op-i386.c

   1 /*
   2  *  i386 micro operations
   3  *
   4  *  Copyright (c) 2003 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 #include "exec-i386.h"
  21
  22 /* NOTE: data are not static to force relocation generation by GCC */
  23
  24 uint8_t parity_table[256] = {
  25     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  26     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  27     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  28     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  29     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  30     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  31     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  32     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  33     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  34     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  35     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  36     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  37     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  38     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  39     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  40     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  41     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  42     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  43     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  44     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  45     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  46     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  47     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  48     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  49     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  50     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  51     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  52     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  53     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  54     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  55     CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0,
  56     0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P,
  57 };
  58
  59 /* modulo 17 table */
  60 const uint8_t rclw_table[32] = {
  61     0, 1, 2, 3, 4, 5, 6, 7,
  62     8, 9,10,11,12,13,14,15,
  63    16, 0, 1, 2, 3, 4, 5, 6,
  64     7, 8, 9,10,11,12,13,14,
  65 };
  66
  67 /* modulo 9 table */
  68 const uint8_t rclb_table[32] = {
  69     0, 1, 2, 3, 4, 5, 6, 7,
  70     8, 0, 1, 2, 3, 4, 5, 6,
  71     7, 8, 0, 1, 2, 3, 4, 5,
  72     6, 7, 8, 0, 1, 2, 3, 4,
  73 };
  74
  75 #ifdef USE_X86LDOUBLE
  76 /* an array of Intel 80-bit FP constants, to be loaded via integer ops */
  77 typedef unsigned short f15ld[5];
  78 const f15ld f15rk[] =
  79 {
  80 /*0*/   {0x0000,0x0000,0x0000,0x0000,0x0000},
  81 /*1*/   {0x0000,0x0000,0x0000,0x8000,0x3fff},
  82 /*pi*/  {0xc235,0x2168,0xdaa2,0xc90f,0x4000},
  83 /*lg2*/ {0xf799,0xfbcf,0x9a84,0x9a20,0x3ffd},
  84 /*ln2*/ {0x79ac,0xd1cf,0x17f7,0xb172,0x3ffe},
  85 /*l2e*/ {0xf0bc,0x5c17,0x3b29,0xb8aa,0x3fff},
  86 /*l2t*/ {0x8afe,0xcd1b,0x784b,0xd49a,0x4000}
  87 };
  88 #else
  89 /* the same, 64-bit version */
  90 typedef unsigned short f15ld[4];
  91 const f15ld f15rk[] =
  92 {
  93 #ifndef WORDS_BIGENDIAN
  94 /*0*/   {0x0000,0x0000,0x0000,0x0000},
  95 /*1*/   {0x0000,0x0000,0x0000,0x3ff0},
  96 /*pi*/  {0x2d18,0x5444,0x21fb,0x4009},
  97 /*lg2*/ {0x79ff,0x509f,0x4413,0x3fd3},
  98 /*ln2*/ {0x39ef,0xfefa,0x2e42,0x3fe6},
  99 /*l2e*/ {0x82fe,0x652b,0x1547,0x3ff7},
 100 /*l2t*/ {0xa371,0x0979,0x934f,0x400a}
 101 #else
 102 /*0*/   {0x0000,0x0000,0x0000,0x0000},
 103 /*1*/   {0x3ff0,0x0000,0x0000,0x0000},
 104 /*pi*/  {0x4009,0x21fb,0x5444,0x2d18},
 105 /*lg2*/ {0x3fd3,0x4413,0x509f,0x79ff},
 106 /*ln2*/ {0x3fe6,0x2e42,0xfefa,0x39ef},
 107 /*l2e*/ {0x3ff7,0x1547,0x652b,0x82fe},
 108 /*l2t*/ {0x400a,0x934f,0x0979,0xa371}
 109 #endif
 110 };
 111 #endif
 112
 113 /* n must be a constant to be efficient */
 114 static inline int lshift(int x, int n)
 115 {
 116     if (n >= 0)
 117         return x << n;
 118     else
 119         return x >> (-n);
 120 }
 121
 122 /* we define the various pieces of code used by the JIT */
 123
 124 #define REG EAX
 125 #define REGNAME _EAX
 126 #include "opreg_template.h"
 127 #undef REG
 128 #undef REGNAME
 129
 130 #define REG ECX
 131 #define REGNAME _ECX
 132 #include "opreg_template.h"
 133 #undef REG
 134 #undef REGNAME
 135
 136 #define REG EDX
 137 #define REGNAME _EDX
 138 #include "opreg_template.h"
 139 #undef REG
 140 #undef REGNAME
 141
 142 #define REG EBX
 143 #define REGNAME _EBX
 144 #include "opreg_template.h"
 145 #undef REG
 146 #undef REGNAME
 147
 148 #define REG ESP
 149 #define REGNAME _ESP
 150 #include "opreg_template.h"
 151 #undef REG
 152 #undef REGNAME
 153
 154 #define REG EBP
 155 #define REGNAME _EBP
 156 #include "opreg_template.h"
 157 #undef REG
 158 #undef REGNAME
 159
 160 #define REG ESI
 161 #define REGNAME _ESI
 162 #include "opreg_template.h"
 163 #undef REG
 164 #undef REGNAME
 165
 166 #define REG EDI
 167 #define REGNAME _EDI
 168 #include "opreg_template.h"
 169 #undef REG
 170 #undef REGNAME
 171
 172 /* operations with flags */
 173
 174 void OPPROTO op_addl_T0_T1_cc(void)
 175 {
 176     CC_SRC = T0;
 177     T0 += T1;
 178     CC_DST = T0;
 179 }
 180
 181 void OPPROTO op_orl_T0_T1_cc(void)
 182 {
 183     T0 |= T1;
 184     CC_DST = T0;
 185 }
 186
 187 void OPPROTO op_andl_T0_T1_cc(void)
 188 {
 189     T0 &= T1;
 190     CC_DST = T0;
 191 }
 192
 193 void OPPROTO op_subl_T0_T1_cc(void)
 194 {
 195     CC_SRC = T0;
 196     T0 -= T1;
 197     CC_DST = T0;
 198 }
 199
 200 void OPPROTO op_xorl_T0_T1_cc(void)
 201 {
 202     T0 ^= T1;
 203     CC_DST = T0;
 204 }
 205
 206 void OPPROTO op_cmpl_T0_T1_cc(void)
 207 {
 208     CC_SRC = T0;
 209     CC_DST = T0 - T1;
 210 }
 211
 212 void OPPROTO op_negl_T0_cc(void)
 213 {
 214     CC_SRC = 0;
 215     T0 = -T0;
 216     CC_DST = T0;
 217 }
 218
 219 void OPPROTO op_incl_T0_cc(void)
 220 {
 221     CC_SRC = cc_table[CC_OP].compute_c();
 222     T0++;
 223     CC_DST = T0;
 224 }
 225
 226 void OPPROTO op_decl_T0_cc(void)
 227 {
 228     CC_SRC = cc_table[CC_OP].compute_c();
 229     T0--;
 230     CC_DST = T0;
 231 }
 232
 233 void OPPROTO op_testl_T0_T1_cc(void)
 234 {
 235     CC_DST = T0 & T1;
 236 }
 237
 238 /* operations without flags */
 239
 240 void OPPROTO op_addl_T0_T1(void)
 241 {
 242     T0 += T1;
 243 }
 244
 245 void OPPROTO op_orl_T0_T1(void)
 246 {
 247     T0 |= T1;
 248 }
 249
 250 void OPPROTO op_andl_T0_T1(void)
 251 {
 252     T0 &= T1;
 253 }
 254
 255 void OPPROTO op_subl_T0_T1(void)
 256 {
 257     T0 -= T1;
 258 }
 259
 260 void OPPROTO op_xorl_T0_T1(void)
 261 {
 262     T0 ^= T1;
 263 }
 264
 265 void OPPROTO op_negl_T0(void)
 266 {
 267     T0 = -T0;
 268 }
 269
 270 void OPPROTO op_incl_T0(void)
 271 {
 272     T0++;
 273 }
 274
 275 void OPPROTO op_decl_T0(void)
 276 {
 277     T0--;
 278 }
 279
 280 void OPPROTO op_notl_T0(void)
 281 {
 282     T0 = ~T0;
 283 }
 284
 285 void OPPROTO op_bswapl_T0(void)
 286 {
 287     T0 = bswap32(T0);
 288 }
 289
 290 /* multiply/divide */
 291 void OPPROTO op_mulb_AL_T0(void)
 292 {
 293     unsigned int res;
 294     res = (uint8_t)EAX * (uint8_t)T0;
 295     EAX = (EAX & 0xffff0000) | res;
 296     CC_SRC = (res & 0xff00);
 297 }
 298
 299 void OPPROTO op_imulb_AL_T0(void)
 300 {
 301     int res;
 302     res = (int8_t)EAX * (int8_t)T0;
 303     EAX = (EAX & 0xffff0000) | (res & 0xffff);
 304     CC_SRC = (res != (int8_t)res);
 305 }
 306
 307 void OPPROTO op_mulw_AX_T0(void)
 308 {
 309     unsigned int res;
 310     res = (uint16_t)EAX * (uint16_t)T0;
 311     EAX = (EAX & 0xffff0000) | (res & 0xffff);
 312     EDX = (EDX & 0xffff0000) | ((res >> 16) & 0xffff);
 313     CC_SRC = res >> 16;
 314 }
 315
 316 void OPPROTO op_imulw_AX_T0(void)
 317 {
 318     int res;
 319     res = (int16_t)EAX * (int16_t)T0;
 320     EAX = (EAX & 0xffff0000) | (res & 0xffff);
 321     EDX = (EDX & 0xffff0000) | ((res >> 16) & 0xffff);
 322     CC_SRC = (res != (int16_t)res);
 323 }
 324
 325 void OPPROTO op_mull_EAX_T0(void)
 326 {
 327     uint64_t res;
 328     res = (uint64_t)((uint32_t)EAX) * (uint64_t)((uint32_t)T0);
 329     EAX = res;
 330     EDX = res >> 32;
 331     CC_SRC = res >> 32;
 332 }
 333
 334 void OPPROTO op_imull_EAX_T0(void)
 335 {
 336     int64_t res;
 337     res = (int64_t)((int32_t)EAX) * (int64_t)((int32_t)T0);
 338     EAX = res;
 339     EDX = res >> 32;
 340     CC_SRC = (res != (int32_t)res);
 341 }
 342
 343 void OPPROTO op_imulw_T0_T1(void)
 344 {
 345     int res;
 346     res = (int16_t)T0 * (int16_t)T1;
 347     T0 = res;
 348     CC_SRC = (res != (int16_t)res);
 349 }
 350
 351 void OPPROTO op_imull_T0_T1(void)
 352 {
 353     int64_t res;
 354     res = (int64_t)((int32_t)T0) * (int64_t)((int32_t)T1);
 355     T0 = res;
 356     CC_SRC = (res != (int32_t)res);
 357 }
 358
 359 /* division, flags are undefined */
 360 /* XXX: add exceptions for overflow */
 361 void OPPROTO op_divb_AL_T0(void)
 362 {
 363     unsigned int num, den, q, r;
 364
 365     num = (EAX & 0xffff);
 366     den = (T0 & 0xff);
 367     if (den == 0)
 368         raise_exception(EXCP00_DIVZ);
 369     q = (num / den) & 0xff;
 370     r = (num % den) & 0xff;
 371     EAX = (EAX & 0xffff0000) | (r << 8) | q;
 372 }
 373
 374 void OPPROTO op_idivb_AL_T0(void)
 375 {
 376     int num, den, q, r;
 377
 378     num = (int16_t)EAX;
 379     den = (int8_t)T0;
 380     if (den == 0)
 381         raise_exception(EXCP00_DIVZ);
 382     q = (num / den) & 0xff;
 383     r = (num % den) & 0xff;
 384     EAX = (EAX & 0xffff0000) | (r << 8) | q;
 385 }
 386
 387 void OPPROTO op_divw_AX_T0(void)
 388 {
 389     unsigned int num, den, q, r;
 390
 391     num = (EAX & 0xffff) | ((EDX & 0xffff) << 16);
 392     den = (T0 & 0xffff);
 393     if (den == 0)
 394         raise_exception(EXCP00_DIVZ);
 395     q = (num / den) & 0xffff;
 396     r = (num % den) & 0xffff;
 397     EAX = (EAX & 0xffff0000) | q;
 398     EDX = (EDX & 0xffff0000) | r;
 399 }
 400
 401 void OPPROTO op_idivw_AX_T0(void)
 402 {
 403     int num, den, q, r;
 404
 405     num = (EAX & 0xffff) | ((EDX & 0xffff) << 16);
 406     den = (int16_t)T0;
 407     if (den == 0)
 408         raise_exception(EXCP00_DIVZ);
 409     q = (num / den) & 0xffff;
 410     r = (num % den) & 0xffff;
 411     EAX = (EAX & 0xffff0000) | q;
 412     EDX = (EDX & 0xffff0000) | r;
 413 }
 414
 415 void OPPROTO op_divl_EAX_T0(void)
 416 {
 417     unsigned int den, q, r;
 418     uint64_t num;
 419
 420     num = EAX | ((uint64_t)EDX << 32);
 421     den = T0;
 422     if (den == 0)
 423         raise_exception(EXCP00_DIVZ);
 424     q = (num / den);
 425     r = (num % den);
 426     EAX = q;
 427     EDX = r;
 428 }
 429
 430 void OPPROTO op_idivl_EAX_T0(void)
 431 {
 432     int den, q, r;
 433     int64_t num;
 434
 435     num = EAX | ((uint64_t)EDX << 32);
 436     den = T0;
 437     if (den == 0)
 438         raise_exception(EXCP00_DIVZ);
 439     q = (num / den);
 440     r = (num % den);
 441     EAX = q;
 442     EDX = r;
 443 }
 444
 445 /* constant load & misc op */
 446
 447 void OPPROTO op_movl_T0_im(void)
 448 {
 449     T0 = PARAM1;
 450 }
 451
 452 void OPPROTO op_addl_T0_im(void)
 453 {
 454     T0 += PARAM1;
 455 }
 456
 457 void OPPROTO op_andl_T0_ffff(void)
 458 {
 459     T0 = T0 & 0xffff;
 460 }
 461
 462 void OPPROTO op_movl_T0_T1(void)
 463 {
 464     T0 = T1;
 465 }
 466
 467 void OPPROTO op_movl_T1_im(void)
 468 {
 469     T1 = PARAM1;
 470 }
 471
 472 void OPPROTO op_addl_T1_im(void)
 473 {
 474     T1 += PARAM1;
 475 }
 476
 477 void OPPROTO op_movl_T1_A0(void)
 478 {
 479     T1 = A0;
 480 }
 481
 482 void OPPROTO op_movl_A0_im(void)
 483 {
 484     A0 = PARAM1;
 485 }
 486
 487 void OPPROTO op_addl_A0_im(void)
 488 {
 489     A0 += PARAM1;
 490 }
 491
 492 void OPPROTO op_andl_A0_ffff(void)
 493 {
 494     A0 = A0 & 0xffff;
 495 }
 496
 497 /* memory access */
 498
 499 void OPPROTO op_ldub_T0_A0(void)
 500 {
 501     T0 = ldub((uint8_t *)A0);
 502 }
 503
 504 void OPPROTO op_ldsb_T0_A0(void)
 505 {
 506     T0 = ldsb((int8_t *)A0);
 507 }
 508
 509 void OPPROTO op_lduw_T0_A0(void)
 510 {
 511     T0 = lduw((uint8_t *)A0);
 512 }
 513
 514 void OPPROTO op_ldsw_T0_A0(void)
 515 {
 516     T0 = ldsw((int8_t *)A0);
 517 }
 518
 519 void OPPROTO op_ldl_T0_A0(void)
 520 {
 521     T0 = ldl((uint8_t *)A0);
 522 }
 523
 524 void OPPROTO op_ldub_T1_A0(void)
 525 {
 526     T1 = ldub((uint8_t *)A0);
 527 }
 528
 529 void OPPROTO op_ldsb_T1_A0(void)
 530 {
 531     T1 = ldsb((int8_t *)A0);
 532 }
 533
 534 void OPPROTO op_lduw_T1_A0(void)
 535 {
 536     T1 = lduw((uint8_t *)A0);
 537 }
 538
 539 void OPPROTO op_ldsw_T1_A0(void)
 540 {
 541     T1 = ldsw((int8_t *)A0);
 542 }
 543
 544 void OPPROTO op_ldl_T1_A0(void)
 545 {
 546     T1 = ldl((uint8_t *)A0);
 547 }
 548
 549 void OPPROTO op_stb_T0_A0(void)
 550 {
 551     stb((uint8_t *)A0, T0);
 552 }
 553
 554 void OPPROTO op_stw_T0_A0(void)
 555 {
 556     stw((uint8_t *)A0, T0);
 557 }
 558
 559 void OPPROTO op_stl_T0_A0(void)
 560 {
 561     stl((uint8_t *)A0, T0);
 562 }
 563
 564 /* used for bit operations */
 565
 566 void OPPROTO op_add_bitw_A0_T1(void)
 567 {
 568     A0 += ((int32_t)T1 >> 4) << 1;
 569 }
 570
 571 void OPPROTO op_add_bitl_A0_T1(void)
 572 {
 573     A0 += ((int32_t)T1 >> 5) << 2;
 574 }
 575
 576 /* indirect jump */
 577
 578 void OPPROTO op_jmp_T0(void)
 579 {
 580     EIP = T0;
 581 }
 582
 583 void OPPROTO op_jmp_im(void)
 584 {
 585     EIP = PARAM1;
 586 }
 587
 588 void OPPROTO op_int_im(void)
 589 {
 590     EIP = PARAM1;
 591     raise_exception(EXCP0D_GPF);
 592 }
 593
 594 void OPPROTO op_int3(void)
 595 {
 596     EIP = PARAM1;
 597     raise_exception(EXCP03_INT3);
 598 }
 599
 600 void OPPROTO op_into(void)
 601 {
 602     int eflags;
 603     eflags = cc_table[CC_OP].compute_all();
 604     if (eflags & CC_O) {
 605         EIP = PARAM1;
 606         raise_exception(EXCP04_INTO);
 607     } else {
 608         EIP = PARAM2;
 609     }
 610 }
 611
 612 /* string ops */
 613
 614 #define ldul ldl
 615
 616 #define SHIFT 0
 617 #include "ops_template.h"
 618 #undef SHIFT
 619
 620 #define SHIFT 1
 621 #include "ops_template.h"
 622 #undef SHIFT
 623
 624 #define SHIFT 2
 625 #include "ops_template.h"
 626 #undef SHIFT
 627
 628 /* sign extend */
 629
 630 void OPPROTO op_movsbl_T0_T0(void)
 631 {
 632     T0 = (int8_t)T0;
 633 }
 634
 635 void OPPROTO op_movzbl_T0_T0(void)
 636 {
 637     T0 = (uint8_t)T0;
 638 }
 639
 640 void OPPROTO op_movswl_T0_T0(void)
 641 {
 642     T0 = (int16_t)T0;
 643 }
 644
 645 void OPPROTO op_movzwl_T0_T0(void)
 646 {
 647     T0 = (uint16_t)T0;
 648 }
 649
 650 void OPPROTO op_movswl_EAX_AX(void)
 651 {
 652     EAX = (int16_t)EAX;
 653 }
 654
 655 void OPPROTO op_movsbw_AX_AL(void)
 656 {
 657     EAX = (EAX & 0xffff0000) | ((int8_t)EAX & 0xffff);
 658 }
 659
 660 void OPPROTO op_movslq_EDX_EAX(void)
 661 {
 662     EDX = (int32_t)EAX >> 31;
 663 }
 664
 665 void OPPROTO op_movswl_DX_AX(void)
 666 {
 667     EDX = (EDX & 0xffff0000) | (((int16_t)EAX >> 15) & 0xffff);
 668 }
 669
 670 /* push/pop */
 671
 672 void op_pushl_T0(void)
 673 {
 674     uint32_t offset;
 675     offset = ESP - 4;
 676     stl((void *)offset, T0);
 677     /* modify ESP after to handle exceptions correctly */
 678     ESP = offset;
 679 }
 680
 681 void op_pushw_T0(void)
 682 {
 683     uint32_t offset;
 684     offset = ESP - 2;
 685     stw((void *)offset, T0);
 686     /* modify ESP after to handle exceptions correctly */
 687     ESP = offset;
 688 }
 689
 690 void op_pushl_ss32_T0(void)
 691 {
 692     uint32_t offset;
 693     offset = ESP - 4;
 694     stl(env->seg_cache[R_SS].base + offset, T0);
 695     /* modify ESP after to handle exceptions correctly */
 696     ESP = offset;
 697 }
 698
 699 void op_pushw_ss32_T0(void)
 700 {
 701     uint32_t offset;
 702     offset = ESP - 2;
 703     stw(env->seg_cache[R_SS].base + offset, T0);
 704     /* modify ESP after to handle exceptions correctly */
 705     ESP = offset;
 706 }
 707
 708 void op_pushl_ss16_T0(void)
 709 {
 710     uint32_t offset;
 711     offset = (ESP - 4) & 0xffff;
 712     stl(env->seg_cache[R_SS].base + offset, T0);
 713     /* modify ESP after to handle exceptions correctly */
 714     ESP = (ESP & ~0xffff) | offset;
 715 }
 716
 717 void op_pushw_ss16_T0(void)
 718 {
 719     uint32_t offset;
 720     offset = (ESP - 2) & 0xffff;
 721     stw(env->seg_cache[R_SS].base + offset, T0);
 722     /* modify ESP after to handle exceptions correctly */
 723     ESP = (ESP & ~0xffff) | offset;
 724 }
 725
 726 /* NOTE: ESP update is done after */
 727 void op_popl_T0(void)
 728 {
 729     T0 = ldl((void *)ESP);
 730 }
 731
 732 void op_popw_T0(void)
 733 {
 734     T0 = lduw((void *)ESP);
 735 }
 736
 737 void op_popl_ss32_T0(void)
 738 {
 739     T0 = ldl(env->seg_cache[R_SS].base + ESP);
 740 }
 741
 742 void op_popw_ss32_T0(void)
 743 {
 744     T0 = lduw(env->seg_cache[R_SS].base + ESP);
 745 }
 746
 747 void op_popl_ss16_T0(void)
 748 {
 749     T0 = ldl(env->seg_cache[R_SS].base + (ESP & 0xffff));
 750 }
 751
 752 void op_popw_ss16_T0(void)
 753 {
 754     T0 = lduw(env->seg_cache[R_SS].base + (ESP & 0xffff));
 755 }
 756
 757 void op_addl_ESP_4(void)
 758 {
 759     ESP += 4;
 760 }
 761
 762 void op_addl_ESP_2(void)
 763 {
 764     ESP += 2;
 765 }
 766
 767 void op_addw_ESP_4(void)
 768 {
 769     ESP = (ESP & ~0xffff) | ((ESP + 4) & 0xffff);
 770 }
 771
 772 void op_addw_ESP_2(void)
 773 {
 774     ESP = (ESP & ~0xffff) | ((ESP + 2) & 0xffff);
 775 }
 776
 777 void op_addl_ESP_im(void)
 778 {
 779     ESP += PARAM1;
 780 }
 781
 782 void op_addw_ESP_im(void)
 783 {
 784     ESP = (ESP & ~0xffff) | ((ESP + PARAM1) & 0xffff);
 785 }
 786
 787 /* rdtsc */
 788 #ifndef __i386__
 789 uint64_t emu_time;
 790 #endif
 791 void op_rdtsc(void)
 792 {
 793     uint64_t val;
 794 #ifdef __i386__
 795     asm("rdtsc" : "=A" (val));
 796 #else
 797     /* better than nothing: the time increases */
 798     val = emu_time++;
 799 #endif
 800     EAX = val;
 801     EDX = val >> 32;
 802 }
 803
 804 /* bcd */
 805
 806 /* XXX: exception */
 807 void OPPROTO op_aam(void)
 808 {
 809     int base = PARAM1;
 810     int al, ah;
 811     al = EAX & 0xff;
 812     ah = al / base;
 813     al = al % base;
 814     EAX = (EAX & ~0xffff) | al | (ah << 8);
 815     CC_DST = al;
 816 }
 817
 818 void OPPROTO op_aad(void)
 819 {
 820     int base = PARAM1;
 821     int al, ah;
 822     al = EAX & 0xff;
 823     ah = (EAX >> 8) & 0xff;
 824     al = ((ah * base) + al) & 0xff;
 825     EAX = (EAX & ~0xffff) | al;
 826     CC_DST = al;
 827 }
 828
 829 void OPPROTO op_aaa(void)
 830 {
 831     int icarry;
 832     int al, ah, af;
 833     int eflags;
 834
 835     eflags = cc_table[CC_OP].compute_all();
 836     af = eflags & CC_A;
 837     al = EAX & 0xff;
 838     ah = (EAX >> 8) & 0xff;
 839
 840     icarry = (al > 0xf9);
 841     if (((al & 0x0f) > 9 ) || af) {
 842         al = (al + 6) & 0x0f;
 843         ah = (ah + 1 + icarry) & 0xff;
 844         eflags |= CC_C | CC_A;
 845     } else {
 846         eflags &= ~(CC_C | CC_A);
 847         al &= 0x0f;
 848     }
 849     EAX = (EAX & ~0xffff) | al | (ah << 8);
 850     CC_SRC = eflags;
 851 }
 852
 853 void OPPROTO op_aas(void)
 854 {
 855     int icarry;
 856     int al, ah, af;
 857     int eflags;
 858
 859     eflags = cc_table[CC_OP].compute_all();
 860     af = eflags & CC_A;
 861     al = EAX & 0xff;
 862     ah = (EAX >> 8) & 0xff;
 863
 864     icarry = (al < 6);
 865     if (((al & 0x0f) > 9 ) || af) {
 866         al = (al - 6) & 0x0f;
 867         ah = (ah - 1 - icarry) & 0xff;
 868         eflags |= CC_C | CC_A;
 869     } else {
 870         eflags &= ~(CC_C | CC_A);
 871         al &= 0x0f;
 872     }
 873     EAX = (EAX & ~0xffff) | al | (ah << 8);
 874     CC_SRC = eflags;
 875 }
 876
 877 void OPPROTO op_daa(void)
 878 {
 879     int al, af, cf;
 880     int eflags;
 881
 882     eflags = cc_table[CC_OP].compute_all();
 883     cf = eflags & CC_C;
 884     af = eflags & CC_A;
 885     al = EAX & 0xff;
 886
 887     eflags = 0;
 888     if (((al & 0x0f) > 9 ) || af) {
 889         al = (al + 6) & 0xff;
 890         eflags |= CC_A;
 891     }
 892     if ((al > 0x9f) || cf) {
 893         al = (al + 0x60) & 0xff;
 894         eflags |= CC_C;
 895     }
 896     EAX = (EAX & ~0xff) | al;
 897     /* well, speed is not an issue here, so we compute the flags by hand */
 898     eflags |= (al == 0) << 6; /* zf */
 899     eflags |= parity_table[al]; /* pf */
 900     eflags |= (al & 0x80); /* sf */
 901     CC_SRC = eflags;
 902 }
 903
 904 void OPPROTO op_das(void)
 905 {
 906     int al, al1, af, cf;
 907     int eflags;
 908
 909     eflags = cc_table[CC_OP].compute_all();
 910     cf = eflags & CC_C;
 911     af = eflags & CC_A;
 912     al = EAX & 0xff;
 913
 914     eflags = 0;
 915     al1 = al;
 916     if (((al & 0x0f) > 9 ) || af) {
 917         eflags |= CC_A;
 918         if (al < 6 || cf)
 919             eflags |= CC_C;
 920         al = (al - 6) & 0xff;
 921     }
 922     if ((al1 > 0x99) || cf) {
 923         al = (al - 0x60) & 0xff;
 924         eflags |= CC_C;
 925     }
 926     EAX = (EAX & ~0xff) | al;
 927     /* well, speed is not an issue here, so we compute the flags by hand */
 928     eflags |= (al == 0) << 6; /* zf */
 929     eflags |= parity_table[al]; /* pf */
 930     eflags |= (al & 0x80); /* sf */
 931     CC_SRC = eflags;
 932 }
 933
 934 /* segment handling */
 935
 936 void load_seg(int seg_reg, int selector)
 937 {
 938     SegmentCache *sc;
 939     SegmentDescriptorTable *dt;
 940     int index;
 941     uint32_t e1, e2;
 942     uint8_t *ptr;
 943
 944     env->segs[seg_reg] = selector;
 945     sc = &env->seg_cache[seg_reg];
 946     if (env->vm86) {
 947         sc->base = (void *)(selector << 4);
 948         sc->limit = 0xffff;
 949         sc->seg_32bit = 0;
 950     } else {
 951         if (selector & 0x4)
 952             dt = &env->ldt;
 953         else
 954             dt = &env->gdt;
 955         index = selector & ~7;
 956         if ((index + 7) > dt->limit)
 957             raise_exception(EXCP0D_GPF);
 958         ptr = dt->base + index;
 959         e1 = ldl(ptr);
 960         e2 = ldl(ptr + 4);
 961         sc->base = (void *)((e1 >> 16) | ((e2 & 0xff) << 16) | (e2 & 0xff000000));
 962         sc->limit = (e1 & 0xffff) | (e2 & 0x000f0000);
 963         if (e2 & (1 << 23))
 964             sc->limit = (sc->limit << 12) | 0xfff;
 965         sc->seg_32bit = (e2 >> 22) & 1;
 966 #if 0
 967         fprintf(logfile, "load_seg: sel=0x%04x base=0x%08lx limit=0x%08lx seg_32bit=%d\n",
 968                 selector, (unsigned long)sc->base, sc->limit, sc->seg_32bit);
 969 #endif
 970     }
 971 }
 972
 973 void OPPROTO op_movl_seg_T0(void)
 974 {
 975     load_seg(PARAM1, T0 & 0xffff);
 976 }
 977
 978 void OPPROTO op_movl_T0_seg(void)
 979 {
 980     T0 = env->segs[PARAM1];
 981 }
 982
 983 void OPPROTO op_addl_A0_seg(void)
 984 {
 985     A0 += *(unsigned long *)((char *)env + PARAM1);
 986 }
 987
 988 /* flags handling */
 989
 990 /* slow jumps cases (compute x86 flags) */
 991 void OPPROTO op_jo_cc(void)
 992 {
 993     int eflags;
 994     eflags = cc_table[CC_OP].compute_all();
 995     if (eflags & CC_O)
 996         EIP = PARAM1;
 997     else
 998         EIP = PARAM2;
 999     FORCE_RET();
1000 }
1001
1002 void OPPROTO op_jb_cc(void)
1003 {
1004     if (cc_table[CC_OP].compute_c())
1005         EIP = PARAM1;
1006     else
1007         EIP = PARAM2;
1008     FORCE_RET();
1009 }
1010
1011 void OPPROTO op_jz_cc(void)
1012 {
1013     int eflags;
1014     eflags = cc_table[CC_OP].compute_all();
1015     if (eflags & CC_Z)
1016         EIP = PARAM1;
1017     else
1018         EIP = PARAM2;
1019     FORCE_RET();
1020 }
1021
1022 void OPPROTO op_jbe_cc(void)
1023 {
1024     int eflags;
1025     eflags = cc_table[CC_OP].compute_all();
1026     if (eflags & (CC_Z | CC_C))
1027         EIP = PARAM1;
1028     else
1029         EIP = PARAM2;
1030     FORCE_RET();
1031 }
1032
1033 void OPPROTO op_js_cc(void)
1034 {
1035     int eflags;
1036     eflags = cc_table[CC_OP].compute_all();
1037     if (eflags & CC_S)
1038         EIP = PARAM1;
1039     else
1040         EIP = PARAM2;
1041     FORCE_RET();
1042 }
1043
1044 void OPPROTO op_jp_cc(void)
1045 {
1046     int eflags;
1047     eflags = cc_table[CC_OP].compute_all();
1048     if (eflags & CC_P)
1049         EIP = PARAM1;
1050     else
1051         EIP = PARAM2;
1052     FORCE_RET();
1053 }
1054
1055 void OPPROTO op_jl_cc(void)
1056 {
1057     int eflags;
1058     eflags = cc_table[CC_OP].compute_all();
1059     if ((eflags ^ (eflags >> 4)) & 0x80)
1060         EIP = PARAM1;
1061     else
1062         EIP = PARAM2;
1063     FORCE_RET();
1064 }
1065
1066 void OPPROTO op_jle_cc(void)
1067 {
1068     int eflags;
1069     eflags = cc_table[CC_OP].compute_all();
1070     if (((eflags ^ (eflags >> 4)) & 0x80) || (eflags & CC_Z))
1071         EIP = PARAM1;
1072     else
1073         EIP = PARAM2;
1074     FORCE_RET();
1075 }
1076
1077 /* slow set cases (compute x86 flags) */
1078 void OPPROTO op_seto_T0_cc(void)
1079 {
1080     int eflags;
1081     eflags = cc_table[CC_OP].compute_all();
1082     T0 = (eflags >> 11) & 1;
1083 }
1084
1085 void OPPROTO op_setb_T0_cc(void)
1086 {
1087     T0 = cc_table[CC_OP].compute_c();
1088 }
1089
1090 void OPPROTO op_setz_T0_cc(void)
1091 {
1092     int eflags;
1093     eflags = cc_table[CC_OP].compute_all();
1094     T0 = (eflags >> 6) & 1;
1095 }
1096
1097 void OPPROTO op_setbe_T0_cc(void)
1098 {
1099     int eflags;
1100     eflags = cc_table[CC_OP].compute_all();
1101     T0 = (eflags & (CC_Z | CC_C)) != 0;
1102 }
1103
1104 void OPPROTO op_sets_T0_cc(void)
1105 {
1106     int eflags;
1107     eflags = cc_table[CC_OP].compute_all();
1108     T0 = (eflags >> 7) & 1;
1109 }
1110
1111 void OPPROTO op_setp_T0_cc(void)
1112 {
1113     int eflags;
1114     eflags = cc_table[CC_OP].compute_all();
1115     T0 = (eflags >> 2) & 1;
1116 }
1117
1118 void OPPROTO op_setl_T0_cc(void)
1119 {
1120     int eflags;
1121     eflags = cc_table[CC_OP].compute_all();
1122     T0 = ((eflags ^ (eflags >> 4)) >> 7) & 1;
1123 }
1124
1125 void OPPROTO op_setle_T0_cc(void)
1126 {
1127     int eflags;
1128     eflags = cc_table[CC_OP].compute_all();
1129     T0 = (((eflags ^ (eflags >> 4)) & 0x80) || (eflags & CC_Z)) != 0;
1130 }
1131
1132 void OPPROTO op_xor_T0_1(void)
1133 {
1134     T0 ^= 1;
1135 }
1136
1137 void OPPROTO op_set_cc_op(void)
1138 {
1139     CC_OP = PARAM1;
1140 }
1141
1142 void OPPROTO op_movl_eflags_T0(void)
1143 {
1144     CC_SRC = T0;
1145     DF = 1 - (2 * ((T0 >> 10) & 1));
1146 }
1147
1148 /* XXX: compute only O flag */
1149 void OPPROTO op_movb_eflags_T0(void)
1150 {
1151     int of;
1152     of = cc_table[CC_OP].compute_all() & CC_O;
1153     CC_SRC = T0 | of;
1154 }
1155
1156 void OPPROTO op_movl_T0_eflags(void)
1157 {
1158     T0 = cc_table[CC_OP].compute_all();
1159     T0 |= (DF & DIRECTION_FLAG);
1160 }
1161
1162 void OPPROTO op_cld(void)
1163 {
1164     DF = 1;
1165 }
1166
1167 void OPPROTO op_std(void)
1168 {
1169     DF = -1;
1170 }
1171
1172 void OPPROTO op_clc(void)
1173 {
1174     int eflags;
1175     eflags = cc_table[CC_OP].compute_all();
1176     eflags &= ~CC_C;
1177     CC_SRC = eflags;
1178 }
1179
1180 void OPPROTO op_stc(void)
1181 {
1182     int eflags;
1183     eflags = cc_table[CC_OP].compute_all();
1184     eflags |= CC_C;
1185     CC_SRC = eflags;
1186 }
1187
1188 void OPPROTO op_cmc(void)
1189 {
1190     int eflags;
1191     eflags = cc_table[CC_OP].compute_all();
1192     eflags ^= CC_C;
1193     CC_SRC = eflags;
1194 }
1195
1196 void OPPROTO op_salc(void)
1197 {
1198     int cf;
1199     cf = cc_table[CC_OP].compute_c();
1200     EAX = (EAX & ~0xff) | ((-cf) & 0xff);
1201 }
1202
1203 static int compute_all_eflags(void)
1204 {
1205     return CC_SRC;
1206 }
1207
1208 static int compute_c_eflags(void)
1209 {
1210     return CC_SRC & CC_C;
1211 }
1212
1213 static int compute_c_mul(void)
1214 {
1215     int cf;
1216     cf = (CC_SRC != 0);
1217     return cf;
1218 }
1219
1220 static int compute_all_mul(void)
1221 {
1222     int cf, pf, af, zf, sf, of;
1223     cf = (CC_SRC != 0);
1224     pf = 0; /* undefined */
1225     af = 0; /* undefined */
1226     zf = 0; /* undefined */
1227     sf = 0; /* undefined */
1228     of = cf << 11;
1229     return cf | pf | af | zf | sf | of;
1230 }
1231
1232 CCTable cc_table[CC_OP_NB] = {
1233     [CC_OP_DYNAMIC] = { /* should never happen */ },
1234
1235     [CC_OP_EFLAGS] = { compute_all_eflags, compute_c_eflags },
1236
1237     [CC_OP_MUL] = { compute_all_mul, compute_c_mul },
1238
1239     [CC_OP_ADDB] = { compute_all_addb, compute_c_addb },
1240     [CC_OP_ADDW] = { compute_all_addw, compute_c_addw  },
1241     [CC_OP_ADDL] = { compute_all_addl, compute_c_addl  },
1242
1243     [CC_OP_ADCB] = { compute_all_adcb, compute_c_adcb },
1244     [CC_OP_ADCW] = { compute_all_adcw, compute_c_adcw  },
1245     [CC_OP_ADCL] = { compute_all_adcl, compute_c_adcl  },
1246
1247     [CC_OP_SUBB] = { compute_all_subb, compute_c_subb  },
1248     [CC_OP_SUBW] = { compute_all_subw, compute_c_subw  },
1249     [CC_OP_SUBL] = { compute_all_subl, compute_c_subl  },
1250
1251     [CC_OP_SBBB] = { compute_all_sbbb, compute_c_sbbb  },
1252     [CC_OP_SBBW] = { compute_all_sbbw, compute_c_sbbw  },
1253     [CC_OP_SBBL] = { compute_all_sbbl, compute_c_sbbl  },
1254
1255     [CC_OP_LOGICB] = { compute_all_logicb, compute_c_logicb },
1256     [CC_OP_LOGICW] = { compute_all_logicw, compute_c_logicw },
1257     [CC_OP_LOGICL] = { compute_all_logicl, compute_c_logicl },
1258
1259     [CC_OP_INCB] = { compute_all_incb, compute_c_incl },
1260     [CC_OP_INCW] = { compute_all_incw, compute_c_incl },
1261     [CC_OP_INCL] = { compute_all_incl, compute_c_incl },
1262
1263     [CC_OP_DECB] = { compute_all_decb, compute_c_incl },
1264     [CC_OP_DECW] = { compute_all_decw, compute_c_incl },
1265     [CC_OP_DECL] = { compute_all_decl, compute_c_incl },
1266
1267     [CC_OP_SHLB] = { compute_all_shlb, compute_c_shll },
1268     [CC_OP_SHLW] = { compute_all_shlw, compute_c_shll },
1269     [CC_OP_SHLL] = { compute_all_shll, compute_c_shll },
1270
1271     [CC_OP_SARB] = { compute_all_sarb, compute_c_shll },
1272     [CC_OP_SARW] = { compute_all_sarw, compute_c_shll },
1273     [CC_OP_SARL] = { compute_all_sarl, compute_c_shll },
1274 };
1275
1276 /* floating point support */
1277
1278 #ifdef USE_X86LDOUBLE
1279 /* use long double functions */
1280 #define lrint lrintl
1281 #define llrint llrintl
1282 #define fabs fabsl
1283 #define sin sinl
1284 #define cos cosl
1285 #define sqrt sqrtl
1286 #define pow powl
1287 #define log logl
1288 #define tan tanl
1289 #define atan2 atan2l
1290 #define floor floorl
1291 #define ceil ceill
1292 #define rint rintl
1293 #endif
1294
1295 extern int lrint(CPU86_LDouble x);
1296 extern int64_t llrint(CPU86_LDouble x);
1297 extern CPU86_LDouble fabs(CPU86_LDouble x);
1298 extern CPU86_LDouble sin(CPU86_LDouble x);
1299 extern CPU86_LDouble cos(CPU86_LDouble x);
1300 extern CPU86_LDouble sqrt(CPU86_LDouble x);
1301 extern CPU86_LDouble pow(CPU86_LDouble, CPU86_LDouble);
1302 extern CPU86_LDouble log(CPU86_LDouble x);
1303 extern CPU86_LDouble tan(CPU86_LDouble x);
1304 extern CPU86_LDouble atan2(CPU86_LDouble, CPU86_LDouble);
1305 extern CPU86_LDouble floor(CPU86_LDouble x);
1306 extern CPU86_LDouble ceil(CPU86_LDouble x);
1307 extern CPU86_LDouble rint(CPU86_LDouble x);
1308
1309 #define RC_MASK         0xc00
1310 #define RC_NEAR         0x000
1311 #define RC_DOWN         0x400
1312 #define RC_UP           0x800
1313 #define RC_CHOP         0xc00
1314
1315 #define MAXTAN 9223372036854775808.0
1316
1317 #ifdef USE_X86LDOUBLE
1318
1319 /* only for x86 */
1320 typedef union {
1321     long double d;
1322     struct {
1323         unsigned long long lower;
1324         unsigned short upper;
1325     } l;
1326 } CPU86_LDoubleU;
1327
1328 /* the following deal with x86 long double-precision numbers */
1329 #define MAXEXPD 0x7fff
1330 #define EXPBIAS 16383
1331 #define EXPD(fp)        (fp.l.upper & 0x7fff)
1332 #define SIGND(fp)       ((fp.l.upper) & 0x8000)
1333 #define MANTD(fp)       (fp.l.lower)
1334 #define BIASEXPONENT(fp) fp.l.upper = (fp.l.upper & ~(0x7fff)) | EXPBIAS
1335
1336 #else
1337
1338 typedef union {
1339     double d;
1340 #ifndef WORDS_BIGENDIAN
1341     struct {
1342         unsigned long lower;
1343         long upper;
1344     } l;
1345 #else
1346     struct {
1347         long upper;
1348         unsigned long lower;
1349     } l;
1350 #endif
1351     long long ll;
1352 } CPU86_LDoubleU;
1353
1354 /* the following deal with IEEE double-precision numbers */
1355 #define MAXEXPD 0x7ff
1356 #define EXPBIAS 1023
1357 #define EXPD(fp)        (((fp.l.upper) >> 20) & 0x7FF)
1358 #define SIGND(fp)       ((fp.l.upper) & 0x80000000)
1359 #define MANTD(fp)       (fp.ll & ((1LL << 52) - 1))
1360 #define BIASEXPONENT(fp) fp.l.upper = (fp.l.upper & ~(0x7ff << 20)) | (EXPBIAS << 20)
1361 #endif
1362
1363 /* fp load FT0 */
1364
1365 void OPPROTO op_flds_FT0_A0(void)
1366 {
1367     FT0 = ldfl((void *)A0);
1368 }
1369
1370 void OPPROTO op_fldl_FT0_A0(void)
1371 {
1372     FT0 = ldfq((void *)A0);
1373 }
1374
1375 /* helpers are needed to avoid static constant reference. XXX: find a better way */
1376 #ifdef USE_INT_TO_FLOAT_HELPERS
1377
1378 void helper_fild_FT0_A0(void)
1379 {
1380     FT0 = (CPU86_LDouble)ldsw((void *)A0);
1381 }
1382
1383 void helper_fildl_FT0_A0(void)
1384 {
1385     FT0 = (CPU86_LDouble)((int32_t)ldl((void *)A0));
1386 }
1387
1388 void helper_fildll_FT0_A0(void)
1389 {
1390     FT0 = (CPU86_LDouble)((int64_t)ldq((void *)A0));
1391 }
1392
1393 void OPPROTO op_fild_FT0_A0(void)
1394 {
1395     helper_fild_FT0_A0();
1396 }
1397
1398 void OPPROTO op_fildl_FT0_A0(void)
1399 {
1400     helper_fildl_FT0_A0();
1401 }
1402
1403 void OPPROTO op_fildll_FT0_A0(void)
1404 {
1405     helper_fildll_FT0_A0();
1406 }
1407
1408 #else
1409
1410 void OPPROTO op_fild_FT0_A0(void)
1411 {
1412     FT0 = (CPU86_LDouble)ldsw((void *)A0);
1413 }
1414
1415 void OPPROTO op_fildl_FT0_A0(void)
1416 {
1417     FT0 = (CPU86_LDouble)((int32_t)ldl((void *)A0));
1418 }
1419
1420 void OPPROTO op_fildll_FT0_A0(void)
1421 {
1422     FT0 = (CPU86_LDouble)((int64_t)ldq((void *)A0));
1423 }
1424 #endif
1425
1426 /* fp load ST0 */
1427
1428 void OPPROTO op_flds_ST0_A0(void)
1429 {
1430     ST0 = ldfl((void *)A0);
1431 }
1432
1433 void OPPROTO op_fldl_ST0_A0(void)
1434 {
1435     ST0 = ldfq((void *)A0);
1436 }
1437
1438 #ifdef USE_X86LDOUBLE
1439 void OPPROTO op_fldt_ST0_A0(void)
1440 {
1441     ST0 = *(long double *)A0;
1442 }
1443 #else
1444 void helper_fldt_ST0_A0(void)
1445 {
1446     CPU86_LDoubleU temp;
1447     int upper, e;
1448     /* mantissa */
1449     upper = lduw((uint8_t *)A0 + 8);
1450     /* XXX: handle overflow ? */
1451     e = (upper & 0x7fff) - 16383 + EXPBIAS; /* exponent */
1452     e |= (upper >> 4) & 0x800; /* sign */
1453     temp.ll = ((ldq((void *)A0) >> 11) & ((1LL << 52) - 1)) | ((uint64_t)e << 52);
1454     ST0 = temp.d;
1455 }
1456
1457 void OPPROTO op_fldt_ST0_A0(void)
1458 {
1459     helper_fldt_ST0_A0();
1460 }
1461 #endif
1462
1463 /* helpers are needed to avoid static constant reference. XXX: find a better way */
1464 #ifdef USE_INT_TO_FLOAT_HELPERS
1465
1466 void helper_fild_ST0_A0(void)
1467 {
1468     ST0 = (CPU86_LDouble)ldsw((void *)A0);
1469 }
1470
1471 void helper_fildl_ST0_A0(void)
1472 {
1473     ST0 = (CPU86_LDouble)((int32_t)ldl((void *)A0));
1474 }
1475
1476 void helper_fildll_ST0_A0(void)
1477 {
1478     ST0 = (CPU86_LDouble)((int64_t)ldq((void *)A0));
1479 }
1480
1481 void OPPROTO op_fild_ST0_A0(void)
1482 {
1483     helper_fild_ST0_A0();
1484 }
1485
1486 void OPPROTO op_fildl_ST0_A0(void)
1487 {
1488     helper_fildl_ST0_A0();
1489 }
1490
1491 void OPPROTO op_fildll_ST0_A0(void)
1492 {
1493     helper_fildll_ST0_A0();
1494 }
1495
1496 #else
1497
1498 void OPPROTO op_fild_ST0_A0(void)
1499 {
1500     ST0 = (CPU86_LDouble)ldsw((void *)A0);
1501 }
1502
1503 void OPPROTO op_fildl_ST0_A0(void)
1504 {
1505     ST0 = (CPU86_LDouble)((int32_t)ldl((void *)A0));
1506 }
1507
1508 void OPPROTO op_fildll_ST0_A0(void)
1509 {
1510     ST0 = (CPU86_LDouble)((int64_t)ldq((void *)A0));
1511 }
1512
1513 #endif
1514
1515 /* fp store */
1516
1517 void OPPROTO op_fsts_ST0_A0(void)
1518 {
1519     stfl((void *)A0, (float)ST0);
1520 }
1521
1522 void OPPROTO op_fstl_ST0_A0(void)
1523 {
1524     stfq((void *)A0, (double)ST0);
1525 }
1526
1527 #ifdef USE_X86LDOUBLE
1528 void OPPROTO op_fstt_ST0_A0(void)
1529 {
1530     *(long double *)A0 = ST0;
1531 }
1532 #else
1533 void helper_fstt_ST0_A0(void)
1534 {
1535     CPU86_LDoubleU temp;
1536     int e;
1537     temp.d = ST0;
1538     /* mantissa */
1539     stq((void *)A0, (MANTD(temp) << 11) | (1LL << 63));
1540     /* exponent + sign */
1541     e = EXPD(temp) - EXPBIAS + 16383;
1542     e |= SIGND(temp) >> 16;
1543     stw((uint8_t *)A0 + 8, e);
1544 }
1545
1546 void OPPROTO op_fstt_ST0_A0(void)
1547 {
1548     helper_fstt_ST0_A0();
1549 }
1550 #endif
1551
1552 void OPPROTO op_fist_ST0_A0(void)
1553 {
1554     int val;
1555     val = lrint(ST0);
1556     stw((void *)A0, val);
1557 }
1558
1559 void OPPROTO op_fistl_ST0_A0(void)
1560 {
1561     int val;
1562     val = lrint(ST0);
1563     stl((void *)A0, val);
1564 }
1565
1566 void OPPROTO op_fistll_ST0_A0(void)
1567 {
1568     int64_t val;
1569     val = llrint(ST0);
1570     stq((void *)A0, val);
1571 }
1572
1573 /* BCD ops */
1574
1575 #define MUL10(iv) ( iv + iv + (iv << 3) )
1576
1577 void helper_fbld_ST0_A0(void)
1578 {
1579     uint8_t *seg;
1580     CPU86_LDouble fpsrcop;
1581     int m32i;
1582     unsigned int v;
1583
1584     /* in this code, seg/m32i will be used as temporary ptr/int */
1585     seg = (uint8_t *)A0 + 8;
1586     v = ldub(seg--);
1587     /* XXX: raise exception */
1588     if (v != 0)
1589         return;
1590     v = ldub(seg--);
1591     /* XXX: raise exception */
1592     if ((v & 0xf0) != 0)
1593         return;
1594     m32i = v;  /* <-- d14 */
1595     v = ldub(seg--);
1596     m32i = MUL10(m32i) + (v >> 4);  /* <-- val * 10 + d13 */
1597     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d12 */
1598     v = ldub(seg--);
1599     m32i = MUL10(m32i) + (v >> 4);  /* <-- val * 10 + d11 */
1600     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d10 */
1601     v = ldub(seg--);
1602     m32i = MUL10(m32i) + (v >> 4);  /* <-- val * 10 + d9 */
1603     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d8 */
1604     fpsrcop = ((CPU86_LDouble)m32i) * 100000000.0;
1605
1606     v = ldub(seg--);
1607     m32i = (v >> 4);  /* <-- d7 */
1608     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d6 */
1609     v = ldub(seg--);
1610     m32i = MUL10(m32i) + (v >> 4);  /* <-- val * 10 + d5 */
1611     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d4 */
1612     v = ldub(seg--);
1613     m32i = MUL10(m32i) + (v >> 4);  /* <-- val * 10 + d3 */
1614     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d2 */
1615     v = ldub(seg);
1616     m32i = MUL10(m32i) + (v >> 4);  /* <-- val * 10 + d1 */
1617     m32i = MUL10(m32i) + (v & 0xf); /* <-- val * 10 + d0 */
1618     fpsrcop += ((CPU86_LDouble)m32i);
1619     if ( ldub(seg+9) & 0x80 )
1620         fpsrcop = -fpsrcop;
1621     ST0 = fpsrcop;
1622 }
1623
1624 void OPPROTO op_fbld_ST0_A0(void)
1625 {
1626     helper_fbld_ST0_A0();
1627 }
1628
1629 void helper_fbst_ST0_A0(void)
1630 {
1631     CPU86_LDouble fptemp;
1632     CPU86_LDouble fpsrcop;
1633     int v;
1634     uint8_t *mem_ref, *mem_end;
1635
1636     fpsrcop = rint(ST0);
1637     mem_ref = (uint8_t *)A0;
1638     mem_end = mem_ref + 8;
1639     if ( fpsrcop < 0.0 ) {
1640         stw(mem_end, 0x8000);
1641         fpsrcop = -fpsrcop;
1642     } else {
1643         stw(mem_end, 0x0000);
1644     }
1645     while (mem_ref < mem_end) {
1646         if (fpsrcop == 0.0)
1647             break;
1648         fptemp = floor(fpsrcop/10.0);
1649         v = ((int)(fpsrcop - fptemp*10.0));
1650         if  (fptemp == 0.0)  {
1651             stb(mem_ref++, v);
1652             break;
1653         }
1654         fpsrcop = fptemp;
1655         fptemp = floor(fpsrcop/10.0);
1656         v |= (((int)(fpsrcop - fptemp*10.0)) << 4);
1657         stb(mem_ref++, v);
1658         fpsrcop = fptemp;
1659     }
1660     while (mem_ref < mem_end) {
1661         stb(mem_ref++, 0);
1662     }
1663 }
1664
1665 void OPPROTO op_fbst_ST0_A0(void)
1666 {
1667     helper_fbst_ST0_A0();
1668 }
1669
1670 /* FPU move */
1671
1672 static inline void fpush(void)
1673 {
1674     env->fpstt = (env->fpstt - 1) & 7;
1675     env->fptags[env->fpstt] = 0; /* validate stack entry */
1676 }
1677
1678 static inline void fpop(void)
1679 {
1680     env->fptags[env->fpstt] = 1; /* invvalidate stack entry */
1681     env->fpstt = (env->fpstt + 1) & 7;
1682 }
1683
1684 void OPPROTO op_fpush(void)
1685 {
1686     fpush();
1687 }
1688
1689 void OPPROTO op_fpop(void)
1690 {
1691     fpop();
1692 }
1693
1694 void OPPROTO op_fdecstp(void)
1695 {
1696     env->fpstt = (env->fpstt - 1) & 7;
1697     env->fpus &= (~0x4700);
1698 }
1699
1700 void OPPROTO op_fincstp(void)
1701 {
1702     env->fpstt = (env->fpstt + 1) & 7;
1703     env->fpus &= (~0x4700);
1704 }
1705
1706 void OPPROTO op_fmov_ST0_FT0(void)
1707 {
1708     ST0 = FT0;
1709 }
1710
1711 void OPPROTO op_fmov_FT0_STN(void)
1712 {
1713     FT0 = ST(PARAM1);
1714 }
1715
1716 void OPPROTO op_fmov_ST0_STN(void)
1717 {
1718     ST0 = ST(PARAM1);
1719 }
1720
1721 void OPPROTO op_fmov_STN_ST0(void)
1722 {
1723     ST(PARAM1) = ST0;
1724 }
1725
1726 void OPPROTO op_fxchg_ST0_STN(void)
1727 {
1728     CPU86_LDouble tmp;
1729     tmp = ST(PARAM1);
1730     ST(PARAM1) = ST0;
1731     ST0 = tmp;
1732 }
1733
1734 /* FPU operations */
1735
1736 /* XXX: handle nans */
1737 void OPPROTO op_fcom_ST0_FT0(void)
1738 {
1739     env->fpus &= (~0x4500);     /* (C3,C2,C0) <-- 000 */
1740     if (ST0 < FT0)
1741         env->fpus |= 0x100;     /* (C3,C2,C0) <-- 001 */
1742     else if (ST0 == FT0)
1743         env->fpus |= 0x4000; /* (C3,C2,C0) <-- 100 */
1744     FORCE_RET();
1745 }
1746
1747 /* XXX: handle nans */
1748 void OPPROTO op_fucom_ST0_FT0(void)
1749 {
1750     env->fpus &= (~0x4500);     /* (C3,C2,C0) <-- 000 */
1751     if (ST0 < FT0)
1752         env->fpus |= 0x100;     /* (C3,C2,C0) <-- 001 */
1753     else if (ST0 == FT0)
1754         env->fpus |= 0x4000; /* (C3,C2,C0) <-- 100 */
1755     FORCE_RET();
1756 }
1757
1758 void OPPROTO op_fadd_ST0_FT0(void)
1759 {
1760     ST0 += FT0;
1761 }
1762
1763 void OPPROTO op_fmul_ST0_FT0(void)
1764 {
1765     ST0 *= FT0;
1766 }
1767
1768 void OPPROTO op_fsub_ST0_FT0(void)
1769 {
1770     ST0 -= FT0;
1771 }
1772
1773 void OPPROTO op_fsubr_ST0_FT0(void)
1774 {
1775     ST0 = FT0 - ST0;
1776 }
1777
1778 void OPPROTO op_fdiv_ST0_FT0(void)
1779 {
1780     ST0 /= FT0;
1781 }
1782
1783 void OPPROTO op_fdivr_ST0_FT0(void)
1784 {
1785     ST0 = FT0 / ST0;
1786 }
1787
1788 /* fp operations between STN and ST0 */
1789
1790 void OPPROTO op_fadd_STN_ST0(void)
1791 {
1792     ST(PARAM1) += ST0;
1793 }
1794
1795 void OPPROTO op_fmul_STN_ST0(void)
1796 {
1797     ST(PARAM1) *= ST0;
1798 }
1799
1800 void OPPROTO op_fsub_STN_ST0(void)
1801 {
1802     ST(PARAM1) -= ST0;
1803 }
1804
1805 void OPPROTO op_fsubr_STN_ST0(void)
1806 {
1807     CPU86_LDouble *p;
1808     p = &ST(PARAM1);
1809     *p = ST0 - *p;
1810 }
1811
1812 void OPPROTO op_fdiv_STN_ST0(void)
1813 {
1814     ST(PARAM1) /= ST0;
1815 }
1816
1817 void OPPROTO op_fdivr_STN_ST0(void)
1818 {
1819     CPU86_LDouble *p;
1820     p = &ST(PARAM1);
1821     *p = ST0 / *p;
1822 }
1823
1824 /* misc FPU operations */
1825 void OPPROTO op_fchs_ST0(void)
1826 {
1827     ST0 = -ST0;
1828 }
1829
1830 void OPPROTO op_fabs_ST0(void)
1831 {
1832     ST0 = fabs(ST0);
1833 }
1834
1835 void helper_fxam_ST0(void)
1836 {
1837     CPU86_LDoubleU temp;
1838     int expdif;
1839
1840     temp.d = ST0;
1841
1842     env->fpus &= (~0x4700);  /* (C3,C2,C1,C0) <-- 0000 */
1843     if (SIGND(temp))
1844         env->fpus |= 0x200; /* C1 <-- 1 */
1845
1846     expdif = EXPD(temp);
1847     if (expdif == MAXEXPD) {
1848         if (MANTD(temp) == 0)
1849             env->fpus |=  0x500 /*Infinity*/;
1850         else
1851             env->fpus |=  0x100 /*NaN*/;
1852     } else if (expdif == 0) {
1853         if (MANTD(temp) == 0)
1854             env->fpus |=  0x4000 /*Zero*/;
1855         else
1856             env->fpus |= 0x4400 /*Denormal*/;
1857     } else {
1858         env->fpus |= 0x400;
1859     }
1860 }
1861
1862 void OPPROTO op_fxam_ST0(void)
1863 {
1864     helper_fxam_ST0();
1865 }
1866
1867 void OPPROTO op_fld1_ST0(void)
1868 {
1869     ST0 = *(CPU86_LDouble *)&f15rk[1];
1870 }
1871
1872 void OPPROTO op_fldl2t_ST0(void)
1873 {
1874     ST0 = *(CPU86_LDouble *)&f15rk[6];
1875 }
1876
1877 void OPPROTO op_fldl2e_ST0(void)
1878 {
1879     ST0 = *(CPU86_LDouble *)&f15rk[5];
1880 }
1881
1882 void OPPROTO op_fldpi_ST0(void)
1883 {
1884     ST0 = *(CPU86_LDouble *)&f15rk[2];
1885 }
1886
1887 void OPPROTO op_fldlg2_ST0(void)
1888 {
1889     ST0 = *(CPU86_LDouble *)&f15rk[3];
1890 }
1891
1892 void OPPROTO op_fldln2_ST0(void)
1893 {
1894     ST0 = *(CPU86_LDouble *)&f15rk[4];
1895 }
1896
1897 void OPPROTO op_fldz_ST0(void)
1898 {
1899     ST0 = *(CPU86_LDouble *)&f15rk[0];
1900 }
1901
1902 void OPPROTO op_fldz_FT0(void)
1903 {
1904     ST0 = *(CPU86_LDouble *)&f15rk[0];
1905 }
1906
1907 void helper_f2xm1(void)
1908 {
1909     ST0 = pow(2.0,ST0) - 1.0;
1910 }
1911
1912 void helper_fyl2x(void)
1913 {
1914     CPU86_LDouble fptemp;
1915
1916     fptemp = ST0;
1917     if (fptemp>0.0){
1918         fptemp = log(fptemp)/log(2.0);   /* log2(ST) */
1919         ST1 *= fptemp;
1920         fpop();
1921     } else {
1922         env->fpus &= (~0x4700);
1923         env->fpus |= 0x400;
1924     }
1925 }
1926
1927 void helper_fptan(void)
1928 {
1929     CPU86_LDouble fptemp;
1930
1931     fptemp = ST0;
1932     if((fptemp > MAXTAN)||(fptemp < -MAXTAN)) {
1933         env->fpus |= 0x400;
1934     } else {
1935         ST0 = tan(fptemp);
1936         fpush();
1937         ST0 = 1.0;
1938         env->fpus &= (~0x400);  /* C2 <-- 0 */
1939         /* the above code is for  |arg| < 2**52 only */
1940     }
1941 }
1942
1943 void helper_fpatan(void)
1944 {
1945     CPU86_LDouble fptemp, fpsrcop;
1946
1947     fpsrcop = ST1;
1948     fptemp = ST0;
1949     ST1 = atan2(fpsrcop,fptemp);
1950     fpop();
1951 }
1952
1953 void helper_fxtract(void)
1954 {
1955     CPU86_LDoubleU temp;
1956     unsigned int expdif;
1957
1958     temp.d = ST0;
1959     expdif = EXPD(temp) - EXPBIAS;
1960     /*DP exponent bias*/
1961     ST0 = expdif;
1962     fpush();
1963     BIASEXPONENT(temp);
1964     ST0 = temp.d;
1965 }
1966
1967 void helper_fprem1(void)
1968 {
1969     CPU86_LDouble dblq, fpsrcop, fptemp;
1970     CPU86_LDoubleU fpsrcop1, fptemp1;
1971     int expdif;
1972     int q;
1973
1974     fpsrcop = ST0;
1975     fptemp = ST1;
1976     fpsrcop1.d = fpsrcop;
1977     fptemp1.d = fptemp;
1978     expdif = EXPD(fpsrcop1) - EXPD(fptemp1);
1979     if (expdif < 53) {
1980         dblq = fpsrcop / fptemp;
1981         dblq = (dblq < 0.0)? ceil(dblq): floor(dblq);
1982         ST0 = fpsrcop - fptemp*dblq;
1983         q = (int)dblq; /* cutting off top bits is assumed here */
1984         env->fpus &= (~0x4700); /* (C3,C2,C1,C0) <-- 0000 */
1985                                 /* (C0,C1,C3) <-- (q2,q1,q0) */
1986         env->fpus |= (q&0x4) << 6; /* (C0) <-- q2 */
1987         env->fpus |= (q&0x2) << 8; /* (C1) <-- q1 */
1988         env->fpus |= (q&0x1) << 14; /* (C3) <-- q0 */
1989     } else {
1990         env->fpus |= 0x400;  /* C2 <-- 1 */
1991         fptemp = pow(2.0, expdif-50);
1992         fpsrcop = (ST0 / ST1) / fptemp;
1993         /* fpsrcop = integer obtained by rounding to the nearest */
1994         fpsrcop = (fpsrcop-floor(fpsrcop) < ceil(fpsrcop)-fpsrcop)?
1995             floor(fpsrcop): ceil(fpsrcop);
1996         ST0 -= (ST1 * fpsrcop * fptemp);
1997     }
1998 }
1999
2000 void helper_fprem(void)
2001 {
2002     CPU86_LDouble dblq, fpsrcop, fptemp;
2003     CPU86_LDoubleU fpsrcop1, fptemp1;
2004     int expdif;
2005     int q;
2006
2007     fpsrcop = ST0;
2008     fptemp = ST1;
2009     fpsrcop1.d = fpsrcop;
2010     fptemp1.d = fptemp;
2011     expdif = EXPD(fpsrcop1) - EXPD(fptemp1);
2012     if ( expdif < 53 ) {
2013         dblq = fpsrcop / fptemp;
2014         dblq = (dblq < 0.0)? ceil(dblq): floor(dblq);
2015         ST0 = fpsrcop - fptemp*dblq;
2016         q = (int)dblq; /* cutting off top bits is assumed here */
2017         env->fpus &= (~0x4700); /* (C3,C2,C1,C0) <-- 0000 */
2018                                 /* (C0,C1,C3) <-- (q2,q1,q0) */
2019         env->fpus |= (q&0x4) << 6; /* (C0) <-- q2 */
2020         env->fpus |= (q&0x2) << 8; /* (C1) <-- q1 */
2021         env->fpus |= (q&0x1) << 14; /* (C3) <-- q0 */
2022     } else {
2023         env->fpus |= 0x400;  /* C2 <-- 1 */
2024         fptemp = pow(2.0, expdif-50);
2025         fpsrcop = (ST0 / ST1) / fptemp;
2026         /* fpsrcop = integer obtained by chopping */
2027         fpsrcop = (fpsrcop < 0.0)?
2028             -(floor(fabs(fpsrcop))): floor(fpsrcop);
2029         ST0 -= (ST1 * fpsrcop * fptemp);
2030     }
2031 }
2032
2033 void helper_fyl2xp1(void)
2034 {
2035     CPU86_LDouble fptemp;
2036
2037     fptemp = ST0;
2038     if ((fptemp+1.0)>0.0) {
2039         fptemp = log(fptemp+1.0) / log(2.0); /* log2(ST+1.0) */
2040         ST1 *= fptemp;
2041         fpop();
2042     } else {
2043         env->fpus &= (~0x4700);
2044         env->fpus |= 0x400;
2045     }
2046 }
2047
2048 void helper_fsqrt(void)
2049 {
2050     CPU86_LDouble fptemp;
2051
2052     fptemp = ST0;
2053     if (fptemp<0.0) {
2054         env->fpus &= (~0x4700);  /* (C3,C2,C1,C0) <-- 0000 */
2055         env->fpus |= 0x400;
2056     }
2057     ST0 = sqrt(fptemp);
2058 }
2059
2060 void helper_fsincos(void)
2061 {
2062     CPU86_LDouble fptemp;
2063
2064     fptemp = ST0;
2065     if ((fptemp > MAXTAN)||(fptemp < -MAXTAN)) {
2066         env->fpus |= 0x400;
2067     } else {
2068         ST0 = sin(fptemp);
2069         fpush();
2070         ST0 = cos(fptemp);
2071         env->fpus &= (~0x400);  /* C2 <-- 0 */
2072         /* the above code is for  |arg| < 2**63 only */
2073     }
2074 }
2075
2076 void helper_frndint(void)
2077 {
2078     ST0 = rint(ST0);
2079 }
2080
2081 void helper_fscale(void)
2082 {
2083     CPU86_LDouble fpsrcop, fptemp;
2084
2085     fpsrcop = 2.0;
2086     fptemp = pow(fpsrcop,ST1);
2087     ST0 *= fptemp;
2088 }
2089
2090 void helper_fsin(void)
2091 {
2092     CPU86_LDouble fptemp;
2093
2094     fptemp = ST0;
2095     if ((fptemp > MAXTAN)||(fptemp < -MAXTAN)) {
2096         env->fpus |= 0x400;
2097     } else {
2098         ST0 = sin(fptemp);
2099         env->fpus &= (~0x400);  /* C2 <-- 0 */
2100         /* the above code is for  |arg| < 2**53 only */
2101     }
2102 }
2103
2104 void helper_fcos(void)
2105 {
2106     CPU86_LDouble fptemp;
2107
2108     fptemp = ST0;
2109     if((fptemp > MAXTAN)||(fptemp < -MAXTAN)) {
2110         env->fpus |= 0x400;
2111     } else {
2112         ST0 = cos(fptemp);
2113         env->fpus &= (~0x400);  /* C2 <-- 0 */
2114         /* the above code is for  |arg5 < 2**63 only */
2115     }
2116 }
2117
2118 /* associated heplers to reduce generated code length and to simplify
2119    relocation (FP constants are usually stored in .rodata section) */
2120
2121 void OPPROTO op_f2xm1(void)
2122 {
2123     helper_f2xm1();
2124 }
2125
2126 void OPPROTO op_fyl2x(void)
2127 {
2128     helper_fyl2x();
2129 }
2130
2131 void OPPROTO op_fptan(void)
2132 {
2133     helper_fptan();
2134 }
2135
2136 void OPPROTO op_fpatan(void)
2137 {
2138     helper_fpatan();
2139 }
2140
2141 void OPPROTO op_fxtract(void)
2142 {
2143     helper_fxtract();
2144 }
2145
2146 void OPPROTO op_fprem1(void)
2147 {
2148     helper_fprem1();
2149 }
2150
2151
2152 void OPPROTO op_fprem(void)
2153 {
2154     helper_fprem();
2155 }
2156
2157 void OPPROTO op_fyl2xp1(void)
2158 {
2159     helper_fyl2xp1();
2160 }
2161
2162 void OPPROTO op_fsqrt(void)
2163 {
2164     helper_fsqrt();
2165 }
2166
2167 void OPPROTO op_fsincos(void)
2168 {
2169     helper_fsincos();
2170 }
2171
2172 void OPPROTO op_frndint(void)
2173 {
2174     helper_frndint();
2175 }
2176
2177 void OPPROTO op_fscale(void)
2178 {
2179     helper_fscale();
2180 }
2181
2182 void OPPROTO op_fsin(void)
2183 {
2184     helper_fsin();
2185 }
2186
2187 void OPPROTO op_fcos(void)
2188 {
2189     helper_fcos();
2190 }
2191
2192 void OPPROTO op_fnstsw_A0(void)
2193 {
2194     int fpus;
2195     fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11;
2196     stw((void *)A0, fpus);
2197 }
2198
2199 void OPPROTO op_fnstsw_EAX(void)
2200 {
2201     int fpus;
2202     fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11;
2203     EAX = (EAX & 0xffff0000) | fpus;
2204 }
2205
2206 void OPPROTO op_fnstcw_A0(void)
2207 {
2208     stw((void *)A0, env->fpuc);
2209 }
2210
2211 void OPPROTO op_fldcw_A0(void)
2212 {
2213     int rnd_type;
2214     env->fpuc = lduw((void *)A0);
2215     /* set rounding mode */
2216     switch(env->fpuc & RC_MASK) {
2217     default:
2218     case RC_NEAR:
2219         rnd_type = FE_TONEAREST;
2220         break;
2221     case RC_DOWN:
2222         rnd_type = FE_DOWNWARD;
2223         break;
2224     case RC_UP:
2225         rnd_type = FE_UPWARD;
2226         break;
2227     case RC_CHOP:
2228         rnd_type = FE_TOWARDZERO;
2229         break;
2230     }
2231     fesetround(rnd_type);
2232 }
2233
2234 void OPPROTO op_fclex(void)
2235 {
2236     env->fpus &= 0x7f00;
2237 }
2238
2239 void OPPROTO op_fninit(void)
2240 {
2241     env->fpus = 0;
2242     env->fpstt = 0;
2243     env->fpuc = 0x37f;
2244     env->fptags[0] = 1;
2245     env->fptags[1] = 1;
2246     env->fptags[2] = 1;
2247     env->fptags[3] = 1;
2248     env->fptags[4] = 1;
2249     env->fptags[5] = 1;
2250     env->fptags[6] = 1;
2251     env->fptags[7] = 1;
2252 }
2253
2254 /* threading support */
2255 void OPPROTO op_lock(void)
2256 {
2257     cpu_lock();
2258 }
2259
2260 void OPPROTO op_unlock(void)
2261 {
2262     cpu_unlock();
2263 }