ??? Eventually more and more of this stuff can go to cpu-independent files. 27 Keep that in mind. */ 28 29 #include "sysdep.h" 30 #include <stdio.h> 31 #include "ansidecl.h" 32 #include "bfd.h" 33 #include "symcat.h" 34 #include "mt-desc.h" 35 #include "mt-opc.h" 36 #include "opintl.h" 37 #include "xregex.h" 38 #include "libiberty.h" 39 #include "safe-ctype.h" 40 41 #undef min 42 #define min(a,b) ((a) < (b) ? (a) : (b)) 43 #undef max 44 #define max(a,b) ((a) > (b) ? (a) : (b)) 45 46 static const char * parse_insn_normal 47 (CGEN_CPU_DESC, const CGEN_INSN *, const char **, CGEN_FIELDS *); 48 49 /* -- assembler routines inserted here. */ 50 51 /* -- asm.c */ 52 /* Range checking for signed numbers. Returns 0 if acceptable 53 and 1 if the value is out of bounds for a signed quantity. */ 54 55 static int 56 signed_out_of_bounds (long val) 57 { 58 if ((val < -32768) || (val > 32767)) 59 return 1; 60 return 0; 61 } 62 63 static const char * 64 parse_loopsize (CGEN_CPU_DESC cd, 65 const char **strp, 66 int opindex, 67 void *arg) 68 { 69 signed long * valuep = (signed long *) arg; 70 const char *errmsg; 71 bfd_reloc_code_real_type code = BFD_RELOC_NONE; 72 enum cgen_parse_operand_result result_type; 73 bfd_vma value; 74 75 /* Is it a control transfer instructions? */ 76 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_LOOPSIZE) 77 { 78 code = BFD_RELOC_MT_PCINSN8; 79 errmsg = cgen_parse_address (cd, strp, opindex, code, 80 & result_type, & value); 81 *valuep = value; 82 return errmsg; 83 } 84 85 abort (); 86 } 87 88 static const char * 89 parse_imm16 (CGEN_CPU_DESC cd, 90 const char **strp, 91 int opindex, 92 void *arg) 93 { 94 signed long * valuep = (signed long *) arg; 95 const char *errmsg; 96 enum cgen_parse_operand_result result_type; 97 bfd_reloc_code_real_type code = BFD_RELOC_NONE; 98 bfd_vma value; 99 100 /* Is it a control transfer instructions? */ 101 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_IMM16O) 102 { 103 code = BFD_RELOC_16_PCREL; 104 errmsg = cgen_parse_address (cd, strp, opindex, code, 105 & result_type, & value); 106 if (errmsg == NULL) 107 { 108 if (signed_out_of_bounds (value)) 109 errmsg = _("Operand out of range. Must be between -32768 and 32767."); 110 } 111 *valuep = value; 112 return errmsg; 113 } 114 115 /* If it's not a control transfer instruction, then 116 we have to check for %OP relocating operators. */ 117 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_IMM16L) 118 ; 119 else if (strncmp (*strp, "%hi16", 5) == 0) 120 { 121 *strp += 5; 122 code = BFD_RELOC_HI16; 123 } 124 else if (strncmp (*strp, "%lo16", 5) == 0) 125 { 126 *strp += 5; 127 code = BFD_RELOC_LO16; 128 } 129 130 /* If we found a %OP relocating operator, then parse it as an address. 131 If not, we need to parse it as an integer, either signed or unsigned 132 depending on which operand type we have. */ 133 if (code != BFD_RELOC_NONE) 134 { 135 /* %OP relocating operator found. */ 136 errmsg = cgen_parse_address (cd, strp, opindex, code, 137 & result_type, & value); 138 if (errmsg == NULL) 139 { 140 switch (result_type) 141 { 142 case (CGEN_PARSE_OPERAND_RESULT_NUMBER): 143 if (code == BFD_RELOC_HI16) 144 value = (value >> 16) & 0xFFFF; 145 else if (code == BFD_RELOC_LO16) 146 value = value & 0xFFFF; 147 else 148 errmsg = _("Biiiig Trouble in parse_imm16!"); 149 break; 150 151 case (CGEN_PARSE_OPERAND_RESULT_QUEUED): 152 /* No special processing for this case. */ 153 break; 154 155 default: 156 errmsg = _("%operator operand is not a symbol"); 157 break; 158 } 159 } 160 *valuep = value; 161 } 162 else 163 { 164 /* Parse hex values like 0xffff as unsigned, and sign extend 165 them manually. */ 166 int parse_signed = (opindex == (CGEN_OPERAND_TYPE)MT_OPERAND_IMM16); 167 168 if ((*strp)[0] == '0' 169 && ((*strp)[1] == 'x' || (*strp)[1] == 'X')) 170 parse_signed = 0; 171 172 /* No relocating operator. Parse as an number. */ 173 if (parse_signed) 174 { 175 /* Parse as as signed integer. */ 176 177 errmsg = cgen_parse_signed_integer (cd, strp, opindex, valuep); 178 179 if (errmsg == NULL) 180 { 181 #if 0 182 /* Manual range checking is needed for the signed case. */ 183 if (*valuep & 0x8000) 184 value = 0xffff0000 | *valuep; 185 else 186 value = *valuep; 187 188 if (signed_out_of_bounds (value)) 189 errmsg = _("Operand out of range. Must be between -32768 and 32767."); 190 /* Truncate to 16 bits. This is necessary 191 because cgen will have sign extended *valuep. */ 192 *valuep &= 0xFFFF; 193 #endif 194 } 195 } 196 else 197 { 198 /* MT_OPERAND_IMM16Z. Parse as an unsigned integer. */ 199 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, (unsigned long *) valuep); 200 201 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_IMM16 202 && *valuep >= 0x8000 203 && *valuep <= 0xffff) 204 *valuep -= 0x10000; 205 } 206 } 207 208 return errmsg; 209 } 210 211 212 static const char * 213 parse_dup (CGEN_CPU_DESC cd, 214 const char **strp, 215 int opindex, 216 unsigned long *valuep) 217 { 218 const char *errmsg = NULL; 219 220 if (strncmp (*strp, "dup", 3) == 0 || strncmp (*strp, "DUP", 3) == 0) 221 { 222 *strp += 3; 223 *valuep = 1; 224 } 225 else if (strncmp (*strp, "xx", 2) == 0 || strncmp (*strp, "XX", 2) == 0) 226 { 227 *strp += 2; 228 *valuep = 0; 229 } 230 else 231 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 232 233 return errmsg; 234 } 235 236 237 static const char * 238 parse_ball (CGEN_CPU_DESC cd, 239 const char **strp, 240 int opindex, 241 unsigned long *valuep) 242 { 243 const char *errmsg = NULL; 244 245 if (strncmp (*strp, "all", 3) == 0 || strncmp (*strp, "ALL", 3) == 0) 246 { 247 *strp += 3; 248 *valuep = 1; 249 } 250 else if (strncmp (*strp, "one", 3) == 0 || strncmp (*strp, "ONE", 3) == 0) 251 { 252 *strp += 3; 253 *valuep = 0; 254 } 255 else 256 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 257 258 return errmsg; 259 } 260 261 static const char * 262 parse_xmode (CGEN_CPU_DESC cd, 263 const char **strp, 264 int opindex, 265 unsigned long *valuep) 266 { 267 const char *errmsg = NULL; 268 269 if (strncmp (*strp, "pm", 2) == 0 || strncmp (*strp, "PM", 2) == 0) 270 { 271 *strp += 2; 272 *valuep = 1; 273 } 274 else if (strncmp (*strp, "xm", 2) == 0 || strncmp (*strp, "XM", 2) == 0) 275 { 276 *strp += 2; 277 *valuep = 0; 278 } 279 else 280 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 281 282 return errmsg; 283 } 284 285 static const char * 286 parse_rc (CGEN_CPU_DESC cd, 287 const char **strp, 288 int opindex, 289 unsigned long *valuep) 290 { 291 const char *errmsg = NULL; 292 293 if (strncmp (*strp, "r", 1) == 0 || strncmp (*strp, "R", 1) == 0) 294 { 295 *strp += 1; 296 *valuep = 1; 297 } 298 else if (strncmp (*strp, "c", 1) == 0 || strncmp (*strp, "C", 1) == 0) 299 { 300 *strp += 1; 301 *valuep = 0; 302 } 303 else 304 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 305 306 return errmsg; 307 } 308 309 static const char * 310 parse_cbrb (CGEN_CPU_DESC cd, 311 const char **strp, 312 int opindex, 313 unsigned long *valuep) 314 { 315 const char *errmsg = NULL; 316 317 if (strncmp (*strp, "rb", 2) == 0 || strncmp (*strp, "RB", 2) == 0) 318 { 319 *strp += 2; 320 *valuep = 1; 321 } 322 else if (strncmp (*strp, "cb", 2) == 0 || strncmp (*strp, "CB", 2) == 0) 323 { 324 *strp += 2; 325 *valuep = 0; 326 } 327 else 328 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 329 330 return errmsg; 331 } 332 333 static const char * 334 parse_rbbc (CGEN_CPU_DESC cd, 335 const char **strp, 336 int opindex, 337 unsigned long *valuep) 338 { 339 const char *errmsg = NULL; 340 341 if (strncmp (*strp, "rt", 2) == 0 || strncmp (*strp, "RT", 2) == 0) 342 { 343 *strp += 2; 344 *valuep = 0; 345 } 346 else if (strncmp (*strp, "br1", 3) == 0 || strncmp (*strp, "BR1", 3) == 0) 347 { 348 *strp += 3; 349 *valuep = 1; 350 } 351 else if (strncmp (*strp, "br2", 3) == 0 || strncmp (*strp, "BR2", 3) == 0) 352 { 353 *strp += 3; 354 *valuep = 2; 355 } 356 else if (strncmp (*strp, "cs", 2) == 0 || strncmp (*strp, "CS", 2) == 0) 357 { 358 *strp += 2; 359 *valuep = 3; 360 } 361 else 362 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 363 364 return errmsg; 365 } 366 367 static const char * 368 parse_type (CGEN_CPU_DESC cd, 369 const char **strp, 370 int opindex, 371 unsigned long *valuep) 372 { 373 const char *errmsg = NULL; 374 375 if (strncmp (*strp, "odd", 3) == 0 || strncmp (*strp, "ODD", 3) == 0) 376 { 377 *strp += 3; 378 *valuep = 0; 379 } 380 else if (strncmp (*strp, "even", 4) == 0 || strncmp (*strp, "EVEN", 4) == 0) 381 { 382 *strp += 4; 383 *valuep = 1; 384 } 385 else if (strncmp (*strp, "oe", 2) == 0 || strncmp (*strp, "OE", 2) == 0) 386 { 387 *strp += 2; 388 *valuep = 2; 389 } 390 else 391 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 392 393 if ((errmsg == NULL) && (*valuep == 3)) 394 errmsg = _("invalid operand. type may have values 0,1,2 only."); 395 396 return errmsg; 397 } 398 399 /* -- dis.c */ 400 401 const char * mt_cgen_parse_operand 402 (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *); 403 404 /* Main entry point for operand parsing. 405 406 This function is basically just a big switch statement. Earlier versions 407 used tables to look up the function to use, but 408 - if the table contains both assembler and disassembler functions then 409 the disassembler contains much of the assembler and vice-versa, 410 - there's a lot of inlining possibilities as things grow, 411 - using a switch statement avoids the function call overhead. 412 413 This function could be moved into `parse_insn_normal', but keeping it 414 separate makes clear the interface between `parse_insn_normal' and each of 415 the handlers. */ 416 417 const char * 418 mt_cgen_parse_operand (CGEN_CPU_DESC cd, 419 int opindex, 420 const char ** strp, 421 CGEN_FIELDS * fields) 422 { 423 const char * errmsg = NULL; 424 /* Used by scalar operands that still need to be parsed. */ 425 long junk ATTRIBUTE_UNUSED; 426 427 switch (opindex) 428 { 429 case MT_OPERAND_A23 : 430 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_A23, (unsigned long *) (& fields->f_a23)); 431 break; 432 case MT_OPERAND_BALL : 433 errmsg = parse_ball (cd, strp, MT_OPERAND_BALL, (unsigned long *) (& fields->f_ball)); 434 break; 435 case MT_OPERAND_BALL2 : 436 errmsg = parse_ball (cd, strp, MT_OPERAND_BALL2, (unsigned long *) (& fields->f_ball2)); 437 break; 438 case MT_OPERAND_BANKADDR : 439 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_BANKADDR, (unsigned long *) (& fields->f_bankaddr)); 440 break; 441 case MT_OPERAND_BRC : 442 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_BRC, (unsigned long *) (& fields->f_brc)); 443 break; 444 case MT_OPERAND_BRC2 : 445 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_BRC2, (unsigned long *) (& fields->f_brc2)); 446 break; 447 case MT_OPERAND_CB1INCR : 448 errmsg = cgen_parse_signed_integer (cd, strp, MT_OPERAND_CB1INCR, (long *) (& fields->f_cb1incr)); 449 break; 450 case MT_OPERAND_CB1SEL : 451 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CB1SEL, (unsigned long *) (& fields->f_cb1sel)); 452 break; 453 case MT_OPERAND_CB2INCR : 454 errmsg = cgen_parse_signed_integer (cd, strp, MT_OPERAND_CB2INCR, (long *) (& fields->f_cb2incr)); 455 break; 456 case MT_OPERAND_CB2SEL : 457 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CB2SEL, (unsigned long *) (& fields->f_cb2sel)); 458 break; 459 case MT_OPERAND_CBRB : 460 errmsg = parse_cbrb (cd, strp, MT_OPERAND_CBRB, (unsigned long *) (& fields->f_cbrb)); 461 break; 462 case MT_OPERAND_CBS : 463 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CBS, (unsigned long *) (& fields->f_cbs)); 464 break; 465 case MT_OPERAND_CBX : 466 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CBX, (unsigned long *) (& fields->f_cbx)); 467 break; 468 case MT_OPERAND_CCB : 469 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CCB, (unsigned long *) (& fields->f_ccb)); 470 break; 471 case MT_OPERAND_CDB : 472 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CDB, (unsigned long *) (& fields->f_cdb)); 473 break; 474 case MT_OPERAND_CELL : 475 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CELL, (unsigned long *) (& fields->f_cell)); 476 break; 477 case MT_OPERAND_COLNUM : 478 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_COLNUM, (unsigned long *) (& fields->f_colnum)); 479 break; 480 case MT_OPERAND_CONTNUM : 481 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CONTNUM, (unsigned long *) (& fields->f_contnum)); 482 break; 483 case MT_OPERAND_CR : 484 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CR, (unsigned long *) (& fields->f_cr)); 485 break; 486 case MT_OPERAND_CTXDISP : 487 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CTXDISP, (unsigned long *) (& fields->f_ctxdisp)); 488 break; 489 case MT_OPERAND_DUP : 490 errmsg = parse_dup (cd, strp, MT_OPERAND_DUP, (unsigned long *) (& fields->f_dup)); 491 break; 492 case MT_OPERAND_FBDISP : 493 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_FBDISP, (unsigned long *) (& fields->f_fbdisp)); 494 break; 495 case MT_OPERAND_FBINCR : 496 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_FBINCR, (unsigned long *) (& fields->f_fbincr)); 497 break; 498 case MT_OPERAND_FRDR : 499 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_dr); 500 break; 501 case MT_OPERAND_FRDRRR : 502 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_drrr); 503 break; 504 case MT_OPERAND_FRSR1 : 505 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_sr1); 506 break; 507 case MT_OPERAND_FRSR2 : 508 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_sr2); 509 break; 510 case MT_OPERAND_ID : 511 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ID, (unsigned long *) (& fields->f_id)); 512 break; 513 case MT_OPERAND_IMM16 : 514 errmsg = parse_imm16 (cd, strp, MT_OPERAND_IMM16, (long *) (& fields->f_imm16s)); 515 break; 516 case MT_OPERAND_IMM16L : 517 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_IMM16L, (unsigned long *) (& fields->f_imm16l)); 518 break; 519 case MT_OPERAND_IMM16O : 520 errmsg = parse_imm16 (cd, strp, MT_OPERAND_IMM16O, (unsigned long *) (& fields->f_imm16s)); 521 break; 522 case MT_OPERAND_IMM16Z : 523 errmsg = parse_imm16 (cd, strp, MT_OPERAND_IMM16Z, (unsigned long *) (& fields->f_imm16u)); 524 break; 525 case MT_OPERAND_INCAMT : 526 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_INCAMT, (unsigned long *) (& fields->f_incamt)); 527 break; 528 case MT_OPERAND_INCR : 529 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_INCR, (unsigned long *) (& fields->f_incr)); 530 break; 531 case MT_OPERAND_LENGTH : 532 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_LENGTH, (unsigned long *) (& fields->f_length)); 533 break; 534 case MT_OPERAND_LOOPSIZE : 535 errmsg = parse_loopsize (cd, strp, MT_OPERAND_LOOPSIZE, (unsigned long *) (& fields->f_loopo)); 536 break; 537 case MT_OPERAND_MASK : 538 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_MASK, (unsigned long *) (& fields->f_mask)); 539 break; 540 case MT_OPERAND_MASK1 : 541 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_MASK1, (unsigned long *) (& fields->f_mask1)); 542 break; 543 case MT_OPERAND_MODE : 544 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_MODE, (unsigned long *) (& fields->f_mode)); 545 break; 546 case MT_OPERAND_PERM : 547 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_PERM, (unsigned long *) (& fields->f_perm)); 548 break; 549 case MT_OPERAND_RBBC : 550 errmsg = parse_rbbc (cd, strp, MT_OPERAND_RBBC, (unsigned long *) (& fields->f_rbbc)); 551 break; 552 case MT_OPERAND_RC : 553 errmsg = parse_rc (cd, strp, MT_OPERAND_RC, (unsigned long *) (& fields->f_rc)); 554 break; 555 case MT_OPERAND_RC1 : 556 errmsg = parse_rc (cd, strp, MT_OPERAND_RC1, (unsigned long *) (& fields->f_rc1)); 557 break; 558 case MT_OPERAND_RC2 : 559 errmsg = parse_rc (cd, strp, MT_OPERAND_RC2, (unsigned long *) (& fields->f_rc2)); 560 break; 561 case MT_OPERAND_RC3 : 562 errmsg = parse_rc (cd, strp, MT_OPERAND_RC3, (unsigned long *) (& fields->f_rc3)); 563 break; 564 case MT_OPERAND_RCNUM : 565 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_RCNUM, (unsigned long *) (& fields->f_rcnum)); 566 break; 567 case MT_OPERAND_RDA : 568 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_RDA, (unsigned long *) (& fields->f_rda)); 569 break; 570 case MT_OPERAND_ROWNUM : 571 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ROWNUM, (unsigned long *) (& fields->f_rownum)); 572 break; 573 case MT_OPERAND_ROWNUM1 : 574 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ROWNUM1, (unsigned long *) (& fields->f_rownum1)); 575 break; 576 case MT_OPERAND_ROWNUM2 : 577 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ROWNUM2, (unsigned long *) (& fields->f_rownum2)); 578 break; 579 case MT_OPERAND_SIZE : 580 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_SIZE, (unsigned long *) (& fields->f_size)); 581 break; 582 case MT_OPERAND_TYPE : 583 errmsg = parse_type (cd, strp, MT_OPERAND_TYPE, (unsigned long *) (& fields->f_type)); 584 break; 585 case MT_OPERAND_WR : 586 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_WR, (unsigned long *) (& fields->f_wr)); 587 break; 588 case MT_OPERAND_XMODE : 589 errmsg = parse_xmode (cd, strp, MT_OPERAND_XMODE, (unsigned long *) (& fields->f_xmode)); 590 break; 591 592 default : 593 /* xgettext:c-format */ 594 fprintf (stderr, _("Unrecognized field %d while parsing.\n"), opindex); 595 abort (); 596 } 597 598 return errmsg; 599 } 600 601 cgen_parse_fn * const mt_cgen_parse_handlers[] = 602 { 603 parse_insn_normal, 604 }; 605 606 void 607 mt_cgen_init_asm (CGEN_CPU_DESC cd) 608 { 609 mt_cgen_init_opcode_table (cd); 610 mt_cgen_init_ibld_table (cd); 611 cd->parse_handlers = & mt_cgen_parse_handlers[0]; 612 cd->parse_operand = mt_cgen_parse_operand; 613 } 614 615 616 617 /* Regex construction routine. 618 619 This translates an opcode syntax string into a regex string, 620 by replacing any non-character syntax element (such as an 621 opcode) with the pattern '.*' 622 623 It then compiles the regex and stores it in the opcode, for 624 later use by mt_cgen_assemble_insn 625 626 Returns NULL for success, an error message for failure. */ 627 628 char * 629 mt_cgen_build_insn_regex (CGEN_INSN *insn) 630 { 631 CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn); 632 const char *mnem = CGEN_INSN_MNEMONIC (insn); 633 char rxbuf[CGEN_MAX_RX_ELEMENTS]; 634 char *rx = rxbuf; 635 const CGEN_SYNTAX_CHAR_TYPE *syn; 636 int reg_err; 637 638 syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc)); 639 640 /* Mnemonics come first in the syntax string. */ 641 if (! CGEN_SYNTAX_MNEMONIC_P (* syn)) 642 return _("missing mnemonic in syntax string"); 643 ++syn; 644 645 /* Generate a case sensitive regular expression that emulates case 646 insensitive matching in the "C" locale. We cannot generate a case 647 insensitive regular expression because in Turkish locales, 'i' and 'I' 648 are not equal modulo case conversion. */ 649 650 /* Copy the literal mnemonic out of the insn. */ 651 for (; *mnem; mnem++) 652 { 653 char c = *mnem; 654 655 if (ISALPHA (c)) 656 { 657 *rx++ = '['; 658 *rx++ = TOLOWER (c); 659 *rx++ = TOUPPER (c); 660 *rx++ = ']'; 661 } 662 else 663 *rx++ = c; 664 } 665 666 /* Copy any remaining literals from the syntax string into the rx. */ 667 for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn) 668 { 669 if (CGEN_SYNTAX_CHAR_P (* syn)) 670 { 671 char c = CGEN_SYNTAX_CHAR (* syn); 672 673 switch (c) 674 { 675 /* Escape any regex metacharacters in the syntax. */ 676 case '.': case '[': case '\\': 677 case '*': case '^': case '$': 678 679 #ifdef CGEN_ESCAPE_EXTENDED_REGEX 680 case '?': case '{': case '}': 681 case '(': case ')': case '*': 682 case '|': case '+': case ']': 683 #endif 684 *rx++ = '\\'; 685 *rx++ = c; 686 break; 687 688 default: 689 if (ISALPHA (c)) 690 { 691 *rx++ = '['; 692 *rx++ = TOLOWER (c); 693 *rx++ = TOUPPER (c); 694 *rx++ = ']'; 695 } 696 else 697 *rx++ = c; 698 break; 699 } 700 } 701 else 702 { 703 /* Replace non-syntax fields with globs. */ 704 *rx++ = '.'; 705 *rx++ = '*'; 706 } 707 } 708 709 /* Trailing whitespace ok. */ 710 * rx++ = '['; 711 * rx++ = ' '; 712 * rx++ = '\t'; 713 * rx++ = ']'; 714 * rx++ = '*'; 715 716 /* But anchor it after that. */ 717 * rx++ = '$'; 718 * rx = '\0'; 719 720 CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t)); 721 reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB); 722 723 if (reg_err == 0) 724 return NULL; 725 else 726 { 727 static char msg[80]; 728 729 regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80); 730 regfree ((regex_t *) CGEN_INSN_RX (insn)); 731 free (CGEN_INSN_RX (insn)); 732 (CGEN_INSN_RX (insn)) = NULL; 733 return msg; 734 } 735 } 736 737 738 /* Default insn parser. 739 740 The syntax string is scanned and operands are parsed and stored in FIELDS. 741 Relocs are queued as we go via other callbacks. 742 743 ??? Note that this is currently an all-or-nothing parser. If we fail to 744 parse the instruction, we return 0 and the caller will start over from 745 the beginning. Backtracking will be necessary in parsing subexpressions, 746 but that can be handled there. Not handling backtracking here may get 747 expensive in the case of the m68k. Deal with later. 748 749 Returns NULL for success, an error message for failure. */ 750 751 static const char * 752 parse_insn_normal (CGEN_CPU_DESC cd, 753 const CGEN_INSN *insn, 754 const char **strp, 755 CGEN_FIELDS *fields) 756 { 757 /* ??? Runtime added insns not handled yet. */ 758 const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn); 759 const char *str = *strp; 760 const char *errmsg; 761 const char *p; 762 const CGEN_SYNTAX_CHAR_TYPE * syn; 763 #ifdef CGEN_MNEMONIC_OPERANDS 764 /* FIXME: wip */ 765 int past_opcode_p; 766 #endif 767 768 /* For now we assume the mnemonic is first (there are no leading operands). 769 We can parse it without needing to set up operand parsing. 770 GAS's input scrubber will ensure mnemonics are lowercase, but we may 771 not be called from GAS. */ 772 p = CGEN_INSN_MNEMONIC (insn); 773 while (*p && TOLOWER (*p) == TOLOWER (*str)) 774 ++p, ++str; 775 776 if (* p) 777 return _("unrecognized instruction"); 778 779 #ifndef CGEN_MNEMONIC_OPERANDS 780 if (* str && ! ISSPACE (* str)) 781 return _("unrecognized instruction"); 782 #endif 783 784 CGEN_INIT_PARSE (cd); 785 cgen_init_parse_operand (cd); 786 #ifdef CGEN_MNEMONIC_OPERANDS 787 past_opcode_p = 0; 788 #endif 789 790 /* We don't check for (*str != '\0') here because we want to parse 791 any trailing fake arguments in the syntax string. */ 792 syn = CGEN_SYNTAX_STRING (syntax); 793 794 /* Mnemonics come first for now, ensure valid string. */ 795 if (! CGEN_SYNTAX_MNEMONIC_P (* syn)) 796 abort (); 797 798 ++syn; 799 800 while (* syn != 0) 801 { 802 /* Non operand chars must match exactly. */ 803 if (CGEN_SYNTAX_CHAR_P (* syn)) 804 { 805 /* FIXME: While we allow for non-GAS callers above, we assume the 806 first char after the mnemonic part is a space. */ 807 /* FIXME: We also take inappropriate advantage of the fact that 808 GAS's input scrubber will remove extraneous blanks. */ 809 if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn))) 810 { 811 #ifdef CGEN_MNEMONIC_OPERANDS 812 if (CGEN_SYNTAX_CHAR(* syn) == ' ') 813 past_opcode_p = 1; 814 #endif 815 ++ syn; 816 ++ str; 817 } 818 else if (*str) 819 { 820 /* Syntax char didn't match. Can't be this insn. */ 821 static char msg [80]; 822 823 /* xgettext:c-format */ 824 sprintf (msg, _("syntax error (expected char `%c', found `%c')"), 825 CGEN_SYNTAX_CHAR(*syn), *str); 826 return msg; 827 } 828 else 829 { 830 /* Ran out of input. */ 831 static char msg [80]; 832 833 /* xgettext:c-format */ 834 sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"), 835 CGEN_SYNTAX_CHAR(*syn)); 836 return msg; 837 } 838 continue; 839 } 840 841 /* We have an operand of some sort. */ 842 errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn), 843 &str, fields); 844 if (errmsg) 845 return errmsg; 846 847 /* Done with this operand, continue with next one. */ 848 ++ syn; 849 } 850 851 /* If we're at the end of the syntax string, we're done. */ 852 if (* syn == 0) 853 { 854 /* FIXME: For the moment we assume a valid `str' can only contain 855 blanks now. IE: We needn't try again with a longer version of 856 the insn and it is assumed that longer versions of insns appear 857 before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3). */ 858 while (ISSPACE (* str)) 859 ++ str; 860 861 if (* str != '\0') 862 return _("junk at end of line"); /* FIXME: would like to include `str' */ 863 864 return NULL; 865 } 866 867 /* We couldn't parse it. */ 868 return _("unrecognized instruction"); 869 } 870 871 /* Main entry point. 872 This routine is called for each instruction to be assembled. 873 STR points to the insn to be assembled. 874 We assume all necessary tables have been initialized. 875 The assembled instruction, less any fixups, is stored in BUF. 876 Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value 877 still needs to be converted to target byte order, otherwise BUF is an array 878 of bytes in target byte order. 879 The result is a pointer to the insn's entry in the opcode table, 880 or NULL if an error occured (an error message will have already been 881 printed). 882 883 Note that when processing (non-alias) macro-insns, 884 this function recurses. 885 886 ??? It's possible to make this cpu-independent. 887 One would have to deal with a few minor things. 888 At this point in time doing so would be more of a curiosity than useful 889 [for example this file isn't _that_ big], but keeping the possibility in 890 mind helps keep the design clean. */ 891 892 const CGEN_INSN * 893 mt_cgen_assemble_insn (CGEN_CPU_DESC cd, 894 const char *str, 895 CGEN_FIELDS *fields, 896 CGEN_INSN_BYTES_PTR buf, 897 char **errmsg) 898 { 899 const char *start; 900 CGEN_INSN_LIST *ilist; 901 const char *parse_errmsg = NULL; 902 const char *insert_errmsg = NULL; 903 int recognized_mnemonic = 0; 904 905 /* Skip leading white space. */ 906 while (ISSPACE (* str)) 907 ++ str; 908 909 /* The instructions are stored in hashed lists. 910 Get the first in the list. */ 911 ilist = CGEN_ASM_LOOKUP_INSN (cd, str); 912 913 /* Keep looking until we find a match. */ 914 start = str; 915 for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist)) 916 { 917 const CGEN_INSN *insn = ilist->insn; 918 recognized_mnemonic = 1; 919 920 #ifdef CGEN_VALIDATE_INSN_SUPPORTED 921 /* Not usually needed as unsupported opcodes 922 shouldn't be in the hash lists. */ 923 /* Is this insn supported by the selected cpu? */ 924 if (! mt_cgen_insn_supported (cd, insn)) 925 continue; 926 #endif 927 /* If the RELAXED attribute is set, this is an insn that shouldn't be 928 chosen immediately. Instead, it is used during assembler/linker 929 relaxation if possible. */ 930 if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0) 931 continue; 932 933 str = start; 934 935 /* Skip this insn if str doesn't look right lexically. */ 936 if (CGEN_INSN_RX (insn) != NULL && 937 regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH) 938 continue; 939 940 /* Allow parse/insert handlers to obtain length of insn. */ 941 CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn); 942 943 parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields); 944 if (parse_errmsg != NULL) 945 continue; 946 947 /* ??? 0 is passed for `pc'. */ 948 insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf, 949 (bfd_vma) 0); 950 if (insert_errmsg != NULL) 951 continue; 952 953 /* It is up to the caller to actually output the insn and any 954 queued relocs. */ 955 return insn; 956 } 957 958 { 959 static char errbuf[150]; 960 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS 961 const char *tmp_errmsg; 962 963 /* If requesting verbose error messages, use insert_errmsg. 964 Failing that, use parse_errmsg. */ 965 tmp_errmsg = (insert_errmsg ? insert_errmsg : 966 parse_errmsg ? parse_errmsg : 967 recognized_mnemonic ? 968 _("unrecognized form of instruction") : 969 _("unrecognized instruction")); 970 971 if (strlen (start) > 50) 972 /* xgettext:c-format */ 973 sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start); 974 else 975 /* xgettext:c-format */ 976 sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start); 977 #else 978 if (strlen (start) > 50) 979 /* xgettext:c-format */ 980 sprintf (errbuf, _("bad instruction `%.50s...'"), start); 981 else 982 /* xgettext:c-format */ 983 sprintf (errbuf, _("bad instruction `%.50s'"), start); 984 #endif 985 986 *errmsg = errbuf; 987 return NULL; 988 } 989 } 990