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, 2007 8 Free Software Foundation, Inc. 9 10 This file is part of libopcodes. 11 12 This library 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 3, or (at your option) 15 any later version. 16 17 It is distributed in the hope that it will be useful, but WITHOUT 18 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 19 or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public 20 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 27/* ??? Eventually more and more of this stuff can go to cpu-independent files. 28 Keep that in mind. */ 29 30#include "sysdep.h" 31#include <stdio.h> 32#include "ansidecl.h" 33#include "bfd.h" 34#include "symcat.h" 35#include "mt-desc.h" 36#include "mt-opc.h" 37#include "opintl.h" 38#include "xregex.h" 39#include "libiberty.h" 40#include "safe-ctype.h" 41 42#undef min 43#define min(a,b) ((a) < (b) ? (a) : (b)) 44#undef max 45#define max(a,b) ((a) > (b) ? (a) : (b)) 46 47static const char * parse_insn_normal 48 (CGEN_CPU_DESC, const CGEN_INSN *, const char **, CGEN_FIELDS *); 49 50/* -- assembler routines inserted here. */ 51 52/* -- asm.c */ 53/* Range checking for signed numbers. Returns 0 if acceptable 54 and 1 if the value is out of bounds for a signed quantity. */ 55 56static int 57signed_out_of_bounds (long val) 58{ 59 if ((val < -32768) || (val > 32767)) 60 return 1; 61 return 0; 62} 63 64static const char * 65parse_loopsize (CGEN_CPU_DESC cd, 66 const char **strp, 67 int opindex, 68 void *arg) 69{ 70 signed long * valuep = (signed long *) arg; 71 const char *errmsg; 72 bfd_reloc_code_real_type code = BFD_RELOC_NONE; 73 enum cgen_parse_operand_result result_type; 74 bfd_vma value; 75 76 /* Is it a control transfer instructions? */ 77 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_LOOPSIZE) 78 { 79 code = BFD_RELOC_MT_PCINSN8; 80 errmsg = cgen_parse_address (cd, strp, opindex, code, 81 & result_type, & value); 82 *valuep = value; 83 return errmsg; 84 } 85 86 abort (); 87} 88 89static const char * 90parse_imm16 (CGEN_CPU_DESC cd, 91 const char **strp, 92 int opindex, 93 void *arg) 94{ 95 signed long * valuep = (signed long *) arg; 96 const char *errmsg; 97 enum cgen_parse_operand_result result_type; 98 bfd_reloc_code_real_type code = BFD_RELOC_NONE; 99 bfd_vma value; 100 101 /* Is it a control transfer instructions? */ 102 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_IMM16O) 103 { 104 code = BFD_RELOC_16_PCREL; 105 errmsg = cgen_parse_address (cd, strp, opindex, code, 106 & result_type, & value); 107 if (errmsg == NULL) 108 { 109 if (signed_out_of_bounds (value)) 110 errmsg = _("Operand out of range. Must be between -32768 and 32767."); 111 } 112 *valuep = value; 113 return errmsg; 114 } 115 116 /* If it's not a control transfer instruction, then 117 we have to check for %OP relocating operators. */ 118 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_IMM16L) 119 ; 120 else if (strncmp (*strp, "%hi16", 5) == 0) 121 { 122 *strp += 5; 123 code = BFD_RELOC_HI16; 124 } 125 else if (strncmp (*strp, "%lo16", 5) == 0) 126 { 127 *strp += 5; 128 code = BFD_RELOC_LO16; 129 } 130 131 /* If we found a %OP relocating operator, then parse it as an address. 132 If not, we need to parse it as an integer, either signed or unsigned 133 depending on which operand type we have. */ 134 if (code != BFD_RELOC_NONE) 135 { 136 /* %OP relocating operator found. */ 137 errmsg = cgen_parse_address (cd, strp, opindex, code, 138 & result_type, & value); 139 if (errmsg == NULL) 140 { 141 switch (result_type) 142 { 143 case (CGEN_PARSE_OPERAND_RESULT_NUMBER): 144 if (code == BFD_RELOC_HI16) 145 value = (value >> 16) & 0xFFFF; 146 else if (code == BFD_RELOC_LO16) 147 value = value & 0xFFFF; 148 else 149 errmsg = _("Biiiig Trouble in parse_imm16!"); 150 break; 151 152 case (CGEN_PARSE_OPERAND_RESULT_QUEUED): 153 /* No special processing for this case. */ 154 break; 155 156 default: 157 errmsg = _("%operator operand is not a symbol"); 158 break; 159 } 160 } 161 *valuep = value; 162 } 163 else 164 { 165 /* Parse hex values like 0xffff as unsigned, and sign extend 166 them manually. */ 167 int parse_signed = (opindex == (CGEN_OPERAND_TYPE)MT_OPERAND_IMM16); 168 169 if ((*strp)[0] == '0' 170 && ((*strp)[1] == 'x' || (*strp)[1] == 'X')) 171 parse_signed = 0; 172 173 /* No relocating operator. Parse as an number. */ 174 if (parse_signed) 175 { 176 /* Parse as as signed integer. */ 177 178 errmsg = cgen_parse_signed_integer (cd, strp, opindex, valuep); 179 180 if (errmsg == NULL) 181 { 182#if 0 183 /* Manual range checking is needed for the signed case. */ 184 if (*valuep & 0x8000) 185 value = 0xffff0000 | *valuep; 186 else 187 value = *valuep; 188 189 if (signed_out_of_bounds (value)) 190 errmsg = _("Operand out of range. Must be between -32768 and 32767."); 191 /* Truncate to 16 bits. This is necessary 192 because cgen will have sign extended *valuep. */ 193 *valuep &= 0xFFFF; 194#endif 195 } 196 } 197 else 198 { 199 /* MT_OPERAND_IMM16Z. Parse as an unsigned integer. */ 200 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, (unsigned long *) valuep); 201 202 if (opindex == (CGEN_OPERAND_TYPE) MT_OPERAND_IMM16 203 && *valuep >= 0x8000 204 && *valuep <= 0xffff) 205 *valuep -= 0x10000; 206 } 207 } 208 209 return errmsg; 210} 211 212 213static const char * 214parse_dup (CGEN_CPU_DESC cd, 215 const char **strp, 216 int opindex, 217 unsigned long *valuep) 218{ 219 const char *errmsg = NULL; 220 221 if (strncmp (*strp, "dup", 3) == 0 || strncmp (*strp, "DUP", 3) == 0) 222 { 223 *strp += 3; 224 *valuep = 1; 225 } 226 else if (strncmp (*strp, "xx", 2) == 0 || strncmp (*strp, "XX", 2) == 0) 227 { 228 *strp += 2; 229 *valuep = 0; 230 } 231 else 232 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 233 234 return errmsg; 235} 236 237 238static const char * 239parse_ball (CGEN_CPU_DESC cd, 240 const char **strp, 241 int opindex, 242 unsigned long *valuep) 243{ 244 const char *errmsg = NULL; 245 246 if (strncmp (*strp, "all", 3) == 0 || strncmp (*strp, "ALL", 3) == 0) 247 { 248 *strp += 3; 249 *valuep = 1; 250 } 251 else if (strncmp (*strp, "one", 3) == 0 || strncmp (*strp, "ONE", 3) == 0) 252 { 253 *strp += 3; 254 *valuep = 0; 255 } 256 else 257 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 258 259 return errmsg; 260} 261 262static const char * 263parse_xmode (CGEN_CPU_DESC cd, 264 const char **strp, 265 int opindex, 266 unsigned long *valuep) 267{ 268 const char *errmsg = NULL; 269 270 if (strncmp (*strp, "pm", 2) == 0 || strncmp (*strp, "PM", 2) == 0) 271 { 272 *strp += 2; 273 *valuep = 1; 274 } 275 else if (strncmp (*strp, "xm", 2) == 0 || strncmp (*strp, "XM", 2) == 0) 276 { 277 *strp += 2; 278 *valuep = 0; 279 } 280 else 281 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 282 283 return errmsg; 284} 285 286static const char * 287parse_rc (CGEN_CPU_DESC cd, 288 const char **strp, 289 int opindex, 290 unsigned long *valuep) 291{ 292 const char *errmsg = NULL; 293 294 if (strncmp (*strp, "r", 1) == 0 || strncmp (*strp, "R", 1) == 0) 295 { 296 *strp += 1; 297 *valuep = 1; 298 } 299 else if (strncmp (*strp, "c", 1) == 0 || strncmp (*strp, "C", 1) == 0) 300 { 301 *strp += 1; 302 *valuep = 0; 303 } 304 else 305 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 306 307 return errmsg; 308} 309 310static const char * 311parse_cbrb (CGEN_CPU_DESC cd, 312 const char **strp, 313 int opindex, 314 unsigned long *valuep) 315{ 316 const char *errmsg = NULL; 317 318 if (strncmp (*strp, "rb", 2) == 0 || strncmp (*strp, "RB", 2) == 0) 319 { 320 *strp += 2; 321 *valuep = 1; 322 } 323 else if (strncmp (*strp, "cb", 2) == 0 || strncmp (*strp, "CB", 2) == 0) 324 { 325 *strp += 2; 326 *valuep = 0; 327 } 328 else 329 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 330 331 return errmsg; 332} 333 334static const char * 335parse_rbbc (CGEN_CPU_DESC cd, 336 const char **strp, 337 int opindex, 338 unsigned long *valuep) 339{ 340 const char *errmsg = NULL; 341 342 if (strncmp (*strp, "rt", 2) == 0 || strncmp (*strp, "RT", 2) == 0) 343 { 344 *strp += 2; 345 *valuep = 0; 346 } 347 else if (strncmp (*strp, "br1", 3) == 0 || strncmp (*strp, "BR1", 3) == 0) 348 { 349 *strp += 3; 350 *valuep = 1; 351 } 352 else if (strncmp (*strp, "br2", 3) == 0 || strncmp (*strp, "BR2", 3) == 0) 353 { 354 *strp += 3; 355 *valuep = 2; 356 } 357 else if (strncmp (*strp, "cs", 2) == 0 || strncmp (*strp, "CS", 2) == 0) 358 { 359 *strp += 2; 360 *valuep = 3; 361 } 362 else 363 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 364 365 return errmsg; 366} 367 368static const char * 369parse_type (CGEN_CPU_DESC cd, 370 const char **strp, 371 int opindex, 372 unsigned long *valuep) 373{ 374 const char *errmsg = NULL; 375 376 if (strncmp (*strp, "odd", 3) == 0 || strncmp (*strp, "ODD", 3) == 0) 377 { 378 *strp += 3; 379 *valuep = 0; 380 } 381 else if (strncmp (*strp, "even", 4) == 0 || strncmp (*strp, "EVEN", 4) == 0) 382 { 383 *strp += 4; 384 *valuep = 1; 385 } 386 else if (strncmp (*strp, "oe", 2) == 0 || strncmp (*strp, "OE", 2) == 0) 387 { 388 *strp += 2; 389 *valuep = 2; 390 } 391 else 392 errmsg = cgen_parse_unsigned_integer (cd, strp, opindex, valuep); 393 394 if ((errmsg == NULL) && (*valuep == 3)) 395 errmsg = _("invalid operand. type may have values 0,1,2 only."); 396 397 return errmsg; 398} 399 400/* -- dis.c */ 401 402const char * mt_cgen_parse_operand 403 (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *); 404 405/* Main entry point for operand parsing. 406 407 This function is basically just a big switch statement. Earlier versions 408 used tables to look up the function to use, but 409 - if the table contains both assembler and disassembler functions then 410 the disassembler contains much of the assembler and vice-versa, 411 - there's a lot of inlining possibilities as things grow, 412 - using a switch statement avoids the function call overhead. 413 414 This function could be moved into `parse_insn_normal', but keeping it 415 separate makes clear the interface between `parse_insn_normal' and each of 416 the handlers. */ 417 418const char * 419mt_cgen_parse_operand (CGEN_CPU_DESC cd, 420 int opindex, 421 const char ** strp, 422 CGEN_FIELDS * fields) 423{ 424 const char * errmsg = NULL; 425 /* Used by scalar operands that still need to be parsed. */ 426 long junk ATTRIBUTE_UNUSED; 427 428 switch (opindex) 429 { 430 case MT_OPERAND_A23 : 431 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_A23, (unsigned long *) (& fields->f_a23)); 432 break; 433 case MT_OPERAND_BALL : 434 errmsg = parse_ball (cd, strp, MT_OPERAND_BALL, (unsigned long *) (& fields->f_ball)); 435 break; 436 case MT_OPERAND_BALL2 : 437 errmsg = parse_ball (cd, strp, MT_OPERAND_BALL2, (unsigned long *) (& fields->f_ball2)); 438 break; 439 case MT_OPERAND_BANKADDR : 440 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_BANKADDR, (unsigned long *) (& fields->f_bankaddr)); 441 break; 442 case MT_OPERAND_BRC : 443 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_BRC, (unsigned long *) (& fields->f_brc)); 444 break; 445 case MT_OPERAND_BRC2 : 446 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_BRC2, (unsigned long *) (& fields->f_brc2)); 447 break; 448 case MT_OPERAND_CB1INCR : 449 errmsg = cgen_parse_signed_integer (cd, strp, MT_OPERAND_CB1INCR, (long *) (& fields->f_cb1incr)); 450 break; 451 case MT_OPERAND_CB1SEL : 452 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CB1SEL, (unsigned long *) (& fields->f_cb1sel)); 453 break; 454 case MT_OPERAND_CB2INCR : 455 errmsg = cgen_parse_signed_integer (cd, strp, MT_OPERAND_CB2INCR, (long *) (& fields->f_cb2incr)); 456 break; 457 case MT_OPERAND_CB2SEL : 458 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CB2SEL, (unsigned long *) (& fields->f_cb2sel)); 459 break; 460 case MT_OPERAND_CBRB : 461 errmsg = parse_cbrb (cd, strp, MT_OPERAND_CBRB, (unsigned long *) (& fields->f_cbrb)); 462 break; 463 case MT_OPERAND_CBS : 464 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CBS, (unsigned long *) (& fields->f_cbs)); 465 break; 466 case MT_OPERAND_CBX : 467 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CBX, (unsigned long *) (& fields->f_cbx)); 468 break; 469 case MT_OPERAND_CCB : 470 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CCB, (unsigned long *) (& fields->f_ccb)); 471 break; 472 case MT_OPERAND_CDB : 473 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CDB, (unsigned long *) (& fields->f_cdb)); 474 break; 475 case MT_OPERAND_CELL : 476 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CELL, (unsigned long *) (& fields->f_cell)); 477 break; 478 case MT_OPERAND_COLNUM : 479 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_COLNUM, (unsigned long *) (& fields->f_colnum)); 480 break; 481 case MT_OPERAND_CONTNUM : 482 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CONTNUM, (unsigned long *) (& fields->f_contnum)); 483 break; 484 case MT_OPERAND_CR : 485 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CR, (unsigned long *) (& fields->f_cr)); 486 break; 487 case MT_OPERAND_CTXDISP : 488 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_CTXDISP, (unsigned long *) (& fields->f_ctxdisp)); 489 break; 490 case MT_OPERAND_DUP : 491 errmsg = parse_dup (cd, strp, MT_OPERAND_DUP, (unsigned long *) (& fields->f_dup)); 492 break; 493 case MT_OPERAND_FBDISP : 494 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_FBDISP, (unsigned long *) (& fields->f_fbdisp)); 495 break; 496 case MT_OPERAND_FBINCR : 497 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_FBINCR, (unsigned long *) (& fields->f_fbincr)); 498 break; 499 case MT_OPERAND_FRDR : 500 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_dr); 501 break; 502 case MT_OPERAND_FRDRRR : 503 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_drrr); 504 break; 505 case MT_OPERAND_FRSR1 : 506 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_sr1); 507 break; 508 case MT_OPERAND_FRSR2 : 509 errmsg = cgen_parse_keyword (cd, strp, & mt_cgen_opval_h_spr, & fields->f_sr2); 510 break; 511 case MT_OPERAND_ID : 512 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ID, (unsigned long *) (& fields->f_id)); 513 break; 514 case MT_OPERAND_IMM16 : 515 errmsg = parse_imm16 (cd, strp, MT_OPERAND_IMM16, (long *) (& fields->f_imm16s)); 516 break; 517 case MT_OPERAND_IMM16L : 518 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_IMM16L, (unsigned long *) (& fields->f_imm16l)); 519 break; 520 case MT_OPERAND_IMM16O : 521 errmsg = parse_imm16 (cd, strp, MT_OPERAND_IMM16O, (unsigned long *) (& fields->f_imm16s)); 522 break; 523 case MT_OPERAND_IMM16Z : 524 errmsg = parse_imm16 (cd, strp, MT_OPERAND_IMM16Z, (unsigned long *) (& fields->f_imm16u)); 525 break; 526 case MT_OPERAND_INCAMT : 527 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_INCAMT, (unsigned long *) (& fields->f_incamt)); 528 break; 529 case MT_OPERAND_INCR : 530 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_INCR, (unsigned long *) (& fields->f_incr)); 531 break; 532 case MT_OPERAND_LENGTH : 533 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_LENGTH, (unsigned long *) (& fields->f_length)); 534 break; 535 case MT_OPERAND_LOOPSIZE : 536 errmsg = parse_loopsize (cd, strp, MT_OPERAND_LOOPSIZE, (unsigned long *) (& fields->f_loopo)); 537 break; 538 case MT_OPERAND_MASK : 539 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_MASK, (unsigned long *) (& fields->f_mask)); 540 break; 541 case MT_OPERAND_MASK1 : 542 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_MASK1, (unsigned long *) (& fields->f_mask1)); 543 break; 544 case MT_OPERAND_MODE : 545 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_MODE, (unsigned long *) (& fields->f_mode)); 546 break; 547 case MT_OPERAND_PERM : 548 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_PERM, (unsigned long *) (& fields->f_perm)); 549 break; 550 case MT_OPERAND_RBBC : 551 errmsg = parse_rbbc (cd, strp, MT_OPERAND_RBBC, (unsigned long *) (& fields->f_rbbc)); 552 break; 553 case MT_OPERAND_RC : 554 errmsg = parse_rc (cd, strp, MT_OPERAND_RC, (unsigned long *) (& fields->f_rc)); 555 break; 556 case MT_OPERAND_RC1 : 557 errmsg = parse_rc (cd, strp, MT_OPERAND_RC1, (unsigned long *) (& fields->f_rc1)); 558 break; 559 case MT_OPERAND_RC2 : 560 errmsg = parse_rc (cd, strp, MT_OPERAND_RC2, (unsigned long *) (& fields->f_rc2)); 561 break; 562 case MT_OPERAND_RC3 : 563 errmsg = parse_rc (cd, strp, MT_OPERAND_RC3, (unsigned long *) (& fields->f_rc3)); 564 break; 565 case MT_OPERAND_RCNUM : 566 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_RCNUM, (unsigned long *) (& fields->f_rcnum)); 567 break; 568 case MT_OPERAND_RDA : 569 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_RDA, (unsigned long *) (& fields->f_rda)); 570 break; 571 case MT_OPERAND_ROWNUM : 572 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ROWNUM, (unsigned long *) (& fields->f_rownum)); 573 break; 574 case MT_OPERAND_ROWNUM1 : 575 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ROWNUM1, (unsigned long *) (& fields->f_rownum1)); 576 break; 577 case MT_OPERAND_ROWNUM2 : 578 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_ROWNUM2, (unsigned long *) (& fields->f_rownum2)); 579 break; 580 case MT_OPERAND_SIZE : 581 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_SIZE, (unsigned long *) (& fields->f_size)); 582 break; 583 case MT_OPERAND_TYPE : 584 errmsg = parse_type (cd, strp, MT_OPERAND_TYPE, (unsigned long *) (& fields->f_type)); 585 break; 586 case MT_OPERAND_WR : 587 errmsg = cgen_parse_unsigned_integer (cd, strp, MT_OPERAND_WR, (unsigned long *) (& fields->f_wr)); 588 break; 589 case MT_OPERAND_XMODE : 590 errmsg = parse_xmode (cd, strp, MT_OPERAND_XMODE, (unsigned long *) (& fields->f_xmode)); 591 break; 592 593 default : 594 /* xgettext:c-format */ 595 fprintf (stderr, _("Unrecognized field %d while parsing.\n"), opindex); 596 abort (); 597 } 598 599 return errmsg; 600} 601 602cgen_parse_fn * const mt_cgen_parse_handlers[] = 603{ 604 parse_insn_normal, 605}; 606 607void 608mt_cgen_init_asm (CGEN_CPU_DESC cd) 609{ 610 mt_cgen_init_opcode_table (cd); 611 mt_cgen_init_ibld_table (cd); 612 cd->parse_handlers = & mt_cgen_parse_handlers[0]; 613 cd->parse_operand = mt_cgen_parse_operand; 614#ifdef CGEN_ASM_INIT_HOOK 615CGEN_ASM_INIT_HOOK 616#endif 617} 618 619 620 621/* Regex construction routine. 622 623 This translates an opcode syntax string into a regex string, 624 by replacing any non-character syntax element (such as an 625 opcode) with the pattern '.*' 626 627 It then compiles the regex and stores it in the opcode, for 628 later use by mt_cgen_assemble_insn 629 630 Returns NULL for success, an error message for failure. */ 631 632char * 633mt_cgen_build_insn_regex (CGEN_INSN *insn) 634{ 635 CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn); 636 const char *mnem = CGEN_INSN_MNEMONIC (insn); 637 char rxbuf[CGEN_MAX_RX_ELEMENTS]; 638 char *rx = rxbuf; 639 const CGEN_SYNTAX_CHAR_TYPE *syn; 640 int reg_err; 641 642 syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc)); 643 644 /* Mnemonics come first in the syntax string. */ 645 if (! CGEN_SYNTAX_MNEMONIC_P (* syn)) 646 return _("missing mnemonic in syntax string"); 647 ++syn; 648 649 /* Generate a case sensitive regular expression that emulates case 650 insensitive matching in the "C" locale. We cannot generate a case 651 insensitive regular expression because in Turkish locales, 'i' and 'I' 652 are not equal modulo case conversion. */ 653 654 /* Copy the literal mnemonic out of the insn. */ 655 for (; *mnem; mnem++) 656 { 657 char c = *mnem; 658 659 if (ISALPHA (c)) 660 { 661 *rx++ = '['; 662 *rx++ = TOLOWER (c); 663 *rx++ = TOUPPER (c); 664 *rx++ = ']'; 665 } 666 else 667 *rx++ = c; 668 } 669 670 /* Copy any remaining literals from the syntax string into the rx. */ 671 for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn) 672 { 673 if (CGEN_SYNTAX_CHAR_P (* syn)) 674 { 675 char c = CGEN_SYNTAX_CHAR (* syn); 676 677 switch (c) 678 { 679 /* Escape any regex metacharacters in the syntax. */ 680 case '.': case '[': case '\\': 681 case '*': case '^': case '$': 682 683#ifdef CGEN_ESCAPE_EXTENDED_REGEX 684 case '?': case '{': case '}': 685 case '(': case ')': case '*': 686 case '|': case '+': case ']': 687#endif 688 *rx++ = '\\'; 689 *rx++ = c; 690 break; 691 692 default: 693 if (ISALPHA (c)) 694 { 695 *rx++ = '['; 696 *rx++ = TOLOWER (c); 697 *rx++ = TOUPPER (c); 698 *rx++ = ']'; 699 } 700 else 701 *rx++ = c; 702 break; 703 } 704 } 705 else 706 { 707 /* Replace non-syntax fields with globs. */ 708 *rx++ = '.'; 709 *rx++ = '*'; 710 } 711 } 712 713 /* Trailing whitespace ok. */ 714 * rx++ = '['; 715 * rx++ = ' '; 716 * rx++ = '\t'; 717 * rx++ = ']'; 718 * rx++ = '*'; 719 720 /* But anchor it after that. */ 721 * rx++ = '$'; 722 * rx = '\0'; 723 724 CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t)); 725 reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB); 726 727 if (reg_err == 0) 728 return NULL; 729 else 730 { 731 static char msg[80]; 732 733 regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80); 734 regfree ((regex_t *) CGEN_INSN_RX (insn)); 735 free (CGEN_INSN_RX (insn)); 736 (CGEN_INSN_RX (insn)) = NULL; 737 return msg; 738 } 739} 740 741 742/* Default insn parser. 743 744 The syntax string is scanned and operands are parsed and stored in FIELDS. 745 Relocs are queued as we go via other callbacks. 746 747 ??? Note that this is currently an all-or-nothing parser. If we fail to 748 parse the instruction, we return 0 and the caller will start over from 749 the beginning. Backtracking will be necessary in parsing subexpressions, 750 but that can be handled there. Not handling backtracking here may get 751 expensive in the case of the m68k. Deal with later. 752 753 Returns NULL for success, an error message for failure. */ 754 755static const char * 756parse_insn_normal (CGEN_CPU_DESC cd, 757 const CGEN_INSN *insn, 758 const char **strp, 759 CGEN_FIELDS *fields) 760{ 761 /* ??? Runtime added insns not handled yet. */ 762 const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn); 763 const char *str = *strp; 764 const char *errmsg; 765 const char *p; 766 const CGEN_SYNTAX_CHAR_TYPE * syn; 767#ifdef CGEN_MNEMONIC_OPERANDS 768 /* FIXME: wip */ 769 int past_opcode_p; 770#endif 771 772 /* For now we assume the mnemonic is first (there are no leading operands). 773 We can parse it without needing to set up operand parsing. 774 GAS's input scrubber will ensure mnemonics are lowercase, but we may 775 not be called from GAS. */ 776 p = CGEN_INSN_MNEMONIC (insn); 777 while (*p && TOLOWER (*p) == TOLOWER (*str)) 778 ++p, ++str; 779 780 if (* p) 781 return _("unrecognized instruction"); 782 783#ifndef CGEN_MNEMONIC_OPERANDS 784 if (* str && ! ISSPACE (* str)) 785 return _("unrecognized instruction"); 786#endif 787 788 CGEN_INIT_PARSE (cd); 789 cgen_init_parse_operand (cd); 790#ifdef CGEN_MNEMONIC_OPERANDS 791 past_opcode_p = 0; 792#endif 793 794 /* We don't check for (*str != '\0') here because we want to parse 795 any trailing fake arguments in the syntax string. */ 796 syn = CGEN_SYNTAX_STRING (syntax); 797 798 /* Mnemonics come first for now, ensure valid string. */ 799 if (! CGEN_SYNTAX_MNEMONIC_P (* syn)) 800 abort (); 801 802 ++syn; 803 804 while (* syn != 0) 805 { 806 /* Non operand chars must match exactly. */ 807 if (CGEN_SYNTAX_CHAR_P (* syn)) 808 { 809 /* FIXME: While we allow for non-GAS callers above, we assume the 810 first char after the mnemonic part is a space. */ 811 /* FIXME: We also take inappropriate advantage of the fact that 812 GAS's input scrubber will remove extraneous blanks. */ 813 if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn))) 814 { 815#ifdef CGEN_MNEMONIC_OPERANDS 816 if (CGEN_SYNTAX_CHAR(* syn) == ' ') 817 past_opcode_p = 1; 818#endif 819 ++ syn; 820 ++ str; 821 } 822 else if (*str) 823 { 824 /* Syntax char didn't match. Can't be this insn. */ 825 static char msg [80]; 826 827 /* xgettext:c-format */ 828 sprintf (msg, _("syntax error (expected char `%c', found `%c')"), 829 CGEN_SYNTAX_CHAR(*syn), *str); 830 return msg; 831 } 832 else 833 { 834 /* Ran out of input. */ 835 static char msg [80]; 836 837 /* xgettext:c-format */ 838 sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"), 839 CGEN_SYNTAX_CHAR(*syn)); 840 return msg; 841 } 842 continue; 843 } 844 845 /* We have an operand of some sort. */ 846 errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn), 847 &str, fields); 848 if (errmsg) 849 return errmsg; 850 851 /* Done with this operand, continue with next one. */ 852 ++ syn; 853 } 854 855 /* If we're at the end of the syntax string, we're done. */ 856 if (* syn == 0) 857 { 858 /* FIXME: For the moment we assume a valid `str' can only contain 859 blanks now. IE: We needn't try again with a longer version of 860 the insn and it is assumed that longer versions of insns appear 861 before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3). */ 862 while (ISSPACE (* str)) 863 ++ str; 864 865 if (* str != '\0') 866 return _("junk at end of line"); /* FIXME: would like to include `str' */ 867 868 return NULL; 869 } 870 871 /* We couldn't parse it. */ 872 return _("unrecognized instruction"); 873} 874 875/* Main entry point. 876 This routine is called for each instruction to be assembled. 877 STR points to the insn to be assembled. 878 We assume all necessary tables have been initialized. 879 The assembled instruction, less any fixups, is stored in BUF. 880 Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value 881 still needs to be converted to target byte order, otherwise BUF is an array 882 of bytes in target byte order. 883 The result is a pointer to the insn's entry in the opcode table, 884 or NULL if an error occured (an error message will have already been 885 printed). 886 887 Note that when processing (non-alias) macro-insns, 888 this function recurses. 889 890 ??? It's possible to make this cpu-independent. 891 One would have to deal with a few minor things. 892 At this point in time doing so would be more of a curiosity than useful 893 [for example this file isn't _that_ big], but keeping the possibility in 894 mind helps keep the design clean. */ 895 896const CGEN_INSN * 897mt_cgen_assemble_insn (CGEN_CPU_DESC cd, 898 const char *str, 899 CGEN_FIELDS *fields, 900 CGEN_INSN_BYTES_PTR buf, 901 char **errmsg) 902{ 903 const char *start; 904 CGEN_INSN_LIST *ilist; 905 const char *parse_errmsg = NULL; 906 const char *insert_errmsg = NULL; 907 int recognized_mnemonic = 0; 908 909 /* Skip leading white space. */ 910 while (ISSPACE (* str)) 911 ++ str; 912 913 /* The instructions are stored in hashed lists. 914 Get the first in the list. */ 915 ilist = CGEN_ASM_LOOKUP_INSN (cd, str); 916 917 /* Keep looking until we find a match. */ 918 start = str; 919 for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist)) 920 { 921 const CGEN_INSN *insn = ilist->insn; 922 recognized_mnemonic = 1; 923 924#ifdef CGEN_VALIDATE_INSN_SUPPORTED 925 /* Not usually needed as unsupported opcodes 926 shouldn't be in the hash lists. */ 927 /* Is this insn supported by the selected cpu? */ 928 if (! mt_cgen_insn_supported (cd, insn)) 929 continue; 930#endif 931 /* If the RELAXED attribute is set, this is an insn that shouldn't be 932 chosen immediately. Instead, it is used during assembler/linker 933 relaxation if possible. */ 934 if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0) 935 continue; 936 937 str = start; 938 939 /* Skip this insn if str doesn't look right lexically. */ 940 if (CGEN_INSN_RX (insn) != NULL && 941 regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH) 942 continue; 943 944 /* Allow parse/insert handlers to obtain length of insn. */ 945 CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn); 946 947 parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields); 948 if (parse_errmsg != NULL) 949 continue; 950 951 /* ??? 0 is passed for `pc'. */ 952 insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf, 953 (bfd_vma) 0); 954 if (insert_errmsg != NULL) 955 continue; 956 957 /* It is up to the caller to actually output the insn and any 958 queued relocs. */ 959 return insn; 960 } 961 962 { 963 static char errbuf[150]; 964#ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS 965 const char *tmp_errmsg; 966 967 /* If requesting verbose error messages, use insert_errmsg. 968 Failing that, use parse_errmsg. */ 969 tmp_errmsg = (insert_errmsg ? insert_errmsg : 970 parse_errmsg ? parse_errmsg : 971 recognized_mnemonic ? 972 _("unrecognized form of instruction") : 973 _("unrecognized instruction")); 974 975 if (strlen (start) > 50) 976 /* xgettext:c-format */ 977 sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start); 978 else 979 /* xgettext:c-format */ 980 sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start); 981#else 982 if (strlen (start) > 50) 983 /* xgettext:c-format */ 984 sprintf (errbuf, _("bad instruction `%.50s...'"), start); 985 else 986 /* xgettext:c-format */ 987 sprintf (errbuf, _("bad instruction `%.50s'"), start); 988#endif 989 990 *errmsg = errbuf; 991 return NULL; 992 } 993} 994