root / target-i386 / translate.c @ 321c5351
History | View | Annotate | Download (283.2 kB)
| 1 |
/*
|
|---|---|
| 2 |
* i386 translation
|
| 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, see <http://www.gnu.org/licenses/>.
|
| 18 |
*/
|
| 19 |
#include <stdarg.h> |
| 20 |
#include <stdlib.h> |
| 21 |
#include <stdio.h> |
| 22 |
#include <string.h> |
| 23 |
#include <inttypes.h> |
| 24 |
#include <signal.h> |
| 25 |
|
| 26 |
#include "qemu/host-utils.h" |
| 27 |
#include "cpu.h" |
| 28 |
#include "disas/disas.h" |
| 29 |
#include "tcg-op.h" |
| 30 |
|
| 31 |
#include "helper.h" |
| 32 |
#define GEN_HELPER 1 |
| 33 |
#include "helper.h" |
| 34 |
|
| 35 |
#define PREFIX_REPZ 0x01 |
| 36 |
#define PREFIX_REPNZ 0x02 |
| 37 |
#define PREFIX_LOCK 0x04 |
| 38 |
#define PREFIX_DATA 0x08 |
| 39 |
#define PREFIX_ADR 0x10 |
| 40 |
#define PREFIX_VEX 0x20 |
| 41 |
|
| 42 |
#ifdef TARGET_X86_64
|
| 43 |
#define CODE64(s) ((s)->code64)
|
| 44 |
#define REX_X(s) ((s)->rex_x)
|
| 45 |
#define REX_B(s) ((s)->rex_b)
|
| 46 |
#else
|
| 47 |
#define CODE64(s) 0 |
| 48 |
#define REX_X(s) 0 |
| 49 |
#define REX_B(s) 0 |
| 50 |
#endif
|
| 51 |
|
| 52 |
#ifdef TARGET_X86_64
|
| 53 |
# define ctztl ctz64
|
| 54 |
# define clztl clz64
|
| 55 |
#else
|
| 56 |
# define ctztl ctz32
|
| 57 |
# define clztl clz32
|
| 58 |
#endif
|
| 59 |
|
| 60 |
//#define MACRO_TEST 1
|
| 61 |
|
| 62 |
/* global register indexes */
|
| 63 |
static TCGv_ptr cpu_env;
|
| 64 |
static TCGv cpu_A0;
|
| 65 |
static TCGv cpu_cc_dst, cpu_cc_src, cpu_cc_src2, cpu_cc_srcT;
|
| 66 |
static TCGv_i32 cpu_cc_op;
|
| 67 |
static TCGv cpu_regs[CPU_NB_REGS];
|
| 68 |
/* local temps */
|
| 69 |
static TCGv cpu_T[2]; |
| 70 |
/* local register indexes (only used inside old micro ops) */
|
| 71 |
static TCGv cpu_tmp0, cpu_tmp4;
|
| 72 |
static TCGv_ptr cpu_ptr0, cpu_ptr1;
|
| 73 |
static TCGv_i32 cpu_tmp2_i32, cpu_tmp3_i32;
|
| 74 |
static TCGv_i64 cpu_tmp1_i64;
|
| 75 |
static TCGv cpu_tmp5;
|
| 76 |
|
| 77 |
static uint8_t gen_opc_cc_op[OPC_BUF_SIZE];
|
| 78 |
|
| 79 |
#include "exec/gen-icount.h" |
| 80 |
|
| 81 |
#ifdef TARGET_X86_64
|
| 82 |
static int x86_64_hregs; |
| 83 |
#endif
|
| 84 |
|
| 85 |
typedef struct DisasContext { |
| 86 |
/* current insn context */
|
| 87 |
int override; /* -1 if no override */ |
| 88 |
int prefix;
|
| 89 |
int aflag, dflag;
|
| 90 |
target_ulong pc; /* pc = eip + cs_base */
|
| 91 |
int is_jmp; /* 1 = means jump (stop translation), 2 means CPU |
| 92 |
static state change (stop translation) */
|
| 93 |
/* current block context */
|
| 94 |
target_ulong cs_base; /* base of CS segment */
|
| 95 |
int pe; /* protected mode */ |
| 96 |
int code32; /* 32 bit code segment */ |
| 97 |
#ifdef TARGET_X86_64
|
| 98 |
int lma; /* long mode active */ |
| 99 |
int code64; /* 64 bit code segment */ |
| 100 |
int rex_x, rex_b;
|
| 101 |
#endif
|
| 102 |
int vex_l; /* vex vector length */ |
| 103 |
int vex_v; /* vex vvvv register, without 1's compliment. */ |
| 104 |
int ss32; /* 32 bit stack segment */ |
| 105 |
CCOp cc_op; /* current CC operation */
|
| 106 |
bool cc_op_dirty;
|
| 107 |
int addseg; /* non zero if either DS/ES/SS have a non zero base */ |
| 108 |
int f_st; /* currently unused */ |
| 109 |
int vm86; /* vm86 mode */ |
| 110 |
int cpl;
|
| 111 |
int iopl;
|
| 112 |
int tf; /* TF cpu flag */ |
| 113 |
int singlestep_enabled; /* "hardware" single step enabled */ |
| 114 |
int jmp_opt; /* use direct block chaining for direct jumps */ |
| 115 |
int mem_index; /* select memory access functions */ |
| 116 |
uint64_t flags; /* all execution flags */
|
| 117 |
struct TranslationBlock *tb;
|
| 118 |
int popl_esp_hack; /* for correct popl with esp base handling */ |
| 119 |
int rip_offset; /* only used in x86_64, but left for simplicity */ |
| 120 |
int cpuid_features;
|
| 121 |
int cpuid_ext_features;
|
| 122 |
int cpuid_ext2_features;
|
| 123 |
int cpuid_ext3_features;
|
| 124 |
int cpuid_7_0_ebx_features;
|
| 125 |
} DisasContext; |
| 126 |
|
| 127 |
static void gen_eob(DisasContext *s); |
| 128 |
static void gen_jmp(DisasContext *s, target_ulong eip); |
| 129 |
static void gen_jmp_tb(DisasContext *s, target_ulong eip, int tb_num); |
| 130 |
static void gen_op(DisasContext *s1, int op, int ot, int d); |
| 131 |
|
| 132 |
/* i386 arith/logic operations */
|
| 133 |
enum {
|
| 134 |
OP_ADDL, |
| 135 |
OP_ORL, |
| 136 |
OP_ADCL, |
| 137 |
OP_SBBL, |
| 138 |
OP_ANDL, |
| 139 |
OP_SUBL, |
| 140 |
OP_XORL, |
| 141 |
OP_CMPL, |
| 142 |
}; |
| 143 |
|
| 144 |
/* i386 shift ops */
|
| 145 |
enum {
|
| 146 |
OP_ROL, |
| 147 |
OP_ROR, |
| 148 |
OP_RCL, |
| 149 |
OP_RCR, |
| 150 |
OP_SHL, |
| 151 |
OP_SHR, |
| 152 |
OP_SHL1, /* undocumented */
|
| 153 |
OP_SAR = 7,
|
| 154 |
}; |
| 155 |
|
| 156 |
enum {
|
| 157 |
JCC_O, |
| 158 |
JCC_B, |
| 159 |
JCC_Z, |
| 160 |
JCC_BE, |
| 161 |
JCC_S, |
| 162 |
JCC_P, |
| 163 |
JCC_L, |
| 164 |
JCC_LE, |
| 165 |
}; |
| 166 |
|
| 167 |
/* operand size */
|
| 168 |
enum {
|
| 169 |
OT_BYTE = 0,
|
| 170 |
OT_WORD, |
| 171 |
OT_LONG, |
| 172 |
OT_QUAD, |
| 173 |
}; |
| 174 |
|
| 175 |
enum {
|
| 176 |
/* I386 int registers */
|
| 177 |
OR_EAX, /* MUST be even numbered */
|
| 178 |
OR_ECX, |
| 179 |
OR_EDX, |
| 180 |
OR_EBX, |
| 181 |
OR_ESP, |
| 182 |
OR_EBP, |
| 183 |
OR_ESI, |
| 184 |
OR_EDI, |
| 185 |
|
| 186 |
OR_TMP0 = 16, /* temporary operand register */ |
| 187 |
OR_TMP1, |
| 188 |
OR_A0, /* temporary register used when doing address evaluation */
|
| 189 |
}; |
| 190 |
|
| 191 |
enum {
|
| 192 |
USES_CC_DST = 1,
|
| 193 |
USES_CC_SRC = 2,
|
| 194 |
USES_CC_SRC2 = 4,
|
| 195 |
USES_CC_SRCT = 8,
|
| 196 |
}; |
| 197 |
|
| 198 |
/* Bit set if the global variable is live after setting CC_OP to X. */
|
| 199 |
static const uint8_t cc_op_live[CC_OP_NB] = { |
| 200 |
[CC_OP_DYNAMIC] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2, |
| 201 |
[CC_OP_EFLAGS] = USES_CC_SRC, |
| 202 |
[CC_OP_MULB ... CC_OP_MULQ] = USES_CC_DST | USES_CC_SRC, |
| 203 |
[CC_OP_ADDB ... CC_OP_ADDQ] = USES_CC_DST | USES_CC_SRC, |
| 204 |
[CC_OP_ADCB ... CC_OP_ADCQ] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2, |
| 205 |
[CC_OP_SUBB ... CC_OP_SUBQ] = USES_CC_DST | USES_CC_SRC | USES_CC_SRCT, |
| 206 |
[CC_OP_SBBB ... CC_OP_SBBQ] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2, |
| 207 |
[CC_OP_LOGICB ... CC_OP_LOGICQ] = USES_CC_DST, |
| 208 |
[CC_OP_INCB ... CC_OP_INCQ] = USES_CC_DST | USES_CC_SRC, |
| 209 |
[CC_OP_DECB ... CC_OP_DECQ] = USES_CC_DST | USES_CC_SRC, |
| 210 |
[CC_OP_SHLB ... CC_OP_SHLQ] = USES_CC_DST | USES_CC_SRC, |
| 211 |
[CC_OP_SARB ... CC_OP_SARQ] = USES_CC_DST | USES_CC_SRC, |
| 212 |
[CC_OP_BMILGB ... CC_OP_BMILGQ] = USES_CC_DST | USES_CC_SRC, |
| 213 |
[CC_OP_ADCX] = USES_CC_DST | USES_CC_SRC, |
| 214 |
[CC_OP_ADOX] = USES_CC_SRC | USES_CC_SRC2, |
| 215 |
[CC_OP_ADCOX] = USES_CC_DST | USES_CC_SRC | USES_CC_SRC2, |
| 216 |
}; |
| 217 |
|
| 218 |
static void set_cc_op(DisasContext *s, CCOp op) |
| 219 |
{
|
| 220 |
int dead;
|
| 221 |
|
| 222 |
if (s->cc_op == op) {
|
| 223 |
return;
|
| 224 |
} |
| 225 |
|
| 226 |
/* Discard CC computation that will no longer be used. */
|
| 227 |
dead = cc_op_live[s->cc_op] & ~cc_op_live[op]; |
| 228 |
if (dead & USES_CC_DST) {
|
| 229 |
tcg_gen_discard_tl(cpu_cc_dst); |
| 230 |
} |
| 231 |
if (dead & USES_CC_SRC) {
|
| 232 |
tcg_gen_discard_tl(cpu_cc_src); |
| 233 |
} |
| 234 |
if (dead & USES_CC_SRC2) {
|
| 235 |
tcg_gen_discard_tl(cpu_cc_src2); |
| 236 |
} |
| 237 |
if (dead & USES_CC_SRCT) {
|
| 238 |
tcg_gen_discard_tl(cpu_cc_srcT); |
| 239 |
} |
| 240 |
|
| 241 |
s->cc_op = op; |
| 242 |
/* The DYNAMIC setting is translator only, and should never be
|
| 243 |
stored. Thus we always consider it clean. */
|
| 244 |
s->cc_op_dirty = (op != CC_OP_DYNAMIC); |
| 245 |
} |
| 246 |
|
| 247 |
static void gen_update_cc_op(DisasContext *s) |
| 248 |
{
|
| 249 |
if (s->cc_op_dirty) {
|
| 250 |
tcg_gen_movi_i32(cpu_cc_op, s->cc_op); |
| 251 |
s->cc_op_dirty = false;
|
| 252 |
} |
| 253 |
} |
| 254 |
|
| 255 |
static inline void gen_op_movl_T0_0(void) |
| 256 |
{
|
| 257 |
tcg_gen_movi_tl(cpu_T[0], 0); |
| 258 |
} |
| 259 |
|
| 260 |
static inline void gen_op_movl_T0_im(int32_t val) |
| 261 |
{
|
| 262 |
tcg_gen_movi_tl(cpu_T[0], val);
|
| 263 |
} |
| 264 |
|
| 265 |
static inline void gen_op_movl_T0_imu(uint32_t val) |
| 266 |
{
|
| 267 |
tcg_gen_movi_tl(cpu_T[0], val);
|
| 268 |
} |
| 269 |
|
| 270 |
static inline void gen_op_movl_T1_im(int32_t val) |
| 271 |
{
|
| 272 |
tcg_gen_movi_tl(cpu_T[1], val);
|
| 273 |
} |
| 274 |
|
| 275 |
static inline void gen_op_movl_T1_imu(uint32_t val) |
| 276 |
{
|
| 277 |
tcg_gen_movi_tl(cpu_T[1], val);
|
| 278 |
} |
| 279 |
|
| 280 |
static inline void gen_op_movl_A0_im(uint32_t val) |
| 281 |
{
|
| 282 |
tcg_gen_movi_tl(cpu_A0, val); |
| 283 |
} |
| 284 |
|
| 285 |
#ifdef TARGET_X86_64
|
| 286 |
static inline void gen_op_movq_A0_im(int64_t val) |
| 287 |
{
|
| 288 |
tcg_gen_movi_tl(cpu_A0, val); |
| 289 |
} |
| 290 |
#endif
|
| 291 |
|
| 292 |
static inline void gen_movtl_T0_im(target_ulong val) |
| 293 |
{
|
| 294 |
tcg_gen_movi_tl(cpu_T[0], val);
|
| 295 |
} |
| 296 |
|
| 297 |
static inline void gen_movtl_T1_im(target_ulong val) |
| 298 |
{
|
| 299 |
tcg_gen_movi_tl(cpu_T[1], val);
|
| 300 |
} |
| 301 |
|
| 302 |
static inline void gen_op_andl_T0_ffff(void) |
| 303 |
{
|
| 304 |
tcg_gen_andi_tl(cpu_T[0], cpu_T[0], 0xffff); |
| 305 |
} |
| 306 |
|
| 307 |
static inline void gen_op_andl_T0_im(uint32_t val) |
| 308 |
{
|
| 309 |
tcg_gen_andi_tl(cpu_T[0], cpu_T[0], val); |
| 310 |
} |
| 311 |
|
| 312 |
static inline void gen_op_movl_T0_T1(void) |
| 313 |
{
|
| 314 |
tcg_gen_mov_tl(cpu_T[0], cpu_T[1]); |
| 315 |
} |
| 316 |
|
| 317 |
static inline void gen_op_andl_A0_ffff(void) |
| 318 |
{
|
| 319 |
tcg_gen_andi_tl(cpu_A0, cpu_A0, 0xffff);
|
| 320 |
} |
| 321 |
|
| 322 |
#ifdef TARGET_X86_64
|
| 323 |
|
| 324 |
#define NB_OP_SIZES 4 |
| 325 |
|
| 326 |
#else /* !TARGET_X86_64 */ |
| 327 |
|
| 328 |
#define NB_OP_SIZES 3 |
| 329 |
|
| 330 |
#endif /* !TARGET_X86_64 */ |
| 331 |
|
| 332 |
#if defined(HOST_WORDS_BIGENDIAN)
|
| 333 |
#define REG_B_OFFSET (sizeof(target_ulong) - 1) |
| 334 |
#define REG_H_OFFSET (sizeof(target_ulong) - 2) |
| 335 |
#define REG_W_OFFSET (sizeof(target_ulong) - 2) |
| 336 |
#define REG_L_OFFSET (sizeof(target_ulong) - 4) |
| 337 |
#define REG_LH_OFFSET (sizeof(target_ulong) - 8) |
| 338 |
#else
|
| 339 |
#define REG_B_OFFSET 0 |
| 340 |
#define REG_H_OFFSET 1 |
| 341 |
#define REG_W_OFFSET 0 |
| 342 |
#define REG_L_OFFSET 0 |
| 343 |
#define REG_LH_OFFSET 4 |
| 344 |
#endif
|
| 345 |
|
| 346 |
/* In instruction encodings for byte register accesses the
|
| 347 |
* register number usually indicates "low 8 bits of register N";
|
| 348 |
* however there are some special cases where N 4..7 indicates
|
| 349 |
* [AH, CH, DH, BH], ie "bits 15..8 of register N-4". Return
|
| 350 |
* true for this special case, false otherwise.
|
| 351 |
*/
|
| 352 |
static inline bool byte_reg_is_xH(int reg) |
| 353 |
{
|
| 354 |
if (reg < 4) { |
| 355 |
return false; |
| 356 |
} |
| 357 |
#ifdef TARGET_X86_64
|
| 358 |
if (reg >= 8 || x86_64_hregs) { |
| 359 |
return false; |
| 360 |
} |
| 361 |
#endif
|
| 362 |
return true; |
| 363 |
} |
| 364 |
|
| 365 |
static inline void gen_op_mov_reg_v(int ot, int reg, TCGv t0) |
| 366 |
{
|
| 367 |
switch(ot) {
|
| 368 |
case OT_BYTE:
|
| 369 |
if (!byte_reg_is_xH(reg)) {
|
| 370 |
tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], t0, 0, 8); |
| 371 |
} else {
|
| 372 |
tcg_gen_deposit_tl(cpu_regs[reg - 4], cpu_regs[reg - 4], t0, 8, 8); |
| 373 |
} |
| 374 |
break;
|
| 375 |
case OT_WORD:
|
| 376 |
tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], t0, 0, 16); |
| 377 |
break;
|
| 378 |
default: /* XXX this shouldn't be reached; abort? */ |
| 379 |
case OT_LONG:
|
| 380 |
/* For x86_64, this sets the higher half of register to zero.
|
| 381 |
For i386, this is equivalent to a mov. */
|
| 382 |
tcg_gen_ext32u_tl(cpu_regs[reg], t0); |
| 383 |
break;
|
| 384 |
#ifdef TARGET_X86_64
|
| 385 |
case OT_QUAD:
|
| 386 |
tcg_gen_mov_tl(cpu_regs[reg], t0); |
| 387 |
break;
|
| 388 |
#endif
|
| 389 |
} |
| 390 |
} |
| 391 |
|
| 392 |
static inline void gen_op_mov_reg_T0(int ot, int reg) |
| 393 |
{
|
| 394 |
gen_op_mov_reg_v(ot, reg, cpu_T[0]);
|
| 395 |
} |
| 396 |
|
| 397 |
static inline void gen_op_mov_reg_T1(int ot, int reg) |
| 398 |
{
|
| 399 |
gen_op_mov_reg_v(ot, reg, cpu_T[1]);
|
| 400 |
} |
| 401 |
|
| 402 |
static inline void gen_op_mov_reg_A0(int size, int reg) |
| 403 |
{
|
| 404 |
switch(size) {
|
| 405 |
case OT_BYTE:
|
| 406 |
tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], cpu_A0, 0, 16); |
| 407 |
break;
|
| 408 |
default: /* XXX this shouldn't be reached; abort? */ |
| 409 |
case OT_WORD:
|
| 410 |
/* For x86_64, this sets the higher half of register to zero.
|
| 411 |
For i386, this is equivalent to a mov. */
|
| 412 |
tcg_gen_ext32u_tl(cpu_regs[reg], cpu_A0); |
| 413 |
break;
|
| 414 |
#ifdef TARGET_X86_64
|
| 415 |
case OT_LONG:
|
| 416 |
tcg_gen_mov_tl(cpu_regs[reg], cpu_A0); |
| 417 |
break;
|
| 418 |
#endif
|
| 419 |
} |
| 420 |
} |
| 421 |
|
| 422 |
static inline void gen_op_mov_v_reg(int ot, TCGv t0, int reg) |
| 423 |
{
|
| 424 |
if (ot == OT_BYTE && byte_reg_is_xH(reg)) {
|
| 425 |
tcg_gen_shri_tl(t0, cpu_regs[reg - 4], 8); |
| 426 |
tcg_gen_ext8u_tl(t0, t0); |
| 427 |
} else {
|
| 428 |
tcg_gen_mov_tl(t0, cpu_regs[reg]); |
| 429 |
} |
| 430 |
} |
| 431 |
|
| 432 |
static inline void gen_op_mov_TN_reg(int ot, int t_index, int reg) |
| 433 |
{
|
| 434 |
gen_op_mov_v_reg(ot, cpu_T[t_index], reg); |
| 435 |
} |
| 436 |
|
| 437 |
static inline void gen_op_movl_A0_reg(int reg) |
| 438 |
{
|
| 439 |
tcg_gen_mov_tl(cpu_A0, cpu_regs[reg]); |
| 440 |
} |
| 441 |
|
| 442 |
static inline void gen_op_addl_A0_im(int32_t val) |
| 443 |
{
|
| 444 |
tcg_gen_addi_tl(cpu_A0, cpu_A0, val); |
| 445 |
#ifdef TARGET_X86_64
|
| 446 |
tcg_gen_andi_tl(cpu_A0, cpu_A0, 0xffffffff);
|
| 447 |
#endif
|
| 448 |
} |
| 449 |
|
| 450 |
#ifdef TARGET_X86_64
|
| 451 |
static inline void gen_op_addq_A0_im(int64_t val) |
| 452 |
{
|
| 453 |
tcg_gen_addi_tl(cpu_A0, cpu_A0, val); |
| 454 |
} |
| 455 |
#endif
|
| 456 |
|
| 457 |
static void gen_add_A0_im(DisasContext *s, int val) |
| 458 |
{
|
| 459 |
#ifdef TARGET_X86_64
|
| 460 |
if (CODE64(s))
|
| 461 |
gen_op_addq_A0_im(val); |
| 462 |
else
|
| 463 |
#endif
|
| 464 |
gen_op_addl_A0_im(val); |
| 465 |
} |
| 466 |
|
| 467 |
static inline void gen_op_addl_T0_T1(void) |
| 468 |
{
|
| 469 |
tcg_gen_add_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 470 |
} |
| 471 |
|
| 472 |
static inline void gen_op_jmp_T0(void) |
| 473 |
{
|
| 474 |
tcg_gen_st_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, eip));
|
| 475 |
} |
| 476 |
|
| 477 |
static inline void gen_op_add_reg_im(int size, int reg, int32_t val) |
| 478 |
{
|
| 479 |
switch(size) {
|
| 480 |
case OT_BYTE:
|
| 481 |
tcg_gen_addi_tl(cpu_tmp0, cpu_regs[reg], val); |
| 482 |
tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], cpu_tmp0, 0, 16); |
| 483 |
break;
|
| 484 |
case OT_WORD:
|
| 485 |
tcg_gen_addi_tl(cpu_tmp0, cpu_regs[reg], val); |
| 486 |
/* For x86_64, this sets the higher half of register to zero.
|
| 487 |
For i386, this is equivalent to a nop. */
|
| 488 |
tcg_gen_ext32u_tl(cpu_tmp0, cpu_tmp0); |
| 489 |
tcg_gen_mov_tl(cpu_regs[reg], cpu_tmp0); |
| 490 |
break;
|
| 491 |
#ifdef TARGET_X86_64
|
| 492 |
case OT_LONG:
|
| 493 |
tcg_gen_addi_tl(cpu_regs[reg], cpu_regs[reg], val); |
| 494 |
break;
|
| 495 |
#endif
|
| 496 |
} |
| 497 |
} |
| 498 |
|
| 499 |
static inline void gen_op_add_reg_T0(int size, int reg) |
| 500 |
{
|
| 501 |
switch(size) {
|
| 502 |
case OT_BYTE:
|
| 503 |
tcg_gen_add_tl(cpu_tmp0, cpu_regs[reg], cpu_T[0]);
|
| 504 |
tcg_gen_deposit_tl(cpu_regs[reg], cpu_regs[reg], cpu_tmp0, 0, 16); |
| 505 |
break;
|
| 506 |
case OT_WORD:
|
| 507 |
tcg_gen_add_tl(cpu_tmp0, cpu_regs[reg], cpu_T[0]);
|
| 508 |
/* For x86_64, this sets the higher half of register to zero.
|
| 509 |
For i386, this is equivalent to a nop. */
|
| 510 |
tcg_gen_ext32u_tl(cpu_tmp0, cpu_tmp0); |
| 511 |
tcg_gen_mov_tl(cpu_regs[reg], cpu_tmp0); |
| 512 |
break;
|
| 513 |
#ifdef TARGET_X86_64
|
| 514 |
case OT_LONG:
|
| 515 |
tcg_gen_add_tl(cpu_regs[reg], cpu_regs[reg], cpu_T[0]);
|
| 516 |
break;
|
| 517 |
#endif
|
| 518 |
} |
| 519 |
} |
| 520 |
|
| 521 |
static inline void gen_op_addl_A0_reg_sN(int shift, int reg) |
| 522 |
{
|
| 523 |
tcg_gen_mov_tl(cpu_tmp0, cpu_regs[reg]); |
| 524 |
if (shift != 0) |
| 525 |
tcg_gen_shli_tl(cpu_tmp0, cpu_tmp0, shift); |
| 526 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 527 |
/* For x86_64, this sets the higher half of register to zero.
|
| 528 |
For i386, this is equivalent to a nop. */
|
| 529 |
tcg_gen_ext32u_tl(cpu_A0, cpu_A0); |
| 530 |
} |
| 531 |
|
| 532 |
static inline void gen_op_movl_A0_seg(int reg) |
| 533 |
{
|
| 534 |
tcg_gen_ld32u_tl(cpu_A0, cpu_env, offsetof(CPUX86State, segs[reg].base) + REG_L_OFFSET); |
| 535 |
} |
| 536 |
|
| 537 |
static inline void gen_op_addl_A0_seg(DisasContext *s, int reg) |
| 538 |
{
|
| 539 |
tcg_gen_ld_tl(cpu_tmp0, cpu_env, offsetof(CPUX86State, segs[reg].base)); |
| 540 |
#ifdef TARGET_X86_64
|
| 541 |
if (CODE64(s)) {
|
| 542 |
tcg_gen_andi_tl(cpu_A0, cpu_A0, 0xffffffff);
|
| 543 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 544 |
} else {
|
| 545 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 546 |
tcg_gen_andi_tl(cpu_A0, cpu_A0, 0xffffffff);
|
| 547 |
} |
| 548 |
#else
|
| 549 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 550 |
#endif
|
| 551 |
} |
| 552 |
|
| 553 |
#ifdef TARGET_X86_64
|
| 554 |
static inline void gen_op_movq_A0_seg(int reg) |
| 555 |
{
|
| 556 |
tcg_gen_ld_tl(cpu_A0, cpu_env, offsetof(CPUX86State, segs[reg].base)); |
| 557 |
} |
| 558 |
|
| 559 |
static inline void gen_op_addq_A0_seg(int reg) |
| 560 |
{
|
| 561 |
tcg_gen_ld_tl(cpu_tmp0, cpu_env, offsetof(CPUX86State, segs[reg].base)); |
| 562 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 563 |
} |
| 564 |
|
| 565 |
static inline void gen_op_movq_A0_reg(int reg) |
| 566 |
{
|
| 567 |
tcg_gen_mov_tl(cpu_A0, cpu_regs[reg]); |
| 568 |
} |
| 569 |
|
| 570 |
static inline void gen_op_addq_A0_reg_sN(int shift, int reg) |
| 571 |
{
|
| 572 |
tcg_gen_mov_tl(cpu_tmp0, cpu_regs[reg]); |
| 573 |
if (shift != 0) |
| 574 |
tcg_gen_shli_tl(cpu_tmp0, cpu_tmp0, shift); |
| 575 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 576 |
} |
| 577 |
#endif
|
| 578 |
|
| 579 |
static inline void gen_op_lds_T0_A0(int idx) |
| 580 |
{
|
| 581 |
int mem_index = (idx >> 2) - 1; |
| 582 |
switch(idx & 3) { |
| 583 |
case OT_BYTE:
|
| 584 |
tcg_gen_qemu_ld8s(cpu_T[0], cpu_A0, mem_index);
|
| 585 |
break;
|
| 586 |
case OT_WORD:
|
| 587 |
tcg_gen_qemu_ld16s(cpu_T[0], cpu_A0, mem_index);
|
| 588 |
break;
|
| 589 |
default:
|
| 590 |
case OT_LONG:
|
| 591 |
tcg_gen_qemu_ld32s(cpu_T[0], cpu_A0, mem_index);
|
| 592 |
break;
|
| 593 |
} |
| 594 |
} |
| 595 |
|
| 596 |
static inline void gen_op_ld_v(int idx, TCGv t0, TCGv a0) |
| 597 |
{
|
| 598 |
int mem_index = (idx >> 2) - 1; |
| 599 |
switch(idx & 3) { |
| 600 |
case OT_BYTE:
|
| 601 |
tcg_gen_qemu_ld8u(t0, a0, mem_index); |
| 602 |
break;
|
| 603 |
case OT_WORD:
|
| 604 |
tcg_gen_qemu_ld16u(t0, a0, mem_index); |
| 605 |
break;
|
| 606 |
case OT_LONG:
|
| 607 |
tcg_gen_qemu_ld32u(t0, a0, mem_index); |
| 608 |
break;
|
| 609 |
default:
|
| 610 |
case OT_QUAD:
|
| 611 |
/* Should never happen on 32-bit targets. */
|
| 612 |
#ifdef TARGET_X86_64
|
| 613 |
tcg_gen_qemu_ld64(t0, a0, mem_index); |
| 614 |
#endif
|
| 615 |
break;
|
| 616 |
} |
| 617 |
} |
| 618 |
|
| 619 |
/* XXX: always use ldu or lds */
|
| 620 |
static inline void gen_op_ld_T0_A0(int idx) |
| 621 |
{
|
| 622 |
gen_op_ld_v(idx, cpu_T[0], cpu_A0);
|
| 623 |
} |
| 624 |
|
| 625 |
static inline void gen_op_ldu_T0_A0(int idx) |
| 626 |
{
|
| 627 |
gen_op_ld_v(idx, cpu_T[0], cpu_A0);
|
| 628 |
} |
| 629 |
|
| 630 |
static inline void gen_op_ld_T1_A0(int idx) |
| 631 |
{
|
| 632 |
gen_op_ld_v(idx, cpu_T[1], cpu_A0);
|
| 633 |
} |
| 634 |
|
| 635 |
static inline void gen_op_st_v(int idx, TCGv t0, TCGv a0) |
| 636 |
{
|
| 637 |
int mem_index = (idx >> 2) - 1; |
| 638 |
switch(idx & 3) { |
| 639 |
case OT_BYTE:
|
| 640 |
tcg_gen_qemu_st8(t0, a0, mem_index); |
| 641 |
break;
|
| 642 |
case OT_WORD:
|
| 643 |
tcg_gen_qemu_st16(t0, a0, mem_index); |
| 644 |
break;
|
| 645 |
case OT_LONG:
|
| 646 |
tcg_gen_qemu_st32(t0, a0, mem_index); |
| 647 |
break;
|
| 648 |
default:
|
| 649 |
case OT_QUAD:
|
| 650 |
/* Should never happen on 32-bit targets. */
|
| 651 |
#ifdef TARGET_X86_64
|
| 652 |
tcg_gen_qemu_st64(t0, a0, mem_index); |
| 653 |
#endif
|
| 654 |
break;
|
| 655 |
} |
| 656 |
} |
| 657 |
|
| 658 |
static inline void gen_op_st_T0_A0(int idx) |
| 659 |
{
|
| 660 |
gen_op_st_v(idx, cpu_T[0], cpu_A0);
|
| 661 |
} |
| 662 |
|
| 663 |
static inline void gen_op_st_T1_A0(int idx) |
| 664 |
{
|
| 665 |
gen_op_st_v(idx, cpu_T[1], cpu_A0);
|
| 666 |
} |
| 667 |
|
| 668 |
static inline void gen_jmp_im(target_ulong pc) |
| 669 |
{
|
| 670 |
tcg_gen_movi_tl(cpu_tmp0, pc); |
| 671 |
tcg_gen_st_tl(cpu_tmp0, cpu_env, offsetof(CPUX86State, eip)); |
| 672 |
} |
| 673 |
|
| 674 |
static inline void gen_string_movl_A0_ESI(DisasContext *s) |
| 675 |
{
|
| 676 |
int override;
|
| 677 |
|
| 678 |
override = s->override; |
| 679 |
#ifdef TARGET_X86_64
|
| 680 |
if (s->aflag == 2) { |
| 681 |
if (override >= 0) { |
| 682 |
gen_op_movq_A0_seg(override); |
| 683 |
gen_op_addq_A0_reg_sN(0, R_ESI);
|
| 684 |
} else {
|
| 685 |
gen_op_movq_A0_reg(R_ESI); |
| 686 |
} |
| 687 |
} else
|
| 688 |
#endif
|
| 689 |
if (s->aflag) {
|
| 690 |
/* 32 bit address */
|
| 691 |
if (s->addseg && override < 0) |
| 692 |
override = R_DS; |
| 693 |
if (override >= 0) { |
| 694 |
gen_op_movl_A0_seg(override); |
| 695 |
gen_op_addl_A0_reg_sN(0, R_ESI);
|
| 696 |
} else {
|
| 697 |
gen_op_movl_A0_reg(R_ESI); |
| 698 |
} |
| 699 |
} else {
|
| 700 |
/* 16 address, always override */
|
| 701 |
if (override < 0) |
| 702 |
override = R_DS; |
| 703 |
gen_op_movl_A0_reg(R_ESI); |
| 704 |
gen_op_andl_A0_ffff(); |
| 705 |
gen_op_addl_A0_seg(s, override); |
| 706 |
} |
| 707 |
} |
| 708 |
|
| 709 |
static inline void gen_string_movl_A0_EDI(DisasContext *s) |
| 710 |
{
|
| 711 |
#ifdef TARGET_X86_64
|
| 712 |
if (s->aflag == 2) { |
| 713 |
gen_op_movq_A0_reg(R_EDI); |
| 714 |
} else
|
| 715 |
#endif
|
| 716 |
if (s->aflag) {
|
| 717 |
if (s->addseg) {
|
| 718 |
gen_op_movl_A0_seg(R_ES); |
| 719 |
gen_op_addl_A0_reg_sN(0, R_EDI);
|
| 720 |
} else {
|
| 721 |
gen_op_movl_A0_reg(R_EDI); |
| 722 |
} |
| 723 |
} else {
|
| 724 |
gen_op_movl_A0_reg(R_EDI); |
| 725 |
gen_op_andl_A0_ffff(); |
| 726 |
gen_op_addl_A0_seg(s, R_ES); |
| 727 |
} |
| 728 |
} |
| 729 |
|
| 730 |
static inline void gen_op_movl_T0_Dshift(int ot) |
| 731 |
{
|
| 732 |
tcg_gen_ld32s_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, df));
|
| 733 |
tcg_gen_shli_tl(cpu_T[0], cpu_T[0], ot); |
| 734 |
}; |
| 735 |
|
| 736 |
static TCGv gen_ext_tl(TCGv dst, TCGv src, int size, bool sign) |
| 737 |
{
|
| 738 |
switch (size) {
|
| 739 |
case OT_BYTE:
|
| 740 |
if (sign) {
|
| 741 |
tcg_gen_ext8s_tl(dst, src); |
| 742 |
} else {
|
| 743 |
tcg_gen_ext8u_tl(dst, src); |
| 744 |
} |
| 745 |
return dst;
|
| 746 |
case OT_WORD:
|
| 747 |
if (sign) {
|
| 748 |
tcg_gen_ext16s_tl(dst, src); |
| 749 |
} else {
|
| 750 |
tcg_gen_ext16u_tl(dst, src); |
| 751 |
} |
| 752 |
return dst;
|
| 753 |
#ifdef TARGET_X86_64
|
| 754 |
case OT_LONG:
|
| 755 |
if (sign) {
|
| 756 |
tcg_gen_ext32s_tl(dst, src); |
| 757 |
} else {
|
| 758 |
tcg_gen_ext32u_tl(dst, src); |
| 759 |
} |
| 760 |
return dst;
|
| 761 |
#endif
|
| 762 |
default:
|
| 763 |
return src;
|
| 764 |
} |
| 765 |
} |
| 766 |
|
| 767 |
static void gen_extu(int ot, TCGv reg) |
| 768 |
{
|
| 769 |
gen_ext_tl(reg, reg, ot, false);
|
| 770 |
} |
| 771 |
|
| 772 |
static void gen_exts(int ot, TCGv reg) |
| 773 |
{
|
| 774 |
gen_ext_tl(reg, reg, ot, true);
|
| 775 |
} |
| 776 |
|
| 777 |
static inline void gen_op_jnz_ecx(int size, int label1) |
| 778 |
{
|
| 779 |
tcg_gen_mov_tl(cpu_tmp0, cpu_regs[R_ECX]); |
| 780 |
gen_extu(size + 1, cpu_tmp0);
|
| 781 |
tcg_gen_brcondi_tl(TCG_COND_NE, cpu_tmp0, 0, label1);
|
| 782 |
} |
| 783 |
|
| 784 |
static inline void gen_op_jz_ecx(int size, int label1) |
| 785 |
{
|
| 786 |
tcg_gen_mov_tl(cpu_tmp0, cpu_regs[R_ECX]); |
| 787 |
gen_extu(size + 1, cpu_tmp0);
|
| 788 |
tcg_gen_brcondi_tl(TCG_COND_EQ, cpu_tmp0, 0, label1);
|
| 789 |
} |
| 790 |
|
| 791 |
static void gen_helper_in_func(int ot, TCGv v, TCGv_i32 n) |
| 792 |
{
|
| 793 |
switch (ot) {
|
| 794 |
case OT_BYTE:
|
| 795 |
gen_helper_inb(v, n); |
| 796 |
break;
|
| 797 |
case OT_WORD:
|
| 798 |
gen_helper_inw(v, n); |
| 799 |
break;
|
| 800 |
case OT_LONG:
|
| 801 |
gen_helper_inl(v, n); |
| 802 |
break;
|
| 803 |
} |
| 804 |
} |
| 805 |
|
| 806 |
static void gen_helper_out_func(int ot, TCGv_i32 v, TCGv_i32 n) |
| 807 |
{
|
| 808 |
switch (ot) {
|
| 809 |
case OT_BYTE:
|
| 810 |
gen_helper_outb(v, n); |
| 811 |
break;
|
| 812 |
case OT_WORD:
|
| 813 |
gen_helper_outw(v, n); |
| 814 |
break;
|
| 815 |
case OT_LONG:
|
| 816 |
gen_helper_outl(v, n); |
| 817 |
break;
|
| 818 |
} |
| 819 |
} |
| 820 |
|
| 821 |
static void gen_check_io(DisasContext *s, int ot, target_ulong cur_eip, |
| 822 |
uint32_t svm_flags) |
| 823 |
{
|
| 824 |
int state_saved;
|
| 825 |
target_ulong next_eip; |
| 826 |
|
| 827 |
state_saved = 0;
|
| 828 |
if (s->pe && (s->cpl > s->iopl || s->vm86)) {
|
| 829 |
gen_update_cc_op(s); |
| 830 |
gen_jmp_im(cur_eip); |
| 831 |
state_saved = 1;
|
| 832 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 833 |
switch (ot) {
|
| 834 |
case OT_BYTE:
|
| 835 |
gen_helper_check_iob(cpu_env, cpu_tmp2_i32); |
| 836 |
break;
|
| 837 |
case OT_WORD:
|
| 838 |
gen_helper_check_iow(cpu_env, cpu_tmp2_i32); |
| 839 |
break;
|
| 840 |
case OT_LONG:
|
| 841 |
gen_helper_check_iol(cpu_env, cpu_tmp2_i32); |
| 842 |
break;
|
| 843 |
} |
| 844 |
} |
| 845 |
if(s->flags & HF_SVMI_MASK) {
|
| 846 |
if (!state_saved) {
|
| 847 |
gen_update_cc_op(s); |
| 848 |
gen_jmp_im(cur_eip); |
| 849 |
} |
| 850 |
svm_flags |= (1 << (4 + ot)); |
| 851 |
next_eip = s->pc - s->cs_base; |
| 852 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 853 |
gen_helper_svm_check_io(cpu_env, cpu_tmp2_i32, |
| 854 |
tcg_const_i32(svm_flags), |
| 855 |
tcg_const_i32(next_eip - cur_eip)); |
| 856 |
} |
| 857 |
} |
| 858 |
|
| 859 |
static inline void gen_movs(DisasContext *s, int ot) |
| 860 |
{
|
| 861 |
gen_string_movl_A0_ESI(s); |
| 862 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 863 |
gen_string_movl_A0_EDI(s); |
| 864 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 865 |
gen_op_movl_T0_Dshift(ot); |
| 866 |
gen_op_add_reg_T0(s->aflag, R_ESI); |
| 867 |
gen_op_add_reg_T0(s->aflag, R_EDI); |
| 868 |
} |
| 869 |
|
| 870 |
static void gen_op_update1_cc(void) |
| 871 |
{
|
| 872 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 873 |
} |
| 874 |
|
| 875 |
static void gen_op_update2_cc(void) |
| 876 |
{
|
| 877 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[1]);
|
| 878 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 879 |
} |
| 880 |
|
| 881 |
static void gen_op_update3_cc(TCGv reg) |
| 882 |
{
|
| 883 |
tcg_gen_mov_tl(cpu_cc_src2, reg); |
| 884 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[1]);
|
| 885 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 886 |
} |
| 887 |
|
| 888 |
static inline void gen_op_testl_T0_T1_cc(void) |
| 889 |
{
|
| 890 |
tcg_gen_and_tl(cpu_cc_dst, cpu_T[0], cpu_T[1]); |
| 891 |
} |
| 892 |
|
| 893 |
static void gen_op_update_neg_cc(void) |
| 894 |
{
|
| 895 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 896 |
tcg_gen_neg_tl(cpu_cc_src, cpu_T[0]);
|
| 897 |
tcg_gen_movi_tl(cpu_cc_srcT, 0);
|
| 898 |
} |
| 899 |
|
| 900 |
/* compute all eflags to cc_src */
|
| 901 |
static void gen_compute_eflags(DisasContext *s) |
| 902 |
{
|
| 903 |
TCGv zero, dst, src1, src2; |
| 904 |
int live, dead;
|
| 905 |
|
| 906 |
if (s->cc_op == CC_OP_EFLAGS) {
|
| 907 |
return;
|
| 908 |
} |
| 909 |
|
| 910 |
TCGV_UNUSED(zero); |
| 911 |
dst = cpu_cc_dst; |
| 912 |
src1 = cpu_cc_src; |
| 913 |
src2 = cpu_cc_src2; |
| 914 |
|
| 915 |
/* Take care to not read values that are not live. */
|
| 916 |
live = cc_op_live[s->cc_op] & ~USES_CC_SRCT; |
| 917 |
dead = live ^ (USES_CC_DST | USES_CC_SRC | USES_CC_SRC2); |
| 918 |
if (dead) {
|
| 919 |
zero = tcg_const_tl(0);
|
| 920 |
if (dead & USES_CC_DST) {
|
| 921 |
dst = zero; |
| 922 |
} |
| 923 |
if (dead & USES_CC_SRC) {
|
| 924 |
src1 = zero; |
| 925 |
} |
| 926 |
if (dead & USES_CC_SRC2) {
|
| 927 |
src2 = zero; |
| 928 |
} |
| 929 |
} |
| 930 |
|
| 931 |
gen_update_cc_op(s); |
| 932 |
gen_helper_cc_compute_all(cpu_cc_src, dst, src1, src2, cpu_cc_op); |
| 933 |
set_cc_op(s, CC_OP_EFLAGS); |
| 934 |
|
| 935 |
if (dead) {
|
| 936 |
tcg_temp_free(zero); |
| 937 |
} |
| 938 |
} |
| 939 |
|
| 940 |
typedef struct CCPrepare { |
| 941 |
TCGCond cond; |
| 942 |
TCGv reg; |
| 943 |
TCGv reg2; |
| 944 |
target_ulong imm; |
| 945 |
target_ulong mask; |
| 946 |
bool use_reg2;
|
| 947 |
bool no_setcond;
|
| 948 |
} CCPrepare; |
| 949 |
|
| 950 |
/* compute eflags.C to reg */
|
| 951 |
static CCPrepare gen_prepare_eflags_c(DisasContext *s, TCGv reg)
|
| 952 |
{
|
| 953 |
TCGv t0, t1; |
| 954 |
int size, shift;
|
| 955 |
|
| 956 |
switch (s->cc_op) {
|
| 957 |
case CC_OP_SUBB ... CC_OP_SUBQ:
|
| 958 |
/* (DATA_TYPE)CC_SRCT < (DATA_TYPE)CC_SRC */
|
| 959 |
size = s->cc_op - CC_OP_SUBB; |
| 960 |
t1 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false);
|
| 961 |
/* If no temporary was used, be careful not to alias t1 and t0. */
|
| 962 |
t0 = TCGV_EQUAL(t1, cpu_cc_src) ? cpu_tmp0 : reg; |
| 963 |
tcg_gen_mov_tl(t0, cpu_cc_srcT); |
| 964 |
gen_extu(size, t0); |
| 965 |
goto add_sub;
|
| 966 |
|
| 967 |
case CC_OP_ADDB ... CC_OP_ADDQ:
|
| 968 |
/* (DATA_TYPE)CC_DST < (DATA_TYPE)CC_SRC */
|
| 969 |
size = s->cc_op - CC_OP_ADDB; |
| 970 |
t1 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false);
|
| 971 |
t0 = gen_ext_tl(reg, cpu_cc_dst, size, false);
|
| 972 |
add_sub:
|
| 973 |
return (CCPrepare) { .cond = TCG_COND_LTU, .reg = t0,
|
| 974 |
.reg2 = t1, .mask = -1, .use_reg2 = true }; |
| 975 |
|
| 976 |
case CC_OP_LOGICB ... CC_OP_LOGICQ:
|
| 977 |
return (CCPrepare) { .cond = TCG_COND_NEVER, .mask = -1 }; |
| 978 |
|
| 979 |
case CC_OP_INCB ... CC_OP_INCQ:
|
| 980 |
case CC_OP_DECB ... CC_OP_DECQ:
|
| 981 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 982 |
.mask = -1, .no_setcond = true }; |
| 983 |
|
| 984 |
case CC_OP_SHLB ... CC_OP_SHLQ:
|
| 985 |
/* (CC_SRC >> (DATA_BITS - 1)) & 1 */
|
| 986 |
size = s->cc_op - CC_OP_SHLB; |
| 987 |
shift = (8 << size) - 1; |
| 988 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 989 |
.mask = (target_ulong)1 << shift };
|
| 990 |
|
| 991 |
case CC_OP_MULB ... CC_OP_MULQ:
|
| 992 |
return (CCPrepare) { .cond = TCG_COND_NE,
|
| 993 |
.reg = cpu_cc_src, .mask = -1 };
|
| 994 |
|
| 995 |
case CC_OP_BMILGB ... CC_OP_BMILGQ:
|
| 996 |
size = s->cc_op - CC_OP_BMILGB; |
| 997 |
t0 = gen_ext_tl(reg, cpu_cc_src, size, false);
|
| 998 |
return (CCPrepare) { .cond = TCG_COND_EQ, .reg = t0, .mask = -1 }; |
| 999 |
|
| 1000 |
case CC_OP_ADCX:
|
| 1001 |
case CC_OP_ADCOX:
|
| 1002 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_dst,
|
| 1003 |
.mask = -1, .no_setcond = true }; |
| 1004 |
|
| 1005 |
case CC_OP_EFLAGS:
|
| 1006 |
case CC_OP_SARB ... CC_OP_SARQ:
|
| 1007 |
/* CC_SRC & 1 */
|
| 1008 |
return (CCPrepare) { .cond = TCG_COND_NE,
|
| 1009 |
.reg = cpu_cc_src, .mask = CC_C }; |
| 1010 |
|
| 1011 |
default:
|
| 1012 |
/* The need to compute only C from CC_OP_DYNAMIC is important
|
| 1013 |
in efficiently implementing e.g. INC at the start of a TB. */
|
| 1014 |
gen_update_cc_op(s); |
| 1015 |
gen_helper_cc_compute_c(reg, cpu_cc_dst, cpu_cc_src, |
| 1016 |
cpu_cc_src2, cpu_cc_op); |
| 1017 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = reg,
|
| 1018 |
.mask = -1, .no_setcond = true }; |
| 1019 |
} |
| 1020 |
} |
| 1021 |
|
| 1022 |
/* compute eflags.P to reg */
|
| 1023 |
static CCPrepare gen_prepare_eflags_p(DisasContext *s, TCGv reg)
|
| 1024 |
{
|
| 1025 |
gen_compute_eflags(s); |
| 1026 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 1027 |
.mask = CC_P }; |
| 1028 |
} |
| 1029 |
|
| 1030 |
/* compute eflags.S to reg */
|
| 1031 |
static CCPrepare gen_prepare_eflags_s(DisasContext *s, TCGv reg)
|
| 1032 |
{
|
| 1033 |
switch (s->cc_op) {
|
| 1034 |
case CC_OP_DYNAMIC:
|
| 1035 |
gen_compute_eflags(s); |
| 1036 |
/* FALLTHRU */
|
| 1037 |
case CC_OP_EFLAGS:
|
| 1038 |
case CC_OP_ADCX:
|
| 1039 |
case CC_OP_ADOX:
|
| 1040 |
case CC_OP_ADCOX:
|
| 1041 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 1042 |
.mask = CC_S }; |
| 1043 |
default:
|
| 1044 |
{
|
| 1045 |
int size = (s->cc_op - CC_OP_ADDB) & 3; |
| 1046 |
TCGv t0 = gen_ext_tl(reg, cpu_cc_dst, size, true);
|
| 1047 |
return (CCPrepare) { .cond = TCG_COND_LT, .reg = t0, .mask = -1 }; |
| 1048 |
} |
| 1049 |
} |
| 1050 |
} |
| 1051 |
|
| 1052 |
/* compute eflags.O to reg */
|
| 1053 |
static CCPrepare gen_prepare_eflags_o(DisasContext *s, TCGv reg)
|
| 1054 |
{
|
| 1055 |
switch (s->cc_op) {
|
| 1056 |
case CC_OP_ADOX:
|
| 1057 |
case CC_OP_ADCOX:
|
| 1058 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src2,
|
| 1059 |
.mask = -1, .no_setcond = true }; |
| 1060 |
|
| 1061 |
default:
|
| 1062 |
gen_compute_eflags(s); |
| 1063 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 1064 |
.mask = CC_O }; |
| 1065 |
} |
| 1066 |
} |
| 1067 |
|
| 1068 |
/* compute eflags.Z to reg */
|
| 1069 |
static CCPrepare gen_prepare_eflags_z(DisasContext *s, TCGv reg)
|
| 1070 |
{
|
| 1071 |
switch (s->cc_op) {
|
| 1072 |
case CC_OP_DYNAMIC:
|
| 1073 |
gen_compute_eflags(s); |
| 1074 |
/* FALLTHRU */
|
| 1075 |
case CC_OP_EFLAGS:
|
| 1076 |
case CC_OP_ADCX:
|
| 1077 |
case CC_OP_ADOX:
|
| 1078 |
case CC_OP_ADCOX:
|
| 1079 |
return (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 1080 |
.mask = CC_Z }; |
| 1081 |
default:
|
| 1082 |
{
|
| 1083 |
int size = (s->cc_op - CC_OP_ADDB) & 3; |
| 1084 |
TCGv t0 = gen_ext_tl(reg, cpu_cc_dst, size, false);
|
| 1085 |
return (CCPrepare) { .cond = TCG_COND_EQ, .reg = t0, .mask = -1 }; |
| 1086 |
} |
| 1087 |
} |
| 1088 |
} |
| 1089 |
|
| 1090 |
/* perform a conditional store into register 'reg' according to jump opcode
|
| 1091 |
value 'b'. In the fast case, T0 is guaranted not to be used. */
|
| 1092 |
static CCPrepare gen_prepare_cc(DisasContext *s, int b, TCGv reg) |
| 1093 |
{
|
| 1094 |
int inv, jcc_op, size, cond;
|
| 1095 |
CCPrepare cc; |
| 1096 |
TCGv t0; |
| 1097 |
|
| 1098 |
inv = b & 1;
|
| 1099 |
jcc_op = (b >> 1) & 7; |
| 1100 |
|
| 1101 |
switch (s->cc_op) {
|
| 1102 |
case CC_OP_SUBB ... CC_OP_SUBQ:
|
| 1103 |
/* We optimize relational operators for the cmp/jcc case. */
|
| 1104 |
size = s->cc_op - CC_OP_SUBB; |
| 1105 |
switch (jcc_op) {
|
| 1106 |
case JCC_BE:
|
| 1107 |
tcg_gen_mov_tl(cpu_tmp4, cpu_cc_srcT); |
| 1108 |
gen_extu(size, cpu_tmp4); |
| 1109 |
t0 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false);
|
| 1110 |
cc = (CCPrepare) { .cond = TCG_COND_LEU, .reg = cpu_tmp4,
|
| 1111 |
.reg2 = t0, .mask = -1, .use_reg2 = true }; |
| 1112 |
break;
|
| 1113 |
|
| 1114 |
case JCC_L:
|
| 1115 |
cond = TCG_COND_LT; |
| 1116 |
goto fast_jcc_l;
|
| 1117 |
case JCC_LE:
|
| 1118 |
cond = TCG_COND_LE; |
| 1119 |
fast_jcc_l:
|
| 1120 |
tcg_gen_mov_tl(cpu_tmp4, cpu_cc_srcT); |
| 1121 |
gen_exts(size, cpu_tmp4); |
| 1122 |
t0 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, true);
|
| 1123 |
cc = (CCPrepare) { .cond = cond, .reg = cpu_tmp4,
|
| 1124 |
.reg2 = t0, .mask = -1, .use_reg2 = true }; |
| 1125 |
break;
|
| 1126 |
|
| 1127 |
default:
|
| 1128 |
goto slow_jcc;
|
| 1129 |
} |
| 1130 |
break;
|
| 1131 |
|
| 1132 |
default:
|
| 1133 |
slow_jcc:
|
| 1134 |
/* This actually generates good code for JC, JZ and JS. */
|
| 1135 |
switch (jcc_op) {
|
| 1136 |
case JCC_O:
|
| 1137 |
cc = gen_prepare_eflags_o(s, reg); |
| 1138 |
break;
|
| 1139 |
case JCC_B:
|
| 1140 |
cc = gen_prepare_eflags_c(s, reg); |
| 1141 |
break;
|
| 1142 |
case JCC_Z:
|
| 1143 |
cc = gen_prepare_eflags_z(s, reg); |
| 1144 |
break;
|
| 1145 |
case JCC_BE:
|
| 1146 |
gen_compute_eflags(s); |
| 1147 |
cc = (CCPrepare) { .cond = TCG_COND_NE, .reg = cpu_cc_src,
|
| 1148 |
.mask = CC_Z | CC_C }; |
| 1149 |
break;
|
| 1150 |
case JCC_S:
|
| 1151 |
cc = gen_prepare_eflags_s(s, reg); |
| 1152 |
break;
|
| 1153 |
case JCC_P:
|
| 1154 |
cc = gen_prepare_eflags_p(s, reg); |
| 1155 |
break;
|
| 1156 |
case JCC_L:
|
| 1157 |
gen_compute_eflags(s); |
| 1158 |
if (TCGV_EQUAL(reg, cpu_cc_src)) {
|
| 1159 |
reg = cpu_tmp0; |
| 1160 |
} |
| 1161 |
tcg_gen_shri_tl(reg, cpu_cc_src, 4); /* CC_O -> CC_S */ |
| 1162 |
tcg_gen_xor_tl(reg, reg, cpu_cc_src); |
| 1163 |
cc = (CCPrepare) { .cond = TCG_COND_NE, .reg = reg,
|
| 1164 |
.mask = CC_S }; |
| 1165 |
break;
|
| 1166 |
default:
|
| 1167 |
case JCC_LE:
|
| 1168 |
gen_compute_eflags(s); |
| 1169 |
if (TCGV_EQUAL(reg, cpu_cc_src)) {
|
| 1170 |
reg = cpu_tmp0; |
| 1171 |
} |
| 1172 |
tcg_gen_shri_tl(reg, cpu_cc_src, 4); /* CC_O -> CC_S */ |
| 1173 |
tcg_gen_xor_tl(reg, reg, cpu_cc_src); |
| 1174 |
cc = (CCPrepare) { .cond = TCG_COND_NE, .reg = reg,
|
| 1175 |
.mask = CC_S | CC_Z }; |
| 1176 |
break;
|
| 1177 |
} |
| 1178 |
break;
|
| 1179 |
} |
| 1180 |
|
| 1181 |
if (inv) {
|
| 1182 |
cc.cond = tcg_invert_cond(cc.cond); |
| 1183 |
} |
| 1184 |
return cc;
|
| 1185 |
} |
| 1186 |
|
| 1187 |
static void gen_setcc1(DisasContext *s, int b, TCGv reg) |
| 1188 |
{
|
| 1189 |
CCPrepare cc = gen_prepare_cc(s, b, reg); |
| 1190 |
|
| 1191 |
if (cc.no_setcond) {
|
| 1192 |
if (cc.cond == TCG_COND_EQ) {
|
| 1193 |
tcg_gen_xori_tl(reg, cc.reg, 1);
|
| 1194 |
} else {
|
| 1195 |
tcg_gen_mov_tl(reg, cc.reg); |
| 1196 |
} |
| 1197 |
return;
|
| 1198 |
} |
| 1199 |
|
| 1200 |
if (cc.cond == TCG_COND_NE && !cc.use_reg2 && cc.imm == 0 && |
| 1201 |
cc.mask != 0 && (cc.mask & (cc.mask - 1)) == 0) { |
| 1202 |
tcg_gen_shri_tl(reg, cc.reg, ctztl(cc.mask)); |
| 1203 |
tcg_gen_andi_tl(reg, reg, 1);
|
| 1204 |
return;
|
| 1205 |
} |
| 1206 |
if (cc.mask != -1) { |
| 1207 |
tcg_gen_andi_tl(reg, cc.reg, cc.mask); |
| 1208 |
cc.reg = reg; |
| 1209 |
} |
| 1210 |
if (cc.use_reg2) {
|
| 1211 |
tcg_gen_setcond_tl(cc.cond, reg, cc.reg, cc.reg2); |
| 1212 |
} else {
|
| 1213 |
tcg_gen_setcondi_tl(cc.cond, reg, cc.reg, cc.imm); |
| 1214 |
} |
| 1215 |
} |
| 1216 |
|
| 1217 |
static inline void gen_compute_eflags_c(DisasContext *s, TCGv reg) |
| 1218 |
{
|
| 1219 |
gen_setcc1(s, JCC_B << 1, reg);
|
| 1220 |
} |
| 1221 |
|
| 1222 |
/* generate a conditional jump to label 'l1' according to jump opcode
|
| 1223 |
value 'b'. In the fast case, T0 is guaranted not to be used. */
|
| 1224 |
static inline void gen_jcc1_noeob(DisasContext *s, int b, int l1) |
| 1225 |
{
|
| 1226 |
CCPrepare cc = gen_prepare_cc(s, b, cpu_T[0]);
|
| 1227 |
|
| 1228 |
if (cc.mask != -1) { |
| 1229 |
tcg_gen_andi_tl(cpu_T[0], cc.reg, cc.mask);
|
| 1230 |
cc.reg = cpu_T[0];
|
| 1231 |
} |
| 1232 |
if (cc.use_reg2) {
|
| 1233 |
tcg_gen_brcond_tl(cc.cond, cc.reg, cc.reg2, l1); |
| 1234 |
} else {
|
| 1235 |
tcg_gen_brcondi_tl(cc.cond, cc.reg, cc.imm, l1); |
| 1236 |
} |
| 1237 |
} |
| 1238 |
|
| 1239 |
/* Generate a conditional jump to label 'l1' according to jump opcode
|
| 1240 |
value 'b'. In the fast case, T0 is guaranted not to be used.
|
| 1241 |
A translation block must end soon. */
|
| 1242 |
static inline void gen_jcc1(DisasContext *s, int b, int l1) |
| 1243 |
{
|
| 1244 |
CCPrepare cc = gen_prepare_cc(s, b, cpu_T[0]);
|
| 1245 |
|
| 1246 |
gen_update_cc_op(s); |
| 1247 |
if (cc.mask != -1) { |
| 1248 |
tcg_gen_andi_tl(cpu_T[0], cc.reg, cc.mask);
|
| 1249 |
cc.reg = cpu_T[0];
|
| 1250 |
} |
| 1251 |
set_cc_op(s, CC_OP_DYNAMIC); |
| 1252 |
if (cc.use_reg2) {
|
| 1253 |
tcg_gen_brcond_tl(cc.cond, cc.reg, cc.reg2, l1); |
| 1254 |
} else {
|
| 1255 |
tcg_gen_brcondi_tl(cc.cond, cc.reg, cc.imm, l1); |
| 1256 |
} |
| 1257 |
} |
| 1258 |
|
| 1259 |
/* XXX: does not work with gdbstub "ice" single step - not a
|
| 1260 |
serious problem */
|
| 1261 |
static int gen_jz_ecx_string(DisasContext *s, target_ulong next_eip) |
| 1262 |
{
|
| 1263 |
int l1, l2;
|
| 1264 |
|
| 1265 |
l1 = gen_new_label(); |
| 1266 |
l2 = gen_new_label(); |
| 1267 |
gen_op_jnz_ecx(s->aflag, l1); |
| 1268 |
gen_set_label(l2); |
| 1269 |
gen_jmp_tb(s, next_eip, 1);
|
| 1270 |
gen_set_label(l1); |
| 1271 |
return l2;
|
| 1272 |
} |
| 1273 |
|
| 1274 |
static inline void gen_stos(DisasContext *s, int ot) |
| 1275 |
{
|
| 1276 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EAX);
|
| 1277 |
gen_string_movl_A0_EDI(s); |
| 1278 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 1279 |
gen_op_movl_T0_Dshift(ot); |
| 1280 |
gen_op_add_reg_T0(s->aflag, R_EDI); |
| 1281 |
} |
| 1282 |
|
| 1283 |
static inline void gen_lods(DisasContext *s, int ot) |
| 1284 |
{
|
| 1285 |
gen_string_movl_A0_ESI(s); |
| 1286 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 1287 |
gen_op_mov_reg_T0(ot, R_EAX); |
| 1288 |
gen_op_movl_T0_Dshift(ot); |
| 1289 |
gen_op_add_reg_T0(s->aflag, R_ESI); |
| 1290 |
} |
| 1291 |
|
| 1292 |
static inline void gen_scas(DisasContext *s, int ot) |
| 1293 |
{
|
| 1294 |
gen_string_movl_A0_EDI(s); |
| 1295 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 1296 |
gen_op(s, OP_CMPL, ot, R_EAX); |
| 1297 |
gen_op_movl_T0_Dshift(ot); |
| 1298 |
gen_op_add_reg_T0(s->aflag, R_EDI); |
| 1299 |
} |
| 1300 |
|
| 1301 |
static inline void gen_cmps(DisasContext *s, int ot) |
| 1302 |
{
|
| 1303 |
gen_string_movl_A0_EDI(s); |
| 1304 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 1305 |
gen_string_movl_A0_ESI(s); |
| 1306 |
gen_op(s, OP_CMPL, ot, OR_TMP0); |
| 1307 |
gen_op_movl_T0_Dshift(ot); |
| 1308 |
gen_op_add_reg_T0(s->aflag, R_ESI); |
| 1309 |
gen_op_add_reg_T0(s->aflag, R_EDI); |
| 1310 |
} |
| 1311 |
|
| 1312 |
static inline void gen_ins(DisasContext *s, int ot) |
| 1313 |
{
|
| 1314 |
if (use_icount)
|
| 1315 |
gen_io_start(); |
| 1316 |
gen_string_movl_A0_EDI(s); |
| 1317 |
/* Note: we must do this dummy write first to be restartable in
|
| 1318 |
case of page fault. */
|
| 1319 |
gen_op_movl_T0_0(); |
| 1320 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 1321 |
gen_op_mov_TN_reg(OT_WORD, 1, R_EDX);
|
| 1322 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[1]);
|
| 1323 |
tcg_gen_andi_i32(cpu_tmp2_i32, cpu_tmp2_i32, 0xffff);
|
| 1324 |
gen_helper_in_func(ot, cpu_T[0], cpu_tmp2_i32);
|
| 1325 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 1326 |
gen_op_movl_T0_Dshift(ot); |
| 1327 |
gen_op_add_reg_T0(s->aflag, R_EDI); |
| 1328 |
if (use_icount)
|
| 1329 |
gen_io_end(); |
| 1330 |
} |
| 1331 |
|
| 1332 |
static inline void gen_outs(DisasContext *s, int ot) |
| 1333 |
{
|
| 1334 |
if (use_icount)
|
| 1335 |
gen_io_start(); |
| 1336 |
gen_string_movl_A0_ESI(s); |
| 1337 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 1338 |
|
| 1339 |
gen_op_mov_TN_reg(OT_WORD, 1, R_EDX);
|
| 1340 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[1]);
|
| 1341 |
tcg_gen_andi_i32(cpu_tmp2_i32, cpu_tmp2_i32, 0xffff);
|
| 1342 |
tcg_gen_trunc_tl_i32(cpu_tmp3_i32, cpu_T[0]);
|
| 1343 |
gen_helper_out_func(ot, cpu_tmp2_i32, cpu_tmp3_i32); |
| 1344 |
|
| 1345 |
gen_op_movl_T0_Dshift(ot); |
| 1346 |
gen_op_add_reg_T0(s->aflag, R_ESI); |
| 1347 |
if (use_icount)
|
| 1348 |
gen_io_end(); |
| 1349 |
} |
| 1350 |
|
| 1351 |
/* same method as Valgrind : we generate jumps to current or next
|
| 1352 |
instruction */
|
| 1353 |
#define GEN_REPZ(op) \
|
| 1354 |
static inline void gen_repz_ ## op(DisasContext *s, int ot, \ |
| 1355 |
target_ulong cur_eip, target_ulong next_eip) \ |
| 1356 |
{ \
|
| 1357 |
int l2;\
|
| 1358 |
gen_update_cc_op(s); \ |
| 1359 |
l2 = gen_jz_ecx_string(s, next_eip); \ |
| 1360 |
gen_ ## op(s, ot); \ |
| 1361 |
gen_op_add_reg_im(s->aflag, R_ECX, -1); \
|
| 1362 |
/* a loop would cause two single step exceptions if ECX = 1 \
|
| 1363 |
before rep string_insn */ \
|
| 1364 |
if (!s->jmp_opt) \
|
| 1365 |
gen_op_jz_ecx(s->aflag, l2); \ |
| 1366 |
gen_jmp(s, cur_eip); \ |
| 1367 |
} |
| 1368 |
|
| 1369 |
#define GEN_REPZ2(op) \
|
| 1370 |
static inline void gen_repz_ ## op(DisasContext *s, int ot, \ |
| 1371 |
target_ulong cur_eip, \ |
| 1372 |
target_ulong next_eip, \ |
| 1373 |
int nz) \
|
| 1374 |
{ \
|
| 1375 |
int l2;\
|
| 1376 |
gen_update_cc_op(s); \ |
| 1377 |
l2 = gen_jz_ecx_string(s, next_eip); \ |
| 1378 |
gen_ ## op(s, ot); \ |
| 1379 |
gen_op_add_reg_im(s->aflag, R_ECX, -1); \
|
| 1380 |
gen_update_cc_op(s); \ |
| 1381 |
gen_jcc1(s, (JCC_Z << 1) | (nz ^ 1), l2); \ |
| 1382 |
if (!s->jmp_opt) \
|
| 1383 |
gen_op_jz_ecx(s->aflag, l2); \ |
| 1384 |
gen_jmp(s, cur_eip); \ |
| 1385 |
} |
| 1386 |
|
| 1387 |
GEN_REPZ(movs) |
| 1388 |
GEN_REPZ(stos) |
| 1389 |
GEN_REPZ(lods) |
| 1390 |
GEN_REPZ(ins) |
| 1391 |
GEN_REPZ(outs) |
| 1392 |
GEN_REPZ2(scas) |
| 1393 |
GEN_REPZ2(cmps) |
| 1394 |
|
| 1395 |
static void gen_helper_fp_arith_ST0_FT0(int op) |
| 1396 |
{
|
| 1397 |
switch (op) {
|
| 1398 |
case 0: |
| 1399 |
gen_helper_fadd_ST0_FT0(cpu_env); |
| 1400 |
break;
|
| 1401 |
case 1: |
| 1402 |
gen_helper_fmul_ST0_FT0(cpu_env); |
| 1403 |
break;
|
| 1404 |
case 2: |
| 1405 |
gen_helper_fcom_ST0_FT0(cpu_env); |
| 1406 |
break;
|
| 1407 |
case 3: |
| 1408 |
gen_helper_fcom_ST0_FT0(cpu_env); |
| 1409 |
break;
|
| 1410 |
case 4: |
| 1411 |
gen_helper_fsub_ST0_FT0(cpu_env); |
| 1412 |
break;
|
| 1413 |
case 5: |
| 1414 |
gen_helper_fsubr_ST0_FT0(cpu_env); |
| 1415 |
break;
|
| 1416 |
case 6: |
| 1417 |
gen_helper_fdiv_ST0_FT0(cpu_env); |
| 1418 |
break;
|
| 1419 |
case 7: |
| 1420 |
gen_helper_fdivr_ST0_FT0(cpu_env); |
| 1421 |
break;
|
| 1422 |
} |
| 1423 |
} |
| 1424 |
|
| 1425 |
/* NOTE the exception in "r" op ordering */
|
| 1426 |
static void gen_helper_fp_arith_STN_ST0(int op, int opreg) |
| 1427 |
{
|
| 1428 |
TCGv_i32 tmp = tcg_const_i32(opreg); |
| 1429 |
switch (op) {
|
| 1430 |
case 0: |
| 1431 |
gen_helper_fadd_STN_ST0(cpu_env, tmp); |
| 1432 |
break;
|
| 1433 |
case 1: |
| 1434 |
gen_helper_fmul_STN_ST0(cpu_env, tmp); |
| 1435 |
break;
|
| 1436 |
case 4: |
| 1437 |
gen_helper_fsubr_STN_ST0(cpu_env, tmp); |
| 1438 |
break;
|
| 1439 |
case 5: |
| 1440 |
gen_helper_fsub_STN_ST0(cpu_env, tmp); |
| 1441 |
break;
|
| 1442 |
case 6: |
| 1443 |
gen_helper_fdivr_STN_ST0(cpu_env, tmp); |
| 1444 |
break;
|
| 1445 |
case 7: |
| 1446 |
gen_helper_fdiv_STN_ST0(cpu_env, tmp); |
| 1447 |
break;
|
| 1448 |
} |
| 1449 |
} |
| 1450 |
|
| 1451 |
/* if d == OR_TMP0, it means memory operand (address in A0) */
|
| 1452 |
static void gen_op(DisasContext *s1, int op, int ot, int d) |
| 1453 |
{
|
| 1454 |
if (d != OR_TMP0) {
|
| 1455 |
gen_op_mov_TN_reg(ot, 0, d);
|
| 1456 |
} else {
|
| 1457 |
gen_op_ld_T0_A0(ot + s1->mem_index); |
| 1458 |
} |
| 1459 |
switch(op) {
|
| 1460 |
case OP_ADCL:
|
| 1461 |
gen_compute_eflags_c(s1, cpu_tmp4); |
| 1462 |
tcg_gen_add_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 1463 |
tcg_gen_add_tl(cpu_T[0], cpu_T[0], cpu_tmp4); |
| 1464 |
if (d != OR_TMP0)
|
| 1465 |
gen_op_mov_reg_T0(ot, d); |
| 1466 |
else
|
| 1467 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1468 |
gen_op_update3_cc(cpu_tmp4); |
| 1469 |
set_cc_op(s1, CC_OP_ADCB + ot); |
| 1470 |
break;
|
| 1471 |
case OP_SBBL:
|
| 1472 |
gen_compute_eflags_c(s1, cpu_tmp4); |
| 1473 |
tcg_gen_sub_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 1474 |
tcg_gen_sub_tl(cpu_T[0], cpu_T[0], cpu_tmp4); |
| 1475 |
if (d != OR_TMP0)
|
| 1476 |
gen_op_mov_reg_T0(ot, d); |
| 1477 |
else
|
| 1478 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1479 |
gen_op_update3_cc(cpu_tmp4); |
| 1480 |
set_cc_op(s1, CC_OP_SBBB + ot); |
| 1481 |
break;
|
| 1482 |
case OP_ADDL:
|
| 1483 |
gen_op_addl_T0_T1(); |
| 1484 |
if (d != OR_TMP0)
|
| 1485 |
gen_op_mov_reg_T0(ot, d); |
| 1486 |
else
|
| 1487 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1488 |
gen_op_update2_cc(); |
| 1489 |
set_cc_op(s1, CC_OP_ADDB + ot); |
| 1490 |
break;
|
| 1491 |
case OP_SUBL:
|
| 1492 |
tcg_gen_mov_tl(cpu_cc_srcT, cpu_T[0]);
|
| 1493 |
tcg_gen_sub_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 1494 |
if (d != OR_TMP0)
|
| 1495 |
gen_op_mov_reg_T0(ot, d); |
| 1496 |
else
|
| 1497 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1498 |
gen_op_update2_cc(); |
| 1499 |
set_cc_op(s1, CC_OP_SUBB + ot); |
| 1500 |
break;
|
| 1501 |
default:
|
| 1502 |
case OP_ANDL:
|
| 1503 |
tcg_gen_and_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 1504 |
if (d != OR_TMP0)
|
| 1505 |
gen_op_mov_reg_T0(ot, d); |
| 1506 |
else
|
| 1507 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1508 |
gen_op_update1_cc(); |
| 1509 |
set_cc_op(s1, CC_OP_LOGICB + ot); |
| 1510 |
break;
|
| 1511 |
case OP_ORL:
|
| 1512 |
tcg_gen_or_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 1513 |
if (d != OR_TMP0)
|
| 1514 |
gen_op_mov_reg_T0(ot, d); |
| 1515 |
else
|
| 1516 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1517 |
gen_op_update1_cc(); |
| 1518 |
set_cc_op(s1, CC_OP_LOGICB + ot); |
| 1519 |
break;
|
| 1520 |
case OP_XORL:
|
| 1521 |
tcg_gen_xor_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 1522 |
if (d != OR_TMP0)
|
| 1523 |
gen_op_mov_reg_T0(ot, d); |
| 1524 |
else
|
| 1525 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1526 |
gen_op_update1_cc(); |
| 1527 |
set_cc_op(s1, CC_OP_LOGICB + ot); |
| 1528 |
break;
|
| 1529 |
case OP_CMPL:
|
| 1530 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[1]);
|
| 1531 |
tcg_gen_mov_tl(cpu_cc_srcT, cpu_T[0]);
|
| 1532 |
tcg_gen_sub_tl(cpu_cc_dst, cpu_T[0], cpu_T[1]); |
| 1533 |
set_cc_op(s1, CC_OP_SUBB + ot); |
| 1534 |
break;
|
| 1535 |
} |
| 1536 |
} |
| 1537 |
|
| 1538 |
/* if d == OR_TMP0, it means memory operand (address in A0) */
|
| 1539 |
static void gen_inc(DisasContext *s1, int ot, int d, int c) |
| 1540 |
{
|
| 1541 |
if (d != OR_TMP0)
|
| 1542 |
gen_op_mov_TN_reg(ot, 0, d);
|
| 1543 |
else
|
| 1544 |
gen_op_ld_T0_A0(ot + s1->mem_index); |
| 1545 |
gen_compute_eflags_c(s1, cpu_cc_src); |
| 1546 |
if (c > 0) { |
| 1547 |
tcg_gen_addi_tl(cpu_T[0], cpu_T[0], 1); |
| 1548 |
set_cc_op(s1, CC_OP_INCB + ot); |
| 1549 |
} else {
|
| 1550 |
tcg_gen_addi_tl(cpu_T[0], cpu_T[0], -1); |
| 1551 |
set_cc_op(s1, CC_OP_DECB + ot); |
| 1552 |
} |
| 1553 |
if (d != OR_TMP0)
|
| 1554 |
gen_op_mov_reg_T0(ot, d); |
| 1555 |
else
|
| 1556 |
gen_op_st_T0_A0(ot + s1->mem_index); |
| 1557 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 1558 |
} |
| 1559 |
|
| 1560 |
static void gen_shift_rm_T1(DisasContext *s, int ot, int op1, |
| 1561 |
int is_right, int is_arith) |
| 1562 |
{
|
| 1563 |
target_ulong mask; |
| 1564 |
int shift_label;
|
| 1565 |
TCGv t0, t1, t2; |
| 1566 |
|
| 1567 |
if (ot == OT_QUAD) {
|
| 1568 |
mask = 0x3f;
|
| 1569 |
} else {
|
| 1570 |
mask = 0x1f;
|
| 1571 |
} |
| 1572 |
|
| 1573 |
/* load */
|
| 1574 |
if (op1 == OR_TMP0) {
|
| 1575 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 1576 |
} else {
|
| 1577 |
gen_op_mov_TN_reg(ot, 0, op1);
|
| 1578 |
} |
| 1579 |
|
| 1580 |
t0 = tcg_temp_local_new(); |
| 1581 |
t1 = tcg_temp_local_new(); |
| 1582 |
t2 = tcg_temp_local_new(); |
| 1583 |
|
| 1584 |
tcg_gen_andi_tl(t2, cpu_T[1], mask);
|
| 1585 |
|
| 1586 |
if (is_right) {
|
| 1587 |
if (is_arith) {
|
| 1588 |
gen_exts(ot, cpu_T[0]);
|
| 1589 |
tcg_gen_mov_tl(t0, cpu_T[0]);
|
| 1590 |
tcg_gen_sar_tl(cpu_T[0], cpu_T[0], t2); |
| 1591 |
} else {
|
| 1592 |
gen_extu(ot, cpu_T[0]);
|
| 1593 |
tcg_gen_mov_tl(t0, cpu_T[0]);
|
| 1594 |
tcg_gen_shr_tl(cpu_T[0], cpu_T[0], t2); |
| 1595 |
} |
| 1596 |
} else {
|
| 1597 |
tcg_gen_mov_tl(t0, cpu_T[0]);
|
| 1598 |
tcg_gen_shl_tl(cpu_T[0], cpu_T[0], t2); |
| 1599 |
} |
| 1600 |
|
| 1601 |
/* store */
|
| 1602 |
if (op1 == OR_TMP0) {
|
| 1603 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 1604 |
} else {
|
| 1605 |
gen_op_mov_reg_T0(ot, op1); |
| 1606 |
} |
| 1607 |
|
| 1608 |
/* Update eflags data because we cannot predict flags afterward. */
|
| 1609 |
gen_update_cc_op(s); |
| 1610 |
set_cc_op(s, CC_OP_DYNAMIC); |
| 1611 |
|
| 1612 |
tcg_gen_mov_tl(t1, cpu_T[0]);
|
| 1613 |
|
| 1614 |
shift_label = gen_new_label(); |
| 1615 |
tcg_gen_brcondi_tl(TCG_COND_EQ, t2, 0, shift_label);
|
| 1616 |
|
| 1617 |
tcg_gen_addi_tl(t2, t2, -1);
|
| 1618 |
tcg_gen_mov_tl(cpu_cc_dst, t1); |
| 1619 |
|
| 1620 |
if (is_right) {
|
| 1621 |
if (is_arith) {
|
| 1622 |
tcg_gen_sar_tl(cpu_cc_src, t0, t2); |
| 1623 |
} else {
|
| 1624 |
tcg_gen_shr_tl(cpu_cc_src, t0, t2); |
| 1625 |
} |
| 1626 |
} else {
|
| 1627 |
tcg_gen_shl_tl(cpu_cc_src, t0, t2); |
| 1628 |
} |
| 1629 |
|
| 1630 |
if (is_right) {
|
| 1631 |
tcg_gen_movi_i32(cpu_cc_op, CC_OP_SARB + ot); |
| 1632 |
} else {
|
| 1633 |
tcg_gen_movi_i32(cpu_cc_op, CC_OP_SHLB + ot); |
| 1634 |
} |
| 1635 |
|
| 1636 |
gen_set_label(shift_label); |
| 1637 |
|
| 1638 |
tcg_temp_free(t0); |
| 1639 |
tcg_temp_free(t1); |
| 1640 |
tcg_temp_free(t2); |
| 1641 |
} |
| 1642 |
|
| 1643 |
static void gen_shift_rm_im(DisasContext *s, int ot, int op1, int op2, |
| 1644 |
int is_right, int is_arith) |
| 1645 |
{
|
| 1646 |
int mask;
|
| 1647 |
|
| 1648 |
if (ot == OT_QUAD)
|
| 1649 |
mask = 0x3f;
|
| 1650 |
else
|
| 1651 |
mask = 0x1f;
|
| 1652 |
|
| 1653 |
/* load */
|
| 1654 |
if (op1 == OR_TMP0)
|
| 1655 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 1656 |
else
|
| 1657 |
gen_op_mov_TN_reg(ot, 0, op1);
|
| 1658 |
|
| 1659 |
op2 &= mask; |
| 1660 |
if (op2 != 0) { |
| 1661 |
if (is_right) {
|
| 1662 |
if (is_arith) {
|
| 1663 |
gen_exts(ot, cpu_T[0]);
|
| 1664 |
tcg_gen_sari_tl(cpu_tmp4, cpu_T[0], op2 - 1); |
| 1665 |
tcg_gen_sari_tl(cpu_T[0], cpu_T[0], op2); |
| 1666 |
} else {
|
| 1667 |
gen_extu(ot, cpu_T[0]);
|
| 1668 |
tcg_gen_shri_tl(cpu_tmp4, cpu_T[0], op2 - 1); |
| 1669 |
tcg_gen_shri_tl(cpu_T[0], cpu_T[0], op2); |
| 1670 |
} |
| 1671 |
} else {
|
| 1672 |
tcg_gen_shli_tl(cpu_tmp4, cpu_T[0], op2 - 1); |
| 1673 |
tcg_gen_shli_tl(cpu_T[0], cpu_T[0], op2); |
| 1674 |
} |
| 1675 |
} |
| 1676 |
|
| 1677 |
/* store */
|
| 1678 |
if (op1 == OR_TMP0)
|
| 1679 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 1680 |
else
|
| 1681 |
gen_op_mov_reg_T0(ot, op1); |
| 1682 |
|
| 1683 |
/* update eflags if non zero shift */
|
| 1684 |
if (op2 != 0) { |
| 1685 |
tcg_gen_mov_tl(cpu_cc_src, cpu_tmp4); |
| 1686 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 1687 |
set_cc_op(s, (is_right ? CC_OP_SARB : CC_OP_SHLB) + ot); |
| 1688 |
} |
| 1689 |
} |
| 1690 |
|
| 1691 |
static inline void tcg_gen_lshift(TCGv ret, TCGv arg1, target_long arg2) |
| 1692 |
{
|
| 1693 |
if (arg2 >= 0) |
| 1694 |
tcg_gen_shli_tl(ret, arg1, arg2); |
| 1695 |
else
|
| 1696 |
tcg_gen_shri_tl(ret, arg1, -arg2); |
| 1697 |
} |
| 1698 |
|
| 1699 |
static void gen_rot_rm_T1(DisasContext *s, int ot, int op1, |
| 1700 |
int is_right)
|
| 1701 |
{
|
| 1702 |
target_ulong mask; |
| 1703 |
int label1, label2, data_bits;
|
| 1704 |
TCGv t0, t1, t2, a0; |
| 1705 |
|
| 1706 |
/* XXX: inefficient, but we must use local temps */
|
| 1707 |
t0 = tcg_temp_local_new(); |
| 1708 |
t1 = tcg_temp_local_new(); |
| 1709 |
t2 = tcg_temp_local_new(); |
| 1710 |
a0 = tcg_temp_local_new(); |
| 1711 |
|
| 1712 |
if (ot == OT_QUAD)
|
| 1713 |
mask = 0x3f;
|
| 1714 |
else
|
| 1715 |
mask = 0x1f;
|
| 1716 |
|
| 1717 |
/* load */
|
| 1718 |
if (op1 == OR_TMP0) {
|
| 1719 |
tcg_gen_mov_tl(a0, cpu_A0); |
| 1720 |
gen_op_ld_v(ot + s->mem_index, t0, a0); |
| 1721 |
} else {
|
| 1722 |
gen_op_mov_v_reg(ot, t0, op1); |
| 1723 |
} |
| 1724 |
|
| 1725 |
tcg_gen_mov_tl(t1, cpu_T[1]);
|
| 1726 |
|
| 1727 |
tcg_gen_andi_tl(t1, t1, mask); |
| 1728 |
|
| 1729 |
/* Must test zero case to avoid using undefined behaviour in TCG
|
| 1730 |
shifts. */
|
| 1731 |
label1 = gen_new_label(); |
| 1732 |
tcg_gen_brcondi_tl(TCG_COND_EQ, t1, 0, label1);
|
| 1733 |
|
| 1734 |
if (ot <= OT_WORD)
|
| 1735 |
tcg_gen_andi_tl(cpu_tmp0, t1, (1 << (3 + ot)) - 1); |
| 1736 |
else
|
| 1737 |
tcg_gen_mov_tl(cpu_tmp0, t1); |
| 1738 |
|
| 1739 |
gen_extu(ot, t0); |
| 1740 |
tcg_gen_mov_tl(t2, t0); |
| 1741 |
|
| 1742 |
data_bits = 8 << ot;
|
| 1743 |
/* XXX: rely on behaviour of shifts when operand 2 overflows (XXX:
|
| 1744 |
fix TCG definition) */
|
| 1745 |
if (is_right) {
|
| 1746 |
tcg_gen_shr_tl(cpu_tmp4, t0, cpu_tmp0); |
| 1747 |
tcg_gen_subfi_tl(cpu_tmp0, data_bits, cpu_tmp0); |
| 1748 |
tcg_gen_shl_tl(t0, t0, cpu_tmp0); |
| 1749 |
} else {
|
| 1750 |
tcg_gen_shl_tl(cpu_tmp4, t0, cpu_tmp0); |
| 1751 |
tcg_gen_subfi_tl(cpu_tmp0, data_bits, cpu_tmp0); |
| 1752 |
tcg_gen_shr_tl(t0, t0, cpu_tmp0); |
| 1753 |
} |
| 1754 |
tcg_gen_or_tl(t0, t0, cpu_tmp4); |
| 1755 |
|
| 1756 |
gen_set_label(label1); |
| 1757 |
/* store */
|
| 1758 |
if (op1 == OR_TMP0) {
|
| 1759 |
gen_op_st_v(ot + s->mem_index, t0, a0); |
| 1760 |
} else {
|
| 1761 |
gen_op_mov_reg_v(ot, op1, t0); |
| 1762 |
} |
| 1763 |
|
| 1764 |
/* update eflags. It is needed anyway most of the time, do it always. */
|
| 1765 |
gen_compute_eflags(s); |
| 1766 |
assert(s->cc_op == CC_OP_EFLAGS); |
| 1767 |
|
| 1768 |
label2 = gen_new_label(); |
| 1769 |
tcg_gen_brcondi_tl(TCG_COND_EQ, t1, 0, label2);
|
| 1770 |
|
| 1771 |
tcg_gen_andi_tl(cpu_cc_src, cpu_cc_src, ~(CC_O | CC_C)); |
| 1772 |
tcg_gen_xor_tl(cpu_tmp0, t2, t0); |
| 1773 |
tcg_gen_lshift(cpu_tmp0, cpu_tmp0, 11 - (data_bits - 1)); |
| 1774 |
tcg_gen_andi_tl(cpu_tmp0, cpu_tmp0, CC_O); |
| 1775 |
tcg_gen_or_tl(cpu_cc_src, cpu_cc_src, cpu_tmp0); |
| 1776 |
if (is_right) {
|
| 1777 |
tcg_gen_shri_tl(t0, t0, data_bits - 1);
|
| 1778 |
} |
| 1779 |
tcg_gen_andi_tl(t0, t0, CC_C); |
| 1780 |
tcg_gen_or_tl(cpu_cc_src, cpu_cc_src, t0); |
| 1781 |
|
| 1782 |
gen_set_label(label2); |
| 1783 |
|
| 1784 |
tcg_temp_free(t0); |
| 1785 |
tcg_temp_free(t1); |
| 1786 |
tcg_temp_free(t2); |
| 1787 |
tcg_temp_free(a0); |
| 1788 |
} |
| 1789 |
|
| 1790 |
static void gen_rot_rm_im(DisasContext *s, int ot, int op1, int op2, |
| 1791 |
int is_right)
|
| 1792 |
{
|
| 1793 |
int mask;
|
| 1794 |
int data_bits;
|
| 1795 |
TCGv t0, t1, a0; |
| 1796 |
|
| 1797 |
/* XXX: inefficient, but we must use local temps */
|
| 1798 |
t0 = tcg_temp_local_new(); |
| 1799 |
t1 = tcg_temp_local_new(); |
| 1800 |
a0 = tcg_temp_local_new(); |
| 1801 |
|
| 1802 |
if (ot == OT_QUAD)
|
| 1803 |
mask = 0x3f;
|
| 1804 |
else
|
| 1805 |
mask = 0x1f;
|
| 1806 |
|
| 1807 |
/* load */
|
| 1808 |
if (op1 == OR_TMP0) {
|
| 1809 |
tcg_gen_mov_tl(a0, cpu_A0); |
| 1810 |
gen_op_ld_v(ot + s->mem_index, t0, a0); |
| 1811 |
} else {
|
| 1812 |
gen_op_mov_v_reg(ot, t0, op1); |
| 1813 |
} |
| 1814 |
|
| 1815 |
gen_extu(ot, t0); |
| 1816 |
tcg_gen_mov_tl(t1, t0); |
| 1817 |
|
| 1818 |
op2 &= mask; |
| 1819 |
data_bits = 8 << ot;
|
| 1820 |
if (op2 != 0) { |
| 1821 |
int shift = op2 & ((1 << (3 + ot)) - 1); |
| 1822 |
if (is_right) {
|
| 1823 |
tcg_gen_shri_tl(cpu_tmp4, t0, shift); |
| 1824 |
tcg_gen_shli_tl(t0, t0, data_bits - shift); |
| 1825 |
} |
| 1826 |
else {
|
| 1827 |
tcg_gen_shli_tl(cpu_tmp4, t0, shift); |
| 1828 |
tcg_gen_shri_tl(t0, t0, data_bits - shift); |
| 1829 |
} |
| 1830 |
tcg_gen_or_tl(t0, t0, cpu_tmp4); |
| 1831 |
} |
| 1832 |
|
| 1833 |
/* store */
|
| 1834 |
if (op1 == OR_TMP0) {
|
| 1835 |
gen_op_st_v(ot + s->mem_index, t0, a0); |
| 1836 |
} else {
|
| 1837 |
gen_op_mov_reg_v(ot, op1, t0); |
| 1838 |
} |
| 1839 |
|
| 1840 |
if (op2 != 0) { |
| 1841 |
/* update eflags */
|
| 1842 |
gen_compute_eflags(s); |
| 1843 |
assert(s->cc_op == CC_OP_EFLAGS); |
| 1844 |
|
| 1845 |
tcg_gen_andi_tl(cpu_cc_src, cpu_cc_src, ~(CC_O | CC_C)); |
| 1846 |
tcg_gen_xor_tl(cpu_tmp0, t1, t0); |
| 1847 |
tcg_gen_lshift(cpu_tmp0, cpu_tmp0, 11 - (data_bits - 1)); |
| 1848 |
tcg_gen_andi_tl(cpu_tmp0, cpu_tmp0, CC_O); |
| 1849 |
tcg_gen_or_tl(cpu_cc_src, cpu_cc_src, cpu_tmp0); |
| 1850 |
if (is_right) {
|
| 1851 |
tcg_gen_shri_tl(t0, t0, data_bits - 1);
|
| 1852 |
} |
| 1853 |
tcg_gen_andi_tl(t0, t0, CC_C); |
| 1854 |
tcg_gen_or_tl(cpu_cc_src, cpu_cc_src, t0); |
| 1855 |
} |
| 1856 |
|
| 1857 |
tcg_temp_free(t0); |
| 1858 |
tcg_temp_free(t1); |
| 1859 |
tcg_temp_free(a0); |
| 1860 |
} |
| 1861 |
|
| 1862 |
/* XXX: add faster immediate = 1 case */
|
| 1863 |
static void gen_rotc_rm_T1(DisasContext *s, int ot, int op1, |
| 1864 |
int is_right)
|
| 1865 |
{
|
| 1866 |
gen_compute_eflags(s); |
| 1867 |
assert(s->cc_op == CC_OP_EFLAGS); |
| 1868 |
|
| 1869 |
/* load */
|
| 1870 |
if (op1 == OR_TMP0)
|
| 1871 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 1872 |
else
|
| 1873 |
gen_op_mov_TN_reg(ot, 0, op1);
|
| 1874 |
|
| 1875 |
if (is_right) {
|
| 1876 |
switch (ot) {
|
| 1877 |
case OT_BYTE:
|
| 1878 |
gen_helper_rcrb(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1879 |
break;
|
| 1880 |
case OT_WORD:
|
| 1881 |
gen_helper_rcrw(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1882 |
break;
|
| 1883 |
case OT_LONG:
|
| 1884 |
gen_helper_rcrl(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1885 |
break;
|
| 1886 |
#ifdef TARGET_X86_64
|
| 1887 |
case OT_QUAD:
|
| 1888 |
gen_helper_rcrq(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1889 |
break;
|
| 1890 |
#endif
|
| 1891 |
} |
| 1892 |
} else {
|
| 1893 |
switch (ot) {
|
| 1894 |
case OT_BYTE:
|
| 1895 |
gen_helper_rclb(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1896 |
break;
|
| 1897 |
case OT_WORD:
|
| 1898 |
gen_helper_rclw(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1899 |
break;
|
| 1900 |
case OT_LONG:
|
| 1901 |
gen_helper_rcll(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1902 |
break;
|
| 1903 |
#ifdef TARGET_X86_64
|
| 1904 |
case OT_QUAD:
|
| 1905 |
gen_helper_rclq(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 1906 |
break;
|
| 1907 |
#endif
|
| 1908 |
} |
| 1909 |
} |
| 1910 |
/* store */
|
| 1911 |
if (op1 == OR_TMP0)
|
| 1912 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 1913 |
else
|
| 1914 |
gen_op_mov_reg_T0(ot, op1); |
| 1915 |
} |
| 1916 |
|
| 1917 |
/* XXX: add faster immediate case */
|
| 1918 |
static void gen_shiftd_rm_T1(DisasContext *s, int ot, int op1, |
| 1919 |
int is_right, TCGv count)
|
| 1920 |
{
|
| 1921 |
int label1, label2, data_bits;
|
| 1922 |
target_ulong mask; |
| 1923 |
TCGv t0, t1, t2, a0; |
| 1924 |
|
| 1925 |
t0 = tcg_temp_local_new(); |
| 1926 |
t1 = tcg_temp_local_new(); |
| 1927 |
t2 = tcg_temp_local_new(); |
| 1928 |
a0 = tcg_temp_local_new(); |
| 1929 |
|
| 1930 |
if (ot == OT_QUAD)
|
| 1931 |
mask = 0x3f;
|
| 1932 |
else
|
| 1933 |
mask = 0x1f;
|
| 1934 |
|
| 1935 |
/* load */
|
| 1936 |
if (op1 == OR_TMP0) {
|
| 1937 |
tcg_gen_mov_tl(a0, cpu_A0); |
| 1938 |
gen_op_ld_v(ot + s->mem_index, t0, a0); |
| 1939 |
} else {
|
| 1940 |
gen_op_mov_v_reg(ot, t0, op1); |
| 1941 |
} |
| 1942 |
|
| 1943 |
tcg_gen_andi_tl(t2, count, mask); |
| 1944 |
tcg_gen_mov_tl(t1, cpu_T[1]);
|
| 1945 |
|
| 1946 |
/* Must test zero case to avoid using undefined behaviour in TCG
|
| 1947 |
shifts. */
|
| 1948 |
label1 = gen_new_label(); |
| 1949 |
tcg_gen_brcondi_tl(TCG_COND_EQ, t2, 0, label1);
|
| 1950 |
|
| 1951 |
tcg_gen_addi_tl(cpu_tmp5, t2, -1);
|
| 1952 |
if (ot == OT_WORD) {
|
| 1953 |
/* Note: we implement the Intel behaviour for shift count > 16 */
|
| 1954 |
if (is_right) {
|
| 1955 |
tcg_gen_andi_tl(t0, t0, 0xffff);
|
| 1956 |
tcg_gen_shli_tl(cpu_tmp0, t1, 16);
|
| 1957 |
tcg_gen_or_tl(t0, t0, cpu_tmp0); |
| 1958 |
tcg_gen_ext32u_tl(t0, t0); |
| 1959 |
|
| 1960 |
tcg_gen_shr_tl(cpu_tmp4, t0, cpu_tmp5); |
| 1961 |
|
| 1962 |
/* only needed if count > 16, but a test would complicate */
|
| 1963 |
tcg_gen_subfi_tl(cpu_tmp5, 32, t2);
|
| 1964 |
tcg_gen_shl_tl(cpu_tmp0, t0, cpu_tmp5); |
| 1965 |
|
| 1966 |
tcg_gen_shr_tl(t0, t0, t2); |
| 1967 |
|
| 1968 |
tcg_gen_or_tl(t0, t0, cpu_tmp0); |
| 1969 |
} else {
|
| 1970 |
/* XXX: not optimal */
|
| 1971 |
tcg_gen_andi_tl(t0, t0, 0xffff);
|
| 1972 |
tcg_gen_shli_tl(t1, t1, 16);
|
| 1973 |
tcg_gen_or_tl(t1, t1, t0); |
| 1974 |
tcg_gen_ext32u_tl(t1, t1); |
| 1975 |
|
| 1976 |
tcg_gen_shl_tl(cpu_tmp4, t0, cpu_tmp5); |
| 1977 |
tcg_gen_subfi_tl(cpu_tmp0, 32, cpu_tmp5);
|
| 1978 |
tcg_gen_shr_tl(cpu_tmp5, t1, cpu_tmp0); |
| 1979 |
tcg_gen_or_tl(cpu_tmp4, cpu_tmp4, cpu_tmp5); |
| 1980 |
|
| 1981 |
tcg_gen_shl_tl(t0, t0, t2); |
| 1982 |
tcg_gen_subfi_tl(cpu_tmp5, 32, t2);
|
| 1983 |
tcg_gen_shr_tl(t1, t1, cpu_tmp5); |
| 1984 |
tcg_gen_or_tl(t0, t0, t1); |
| 1985 |
} |
| 1986 |
} else {
|
| 1987 |
data_bits = 8 << ot;
|
| 1988 |
if (is_right) {
|
| 1989 |
if (ot == OT_LONG)
|
| 1990 |
tcg_gen_ext32u_tl(t0, t0); |
| 1991 |
|
| 1992 |
tcg_gen_shr_tl(cpu_tmp4, t0, cpu_tmp5); |
| 1993 |
|
| 1994 |
tcg_gen_shr_tl(t0, t0, t2); |
| 1995 |
tcg_gen_subfi_tl(cpu_tmp5, data_bits, t2); |
| 1996 |
tcg_gen_shl_tl(t1, t1, cpu_tmp5); |
| 1997 |
tcg_gen_or_tl(t0, t0, t1); |
| 1998 |
|
| 1999 |
} else {
|
| 2000 |
if (ot == OT_LONG)
|
| 2001 |
tcg_gen_ext32u_tl(t1, t1); |
| 2002 |
|
| 2003 |
tcg_gen_shl_tl(cpu_tmp4, t0, cpu_tmp5); |
| 2004 |
|
| 2005 |
tcg_gen_shl_tl(t0, t0, t2); |
| 2006 |
tcg_gen_subfi_tl(cpu_tmp5, data_bits, t2); |
| 2007 |
tcg_gen_shr_tl(t1, t1, cpu_tmp5); |
| 2008 |
tcg_gen_or_tl(t0, t0, t1); |
| 2009 |
} |
| 2010 |
} |
| 2011 |
tcg_gen_mov_tl(t1, cpu_tmp4); |
| 2012 |
|
| 2013 |
gen_set_label(label1); |
| 2014 |
/* store */
|
| 2015 |
if (op1 == OR_TMP0) {
|
| 2016 |
gen_op_st_v(ot + s->mem_index, t0, a0); |
| 2017 |
} else {
|
| 2018 |
gen_op_mov_reg_v(ot, op1, t0); |
| 2019 |
} |
| 2020 |
|
| 2021 |
/* Update eflags data because we cannot predict flags afterward. */
|
| 2022 |
gen_update_cc_op(s); |
| 2023 |
set_cc_op(s, CC_OP_DYNAMIC); |
| 2024 |
|
| 2025 |
label2 = gen_new_label(); |
| 2026 |
tcg_gen_brcondi_tl(TCG_COND_EQ, t2, 0, label2);
|
| 2027 |
|
| 2028 |
tcg_gen_mov_tl(cpu_cc_src, t1); |
| 2029 |
tcg_gen_mov_tl(cpu_cc_dst, t0); |
| 2030 |
if (is_right) {
|
| 2031 |
tcg_gen_movi_i32(cpu_cc_op, CC_OP_SARB + ot); |
| 2032 |
} else {
|
| 2033 |
tcg_gen_movi_i32(cpu_cc_op, CC_OP_SHLB + ot); |
| 2034 |
} |
| 2035 |
gen_set_label(label2); |
| 2036 |
|
| 2037 |
tcg_temp_free(t0); |
| 2038 |
tcg_temp_free(t1); |
| 2039 |
tcg_temp_free(t2); |
| 2040 |
tcg_temp_free(a0); |
| 2041 |
} |
| 2042 |
|
| 2043 |
static void gen_shift(DisasContext *s1, int op, int ot, int d, int s) |
| 2044 |
{
|
| 2045 |
if (s != OR_TMP1)
|
| 2046 |
gen_op_mov_TN_reg(ot, 1, s);
|
| 2047 |
switch(op) {
|
| 2048 |
case OP_ROL:
|
| 2049 |
gen_rot_rm_T1(s1, ot, d, 0);
|
| 2050 |
break;
|
| 2051 |
case OP_ROR:
|
| 2052 |
gen_rot_rm_T1(s1, ot, d, 1);
|
| 2053 |
break;
|
| 2054 |
case OP_SHL:
|
| 2055 |
case OP_SHL1:
|
| 2056 |
gen_shift_rm_T1(s1, ot, d, 0, 0); |
| 2057 |
break;
|
| 2058 |
case OP_SHR:
|
| 2059 |
gen_shift_rm_T1(s1, ot, d, 1, 0); |
| 2060 |
break;
|
| 2061 |
case OP_SAR:
|
| 2062 |
gen_shift_rm_T1(s1, ot, d, 1, 1); |
| 2063 |
break;
|
| 2064 |
case OP_RCL:
|
| 2065 |
gen_rotc_rm_T1(s1, ot, d, 0);
|
| 2066 |
break;
|
| 2067 |
case OP_RCR:
|
| 2068 |
gen_rotc_rm_T1(s1, ot, d, 1);
|
| 2069 |
break;
|
| 2070 |
} |
| 2071 |
} |
| 2072 |
|
| 2073 |
static void gen_shifti(DisasContext *s1, int op, int ot, int d, int c) |
| 2074 |
{
|
| 2075 |
switch(op) {
|
| 2076 |
case OP_ROL:
|
| 2077 |
gen_rot_rm_im(s1, ot, d, c, 0);
|
| 2078 |
break;
|
| 2079 |
case OP_ROR:
|
| 2080 |
gen_rot_rm_im(s1, ot, d, c, 1);
|
| 2081 |
break;
|
| 2082 |
case OP_SHL:
|
| 2083 |
case OP_SHL1:
|
| 2084 |
gen_shift_rm_im(s1, ot, d, c, 0, 0); |
| 2085 |
break;
|
| 2086 |
case OP_SHR:
|
| 2087 |
gen_shift_rm_im(s1, ot, d, c, 1, 0); |
| 2088 |
break;
|
| 2089 |
case OP_SAR:
|
| 2090 |
gen_shift_rm_im(s1, ot, d, c, 1, 1); |
| 2091 |
break;
|
| 2092 |
default:
|
| 2093 |
/* currently not optimized */
|
| 2094 |
gen_op_movl_T1_im(c); |
| 2095 |
gen_shift(s1, op, ot, d, OR_TMP1); |
| 2096 |
break;
|
| 2097 |
} |
| 2098 |
} |
| 2099 |
|
| 2100 |
static void gen_lea_modrm(CPUX86State *env, DisasContext *s, int modrm, |
| 2101 |
int *reg_ptr, int *offset_ptr) |
| 2102 |
{
|
| 2103 |
target_long disp; |
| 2104 |
int havesib;
|
| 2105 |
int base;
|
| 2106 |
int index;
|
| 2107 |
int scale;
|
| 2108 |
int opreg;
|
| 2109 |
int mod, rm, code, override, must_add_seg;
|
| 2110 |
|
| 2111 |
override = s->override; |
| 2112 |
must_add_seg = s->addseg; |
| 2113 |
if (override >= 0) |
| 2114 |
must_add_seg = 1;
|
| 2115 |
mod = (modrm >> 6) & 3; |
| 2116 |
rm = modrm & 7;
|
| 2117 |
|
| 2118 |
if (s->aflag) {
|
| 2119 |
|
| 2120 |
havesib = 0;
|
| 2121 |
base = rm; |
| 2122 |
index = 0;
|
| 2123 |
scale = 0;
|
| 2124 |
|
| 2125 |
if (base == 4) { |
| 2126 |
havesib = 1;
|
| 2127 |
code = cpu_ldub_code(env, s->pc++); |
| 2128 |
scale = (code >> 6) & 3; |
| 2129 |
index = ((code >> 3) & 7) | REX_X(s); |
| 2130 |
base = (code & 7);
|
| 2131 |
} |
| 2132 |
base |= REX_B(s); |
| 2133 |
|
| 2134 |
switch (mod) {
|
| 2135 |
case 0: |
| 2136 |
if ((base & 7) == 5) { |
| 2137 |
base = -1;
|
| 2138 |
disp = (int32_t)cpu_ldl_code(env, s->pc); |
| 2139 |
s->pc += 4;
|
| 2140 |
if (CODE64(s) && !havesib) {
|
| 2141 |
disp += s->pc + s->rip_offset; |
| 2142 |
} |
| 2143 |
} else {
|
| 2144 |
disp = 0;
|
| 2145 |
} |
| 2146 |
break;
|
| 2147 |
case 1: |
| 2148 |
disp = (int8_t)cpu_ldub_code(env, s->pc++); |
| 2149 |
break;
|
| 2150 |
default:
|
| 2151 |
case 2: |
| 2152 |
disp = (int32_t)cpu_ldl_code(env, s->pc); |
| 2153 |
s->pc += 4;
|
| 2154 |
break;
|
| 2155 |
} |
| 2156 |
|
| 2157 |
if (base >= 0) { |
| 2158 |
/* for correct popl handling with esp */
|
| 2159 |
if (base == 4 && s->popl_esp_hack) |
| 2160 |
disp += s->popl_esp_hack; |
| 2161 |
#ifdef TARGET_X86_64
|
| 2162 |
if (s->aflag == 2) { |
| 2163 |
gen_op_movq_A0_reg(base); |
| 2164 |
if (disp != 0) { |
| 2165 |
gen_op_addq_A0_im(disp); |
| 2166 |
} |
| 2167 |
} else
|
| 2168 |
#endif
|
| 2169 |
{
|
| 2170 |
gen_op_movl_A0_reg(base); |
| 2171 |
if (disp != 0) |
| 2172 |
gen_op_addl_A0_im(disp); |
| 2173 |
} |
| 2174 |
} else {
|
| 2175 |
#ifdef TARGET_X86_64
|
| 2176 |
if (s->aflag == 2) { |
| 2177 |
gen_op_movq_A0_im(disp); |
| 2178 |
} else
|
| 2179 |
#endif
|
| 2180 |
{
|
| 2181 |
gen_op_movl_A0_im(disp); |
| 2182 |
} |
| 2183 |
} |
| 2184 |
/* index == 4 means no index */
|
| 2185 |
if (havesib && (index != 4)) { |
| 2186 |
#ifdef TARGET_X86_64
|
| 2187 |
if (s->aflag == 2) { |
| 2188 |
gen_op_addq_A0_reg_sN(scale, index); |
| 2189 |
} else
|
| 2190 |
#endif
|
| 2191 |
{
|
| 2192 |
gen_op_addl_A0_reg_sN(scale, index); |
| 2193 |
} |
| 2194 |
} |
| 2195 |
if (must_add_seg) {
|
| 2196 |
if (override < 0) { |
| 2197 |
if (base == R_EBP || base == R_ESP)
|
| 2198 |
override = R_SS; |
| 2199 |
else
|
| 2200 |
override = R_DS; |
| 2201 |
} |
| 2202 |
#ifdef TARGET_X86_64
|
| 2203 |
if (s->aflag == 2) { |
| 2204 |
gen_op_addq_A0_seg(override); |
| 2205 |
} else
|
| 2206 |
#endif
|
| 2207 |
{
|
| 2208 |
gen_op_addl_A0_seg(s, override); |
| 2209 |
} |
| 2210 |
} |
| 2211 |
} else {
|
| 2212 |
switch (mod) {
|
| 2213 |
case 0: |
| 2214 |
if (rm == 6) { |
| 2215 |
disp = cpu_lduw_code(env, s->pc); |
| 2216 |
s->pc += 2;
|
| 2217 |
gen_op_movl_A0_im(disp); |
| 2218 |
rm = 0; /* avoid SS override */ |
| 2219 |
goto no_rm;
|
| 2220 |
} else {
|
| 2221 |
disp = 0;
|
| 2222 |
} |
| 2223 |
break;
|
| 2224 |
case 1: |
| 2225 |
disp = (int8_t)cpu_ldub_code(env, s->pc++); |
| 2226 |
break;
|
| 2227 |
default:
|
| 2228 |
case 2: |
| 2229 |
disp = cpu_lduw_code(env, s->pc); |
| 2230 |
s->pc += 2;
|
| 2231 |
break;
|
| 2232 |
} |
| 2233 |
switch(rm) {
|
| 2234 |
case 0: |
| 2235 |
gen_op_movl_A0_reg(R_EBX); |
| 2236 |
gen_op_addl_A0_reg_sN(0, R_ESI);
|
| 2237 |
break;
|
| 2238 |
case 1: |
| 2239 |
gen_op_movl_A0_reg(R_EBX); |
| 2240 |
gen_op_addl_A0_reg_sN(0, R_EDI);
|
| 2241 |
break;
|
| 2242 |
case 2: |
| 2243 |
gen_op_movl_A0_reg(R_EBP); |
| 2244 |
gen_op_addl_A0_reg_sN(0, R_ESI);
|
| 2245 |
break;
|
| 2246 |
case 3: |
| 2247 |
gen_op_movl_A0_reg(R_EBP); |
| 2248 |
gen_op_addl_A0_reg_sN(0, R_EDI);
|
| 2249 |
break;
|
| 2250 |
case 4: |
| 2251 |
gen_op_movl_A0_reg(R_ESI); |
| 2252 |
break;
|
| 2253 |
case 5: |
| 2254 |
gen_op_movl_A0_reg(R_EDI); |
| 2255 |
break;
|
| 2256 |
case 6: |
| 2257 |
gen_op_movl_A0_reg(R_EBP); |
| 2258 |
break;
|
| 2259 |
default:
|
| 2260 |
case 7: |
| 2261 |
gen_op_movl_A0_reg(R_EBX); |
| 2262 |
break;
|
| 2263 |
} |
| 2264 |
if (disp != 0) |
| 2265 |
gen_op_addl_A0_im(disp); |
| 2266 |
gen_op_andl_A0_ffff(); |
| 2267 |
no_rm:
|
| 2268 |
if (must_add_seg) {
|
| 2269 |
if (override < 0) { |
| 2270 |
if (rm == 2 || rm == 3 || rm == 6) |
| 2271 |
override = R_SS; |
| 2272 |
else
|
| 2273 |
override = R_DS; |
| 2274 |
} |
| 2275 |
gen_op_addl_A0_seg(s, override); |
| 2276 |
} |
| 2277 |
} |
| 2278 |
|
| 2279 |
opreg = OR_A0; |
| 2280 |
disp = 0;
|
| 2281 |
*reg_ptr = opreg; |
| 2282 |
*offset_ptr = disp; |
| 2283 |
} |
| 2284 |
|
| 2285 |
static void gen_nop_modrm(CPUX86State *env, DisasContext *s, int modrm) |
| 2286 |
{
|
| 2287 |
int mod, rm, base, code;
|
| 2288 |
|
| 2289 |
mod = (modrm >> 6) & 3; |
| 2290 |
if (mod == 3) |
| 2291 |
return;
|
| 2292 |
rm = modrm & 7;
|
| 2293 |
|
| 2294 |
if (s->aflag) {
|
| 2295 |
|
| 2296 |
base = rm; |
| 2297 |
|
| 2298 |
if (base == 4) { |
| 2299 |
code = cpu_ldub_code(env, s->pc++); |
| 2300 |
base = (code & 7);
|
| 2301 |
} |
| 2302 |
|
| 2303 |
switch (mod) {
|
| 2304 |
case 0: |
| 2305 |
if (base == 5) { |
| 2306 |
s->pc += 4;
|
| 2307 |
} |
| 2308 |
break;
|
| 2309 |
case 1: |
| 2310 |
s->pc++; |
| 2311 |
break;
|
| 2312 |
default:
|
| 2313 |
case 2: |
| 2314 |
s->pc += 4;
|
| 2315 |
break;
|
| 2316 |
} |
| 2317 |
} else {
|
| 2318 |
switch (mod) {
|
| 2319 |
case 0: |
| 2320 |
if (rm == 6) { |
| 2321 |
s->pc += 2;
|
| 2322 |
} |
| 2323 |
break;
|
| 2324 |
case 1: |
| 2325 |
s->pc++; |
| 2326 |
break;
|
| 2327 |
default:
|
| 2328 |
case 2: |
| 2329 |
s->pc += 2;
|
| 2330 |
break;
|
| 2331 |
} |
| 2332 |
} |
| 2333 |
} |
| 2334 |
|
| 2335 |
/* used for LEA and MOV AX, mem */
|
| 2336 |
static void gen_add_A0_ds_seg(DisasContext *s) |
| 2337 |
{
|
| 2338 |
int override, must_add_seg;
|
| 2339 |
must_add_seg = s->addseg; |
| 2340 |
override = R_DS; |
| 2341 |
if (s->override >= 0) { |
| 2342 |
override = s->override; |
| 2343 |
must_add_seg = 1;
|
| 2344 |
} |
| 2345 |
if (must_add_seg) {
|
| 2346 |
#ifdef TARGET_X86_64
|
| 2347 |
if (CODE64(s)) {
|
| 2348 |
gen_op_addq_A0_seg(override); |
| 2349 |
} else
|
| 2350 |
#endif
|
| 2351 |
{
|
| 2352 |
gen_op_addl_A0_seg(s, override); |
| 2353 |
} |
| 2354 |
} |
| 2355 |
} |
| 2356 |
|
| 2357 |
/* generate modrm memory load or store of 'reg'. TMP0 is used if reg ==
|
| 2358 |
OR_TMP0 */
|
| 2359 |
static void gen_ldst_modrm(CPUX86State *env, DisasContext *s, int modrm, |
| 2360 |
int ot, int reg, int is_store) |
| 2361 |
{
|
| 2362 |
int mod, rm, opreg, disp;
|
| 2363 |
|
| 2364 |
mod = (modrm >> 6) & 3; |
| 2365 |
rm = (modrm & 7) | REX_B(s);
|
| 2366 |
if (mod == 3) { |
| 2367 |
if (is_store) {
|
| 2368 |
if (reg != OR_TMP0)
|
| 2369 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 2370 |
gen_op_mov_reg_T0(ot, rm); |
| 2371 |
} else {
|
| 2372 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 2373 |
if (reg != OR_TMP0)
|
| 2374 |
gen_op_mov_reg_T0(ot, reg); |
| 2375 |
} |
| 2376 |
} else {
|
| 2377 |
gen_lea_modrm(env, s, modrm, &opreg, &disp); |
| 2378 |
if (is_store) {
|
| 2379 |
if (reg != OR_TMP0)
|
| 2380 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 2381 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 2382 |
} else {
|
| 2383 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 2384 |
if (reg != OR_TMP0)
|
| 2385 |
gen_op_mov_reg_T0(ot, reg); |
| 2386 |
} |
| 2387 |
} |
| 2388 |
} |
| 2389 |
|
| 2390 |
static inline uint32_t insn_get(CPUX86State *env, DisasContext *s, int ot) |
| 2391 |
{
|
| 2392 |
uint32_t ret; |
| 2393 |
|
| 2394 |
switch(ot) {
|
| 2395 |
case OT_BYTE:
|
| 2396 |
ret = cpu_ldub_code(env, s->pc); |
| 2397 |
s->pc++; |
| 2398 |
break;
|
| 2399 |
case OT_WORD:
|
| 2400 |
ret = cpu_lduw_code(env, s->pc); |
| 2401 |
s->pc += 2;
|
| 2402 |
break;
|
| 2403 |
default:
|
| 2404 |
case OT_LONG:
|
| 2405 |
ret = cpu_ldl_code(env, s->pc); |
| 2406 |
s->pc += 4;
|
| 2407 |
break;
|
| 2408 |
} |
| 2409 |
return ret;
|
| 2410 |
} |
| 2411 |
|
| 2412 |
static inline int insn_const_size(unsigned int ot) |
| 2413 |
{
|
| 2414 |
if (ot <= OT_LONG)
|
| 2415 |
return 1 << ot; |
| 2416 |
else
|
| 2417 |
return 4; |
| 2418 |
} |
| 2419 |
|
| 2420 |
static inline void gen_goto_tb(DisasContext *s, int tb_num, target_ulong eip) |
| 2421 |
{
|
| 2422 |
TranslationBlock *tb; |
| 2423 |
target_ulong pc; |
| 2424 |
|
| 2425 |
pc = s->cs_base + eip; |
| 2426 |
tb = s->tb; |
| 2427 |
/* NOTE: we handle the case where the TB spans two pages here */
|
| 2428 |
if ((pc & TARGET_PAGE_MASK) == (tb->pc & TARGET_PAGE_MASK) ||
|
| 2429 |
(pc & TARGET_PAGE_MASK) == ((s->pc - 1) & TARGET_PAGE_MASK)) {
|
| 2430 |
/* jump to same page: we can use a direct jump */
|
| 2431 |
tcg_gen_goto_tb(tb_num); |
| 2432 |
gen_jmp_im(eip); |
| 2433 |
tcg_gen_exit_tb((tcg_target_long)tb + tb_num); |
| 2434 |
} else {
|
| 2435 |
/* jump to another page: currently not optimized */
|
| 2436 |
gen_jmp_im(eip); |
| 2437 |
gen_eob(s); |
| 2438 |
} |
| 2439 |
} |
| 2440 |
|
| 2441 |
static inline void gen_jcc(DisasContext *s, int b, |
| 2442 |
target_ulong val, target_ulong next_eip) |
| 2443 |
{
|
| 2444 |
int l1, l2;
|
| 2445 |
|
| 2446 |
if (s->jmp_opt) {
|
| 2447 |
l1 = gen_new_label(); |
| 2448 |
gen_jcc1(s, b, l1); |
| 2449 |
|
| 2450 |
gen_goto_tb(s, 0, next_eip);
|
| 2451 |
|
| 2452 |
gen_set_label(l1); |
| 2453 |
gen_goto_tb(s, 1, val);
|
| 2454 |
s->is_jmp = DISAS_TB_JUMP; |
| 2455 |
} else {
|
| 2456 |
l1 = gen_new_label(); |
| 2457 |
l2 = gen_new_label(); |
| 2458 |
gen_jcc1(s, b, l1); |
| 2459 |
|
| 2460 |
gen_jmp_im(next_eip); |
| 2461 |
tcg_gen_br(l2); |
| 2462 |
|
| 2463 |
gen_set_label(l1); |
| 2464 |
gen_jmp_im(val); |
| 2465 |
gen_set_label(l2); |
| 2466 |
gen_eob(s); |
| 2467 |
} |
| 2468 |
} |
| 2469 |
|
| 2470 |
static void gen_cmovcc1(CPUX86State *env, DisasContext *s, int ot, int b, |
| 2471 |
int modrm, int reg) |
| 2472 |
{
|
| 2473 |
CCPrepare cc; |
| 2474 |
|
| 2475 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 2476 |
|
| 2477 |
cc = gen_prepare_cc(s, b, cpu_T[1]);
|
| 2478 |
if (cc.mask != -1) { |
| 2479 |
TCGv t0 = tcg_temp_new(); |
| 2480 |
tcg_gen_andi_tl(t0, cc.reg, cc.mask); |
| 2481 |
cc.reg = t0; |
| 2482 |
} |
| 2483 |
if (!cc.use_reg2) {
|
| 2484 |
cc.reg2 = tcg_const_tl(cc.imm); |
| 2485 |
} |
| 2486 |
|
| 2487 |
tcg_gen_movcond_tl(cc.cond, cpu_T[0], cc.reg, cc.reg2,
|
| 2488 |
cpu_T[0], cpu_regs[reg]);
|
| 2489 |
gen_op_mov_reg_T0(ot, reg); |
| 2490 |
|
| 2491 |
if (cc.mask != -1) { |
| 2492 |
tcg_temp_free(cc.reg); |
| 2493 |
} |
| 2494 |
if (!cc.use_reg2) {
|
| 2495 |
tcg_temp_free(cc.reg2); |
| 2496 |
} |
| 2497 |
} |
| 2498 |
|
| 2499 |
static inline void gen_op_movl_T0_seg(int seg_reg) |
| 2500 |
{
|
| 2501 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env,
|
| 2502 |
offsetof(CPUX86State,segs[seg_reg].selector)); |
| 2503 |
} |
| 2504 |
|
| 2505 |
static inline void gen_op_movl_seg_T0_vm(int seg_reg) |
| 2506 |
{
|
| 2507 |
tcg_gen_andi_tl(cpu_T[0], cpu_T[0], 0xffff); |
| 2508 |
tcg_gen_st32_tl(cpu_T[0], cpu_env,
|
| 2509 |
offsetof(CPUX86State,segs[seg_reg].selector)); |
| 2510 |
tcg_gen_shli_tl(cpu_T[0], cpu_T[0], 4); |
| 2511 |
tcg_gen_st_tl(cpu_T[0], cpu_env,
|
| 2512 |
offsetof(CPUX86State,segs[seg_reg].base)); |
| 2513 |
} |
| 2514 |
|
| 2515 |
/* move T0 to seg_reg and compute if the CPU state may change. Never
|
| 2516 |
call this function with seg_reg == R_CS */
|
| 2517 |
static void gen_movl_seg_T0(DisasContext *s, int seg_reg, target_ulong cur_eip) |
| 2518 |
{
|
| 2519 |
if (s->pe && !s->vm86) {
|
| 2520 |
/* XXX: optimize by finding processor state dynamically */
|
| 2521 |
gen_update_cc_op(s); |
| 2522 |
gen_jmp_im(cur_eip); |
| 2523 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 2524 |
gen_helper_load_seg(cpu_env, tcg_const_i32(seg_reg), cpu_tmp2_i32); |
| 2525 |
/* abort translation because the addseg value may change or
|
| 2526 |
because ss32 may change. For R_SS, translation must always
|
| 2527 |
stop as a special handling must be done to disable hardware
|
| 2528 |
interrupts for the next instruction */
|
| 2529 |
if (seg_reg == R_SS || (s->code32 && seg_reg < R_FS))
|
| 2530 |
s->is_jmp = DISAS_TB_JUMP; |
| 2531 |
} else {
|
| 2532 |
gen_op_movl_seg_T0_vm(seg_reg); |
| 2533 |
if (seg_reg == R_SS)
|
| 2534 |
s->is_jmp = DISAS_TB_JUMP; |
| 2535 |
} |
| 2536 |
} |
| 2537 |
|
| 2538 |
static inline int svm_is_rep(int prefixes) |
| 2539 |
{
|
| 2540 |
return ((prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) ? 8 : 0); |
| 2541 |
} |
| 2542 |
|
| 2543 |
static inline void |
| 2544 |
gen_svm_check_intercept_param(DisasContext *s, target_ulong pc_start, |
| 2545 |
uint32_t type, uint64_t param) |
| 2546 |
{
|
| 2547 |
/* no SVM activated; fast case */
|
| 2548 |
if (likely(!(s->flags & HF_SVMI_MASK)))
|
| 2549 |
return;
|
| 2550 |
gen_update_cc_op(s); |
| 2551 |
gen_jmp_im(pc_start - s->cs_base); |
| 2552 |
gen_helper_svm_check_intercept_param(cpu_env, tcg_const_i32(type), |
| 2553 |
tcg_const_i64(param)); |
| 2554 |
} |
| 2555 |
|
| 2556 |
static inline void |
| 2557 |
gen_svm_check_intercept(DisasContext *s, target_ulong pc_start, uint64_t type) |
| 2558 |
{
|
| 2559 |
gen_svm_check_intercept_param(s, pc_start, type, 0);
|
| 2560 |
} |
| 2561 |
|
| 2562 |
static inline void gen_stack_update(DisasContext *s, int addend) |
| 2563 |
{
|
| 2564 |
#ifdef TARGET_X86_64
|
| 2565 |
if (CODE64(s)) {
|
| 2566 |
gen_op_add_reg_im(2, R_ESP, addend);
|
| 2567 |
} else
|
| 2568 |
#endif
|
| 2569 |
if (s->ss32) {
|
| 2570 |
gen_op_add_reg_im(1, R_ESP, addend);
|
| 2571 |
} else {
|
| 2572 |
gen_op_add_reg_im(0, R_ESP, addend);
|
| 2573 |
} |
| 2574 |
} |
| 2575 |
|
| 2576 |
/* generate a push. It depends on ss32, addseg and dflag */
|
| 2577 |
static void gen_push_T0(DisasContext *s) |
| 2578 |
{
|
| 2579 |
#ifdef TARGET_X86_64
|
| 2580 |
if (CODE64(s)) {
|
| 2581 |
gen_op_movq_A0_reg(R_ESP); |
| 2582 |
if (s->dflag) {
|
| 2583 |
gen_op_addq_A0_im(-8);
|
| 2584 |
gen_op_st_T0_A0(OT_QUAD + s->mem_index); |
| 2585 |
} else {
|
| 2586 |
gen_op_addq_A0_im(-2);
|
| 2587 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 2588 |
} |
| 2589 |
gen_op_mov_reg_A0(2, R_ESP);
|
| 2590 |
} else
|
| 2591 |
#endif
|
| 2592 |
{
|
| 2593 |
gen_op_movl_A0_reg(R_ESP); |
| 2594 |
if (!s->dflag)
|
| 2595 |
gen_op_addl_A0_im(-2);
|
| 2596 |
else
|
| 2597 |
gen_op_addl_A0_im(-4);
|
| 2598 |
if (s->ss32) {
|
| 2599 |
if (s->addseg) {
|
| 2600 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2601 |
gen_op_addl_A0_seg(s, R_SS); |
| 2602 |
} |
| 2603 |
} else {
|
| 2604 |
gen_op_andl_A0_ffff(); |
| 2605 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2606 |
gen_op_addl_A0_seg(s, R_SS); |
| 2607 |
} |
| 2608 |
gen_op_st_T0_A0(s->dflag + 1 + s->mem_index);
|
| 2609 |
if (s->ss32 && !s->addseg)
|
| 2610 |
gen_op_mov_reg_A0(1, R_ESP);
|
| 2611 |
else
|
| 2612 |
gen_op_mov_reg_T1(s->ss32 + 1, R_ESP);
|
| 2613 |
} |
| 2614 |
} |
| 2615 |
|
| 2616 |
/* generate a push. It depends on ss32, addseg and dflag */
|
| 2617 |
/* slower version for T1, only used for call Ev */
|
| 2618 |
static void gen_push_T1(DisasContext *s) |
| 2619 |
{
|
| 2620 |
#ifdef TARGET_X86_64
|
| 2621 |
if (CODE64(s)) {
|
| 2622 |
gen_op_movq_A0_reg(R_ESP); |
| 2623 |
if (s->dflag) {
|
| 2624 |
gen_op_addq_A0_im(-8);
|
| 2625 |
gen_op_st_T1_A0(OT_QUAD + s->mem_index); |
| 2626 |
} else {
|
| 2627 |
gen_op_addq_A0_im(-2);
|
| 2628 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 2629 |
} |
| 2630 |
gen_op_mov_reg_A0(2, R_ESP);
|
| 2631 |
} else
|
| 2632 |
#endif
|
| 2633 |
{
|
| 2634 |
gen_op_movl_A0_reg(R_ESP); |
| 2635 |
if (!s->dflag)
|
| 2636 |
gen_op_addl_A0_im(-2);
|
| 2637 |
else
|
| 2638 |
gen_op_addl_A0_im(-4);
|
| 2639 |
if (s->ss32) {
|
| 2640 |
if (s->addseg) {
|
| 2641 |
gen_op_addl_A0_seg(s, R_SS); |
| 2642 |
} |
| 2643 |
} else {
|
| 2644 |
gen_op_andl_A0_ffff(); |
| 2645 |
gen_op_addl_A0_seg(s, R_SS); |
| 2646 |
} |
| 2647 |
gen_op_st_T1_A0(s->dflag + 1 + s->mem_index);
|
| 2648 |
|
| 2649 |
if (s->ss32 && !s->addseg)
|
| 2650 |
gen_op_mov_reg_A0(1, R_ESP);
|
| 2651 |
else
|
| 2652 |
gen_stack_update(s, (-2) << s->dflag);
|
| 2653 |
} |
| 2654 |
} |
| 2655 |
|
| 2656 |
/* two step pop is necessary for precise exceptions */
|
| 2657 |
static void gen_pop_T0(DisasContext *s) |
| 2658 |
{
|
| 2659 |
#ifdef TARGET_X86_64
|
| 2660 |
if (CODE64(s)) {
|
| 2661 |
gen_op_movq_A0_reg(R_ESP); |
| 2662 |
gen_op_ld_T0_A0((s->dflag ? OT_QUAD : OT_WORD) + s->mem_index); |
| 2663 |
} else
|
| 2664 |
#endif
|
| 2665 |
{
|
| 2666 |
gen_op_movl_A0_reg(R_ESP); |
| 2667 |
if (s->ss32) {
|
| 2668 |
if (s->addseg)
|
| 2669 |
gen_op_addl_A0_seg(s, R_SS); |
| 2670 |
} else {
|
| 2671 |
gen_op_andl_A0_ffff(); |
| 2672 |
gen_op_addl_A0_seg(s, R_SS); |
| 2673 |
} |
| 2674 |
gen_op_ld_T0_A0(s->dflag + 1 + s->mem_index);
|
| 2675 |
} |
| 2676 |
} |
| 2677 |
|
| 2678 |
static void gen_pop_update(DisasContext *s) |
| 2679 |
{
|
| 2680 |
#ifdef TARGET_X86_64
|
| 2681 |
if (CODE64(s) && s->dflag) {
|
| 2682 |
gen_stack_update(s, 8);
|
| 2683 |
} else
|
| 2684 |
#endif
|
| 2685 |
{
|
| 2686 |
gen_stack_update(s, 2 << s->dflag);
|
| 2687 |
} |
| 2688 |
} |
| 2689 |
|
| 2690 |
static void gen_stack_A0(DisasContext *s) |
| 2691 |
{
|
| 2692 |
gen_op_movl_A0_reg(R_ESP); |
| 2693 |
if (!s->ss32)
|
| 2694 |
gen_op_andl_A0_ffff(); |
| 2695 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2696 |
if (s->addseg)
|
| 2697 |
gen_op_addl_A0_seg(s, R_SS); |
| 2698 |
} |
| 2699 |
|
| 2700 |
/* NOTE: wrap around in 16 bit not fully handled */
|
| 2701 |
static void gen_pusha(DisasContext *s) |
| 2702 |
{
|
| 2703 |
int i;
|
| 2704 |
gen_op_movl_A0_reg(R_ESP); |
| 2705 |
gen_op_addl_A0_im(-16 << s->dflag);
|
| 2706 |
if (!s->ss32)
|
| 2707 |
gen_op_andl_A0_ffff(); |
| 2708 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2709 |
if (s->addseg)
|
| 2710 |
gen_op_addl_A0_seg(s, R_SS); |
| 2711 |
for(i = 0;i < 8; i++) { |
| 2712 |
gen_op_mov_TN_reg(OT_LONG, 0, 7 - i); |
| 2713 |
gen_op_st_T0_A0(OT_WORD + s->dflag + s->mem_index); |
| 2714 |
gen_op_addl_A0_im(2 << s->dflag);
|
| 2715 |
} |
| 2716 |
gen_op_mov_reg_T1(OT_WORD + s->ss32, R_ESP); |
| 2717 |
} |
| 2718 |
|
| 2719 |
/* NOTE: wrap around in 16 bit not fully handled */
|
| 2720 |
static void gen_popa(DisasContext *s) |
| 2721 |
{
|
| 2722 |
int i;
|
| 2723 |
gen_op_movl_A0_reg(R_ESP); |
| 2724 |
if (!s->ss32)
|
| 2725 |
gen_op_andl_A0_ffff(); |
| 2726 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2727 |
tcg_gen_addi_tl(cpu_T[1], cpu_T[1], 16 << s->dflag); |
| 2728 |
if (s->addseg)
|
| 2729 |
gen_op_addl_A0_seg(s, R_SS); |
| 2730 |
for(i = 0;i < 8; i++) { |
| 2731 |
/* ESP is not reloaded */
|
| 2732 |
if (i != 3) { |
| 2733 |
gen_op_ld_T0_A0(OT_WORD + s->dflag + s->mem_index); |
| 2734 |
gen_op_mov_reg_T0(OT_WORD + s->dflag, 7 - i);
|
| 2735 |
} |
| 2736 |
gen_op_addl_A0_im(2 << s->dflag);
|
| 2737 |
} |
| 2738 |
gen_op_mov_reg_T1(OT_WORD + s->ss32, R_ESP); |
| 2739 |
} |
| 2740 |
|
| 2741 |
static void gen_enter(DisasContext *s, int esp_addend, int level) |
| 2742 |
{
|
| 2743 |
int ot, opsize;
|
| 2744 |
|
| 2745 |
level &= 0x1f;
|
| 2746 |
#ifdef TARGET_X86_64
|
| 2747 |
if (CODE64(s)) {
|
| 2748 |
ot = s->dflag ? OT_QUAD : OT_WORD; |
| 2749 |
opsize = 1 << ot;
|
| 2750 |
|
| 2751 |
gen_op_movl_A0_reg(R_ESP); |
| 2752 |
gen_op_addq_A0_im(-opsize); |
| 2753 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2754 |
|
| 2755 |
/* push bp */
|
| 2756 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EBP);
|
| 2757 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 2758 |
if (level) {
|
| 2759 |
/* XXX: must save state */
|
| 2760 |
gen_helper_enter64_level(cpu_env, tcg_const_i32(level), |
| 2761 |
tcg_const_i32((ot == OT_QUAD)), |
| 2762 |
cpu_T[1]);
|
| 2763 |
} |
| 2764 |
gen_op_mov_reg_T1(ot, R_EBP); |
| 2765 |
tcg_gen_addi_tl(cpu_T[1], cpu_T[1], -esp_addend + (-opsize * level)); |
| 2766 |
gen_op_mov_reg_T1(OT_QUAD, R_ESP); |
| 2767 |
} else
|
| 2768 |
#endif
|
| 2769 |
{
|
| 2770 |
ot = s->dflag + OT_WORD; |
| 2771 |
opsize = 2 << s->dflag;
|
| 2772 |
|
| 2773 |
gen_op_movl_A0_reg(R_ESP); |
| 2774 |
gen_op_addl_A0_im(-opsize); |
| 2775 |
if (!s->ss32)
|
| 2776 |
gen_op_andl_A0_ffff(); |
| 2777 |
tcg_gen_mov_tl(cpu_T[1], cpu_A0);
|
| 2778 |
if (s->addseg)
|
| 2779 |
gen_op_addl_A0_seg(s, R_SS); |
| 2780 |
/* push bp */
|
| 2781 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EBP);
|
| 2782 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 2783 |
if (level) {
|
| 2784 |
/* XXX: must save state */
|
| 2785 |
gen_helper_enter_level(cpu_env, tcg_const_i32(level), |
| 2786 |
tcg_const_i32(s->dflag), |
| 2787 |
cpu_T[1]);
|
| 2788 |
} |
| 2789 |
gen_op_mov_reg_T1(ot, R_EBP); |
| 2790 |
tcg_gen_addi_tl(cpu_T[1], cpu_T[1], -esp_addend + (-opsize * level)); |
| 2791 |
gen_op_mov_reg_T1(OT_WORD + s->ss32, R_ESP); |
| 2792 |
} |
| 2793 |
} |
| 2794 |
|
| 2795 |
static void gen_exception(DisasContext *s, int trapno, target_ulong cur_eip) |
| 2796 |
{
|
| 2797 |
gen_update_cc_op(s); |
| 2798 |
gen_jmp_im(cur_eip); |
| 2799 |
gen_helper_raise_exception(cpu_env, tcg_const_i32(trapno)); |
| 2800 |
s->is_jmp = DISAS_TB_JUMP; |
| 2801 |
} |
| 2802 |
|
| 2803 |
/* an interrupt is different from an exception because of the
|
| 2804 |
privilege checks */
|
| 2805 |
static void gen_interrupt(DisasContext *s, int intno, |
| 2806 |
target_ulong cur_eip, target_ulong next_eip) |
| 2807 |
{
|
| 2808 |
gen_update_cc_op(s); |
| 2809 |
gen_jmp_im(cur_eip); |
| 2810 |
gen_helper_raise_interrupt(cpu_env, tcg_const_i32(intno), |
| 2811 |
tcg_const_i32(next_eip - cur_eip)); |
| 2812 |
s->is_jmp = DISAS_TB_JUMP; |
| 2813 |
} |
| 2814 |
|
| 2815 |
static void gen_debug(DisasContext *s, target_ulong cur_eip) |
| 2816 |
{
|
| 2817 |
gen_update_cc_op(s); |
| 2818 |
gen_jmp_im(cur_eip); |
| 2819 |
gen_helper_debug(cpu_env); |
| 2820 |
s->is_jmp = DISAS_TB_JUMP; |
| 2821 |
} |
| 2822 |
|
| 2823 |
/* generate a generic end of block. Trace exception is also generated
|
| 2824 |
if needed */
|
| 2825 |
static void gen_eob(DisasContext *s) |
| 2826 |
{
|
| 2827 |
gen_update_cc_op(s); |
| 2828 |
if (s->tb->flags & HF_INHIBIT_IRQ_MASK) {
|
| 2829 |
gen_helper_reset_inhibit_irq(cpu_env); |
| 2830 |
} |
| 2831 |
if (s->tb->flags & HF_RF_MASK) {
|
| 2832 |
gen_helper_reset_rf(cpu_env); |
| 2833 |
} |
| 2834 |
if (s->singlestep_enabled) {
|
| 2835 |
gen_helper_debug(cpu_env); |
| 2836 |
} else if (s->tf) { |
| 2837 |
gen_helper_single_step(cpu_env); |
| 2838 |
} else {
|
| 2839 |
tcg_gen_exit_tb(0);
|
| 2840 |
} |
| 2841 |
s->is_jmp = DISAS_TB_JUMP; |
| 2842 |
} |
| 2843 |
|
| 2844 |
/* generate a jump to eip. No segment change must happen before as a
|
| 2845 |
direct call to the next block may occur */
|
| 2846 |
static void gen_jmp_tb(DisasContext *s, target_ulong eip, int tb_num) |
| 2847 |
{
|
| 2848 |
gen_update_cc_op(s); |
| 2849 |
set_cc_op(s, CC_OP_DYNAMIC); |
| 2850 |
if (s->jmp_opt) {
|
| 2851 |
gen_goto_tb(s, tb_num, eip); |
| 2852 |
s->is_jmp = DISAS_TB_JUMP; |
| 2853 |
} else {
|
| 2854 |
gen_jmp_im(eip); |
| 2855 |
gen_eob(s); |
| 2856 |
} |
| 2857 |
} |
| 2858 |
|
| 2859 |
static void gen_jmp(DisasContext *s, target_ulong eip) |
| 2860 |
{
|
| 2861 |
gen_jmp_tb(s, eip, 0);
|
| 2862 |
} |
| 2863 |
|
| 2864 |
static inline void gen_ldq_env_A0(int idx, int offset) |
| 2865 |
{
|
| 2866 |
int mem_index = (idx >> 2) - 1; |
| 2867 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_A0, mem_index); |
| 2868 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, offset); |
| 2869 |
} |
| 2870 |
|
| 2871 |
static inline void gen_stq_env_A0(int idx, int offset) |
| 2872 |
{
|
| 2873 |
int mem_index = (idx >> 2) - 1; |
| 2874 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, offset); |
| 2875 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_A0, mem_index); |
| 2876 |
} |
| 2877 |
|
| 2878 |
static inline void gen_ldo_env_A0(int idx, int offset) |
| 2879 |
{
|
| 2880 |
int mem_index = (idx >> 2) - 1; |
| 2881 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_A0, mem_index); |
| 2882 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(XMMReg, XMM_Q(0)));
|
| 2883 |
tcg_gen_addi_tl(cpu_tmp0, cpu_A0, 8);
|
| 2884 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_tmp0, mem_index); |
| 2885 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(XMMReg, XMM_Q(1)));
|
| 2886 |
} |
| 2887 |
|
| 2888 |
static inline void gen_sto_env_A0(int idx, int offset) |
| 2889 |
{
|
| 2890 |
int mem_index = (idx >> 2) - 1; |
| 2891 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(XMMReg, XMM_Q(0)));
|
| 2892 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_A0, mem_index); |
| 2893 |
tcg_gen_addi_tl(cpu_tmp0, cpu_A0, 8);
|
| 2894 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, offset + offsetof(XMMReg, XMM_Q(1)));
|
| 2895 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_tmp0, mem_index); |
| 2896 |
} |
| 2897 |
|
| 2898 |
static inline void gen_op_movo(int d_offset, int s_offset) |
| 2899 |
{
|
| 2900 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, s_offset); |
| 2901 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset); |
| 2902 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, s_offset + 8);
|
| 2903 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset + 8);
|
| 2904 |
} |
| 2905 |
|
| 2906 |
static inline void gen_op_movq(int d_offset, int s_offset) |
| 2907 |
{
|
| 2908 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, s_offset); |
| 2909 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset); |
| 2910 |
} |
| 2911 |
|
| 2912 |
static inline void gen_op_movl(int d_offset, int s_offset) |
| 2913 |
{
|
| 2914 |
tcg_gen_ld_i32(cpu_tmp2_i32, cpu_env, s_offset); |
| 2915 |
tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, d_offset); |
| 2916 |
} |
| 2917 |
|
| 2918 |
static inline void gen_op_movq_env_0(int d_offset) |
| 2919 |
{
|
| 2920 |
tcg_gen_movi_i64(cpu_tmp1_i64, 0);
|
| 2921 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, d_offset); |
| 2922 |
} |
| 2923 |
|
| 2924 |
typedef void (*SSEFunc_i_ep)(TCGv_i32 val, TCGv_ptr env, TCGv_ptr reg); |
| 2925 |
typedef void (*SSEFunc_l_ep)(TCGv_i64 val, TCGv_ptr env, TCGv_ptr reg); |
| 2926 |
typedef void (*SSEFunc_0_epi)(TCGv_ptr env, TCGv_ptr reg, TCGv_i32 val); |
| 2927 |
typedef void (*SSEFunc_0_epl)(TCGv_ptr env, TCGv_ptr reg, TCGv_i64 val); |
| 2928 |
typedef void (*SSEFunc_0_epp)(TCGv_ptr env, TCGv_ptr reg_a, TCGv_ptr reg_b); |
| 2929 |
typedef void (*SSEFunc_0_eppi)(TCGv_ptr env, TCGv_ptr reg_a, TCGv_ptr reg_b, |
| 2930 |
TCGv_i32 val); |
| 2931 |
typedef void (*SSEFunc_0_ppi)(TCGv_ptr reg_a, TCGv_ptr reg_b, TCGv_i32 val); |
| 2932 |
typedef void (*SSEFunc_0_eppt)(TCGv_ptr env, TCGv_ptr reg_a, TCGv_ptr reg_b, |
| 2933 |
TCGv val); |
| 2934 |
|
| 2935 |
#define SSE_SPECIAL ((void *)1) |
| 2936 |
#define SSE_DUMMY ((void *)2) |
| 2937 |
|
| 2938 |
#define MMX_OP2(x) { gen_helper_ ## x ## _mmx, gen_helper_ ## x ## _xmm } |
| 2939 |
#define SSE_FOP(x) { gen_helper_ ## x ## ps, gen_helper_ ## x ## pd, \ |
| 2940 |
gen_helper_ ## x ## ss, gen_helper_ ## x ## sd, } |
| 2941 |
|
| 2942 |
static const SSEFunc_0_epp sse_op_table1[256][4] = { |
| 2943 |
/* 3DNow! extensions */
|
| 2944 |
[0x0e] = { SSE_DUMMY }, /* femms */ |
| 2945 |
[0x0f] = { SSE_DUMMY }, /* pf... */ |
| 2946 |
/* pure SSE operations */
|
| 2947 |
[0x10] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movups, movupd, movss, movsd */ |
| 2948 |
[0x11] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movups, movupd, movss, movsd */ |
| 2949 |
[0x12] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movlps, movlpd, movsldup, movddup */ |
| 2950 |
[0x13] = { SSE_SPECIAL, SSE_SPECIAL }, /* movlps, movlpd */ |
| 2951 |
[0x14] = { gen_helper_punpckldq_xmm, gen_helper_punpcklqdq_xmm },
|
| 2952 |
[0x15] = { gen_helper_punpckhdq_xmm, gen_helper_punpckhqdq_xmm },
|
| 2953 |
[0x16] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movhps, movhpd, movshdup */ |
| 2954 |
[0x17] = { SSE_SPECIAL, SSE_SPECIAL }, /* movhps, movhpd */ |
| 2955 |
|
| 2956 |
[0x28] = { SSE_SPECIAL, SSE_SPECIAL }, /* movaps, movapd */ |
| 2957 |
[0x29] = { SSE_SPECIAL, SSE_SPECIAL }, /* movaps, movapd */ |
| 2958 |
[0x2a] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* cvtpi2ps, cvtpi2pd, cvtsi2ss, cvtsi2sd */ |
| 2959 |
[0x2b] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movntps, movntpd, movntss, movntsd */ |
| 2960 |
[0x2c] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* cvttps2pi, cvttpd2pi, cvttsd2si, cvttss2si */ |
| 2961 |
[0x2d] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* cvtps2pi, cvtpd2pi, cvtsd2si, cvtss2si */ |
| 2962 |
[0x2e] = { gen_helper_ucomiss, gen_helper_ucomisd },
|
| 2963 |
[0x2f] = { gen_helper_comiss, gen_helper_comisd },
|
| 2964 |
[0x50] = { SSE_SPECIAL, SSE_SPECIAL }, /* movmskps, movmskpd */ |
| 2965 |
[0x51] = SSE_FOP(sqrt),
|
| 2966 |
[0x52] = { gen_helper_rsqrtps, NULL, gen_helper_rsqrtss, NULL }, |
| 2967 |
[0x53] = { gen_helper_rcpps, NULL, gen_helper_rcpss, NULL }, |
| 2968 |
[0x54] = { gen_helper_pand_xmm, gen_helper_pand_xmm }, /* andps, andpd */ |
| 2969 |
[0x55] = { gen_helper_pandn_xmm, gen_helper_pandn_xmm }, /* andnps, andnpd */ |
| 2970 |
[0x56] = { gen_helper_por_xmm, gen_helper_por_xmm }, /* orps, orpd */ |
| 2971 |
[0x57] = { gen_helper_pxor_xmm, gen_helper_pxor_xmm }, /* xorps, xorpd */ |
| 2972 |
[0x58] = SSE_FOP(add),
|
| 2973 |
[0x59] = SSE_FOP(mul),
|
| 2974 |
[0x5a] = { gen_helper_cvtps2pd, gen_helper_cvtpd2ps,
|
| 2975 |
gen_helper_cvtss2sd, gen_helper_cvtsd2ss }, |
| 2976 |
[0x5b] = { gen_helper_cvtdq2ps, gen_helper_cvtps2dq, gen_helper_cvttps2dq },
|
| 2977 |
[0x5c] = SSE_FOP(sub),
|
| 2978 |
[0x5d] = SSE_FOP(min),
|
| 2979 |
[0x5e] = SSE_FOP(div),
|
| 2980 |
[0x5f] = SSE_FOP(max),
|
| 2981 |
|
| 2982 |
[0xc2] = SSE_FOP(cmpeq),
|
| 2983 |
[0xc6] = { (SSEFunc_0_epp)gen_helper_shufps,
|
| 2984 |
(SSEFunc_0_epp)gen_helper_shufpd }, /* XXX: casts */
|
| 2985 |
|
| 2986 |
/* SSSE3, SSE4, MOVBE, CRC32, BMI1, BMI2, ADX. */
|
| 2987 |
[0x38] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL },
|
| 2988 |
[0x3a] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL },
|
| 2989 |
|
| 2990 |
/* MMX ops and their SSE extensions */
|
| 2991 |
[0x60] = MMX_OP2(punpcklbw),
|
| 2992 |
[0x61] = MMX_OP2(punpcklwd),
|
| 2993 |
[0x62] = MMX_OP2(punpckldq),
|
| 2994 |
[0x63] = MMX_OP2(packsswb),
|
| 2995 |
[0x64] = MMX_OP2(pcmpgtb),
|
| 2996 |
[0x65] = MMX_OP2(pcmpgtw),
|
| 2997 |
[0x66] = MMX_OP2(pcmpgtl),
|
| 2998 |
[0x67] = MMX_OP2(packuswb),
|
| 2999 |
[0x68] = MMX_OP2(punpckhbw),
|
| 3000 |
[0x69] = MMX_OP2(punpckhwd),
|
| 3001 |
[0x6a] = MMX_OP2(punpckhdq),
|
| 3002 |
[0x6b] = MMX_OP2(packssdw),
|
| 3003 |
[0x6c] = { NULL, gen_helper_punpcklqdq_xmm }, |
| 3004 |
[0x6d] = { NULL, gen_helper_punpckhqdq_xmm }, |
| 3005 |
[0x6e] = { SSE_SPECIAL, SSE_SPECIAL }, /* movd mm, ea */ |
| 3006 |
[0x6f] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movq, movdqa, , movqdu */ |
| 3007 |
[0x70] = { (SSEFunc_0_epp)gen_helper_pshufw_mmx,
|
| 3008 |
(SSEFunc_0_epp)gen_helper_pshufd_xmm, |
| 3009 |
(SSEFunc_0_epp)gen_helper_pshufhw_xmm, |
| 3010 |
(SSEFunc_0_epp)gen_helper_pshuflw_xmm }, /* XXX: casts */
|
| 3011 |
[0x71] = { SSE_SPECIAL, SSE_SPECIAL }, /* shiftw */ |
| 3012 |
[0x72] = { SSE_SPECIAL, SSE_SPECIAL }, /* shiftd */ |
| 3013 |
[0x73] = { SSE_SPECIAL, SSE_SPECIAL }, /* shiftq */ |
| 3014 |
[0x74] = MMX_OP2(pcmpeqb),
|
| 3015 |
[0x75] = MMX_OP2(pcmpeqw),
|
| 3016 |
[0x76] = MMX_OP2(pcmpeql),
|
| 3017 |
[0x77] = { SSE_DUMMY }, /* emms */ |
| 3018 |
[0x78] = { NULL, SSE_SPECIAL, NULL, SSE_SPECIAL }, /* extrq_i, insertq_i */ |
| 3019 |
[0x79] = { NULL, gen_helper_extrq_r, NULL, gen_helper_insertq_r }, |
| 3020 |
[0x7c] = { NULL, gen_helper_haddpd, NULL, gen_helper_haddps }, |
| 3021 |
[0x7d] = { NULL, gen_helper_hsubpd, NULL, gen_helper_hsubps }, |
| 3022 |
[0x7e] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movd, movd, , movq */ |
| 3023 |
[0x7f] = { SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, /* movq, movdqa, movdqu */ |
| 3024 |
[0xc4] = { SSE_SPECIAL, SSE_SPECIAL }, /* pinsrw */ |
| 3025 |
[0xc5] = { SSE_SPECIAL, SSE_SPECIAL }, /* pextrw */ |
| 3026 |
[0xd0] = { NULL, gen_helper_addsubpd, NULL, gen_helper_addsubps }, |
| 3027 |
[0xd1] = MMX_OP2(psrlw),
|
| 3028 |
[0xd2] = MMX_OP2(psrld),
|
| 3029 |
[0xd3] = MMX_OP2(psrlq),
|
| 3030 |
[0xd4] = MMX_OP2(paddq),
|
| 3031 |
[0xd5] = MMX_OP2(pmullw),
|
| 3032 |
[0xd6] = { NULL, SSE_SPECIAL, SSE_SPECIAL, SSE_SPECIAL }, |
| 3033 |
[0xd7] = { SSE_SPECIAL, SSE_SPECIAL }, /* pmovmskb */ |
| 3034 |
[0xd8] = MMX_OP2(psubusb),
|
| 3035 |
[0xd9] = MMX_OP2(psubusw),
|
| 3036 |
[0xda] = MMX_OP2(pminub),
|
| 3037 |
[0xdb] = MMX_OP2(pand),
|
| 3038 |
[0xdc] = MMX_OP2(paddusb),
|
| 3039 |
[0xdd] = MMX_OP2(paddusw),
|
| 3040 |
[0xde] = MMX_OP2(pmaxub),
|
| 3041 |
[0xdf] = MMX_OP2(pandn),
|
| 3042 |
[0xe0] = MMX_OP2(pavgb),
|
| 3043 |
[0xe1] = MMX_OP2(psraw),
|
| 3044 |
[0xe2] = MMX_OP2(psrad),
|
| 3045 |
[0xe3] = MMX_OP2(pavgw),
|
| 3046 |
[0xe4] = MMX_OP2(pmulhuw),
|
| 3047 |
[0xe5] = MMX_OP2(pmulhw),
|
| 3048 |
[0xe6] = { NULL, gen_helper_cvttpd2dq, gen_helper_cvtdq2pd, gen_helper_cvtpd2dq }, |
| 3049 |
[0xe7] = { SSE_SPECIAL , SSE_SPECIAL }, /* movntq, movntq */ |
| 3050 |
[0xe8] = MMX_OP2(psubsb),
|
| 3051 |
[0xe9] = MMX_OP2(psubsw),
|
| 3052 |
[0xea] = MMX_OP2(pminsw),
|
| 3053 |
[0xeb] = MMX_OP2(por),
|
| 3054 |
[0xec] = MMX_OP2(paddsb),
|
| 3055 |
[0xed] = MMX_OP2(paddsw),
|
| 3056 |
[0xee] = MMX_OP2(pmaxsw),
|
| 3057 |
[0xef] = MMX_OP2(pxor),
|
| 3058 |
[0xf0] = { NULL, NULL, NULL, SSE_SPECIAL }, /* lddqu */ |
| 3059 |
[0xf1] = MMX_OP2(psllw),
|
| 3060 |
[0xf2] = MMX_OP2(pslld),
|
| 3061 |
[0xf3] = MMX_OP2(psllq),
|
| 3062 |
[0xf4] = MMX_OP2(pmuludq),
|
| 3063 |
[0xf5] = MMX_OP2(pmaddwd),
|
| 3064 |
[0xf6] = MMX_OP2(psadbw),
|
| 3065 |
[0xf7] = { (SSEFunc_0_epp)gen_helper_maskmov_mmx,
|
| 3066 |
(SSEFunc_0_epp)gen_helper_maskmov_xmm }, /* XXX: casts */
|
| 3067 |
[0xf8] = MMX_OP2(psubb),
|
| 3068 |
[0xf9] = MMX_OP2(psubw),
|
| 3069 |
[0xfa] = MMX_OP2(psubl),
|
| 3070 |
[0xfb] = MMX_OP2(psubq),
|
| 3071 |
[0xfc] = MMX_OP2(paddb),
|
| 3072 |
[0xfd] = MMX_OP2(paddw),
|
| 3073 |
[0xfe] = MMX_OP2(paddl),
|
| 3074 |
}; |
| 3075 |
|
| 3076 |
static const SSEFunc_0_epp sse_op_table2[3 * 8][2] = { |
| 3077 |
[0 + 2] = MMX_OP2(psrlw), |
| 3078 |
[0 + 4] = MMX_OP2(psraw), |
| 3079 |
[0 + 6] = MMX_OP2(psllw), |
| 3080 |
[8 + 2] = MMX_OP2(psrld), |
| 3081 |
[8 + 4] = MMX_OP2(psrad), |
| 3082 |
[8 + 6] = MMX_OP2(pslld), |
| 3083 |
[16 + 2] = MMX_OP2(psrlq), |
| 3084 |
[16 + 3] = { NULL, gen_helper_psrldq_xmm }, |
| 3085 |
[16 + 6] = MMX_OP2(psllq), |
| 3086 |
[16 + 7] = { NULL, gen_helper_pslldq_xmm }, |
| 3087 |
}; |
| 3088 |
|
| 3089 |
static const SSEFunc_0_epi sse_op_table3ai[] = { |
| 3090 |
gen_helper_cvtsi2ss, |
| 3091 |
gen_helper_cvtsi2sd |
| 3092 |
}; |
| 3093 |
|
| 3094 |
#ifdef TARGET_X86_64
|
| 3095 |
static const SSEFunc_0_epl sse_op_table3aq[] = { |
| 3096 |
gen_helper_cvtsq2ss, |
| 3097 |
gen_helper_cvtsq2sd |
| 3098 |
}; |
| 3099 |
#endif
|
| 3100 |
|
| 3101 |
static const SSEFunc_i_ep sse_op_table3bi[] = { |
| 3102 |
gen_helper_cvttss2si, |
| 3103 |
gen_helper_cvtss2si, |
| 3104 |
gen_helper_cvttsd2si, |
| 3105 |
gen_helper_cvtsd2si |
| 3106 |
}; |
| 3107 |
|
| 3108 |
#ifdef TARGET_X86_64
|
| 3109 |
static const SSEFunc_l_ep sse_op_table3bq[] = { |
| 3110 |
gen_helper_cvttss2sq, |
| 3111 |
gen_helper_cvtss2sq, |
| 3112 |
gen_helper_cvttsd2sq, |
| 3113 |
gen_helper_cvtsd2sq |
| 3114 |
}; |
| 3115 |
#endif
|
| 3116 |
|
| 3117 |
static const SSEFunc_0_epp sse_op_table4[8][4] = { |
| 3118 |
SSE_FOP(cmpeq), |
| 3119 |
SSE_FOP(cmplt), |
| 3120 |
SSE_FOP(cmple), |
| 3121 |
SSE_FOP(cmpunord), |
| 3122 |
SSE_FOP(cmpneq), |
| 3123 |
SSE_FOP(cmpnlt), |
| 3124 |
SSE_FOP(cmpnle), |
| 3125 |
SSE_FOP(cmpord), |
| 3126 |
}; |
| 3127 |
|
| 3128 |
static const SSEFunc_0_epp sse_op_table5[256] = { |
| 3129 |
[0x0c] = gen_helper_pi2fw,
|
| 3130 |
[0x0d] = gen_helper_pi2fd,
|
| 3131 |
[0x1c] = gen_helper_pf2iw,
|
| 3132 |
[0x1d] = gen_helper_pf2id,
|
| 3133 |
[0x8a] = gen_helper_pfnacc,
|
| 3134 |
[0x8e] = gen_helper_pfpnacc,
|
| 3135 |
[0x90] = gen_helper_pfcmpge,
|
| 3136 |
[0x94] = gen_helper_pfmin,
|
| 3137 |
[0x96] = gen_helper_pfrcp,
|
| 3138 |
[0x97] = gen_helper_pfrsqrt,
|
| 3139 |
[0x9a] = gen_helper_pfsub,
|
| 3140 |
[0x9e] = gen_helper_pfadd,
|
| 3141 |
[0xa0] = gen_helper_pfcmpgt,
|
| 3142 |
[0xa4] = gen_helper_pfmax,
|
| 3143 |
[0xa6] = gen_helper_movq, /* pfrcpit1; no need to actually increase precision */ |
| 3144 |
[0xa7] = gen_helper_movq, /* pfrsqit1 */ |
| 3145 |
[0xaa] = gen_helper_pfsubr,
|
| 3146 |
[0xae] = gen_helper_pfacc,
|
| 3147 |
[0xb0] = gen_helper_pfcmpeq,
|
| 3148 |
[0xb4] = gen_helper_pfmul,
|
| 3149 |
[0xb6] = gen_helper_movq, /* pfrcpit2 */ |
| 3150 |
[0xb7] = gen_helper_pmulhrw_mmx,
|
| 3151 |
[0xbb] = gen_helper_pswapd,
|
| 3152 |
[0xbf] = gen_helper_pavgb_mmx /* pavgusb */ |
| 3153 |
}; |
| 3154 |
|
| 3155 |
struct SSEOpHelper_epp {
|
| 3156 |
SSEFunc_0_epp op[2];
|
| 3157 |
uint32_t ext_mask; |
| 3158 |
}; |
| 3159 |
|
| 3160 |
struct SSEOpHelper_eppi {
|
| 3161 |
SSEFunc_0_eppi op[2];
|
| 3162 |
uint32_t ext_mask; |
| 3163 |
}; |
| 3164 |
|
| 3165 |
#define SSSE3_OP(x) { MMX_OP2(x), CPUID_EXT_SSSE3 }
|
| 3166 |
#define SSE41_OP(x) { { NULL, gen_helper_ ## x ## _xmm }, CPUID_EXT_SSE41 } |
| 3167 |
#define SSE42_OP(x) { { NULL, gen_helper_ ## x ## _xmm }, CPUID_EXT_SSE42 } |
| 3168 |
#define SSE41_SPECIAL { { NULL, SSE_SPECIAL }, CPUID_EXT_SSE41 } |
| 3169 |
|
| 3170 |
static const struct SSEOpHelper_epp sse_op_table6[256] = { |
| 3171 |
[0x00] = SSSE3_OP(pshufb),
|
| 3172 |
[0x01] = SSSE3_OP(phaddw),
|
| 3173 |
[0x02] = SSSE3_OP(phaddd),
|
| 3174 |
[0x03] = SSSE3_OP(phaddsw),
|
| 3175 |
[0x04] = SSSE3_OP(pmaddubsw),
|
| 3176 |
[0x05] = SSSE3_OP(phsubw),
|
| 3177 |
[0x06] = SSSE3_OP(phsubd),
|
| 3178 |
[0x07] = SSSE3_OP(phsubsw),
|
| 3179 |
[0x08] = SSSE3_OP(psignb),
|
| 3180 |
[0x09] = SSSE3_OP(psignw),
|
| 3181 |
[0x0a] = SSSE3_OP(psignd),
|
| 3182 |
[0x0b] = SSSE3_OP(pmulhrsw),
|
| 3183 |
[0x10] = SSE41_OP(pblendvb),
|
| 3184 |
[0x14] = SSE41_OP(blendvps),
|
| 3185 |
[0x15] = SSE41_OP(blendvpd),
|
| 3186 |
[0x17] = SSE41_OP(ptest),
|
| 3187 |
[0x1c] = SSSE3_OP(pabsb),
|
| 3188 |
[0x1d] = SSSE3_OP(pabsw),
|
| 3189 |
[0x1e] = SSSE3_OP(pabsd),
|
| 3190 |
[0x20] = SSE41_OP(pmovsxbw),
|
| 3191 |
[0x21] = SSE41_OP(pmovsxbd),
|
| 3192 |
[0x22] = SSE41_OP(pmovsxbq),
|
| 3193 |
[0x23] = SSE41_OP(pmovsxwd),
|
| 3194 |
[0x24] = SSE41_OP(pmovsxwq),
|
| 3195 |
[0x25] = SSE41_OP(pmovsxdq),
|
| 3196 |
[0x28] = SSE41_OP(pmuldq),
|
| 3197 |
[0x29] = SSE41_OP(pcmpeqq),
|
| 3198 |
[0x2a] = SSE41_SPECIAL, /* movntqda */ |
| 3199 |
[0x2b] = SSE41_OP(packusdw),
|
| 3200 |
[0x30] = SSE41_OP(pmovzxbw),
|
| 3201 |
[0x31] = SSE41_OP(pmovzxbd),
|
| 3202 |
[0x32] = SSE41_OP(pmovzxbq),
|
| 3203 |
[0x33] = SSE41_OP(pmovzxwd),
|
| 3204 |
[0x34] = SSE41_OP(pmovzxwq),
|
| 3205 |
[0x35] = SSE41_OP(pmovzxdq),
|
| 3206 |
[0x37] = SSE42_OP(pcmpgtq),
|
| 3207 |
[0x38] = SSE41_OP(pminsb),
|
| 3208 |
[0x39] = SSE41_OP(pminsd),
|
| 3209 |
[0x3a] = SSE41_OP(pminuw),
|
| 3210 |
[0x3b] = SSE41_OP(pminud),
|
| 3211 |
[0x3c] = SSE41_OP(pmaxsb),
|
| 3212 |
[0x3d] = SSE41_OP(pmaxsd),
|
| 3213 |
[0x3e] = SSE41_OP(pmaxuw),
|
| 3214 |
[0x3f] = SSE41_OP(pmaxud),
|
| 3215 |
[0x40] = SSE41_OP(pmulld),
|
| 3216 |
[0x41] = SSE41_OP(phminposuw),
|
| 3217 |
}; |
| 3218 |
|
| 3219 |
static const struct SSEOpHelper_eppi sse_op_table7[256] = { |
| 3220 |
[0x08] = SSE41_OP(roundps),
|
| 3221 |
[0x09] = SSE41_OP(roundpd),
|
| 3222 |
[0x0a] = SSE41_OP(roundss),
|
| 3223 |
[0x0b] = SSE41_OP(roundsd),
|
| 3224 |
[0x0c] = SSE41_OP(blendps),
|
| 3225 |
[0x0d] = SSE41_OP(blendpd),
|
| 3226 |
[0x0e] = SSE41_OP(pblendw),
|
| 3227 |
[0x0f] = SSSE3_OP(palignr),
|
| 3228 |
[0x14] = SSE41_SPECIAL, /* pextrb */ |
| 3229 |
[0x15] = SSE41_SPECIAL, /* pextrw */ |
| 3230 |
[0x16] = SSE41_SPECIAL, /* pextrd/pextrq */ |
| 3231 |
[0x17] = SSE41_SPECIAL, /* extractps */ |
| 3232 |
[0x20] = SSE41_SPECIAL, /* pinsrb */ |
| 3233 |
[0x21] = SSE41_SPECIAL, /* insertps */ |
| 3234 |
[0x22] = SSE41_SPECIAL, /* pinsrd/pinsrq */ |
| 3235 |
[0x40] = SSE41_OP(dpps),
|
| 3236 |
[0x41] = SSE41_OP(dppd),
|
| 3237 |
[0x42] = SSE41_OP(mpsadbw),
|
| 3238 |
[0x60] = SSE42_OP(pcmpestrm),
|
| 3239 |
[0x61] = SSE42_OP(pcmpestri),
|
| 3240 |
[0x62] = SSE42_OP(pcmpistrm),
|
| 3241 |
[0x63] = SSE42_OP(pcmpistri),
|
| 3242 |
}; |
| 3243 |
|
| 3244 |
static void gen_sse(CPUX86State *env, DisasContext *s, int b, |
| 3245 |
target_ulong pc_start, int rex_r)
|
| 3246 |
{
|
| 3247 |
int b1, op1_offset, op2_offset, is_xmm, val, ot;
|
| 3248 |
int modrm, mod, rm, reg, reg_addr, offset_addr;
|
| 3249 |
SSEFunc_0_epp sse_fn_epp; |
| 3250 |
SSEFunc_0_eppi sse_fn_eppi; |
| 3251 |
SSEFunc_0_ppi sse_fn_ppi; |
| 3252 |
SSEFunc_0_eppt sse_fn_eppt; |
| 3253 |
|
| 3254 |
b &= 0xff;
|
| 3255 |
if (s->prefix & PREFIX_DATA)
|
| 3256 |
b1 = 1;
|
| 3257 |
else if (s->prefix & PREFIX_REPZ) |
| 3258 |
b1 = 2;
|
| 3259 |
else if (s->prefix & PREFIX_REPNZ) |
| 3260 |
b1 = 3;
|
| 3261 |
else
|
| 3262 |
b1 = 0;
|
| 3263 |
sse_fn_epp = sse_op_table1[b][b1]; |
| 3264 |
if (!sse_fn_epp) {
|
| 3265 |
goto illegal_op;
|
| 3266 |
} |
| 3267 |
if ((b <= 0x5f && b >= 0x10) || b == 0xc6 || b == 0xc2) { |
| 3268 |
is_xmm = 1;
|
| 3269 |
} else {
|
| 3270 |
if (b1 == 0) { |
| 3271 |
/* MMX case */
|
| 3272 |
is_xmm = 0;
|
| 3273 |
} else {
|
| 3274 |
is_xmm = 1;
|
| 3275 |
} |
| 3276 |
} |
| 3277 |
/* simple MMX/SSE operation */
|
| 3278 |
if (s->flags & HF_TS_MASK) {
|
| 3279 |
gen_exception(s, EXCP07_PREX, pc_start - s->cs_base); |
| 3280 |
return;
|
| 3281 |
} |
| 3282 |
if (s->flags & HF_EM_MASK) {
|
| 3283 |
illegal_op:
|
| 3284 |
gen_exception(s, EXCP06_ILLOP, pc_start - s->cs_base); |
| 3285 |
return;
|
| 3286 |
} |
| 3287 |
if (is_xmm && !(s->flags & HF_OSFXSR_MASK))
|
| 3288 |
if ((b != 0x38 && b != 0x3a) || (s->prefix & PREFIX_DATA)) |
| 3289 |
goto illegal_op;
|
| 3290 |
if (b == 0x0e) { |
| 3291 |
if (!(s->cpuid_ext2_features & CPUID_EXT2_3DNOW))
|
| 3292 |
goto illegal_op;
|
| 3293 |
/* femms */
|
| 3294 |
gen_helper_emms(cpu_env); |
| 3295 |
return;
|
| 3296 |
} |
| 3297 |
if (b == 0x77) { |
| 3298 |
/* emms */
|
| 3299 |
gen_helper_emms(cpu_env); |
| 3300 |
return;
|
| 3301 |
} |
| 3302 |
/* prepare MMX state (XXX: optimize by storing fptt and fptags in
|
| 3303 |
the static cpu state) */
|
| 3304 |
if (!is_xmm) {
|
| 3305 |
gen_helper_enter_mmx(cpu_env); |
| 3306 |
} |
| 3307 |
|
| 3308 |
modrm = cpu_ldub_code(env, s->pc++); |
| 3309 |
reg = ((modrm >> 3) & 7); |
| 3310 |
if (is_xmm)
|
| 3311 |
reg |= rex_r; |
| 3312 |
mod = (modrm >> 6) & 3; |
| 3313 |
if (sse_fn_epp == SSE_SPECIAL) {
|
| 3314 |
b |= (b1 << 8);
|
| 3315 |
switch(b) {
|
| 3316 |
case 0x0e7: /* movntq */ |
| 3317 |
if (mod == 3) |
| 3318 |
goto illegal_op;
|
| 3319 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3320 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3321 |
break;
|
| 3322 |
case 0x1e7: /* movntdq */ |
| 3323 |
case 0x02b: /* movntps */ |
| 3324 |
case 0x12b: /* movntps */ |
| 3325 |
if (mod == 3) |
| 3326 |
goto illegal_op;
|
| 3327 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3328 |
gen_sto_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg])); |
| 3329 |
break;
|
| 3330 |
case 0x3f0: /* lddqu */ |
| 3331 |
if (mod == 3) |
| 3332 |
goto illegal_op;
|
| 3333 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3334 |
gen_ldo_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg])); |
| 3335 |
break;
|
| 3336 |
case 0x22b: /* movntss */ |
| 3337 |
case 0x32b: /* movntsd */ |
| 3338 |
if (mod == 3) |
| 3339 |
goto illegal_op;
|
| 3340 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3341 |
if (b1 & 1) { |
| 3342 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State, |
| 3343 |
xmm_regs[reg])); |
| 3344 |
} else {
|
| 3345 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,
|
| 3346 |
xmm_regs[reg].XMM_L(0)));
|
| 3347 |
gen_op_st_T0_A0(OT_LONG + s->mem_index); |
| 3348 |
} |
| 3349 |
break;
|
| 3350 |
case 0x6e: /* movd mm, ea */ |
| 3351 |
#ifdef TARGET_X86_64
|
| 3352 |
if (s->dflag == 2) { |
| 3353 |
gen_ldst_modrm(env, s, modrm, OT_QUAD, OR_TMP0, 0);
|
| 3354 |
tcg_gen_st_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,fpregs[reg].mmx));
|
| 3355 |
} else
|
| 3356 |
#endif
|
| 3357 |
{
|
| 3358 |
gen_ldst_modrm(env, s, modrm, OT_LONG, OR_TMP0, 0);
|
| 3359 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, |
| 3360 |
offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3361 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 3362 |
gen_helper_movl_mm_T0_mmx(cpu_ptr0, cpu_tmp2_i32); |
| 3363 |
} |
| 3364 |
break;
|
| 3365 |
case 0x16e: /* movd xmm, ea */ |
| 3366 |
#ifdef TARGET_X86_64
|
| 3367 |
if (s->dflag == 2) { |
| 3368 |
gen_ldst_modrm(env, s, modrm, OT_QUAD, OR_TMP0, 0);
|
| 3369 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, |
| 3370 |
offsetof(CPUX86State,xmm_regs[reg])); |
| 3371 |
gen_helper_movq_mm_T0_xmm(cpu_ptr0, cpu_T[0]);
|
| 3372 |
} else
|
| 3373 |
#endif
|
| 3374 |
{
|
| 3375 |
gen_ldst_modrm(env, s, modrm, OT_LONG, OR_TMP0, 0);
|
| 3376 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, |
| 3377 |
offsetof(CPUX86State,xmm_regs[reg])); |
| 3378 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 3379 |
gen_helper_movl_mm_T0_xmm(cpu_ptr0, cpu_tmp2_i32); |
| 3380 |
} |
| 3381 |
break;
|
| 3382 |
case 0x6f: /* movq mm, ea */ |
| 3383 |
if (mod != 3) { |
| 3384 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3385 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3386 |
} else {
|
| 3387 |
rm = (modrm & 7);
|
| 3388 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, |
| 3389 |
offsetof(CPUX86State,fpregs[rm].mmx)); |
| 3390 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, |
| 3391 |
offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3392 |
} |
| 3393 |
break;
|
| 3394 |
case 0x010: /* movups */ |
| 3395 |
case 0x110: /* movupd */ |
| 3396 |
case 0x028: /* movaps */ |
| 3397 |
case 0x128: /* movapd */ |
| 3398 |
case 0x16f: /* movdqa xmm, ea */ |
| 3399 |
case 0x26f: /* movdqu xmm, ea */ |
| 3400 |
if (mod != 3) { |
| 3401 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3402 |
gen_ldo_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg])); |
| 3403 |
} else {
|
| 3404 |
rm = (modrm & 7) | REX_B(s);
|
| 3405 |
gen_op_movo(offsetof(CPUX86State,xmm_regs[reg]), |
| 3406 |
offsetof(CPUX86State,xmm_regs[rm])); |
| 3407 |
} |
| 3408 |
break;
|
| 3409 |
case 0x210: /* movss xmm, ea */ |
| 3410 |
if (mod != 3) { |
| 3411 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3412 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 3413 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(0))); |
| 3414 |
gen_op_movl_T0_0(); |
| 3415 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(1))); |
| 3416 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(2))); |
| 3417 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(3))); |
| 3418 |
} else {
|
| 3419 |
rm = (modrm & 7) | REX_B(s);
|
| 3420 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(0)),
|
| 3421 |
offsetof(CPUX86State,xmm_regs[rm].XMM_L(0)));
|
| 3422 |
} |
| 3423 |
break;
|
| 3424 |
case 0x310: /* movsd xmm, ea */ |
| 3425 |
if (mod != 3) { |
| 3426 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3427 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3428 |
gen_op_movl_T0_0(); |
| 3429 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(2))); |
| 3430 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(3))); |
| 3431 |
} else {
|
| 3432 |
rm = (modrm & 7) | REX_B(s);
|
| 3433 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)),
|
| 3434 |
offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)));
|
| 3435 |
} |
| 3436 |
break;
|
| 3437 |
case 0x012: /* movlps */ |
| 3438 |
case 0x112: /* movlpd */ |
| 3439 |
if (mod != 3) { |
| 3440 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3441 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3442 |
} else {
|
| 3443 |
/* movhlps */
|
| 3444 |
rm = (modrm & 7) | REX_B(s);
|
| 3445 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)),
|
| 3446 |
offsetof(CPUX86State,xmm_regs[rm].XMM_Q(1)));
|
| 3447 |
} |
| 3448 |
break;
|
| 3449 |
case 0x212: /* movsldup */ |
| 3450 |
if (mod != 3) { |
| 3451 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3452 |
gen_ldo_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg])); |
| 3453 |
} else {
|
| 3454 |
rm = (modrm & 7) | REX_B(s);
|
| 3455 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(0)),
|
| 3456 |
offsetof(CPUX86State,xmm_regs[rm].XMM_L(0)));
|
| 3457 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(2)),
|
| 3458 |
offsetof(CPUX86State,xmm_regs[rm].XMM_L(2)));
|
| 3459 |
} |
| 3460 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(1)),
|
| 3461 |
offsetof(CPUX86State,xmm_regs[reg].XMM_L(0)));
|
| 3462 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(3)),
|
| 3463 |
offsetof(CPUX86State,xmm_regs[reg].XMM_L(2)));
|
| 3464 |
break;
|
| 3465 |
case 0x312: /* movddup */ |
| 3466 |
if (mod != 3) { |
| 3467 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3468 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3469 |
} else {
|
| 3470 |
rm = (modrm & 7) | REX_B(s);
|
| 3471 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)),
|
| 3472 |
offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)));
|
| 3473 |
} |
| 3474 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(1)),
|
| 3475 |
offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3476 |
break;
|
| 3477 |
case 0x016: /* movhps */ |
| 3478 |
case 0x116: /* movhpd */ |
| 3479 |
if (mod != 3) { |
| 3480 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3481 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(1)));
|
| 3482 |
} else {
|
| 3483 |
/* movlhps */
|
| 3484 |
rm = (modrm & 7) | REX_B(s);
|
| 3485 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(1)),
|
| 3486 |
offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)));
|
| 3487 |
} |
| 3488 |
break;
|
| 3489 |
case 0x216: /* movshdup */ |
| 3490 |
if (mod != 3) { |
| 3491 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3492 |
gen_ldo_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg])); |
| 3493 |
} else {
|
| 3494 |
rm = (modrm & 7) | REX_B(s);
|
| 3495 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(1)),
|
| 3496 |
offsetof(CPUX86State,xmm_regs[rm].XMM_L(1)));
|
| 3497 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(3)),
|
| 3498 |
offsetof(CPUX86State,xmm_regs[rm].XMM_L(3)));
|
| 3499 |
} |
| 3500 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(0)),
|
| 3501 |
offsetof(CPUX86State,xmm_regs[reg].XMM_L(1)));
|
| 3502 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[reg].XMM_L(2)),
|
| 3503 |
offsetof(CPUX86State,xmm_regs[reg].XMM_L(3)));
|
| 3504 |
break;
|
| 3505 |
case 0x178: |
| 3506 |
case 0x378: |
| 3507 |
{
|
| 3508 |
int bit_index, field_length;
|
| 3509 |
|
| 3510 |
if (b1 == 1 && reg != 0) |
| 3511 |
goto illegal_op;
|
| 3512 |
field_length = cpu_ldub_code(env, s->pc++) & 0x3F;
|
| 3513 |
bit_index = cpu_ldub_code(env, s->pc++) & 0x3F;
|
| 3514 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, |
| 3515 |
offsetof(CPUX86State,xmm_regs[reg])); |
| 3516 |
if (b1 == 1) |
| 3517 |
gen_helper_extrq_i(cpu_env, cpu_ptr0, |
| 3518 |
tcg_const_i32(bit_index), |
| 3519 |
tcg_const_i32(field_length)); |
| 3520 |
else
|
| 3521 |
gen_helper_insertq_i(cpu_env, cpu_ptr0, |
| 3522 |
tcg_const_i32(bit_index), |
| 3523 |
tcg_const_i32(field_length)); |
| 3524 |
} |
| 3525 |
break;
|
| 3526 |
case 0x7e: /* movd ea, mm */ |
| 3527 |
#ifdef TARGET_X86_64
|
| 3528 |
if (s->dflag == 2) { |
| 3529 |
tcg_gen_ld_i64(cpu_T[0], cpu_env,
|
| 3530 |
offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3531 |
gen_ldst_modrm(env, s, modrm, OT_QUAD, OR_TMP0, 1);
|
| 3532 |
} else
|
| 3533 |
#endif
|
| 3534 |
{
|
| 3535 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env,
|
| 3536 |
offsetof(CPUX86State,fpregs[reg].mmx.MMX_L(0)));
|
| 3537 |
gen_ldst_modrm(env, s, modrm, OT_LONG, OR_TMP0, 1);
|
| 3538 |
} |
| 3539 |
break;
|
| 3540 |
case 0x17e: /* movd ea, xmm */ |
| 3541 |
#ifdef TARGET_X86_64
|
| 3542 |
if (s->dflag == 2) { |
| 3543 |
tcg_gen_ld_i64(cpu_T[0], cpu_env,
|
| 3544 |
offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3545 |
gen_ldst_modrm(env, s, modrm, OT_QUAD, OR_TMP0, 1);
|
| 3546 |
} else
|
| 3547 |
#endif
|
| 3548 |
{
|
| 3549 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env,
|
| 3550 |
offsetof(CPUX86State,xmm_regs[reg].XMM_L(0)));
|
| 3551 |
gen_ldst_modrm(env, s, modrm, OT_LONG, OR_TMP0, 1);
|
| 3552 |
} |
| 3553 |
break;
|
| 3554 |
case 0x27e: /* movq xmm, ea */ |
| 3555 |
if (mod != 3) { |
| 3556 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3557 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3558 |
} else {
|
| 3559 |
rm = (modrm & 7) | REX_B(s);
|
| 3560 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)),
|
| 3561 |
offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)));
|
| 3562 |
} |
| 3563 |
gen_op_movq_env_0(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(1)));
|
| 3564 |
break;
|
| 3565 |
case 0x7f: /* movq ea, mm */ |
| 3566 |
if (mod != 3) { |
| 3567 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3568 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3569 |
} else {
|
| 3570 |
rm = (modrm & 7);
|
| 3571 |
gen_op_movq(offsetof(CPUX86State,fpregs[rm].mmx), |
| 3572 |
offsetof(CPUX86State,fpregs[reg].mmx)); |
| 3573 |
} |
| 3574 |
break;
|
| 3575 |
case 0x011: /* movups */ |
| 3576 |
case 0x111: /* movupd */ |
| 3577 |
case 0x029: /* movaps */ |
| 3578 |
case 0x129: /* movapd */ |
| 3579 |
case 0x17f: /* movdqa ea, xmm */ |
| 3580 |
case 0x27f: /* movdqu ea, xmm */ |
| 3581 |
if (mod != 3) { |
| 3582 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3583 |
gen_sto_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg])); |
| 3584 |
} else {
|
| 3585 |
rm = (modrm & 7) | REX_B(s);
|
| 3586 |
gen_op_movo(offsetof(CPUX86State,xmm_regs[rm]), |
| 3587 |
offsetof(CPUX86State,xmm_regs[reg])); |
| 3588 |
} |
| 3589 |
break;
|
| 3590 |
case 0x211: /* movss ea, xmm */ |
| 3591 |
if (mod != 3) { |
| 3592 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3593 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_regs[reg].XMM_L(0))); |
| 3594 |
gen_op_st_T0_A0(OT_LONG + s->mem_index); |
| 3595 |
} else {
|
| 3596 |
rm = (modrm & 7) | REX_B(s);
|
| 3597 |
gen_op_movl(offsetof(CPUX86State,xmm_regs[rm].XMM_L(0)),
|
| 3598 |
offsetof(CPUX86State,xmm_regs[reg].XMM_L(0)));
|
| 3599 |
} |
| 3600 |
break;
|
| 3601 |
case 0x311: /* movsd ea, xmm */ |
| 3602 |
if (mod != 3) { |
| 3603 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3604 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3605 |
} else {
|
| 3606 |
rm = (modrm & 7) | REX_B(s);
|
| 3607 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)),
|
| 3608 |
offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3609 |
} |
| 3610 |
break;
|
| 3611 |
case 0x013: /* movlps */ |
| 3612 |
case 0x113: /* movlpd */ |
| 3613 |
if (mod != 3) { |
| 3614 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3615 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3616 |
} else {
|
| 3617 |
goto illegal_op;
|
| 3618 |
} |
| 3619 |
break;
|
| 3620 |
case 0x017: /* movhps */ |
| 3621 |
case 0x117: /* movhpd */ |
| 3622 |
if (mod != 3) { |
| 3623 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3624 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(1)));
|
| 3625 |
} else {
|
| 3626 |
goto illegal_op;
|
| 3627 |
} |
| 3628 |
break;
|
| 3629 |
case 0x71: /* shift mm, im */ |
| 3630 |
case 0x72: |
| 3631 |
case 0x73: |
| 3632 |
case 0x171: /* shift xmm, im */ |
| 3633 |
case 0x172: |
| 3634 |
case 0x173: |
| 3635 |
if (b1 >= 2) { |
| 3636 |
goto illegal_op;
|
| 3637 |
} |
| 3638 |
val = cpu_ldub_code(env, s->pc++); |
| 3639 |
if (is_xmm) {
|
| 3640 |
gen_op_movl_T0_im(val); |
| 3641 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_t0.XMM_L(0))); |
| 3642 |
gen_op_movl_T0_0(); |
| 3643 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_t0.XMM_L(1))); |
| 3644 |
op1_offset = offsetof(CPUX86State,xmm_t0); |
| 3645 |
} else {
|
| 3646 |
gen_op_movl_T0_im(val); |
| 3647 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,mmx_t0.MMX_L(0))); |
| 3648 |
gen_op_movl_T0_0(); |
| 3649 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,mmx_t0.MMX_L(1))); |
| 3650 |
op1_offset = offsetof(CPUX86State,mmx_t0); |
| 3651 |
} |
| 3652 |
sse_fn_epp = sse_op_table2[((b - 1) & 3) * 8 + |
| 3653 |
(((modrm >> 3)) & 7)][b1]; |
| 3654 |
if (!sse_fn_epp) {
|
| 3655 |
goto illegal_op;
|
| 3656 |
} |
| 3657 |
if (is_xmm) {
|
| 3658 |
rm = (modrm & 7) | REX_B(s);
|
| 3659 |
op2_offset = offsetof(CPUX86State,xmm_regs[rm]); |
| 3660 |
} else {
|
| 3661 |
rm = (modrm & 7);
|
| 3662 |
op2_offset = offsetof(CPUX86State,fpregs[rm].mmx); |
| 3663 |
} |
| 3664 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op2_offset); |
| 3665 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op1_offset); |
| 3666 |
sse_fn_epp(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3667 |
break;
|
| 3668 |
case 0x050: /* movmskps */ |
| 3669 |
rm = (modrm & 7) | REX_B(s);
|
| 3670 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, |
| 3671 |
offsetof(CPUX86State,xmm_regs[rm])); |
| 3672 |
gen_helper_movmskps(cpu_tmp2_i32, cpu_env, cpu_ptr0); |
| 3673 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 3674 |
gen_op_mov_reg_T0(OT_LONG, reg); |
| 3675 |
break;
|
| 3676 |
case 0x150: /* movmskpd */ |
| 3677 |
rm = (modrm & 7) | REX_B(s);
|
| 3678 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, |
| 3679 |
offsetof(CPUX86State,xmm_regs[rm])); |
| 3680 |
gen_helper_movmskpd(cpu_tmp2_i32, cpu_env, cpu_ptr0); |
| 3681 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 3682 |
gen_op_mov_reg_T0(OT_LONG, reg); |
| 3683 |
break;
|
| 3684 |
case 0x02a: /* cvtpi2ps */ |
| 3685 |
case 0x12a: /* cvtpi2pd */ |
| 3686 |
gen_helper_enter_mmx(cpu_env); |
| 3687 |
if (mod != 3) { |
| 3688 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3689 |
op2_offset = offsetof(CPUX86State,mmx_t0); |
| 3690 |
gen_ldq_env_A0(s->mem_index, op2_offset); |
| 3691 |
} else {
|
| 3692 |
rm = (modrm & 7);
|
| 3693 |
op2_offset = offsetof(CPUX86State,fpregs[rm].mmx); |
| 3694 |
} |
| 3695 |
op1_offset = offsetof(CPUX86State,xmm_regs[reg]); |
| 3696 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 3697 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 3698 |
switch(b >> 8) { |
| 3699 |
case 0x0: |
| 3700 |
gen_helper_cvtpi2ps(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3701 |
break;
|
| 3702 |
default:
|
| 3703 |
case 0x1: |
| 3704 |
gen_helper_cvtpi2pd(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3705 |
break;
|
| 3706 |
} |
| 3707 |
break;
|
| 3708 |
case 0x22a: /* cvtsi2ss */ |
| 3709 |
case 0x32a: /* cvtsi2sd */ |
| 3710 |
ot = (s->dflag == 2) ? OT_QUAD : OT_LONG;
|
| 3711 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 3712 |
op1_offset = offsetof(CPUX86State,xmm_regs[reg]); |
| 3713 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 3714 |
if (ot == OT_LONG) {
|
| 3715 |
SSEFunc_0_epi sse_fn_epi = sse_op_table3ai[(b >> 8) & 1]; |
| 3716 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 3717 |
sse_fn_epi(cpu_env, cpu_ptr0, cpu_tmp2_i32); |
| 3718 |
} else {
|
| 3719 |
#ifdef TARGET_X86_64
|
| 3720 |
SSEFunc_0_epl sse_fn_epl = sse_op_table3aq[(b >> 8) & 1]; |
| 3721 |
sse_fn_epl(cpu_env, cpu_ptr0, cpu_T[0]);
|
| 3722 |
#else
|
| 3723 |
goto illegal_op;
|
| 3724 |
#endif
|
| 3725 |
} |
| 3726 |
break;
|
| 3727 |
case 0x02c: /* cvttps2pi */ |
| 3728 |
case 0x12c: /* cvttpd2pi */ |
| 3729 |
case 0x02d: /* cvtps2pi */ |
| 3730 |
case 0x12d: /* cvtpd2pi */ |
| 3731 |
gen_helper_enter_mmx(cpu_env); |
| 3732 |
if (mod != 3) { |
| 3733 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3734 |
op2_offset = offsetof(CPUX86State,xmm_t0); |
| 3735 |
gen_ldo_env_A0(s->mem_index, op2_offset); |
| 3736 |
} else {
|
| 3737 |
rm = (modrm & 7) | REX_B(s);
|
| 3738 |
op2_offset = offsetof(CPUX86State,xmm_regs[rm]); |
| 3739 |
} |
| 3740 |
op1_offset = offsetof(CPUX86State,fpregs[reg & 7].mmx);
|
| 3741 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 3742 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 3743 |
switch(b) {
|
| 3744 |
case 0x02c: |
| 3745 |
gen_helper_cvttps2pi(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3746 |
break;
|
| 3747 |
case 0x12c: |
| 3748 |
gen_helper_cvttpd2pi(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3749 |
break;
|
| 3750 |
case 0x02d: |
| 3751 |
gen_helper_cvtps2pi(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3752 |
break;
|
| 3753 |
case 0x12d: |
| 3754 |
gen_helper_cvtpd2pi(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3755 |
break;
|
| 3756 |
} |
| 3757 |
break;
|
| 3758 |
case 0x22c: /* cvttss2si */ |
| 3759 |
case 0x32c: /* cvttsd2si */ |
| 3760 |
case 0x22d: /* cvtss2si */ |
| 3761 |
case 0x32d: /* cvtsd2si */ |
| 3762 |
ot = (s->dflag == 2) ? OT_QUAD : OT_LONG;
|
| 3763 |
if (mod != 3) { |
| 3764 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3765 |
if ((b >> 8) & 1) { |
| 3766 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_t0.XMM_Q(0)));
|
| 3767 |
} else {
|
| 3768 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 3769 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_t0.XMM_L(0))); |
| 3770 |
} |
| 3771 |
op2_offset = offsetof(CPUX86State,xmm_t0); |
| 3772 |
} else {
|
| 3773 |
rm = (modrm & 7) | REX_B(s);
|
| 3774 |
op2_offset = offsetof(CPUX86State,xmm_regs[rm]); |
| 3775 |
} |
| 3776 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op2_offset); |
| 3777 |
if (ot == OT_LONG) {
|
| 3778 |
SSEFunc_i_ep sse_fn_i_ep = |
| 3779 |
sse_op_table3bi[((b >> 7) & 2) | (b & 1)]; |
| 3780 |
sse_fn_i_ep(cpu_tmp2_i32, cpu_env, cpu_ptr0); |
| 3781 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 3782 |
} else {
|
| 3783 |
#ifdef TARGET_X86_64
|
| 3784 |
SSEFunc_l_ep sse_fn_l_ep = |
| 3785 |
sse_op_table3bq[((b >> 7) & 2) | (b & 1)]; |
| 3786 |
sse_fn_l_ep(cpu_T[0], cpu_env, cpu_ptr0);
|
| 3787 |
#else
|
| 3788 |
goto illegal_op;
|
| 3789 |
#endif
|
| 3790 |
} |
| 3791 |
gen_op_mov_reg_T0(ot, reg); |
| 3792 |
break;
|
| 3793 |
case 0xc4: /* pinsrw */ |
| 3794 |
case 0x1c4: |
| 3795 |
s->rip_offset = 1;
|
| 3796 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 3797 |
val = cpu_ldub_code(env, s->pc++); |
| 3798 |
if (b1) {
|
| 3799 |
val &= 7;
|
| 3800 |
tcg_gen_st16_tl(cpu_T[0], cpu_env,
|
| 3801 |
offsetof(CPUX86State,xmm_regs[reg].XMM_W(val))); |
| 3802 |
} else {
|
| 3803 |
val &= 3;
|
| 3804 |
tcg_gen_st16_tl(cpu_T[0], cpu_env,
|
| 3805 |
offsetof(CPUX86State,fpregs[reg].mmx.MMX_W(val))); |
| 3806 |
} |
| 3807 |
break;
|
| 3808 |
case 0xc5: /* pextrw */ |
| 3809 |
case 0x1c5: |
| 3810 |
if (mod != 3) |
| 3811 |
goto illegal_op;
|
| 3812 |
ot = (s->dflag == 2) ? OT_QUAD : OT_LONG;
|
| 3813 |
val = cpu_ldub_code(env, s->pc++); |
| 3814 |
if (b1) {
|
| 3815 |
val &= 7;
|
| 3816 |
rm = (modrm & 7) | REX_B(s);
|
| 3817 |
tcg_gen_ld16u_tl(cpu_T[0], cpu_env,
|
| 3818 |
offsetof(CPUX86State,xmm_regs[rm].XMM_W(val))); |
| 3819 |
} else {
|
| 3820 |
val &= 3;
|
| 3821 |
rm = (modrm & 7);
|
| 3822 |
tcg_gen_ld16u_tl(cpu_T[0], cpu_env,
|
| 3823 |
offsetof(CPUX86State,fpregs[rm].mmx.MMX_W(val))); |
| 3824 |
} |
| 3825 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 3826 |
gen_op_mov_reg_T0(ot, reg); |
| 3827 |
break;
|
| 3828 |
case 0x1d6: /* movq ea, xmm */ |
| 3829 |
if (mod != 3) { |
| 3830 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3831 |
gen_stq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3832 |
} else {
|
| 3833 |
rm = (modrm & 7) | REX_B(s);
|
| 3834 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)),
|
| 3835 |
offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)));
|
| 3836 |
gen_op_movq_env_0(offsetof(CPUX86State,xmm_regs[rm].XMM_Q(1)));
|
| 3837 |
} |
| 3838 |
break;
|
| 3839 |
case 0x2d6: /* movq2dq */ |
| 3840 |
gen_helper_enter_mmx(cpu_env); |
| 3841 |
rm = (modrm & 7);
|
| 3842 |
gen_op_movq(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(0)),
|
| 3843 |
offsetof(CPUX86State,fpregs[rm].mmx)); |
| 3844 |
gen_op_movq_env_0(offsetof(CPUX86State,xmm_regs[reg].XMM_Q(1)));
|
| 3845 |
break;
|
| 3846 |
case 0x3d6: /* movdq2q */ |
| 3847 |
gen_helper_enter_mmx(cpu_env); |
| 3848 |
rm = (modrm & 7) | REX_B(s);
|
| 3849 |
gen_op_movq(offsetof(CPUX86State,fpregs[reg & 7].mmx),
|
| 3850 |
offsetof(CPUX86State,xmm_regs[rm].XMM_Q(0)));
|
| 3851 |
break;
|
| 3852 |
case 0xd7: /* pmovmskb */ |
| 3853 |
case 0x1d7: |
| 3854 |
if (mod != 3) |
| 3855 |
goto illegal_op;
|
| 3856 |
if (b1) {
|
| 3857 |
rm = (modrm & 7) | REX_B(s);
|
| 3858 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, offsetof(CPUX86State,xmm_regs[rm])); |
| 3859 |
gen_helper_pmovmskb_xmm(cpu_tmp2_i32, cpu_env, cpu_ptr0); |
| 3860 |
} else {
|
| 3861 |
rm = (modrm & 7);
|
| 3862 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, offsetof(CPUX86State,fpregs[rm].mmx)); |
| 3863 |
gen_helper_pmovmskb_mmx(cpu_tmp2_i32, cpu_env, cpu_ptr0); |
| 3864 |
} |
| 3865 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 3866 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 3867 |
gen_op_mov_reg_T0(OT_LONG, reg); |
| 3868 |
break;
|
| 3869 |
|
| 3870 |
case 0x138: |
| 3871 |
case 0x038: |
| 3872 |
b = modrm; |
| 3873 |
if ((b & 0xf0) == 0xf0) { |
| 3874 |
goto do_0f_38_fx;
|
| 3875 |
} |
| 3876 |
modrm = cpu_ldub_code(env, s->pc++); |
| 3877 |
rm = modrm & 7;
|
| 3878 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 3879 |
mod = (modrm >> 6) & 3; |
| 3880 |
if (b1 >= 2) { |
| 3881 |
goto illegal_op;
|
| 3882 |
} |
| 3883 |
|
| 3884 |
sse_fn_epp = sse_op_table6[b].op[b1]; |
| 3885 |
if (!sse_fn_epp) {
|
| 3886 |
goto illegal_op;
|
| 3887 |
} |
| 3888 |
if (!(s->cpuid_ext_features & sse_op_table6[b].ext_mask))
|
| 3889 |
goto illegal_op;
|
| 3890 |
|
| 3891 |
if (b1) {
|
| 3892 |
op1_offset = offsetof(CPUX86State,xmm_regs[reg]); |
| 3893 |
if (mod == 3) { |
| 3894 |
op2_offset = offsetof(CPUX86State,xmm_regs[rm | REX_B(s)]); |
| 3895 |
} else {
|
| 3896 |
op2_offset = offsetof(CPUX86State,xmm_t0); |
| 3897 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3898 |
switch (b) {
|
| 3899 |
case 0x20: case 0x30: /* pmovsxbw, pmovzxbw */ |
| 3900 |
case 0x23: case 0x33: /* pmovsxwd, pmovzxwd */ |
| 3901 |
case 0x25: case 0x35: /* pmovsxdq, pmovzxdq */ |
| 3902 |
gen_ldq_env_A0(s->mem_index, op2_offset + |
| 3903 |
offsetof(XMMReg, XMM_Q(0)));
|
| 3904 |
break;
|
| 3905 |
case 0x21: case 0x31: /* pmovsxbd, pmovzxbd */ |
| 3906 |
case 0x24: case 0x34: /* pmovsxwq, pmovzxwq */ |
| 3907 |
tcg_gen_qemu_ld32u(cpu_tmp0, cpu_A0, |
| 3908 |
(s->mem_index >> 2) - 1); |
| 3909 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_tmp0); |
| 3910 |
tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, op2_offset + |
| 3911 |
offsetof(XMMReg, XMM_L(0)));
|
| 3912 |
break;
|
| 3913 |
case 0x22: case 0x32: /* pmovsxbq, pmovzxbq */ |
| 3914 |
tcg_gen_qemu_ld16u(cpu_tmp0, cpu_A0, |
| 3915 |
(s->mem_index >> 2) - 1); |
| 3916 |
tcg_gen_st16_tl(cpu_tmp0, cpu_env, op2_offset + |
| 3917 |
offsetof(XMMReg, XMM_W(0)));
|
| 3918 |
break;
|
| 3919 |
case 0x2a: /* movntqda */ |
| 3920 |
gen_ldo_env_A0(s->mem_index, op1_offset); |
| 3921 |
return;
|
| 3922 |
default:
|
| 3923 |
gen_ldo_env_A0(s->mem_index, op2_offset); |
| 3924 |
} |
| 3925 |
} |
| 3926 |
} else {
|
| 3927 |
op1_offset = offsetof(CPUX86State,fpregs[reg].mmx); |
| 3928 |
if (mod == 3) { |
| 3929 |
op2_offset = offsetof(CPUX86State,fpregs[rm].mmx); |
| 3930 |
} else {
|
| 3931 |
op2_offset = offsetof(CPUX86State,mmx_t0); |
| 3932 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 3933 |
gen_ldq_env_A0(s->mem_index, op2_offset); |
| 3934 |
} |
| 3935 |
} |
| 3936 |
if (sse_fn_epp == SSE_SPECIAL) {
|
| 3937 |
goto illegal_op;
|
| 3938 |
} |
| 3939 |
|
| 3940 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 3941 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 3942 |
sse_fn_epp(cpu_env, cpu_ptr0, cpu_ptr1); |
| 3943 |
|
| 3944 |
if (b == 0x17) { |
| 3945 |
set_cc_op(s, CC_OP_EFLAGS); |
| 3946 |
} |
| 3947 |
break;
|
| 3948 |
|
| 3949 |
case 0x238: |
| 3950 |
case 0x338: |
| 3951 |
do_0f_38_fx:
|
| 3952 |
/* Various integer extensions at 0f 38 f[0-f]. */
|
| 3953 |
b = modrm | (b1 << 8);
|
| 3954 |
modrm = cpu_ldub_code(env, s->pc++); |
| 3955 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 3956 |
|
| 3957 |
switch (b) {
|
| 3958 |
case 0x3f0: /* crc32 Gd,Eb */ |
| 3959 |
case 0x3f1: /* crc32 Gd,Ey */ |
| 3960 |
do_crc32:
|
| 3961 |
if (!(s->cpuid_ext_features & CPUID_EXT_SSE42)) {
|
| 3962 |
goto illegal_op;
|
| 3963 |
} |
| 3964 |
if ((b & 0xff) == 0xf0) { |
| 3965 |
ot = OT_BYTE; |
| 3966 |
} else if (s->dflag != 2) { |
| 3967 |
ot = (s->prefix & PREFIX_DATA ? OT_WORD : OT_LONG); |
| 3968 |
} else {
|
| 3969 |
ot = OT_QUAD; |
| 3970 |
} |
| 3971 |
|
| 3972 |
gen_op_mov_TN_reg(OT_LONG, 0, reg);
|
| 3973 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 3974 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 3975 |
gen_helper_crc32(cpu_T[0], cpu_tmp2_i32,
|
| 3976 |
cpu_T[0], tcg_const_i32(8 << ot)); |
| 3977 |
|
| 3978 |
ot = (s->dflag == 2) ? OT_QUAD : OT_LONG;
|
| 3979 |
gen_op_mov_reg_T0(ot, reg); |
| 3980 |
break;
|
| 3981 |
|
| 3982 |
case 0x1f0: /* crc32 or movbe */ |
| 3983 |
case 0x1f1: |
| 3984 |
/* For these insns, the f3 prefix is supposed to have priority
|
| 3985 |
over the 66 prefix, but that's not what we implement above
|
| 3986 |
setting b1. */
|
| 3987 |
if (s->prefix & PREFIX_REPNZ) {
|
| 3988 |
goto do_crc32;
|
| 3989 |
} |
| 3990 |
/* FALLTHRU */
|
| 3991 |
case 0x0f0: /* movbe Gy,My */ |
| 3992 |
case 0x0f1: /* movbe My,Gy */ |
| 3993 |
if (!(s->cpuid_ext_features & CPUID_EXT_MOVBE)) {
|
| 3994 |
goto illegal_op;
|
| 3995 |
} |
| 3996 |
if (s->dflag != 2) { |
| 3997 |
ot = (s->prefix & PREFIX_DATA ? OT_WORD : OT_LONG); |
| 3998 |
} else {
|
| 3999 |
ot = OT_QUAD; |
| 4000 |
} |
| 4001 |
|
| 4002 |
/* Load the data incoming to the bswap. Note that the TCG
|
| 4003 |
implementation of bswap requires the input be zero
|
| 4004 |
extended. In the case of the loads, we simply know that
|
| 4005 |
gen_op_ld_v via gen_ldst_modrm does that already. */
|
| 4006 |
if ((b & 1) == 0) { |
| 4007 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4008 |
} else {
|
| 4009 |
switch (ot) {
|
| 4010 |
case OT_WORD:
|
| 4011 |
tcg_gen_ext16u_tl(cpu_T[0], cpu_regs[reg]);
|
| 4012 |
break;
|
| 4013 |
default:
|
| 4014 |
tcg_gen_ext32u_tl(cpu_T[0], cpu_regs[reg]);
|
| 4015 |
break;
|
| 4016 |
case OT_QUAD:
|
| 4017 |
tcg_gen_mov_tl(cpu_T[0], cpu_regs[reg]);
|
| 4018 |
break;
|
| 4019 |
} |
| 4020 |
} |
| 4021 |
|
| 4022 |
switch (ot) {
|
| 4023 |
case OT_WORD:
|
| 4024 |
tcg_gen_bswap16_tl(cpu_T[0], cpu_T[0]); |
| 4025 |
break;
|
| 4026 |
default:
|
| 4027 |
tcg_gen_bswap32_tl(cpu_T[0], cpu_T[0]); |
| 4028 |
break;
|
| 4029 |
#ifdef TARGET_X86_64
|
| 4030 |
case OT_QUAD:
|
| 4031 |
tcg_gen_bswap64_tl(cpu_T[0], cpu_T[0]); |
| 4032 |
break;
|
| 4033 |
#endif
|
| 4034 |
} |
| 4035 |
|
| 4036 |
if ((b & 1) == 0) { |
| 4037 |
gen_op_mov_reg_T0(ot, reg); |
| 4038 |
} else {
|
| 4039 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 1);
|
| 4040 |
} |
| 4041 |
break;
|
| 4042 |
|
| 4043 |
case 0x0f2: /* andn Gy, By, Ey */ |
| 4044 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI1)
|
| 4045 |
|| !(s->prefix & PREFIX_VEX) |
| 4046 |
|| s->vex_l != 0) {
|
| 4047 |
goto illegal_op;
|
| 4048 |
} |
| 4049 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4050 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4051 |
tcg_gen_andc_tl(cpu_T[0], cpu_regs[s->vex_v], cpu_T[0]); |
| 4052 |
gen_op_mov_reg_T0(ot, reg); |
| 4053 |
gen_op_update1_cc(); |
| 4054 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 4055 |
break;
|
| 4056 |
|
| 4057 |
case 0x0f7: /* bextr Gy, Ey, By */ |
| 4058 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI1)
|
| 4059 |
|| !(s->prefix & PREFIX_VEX) |
| 4060 |
|| s->vex_l != 0) {
|
| 4061 |
goto illegal_op;
|
| 4062 |
} |
| 4063 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4064 |
{
|
| 4065 |
TCGv bound, zero; |
| 4066 |
|
| 4067 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4068 |
/* Extract START, and shift the operand.
|
| 4069 |
Shifts larger than operand size get zeros. */
|
| 4070 |
tcg_gen_ext8u_tl(cpu_A0, cpu_regs[s->vex_v]); |
| 4071 |
tcg_gen_shr_tl(cpu_T[0], cpu_T[0], cpu_A0); |
| 4072 |
|
| 4073 |
bound = tcg_const_tl(ot == OT_QUAD ? 63 : 31); |
| 4074 |
zero = tcg_const_tl(0);
|
| 4075 |
tcg_gen_movcond_tl(TCG_COND_LEU, cpu_T[0], cpu_A0, bound,
|
| 4076 |
cpu_T[0], zero);
|
| 4077 |
tcg_temp_free(zero); |
| 4078 |
|
| 4079 |
/* Extract the LEN into a mask. Lengths larger than
|
| 4080 |
operand size get all ones. */
|
| 4081 |
tcg_gen_shri_tl(cpu_A0, cpu_regs[s->vex_v], 8);
|
| 4082 |
tcg_gen_ext8u_tl(cpu_A0, cpu_A0); |
| 4083 |
tcg_gen_movcond_tl(TCG_COND_LEU, cpu_A0, cpu_A0, bound, |
| 4084 |
cpu_A0, bound); |
| 4085 |
tcg_temp_free(bound); |
| 4086 |
tcg_gen_movi_tl(cpu_T[1], 1); |
| 4087 |
tcg_gen_shl_tl(cpu_T[1], cpu_T[1], cpu_A0); |
| 4088 |
tcg_gen_subi_tl(cpu_T[1], cpu_T[1], 1); |
| 4089 |
tcg_gen_and_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 4090 |
|
| 4091 |
gen_op_mov_reg_T0(ot, reg); |
| 4092 |
gen_op_update1_cc(); |
| 4093 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 4094 |
} |
| 4095 |
break;
|
| 4096 |
|
| 4097 |
case 0x0f5: /* bzhi Gy, Ey, By */ |
| 4098 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI2)
|
| 4099 |
|| !(s->prefix & PREFIX_VEX) |
| 4100 |
|| s->vex_l != 0) {
|
| 4101 |
goto illegal_op;
|
| 4102 |
} |
| 4103 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4104 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4105 |
tcg_gen_ext8u_tl(cpu_T[1], cpu_regs[s->vex_v]);
|
| 4106 |
{
|
| 4107 |
TCGv bound = tcg_const_tl(ot == OT_QUAD ? 63 : 31); |
| 4108 |
/* Note that since we're using BMILG (in order to get O
|
| 4109 |
cleared) we need to store the inverse into C. */
|
| 4110 |
tcg_gen_setcond_tl(TCG_COND_LT, cpu_cc_src, |
| 4111 |
cpu_T[1], bound);
|
| 4112 |
tcg_gen_movcond_tl(TCG_COND_GT, cpu_T[1], cpu_T[1], |
| 4113 |
bound, bound, cpu_T[1]);
|
| 4114 |
tcg_temp_free(bound); |
| 4115 |
} |
| 4116 |
tcg_gen_movi_tl(cpu_A0, -1);
|
| 4117 |
tcg_gen_shl_tl(cpu_A0, cpu_A0, cpu_T[1]);
|
| 4118 |
tcg_gen_andc_tl(cpu_T[0], cpu_T[0], cpu_A0); |
| 4119 |
gen_op_mov_reg_T0(ot, reg); |
| 4120 |
gen_op_update1_cc(); |
| 4121 |
set_cc_op(s, CC_OP_BMILGB + ot); |
| 4122 |
break;
|
| 4123 |
|
| 4124 |
case 0x3f6: /* mulx By, Gy, rdx, Ey */ |
| 4125 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI2)
|
| 4126 |
|| !(s->prefix & PREFIX_VEX) |
| 4127 |
|| s->vex_l != 0) {
|
| 4128 |
goto illegal_op;
|
| 4129 |
} |
| 4130 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4131 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4132 |
switch (ot) {
|
| 4133 |
TCGv_i64 t0, t1; |
| 4134 |
default:
|
| 4135 |
t0 = tcg_temp_new_i64(); |
| 4136 |
t1 = tcg_temp_new_i64(); |
| 4137 |
#ifdef TARGET_X86_64
|
| 4138 |
tcg_gen_ext32u_i64(t0, cpu_T[0]);
|
| 4139 |
tcg_gen_ext32u_i64(t1, cpu_regs[R_EDX]); |
| 4140 |
#else
|
| 4141 |
tcg_gen_extu_i32_i64(t0, cpu_T[0]);
|
| 4142 |
tcg_gen_extu_i32_i64(t0, cpu_regs[R_EDX]); |
| 4143 |
#endif
|
| 4144 |
tcg_gen_mul_i64(t0, t0, t1); |
| 4145 |
tcg_gen_trunc_i64_tl(cpu_T[0], t0);
|
| 4146 |
tcg_gen_shri_i64(t0, t0, 32);
|
| 4147 |
tcg_gen_trunc_i64_tl(cpu_T[1], t0);
|
| 4148 |
tcg_temp_free_i64(t0); |
| 4149 |
tcg_temp_free_i64(t1); |
| 4150 |
gen_op_mov_reg_T0(OT_LONG, s->vex_v); |
| 4151 |
gen_op_mov_reg_T1(OT_LONG, reg); |
| 4152 |
break;
|
| 4153 |
#ifdef TARGET_X86_64
|
| 4154 |
case OT_QUAD:
|
| 4155 |
tcg_gen_mov_tl(cpu_T[1], cpu_regs[R_EDX]);
|
| 4156 |
tcg_gen_mul_tl(cpu_regs[s->vex_v], cpu_T[0], cpu_T[1]); |
| 4157 |
gen_helper_umulh(cpu_regs[reg], cpu_T[0], cpu_T[1]); |
| 4158 |
break;
|
| 4159 |
#endif
|
| 4160 |
} |
| 4161 |
break;
|
| 4162 |
|
| 4163 |
case 0x3f5: /* pdep Gy, By, Ey */ |
| 4164 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI2)
|
| 4165 |
|| !(s->prefix & PREFIX_VEX) |
| 4166 |
|| s->vex_l != 0) {
|
| 4167 |
goto illegal_op;
|
| 4168 |
} |
| 4169 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4170 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4171 |
/* Note that by zero-extending the mask operand, we
|
| 4172 |
automatically handle zero-extending the result. */
|
| 4173 |
if (s->dflag == 2) { |
| 4174 |
tcg_gen_mov_tl(cpu_T[1], cpu_regs[s->vex_v]);
|
| 4175 |
} else {
|
| 4176 |
tcg_gen_ext32u_tl(cpu_T[1], cpu_regs[s->vex_v]);
|
| 4177 |
} |
| 4178 |
gen_helper_pdep(cpu_regs[reg], cpu_T[0], cpu_T[1]); |
| 4179 |
break;
|
| 4180 |
|
| 4181 |
case 0x2f5: /* pext Gy, By, Ey */ |
| 4182 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI2)
|
| 4183 |
|| !(s->prefix & PREFIX_VEX) |
| 4184 |
|| s->vex_l != 0) {
|
| 4185 |
goto illegal_op;
|
| 4186 |
} |
| 4187 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4188 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4189 |
/* Note that by zero-extending the mask operand, we
|
| 4190 |
automatically handle zero-extending the result. */
|
| 4191 |
if (s->dflag == 2) { |
| 4192 |
tcg_gen_mov_tl(cpu_T[1], cpu_regs[s->vex_v]);
|
| 4193 |
} else {
|
| 4194 |
tcg_gen_ext32u_tl(cpu_T[1], cpu_regs[s->vex_v]);
|
| 4195 |
} |
| 4196 |
gen_helper_pext(cpu_regs[reg], cpu_T[0], cpu_T[1]); |
| 4197 |
break;
|
| 4198 |
|
| 4199 |
case 0x1f6: /* adcx Gy, Ey */ |
| 4200 |
case 0x2f6: /* adox Gy, Ey */ |
| 4201 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_ADX)) {
|
| 4202 |
goto illegal_op;
|
| 4203 |
} else {
|
| 4204 |
TCGv carry_in, carry_out; |
| 4205 |
int end_op;
|
| 4206 |
|
| 4207 |
ot = (s->dflag == 2 ? OT_QUAD : OT_LONG);
|
| 4208 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4209 |
|
| 4210 |
/* Re-use the carry-out from a previous round. */
|
| 4211 |
TCGV_UNUSED(carry_in); |
| 4212 |
carry_out = (b == 0x1f6 ? cpu_cc_dst : cpu_cc_src2);
|
| 4213 |
switch (s->cc_op) {
|
| 4214 |
case CC_OP_ADCX:
|
| 4215 |
if (b == 0x1f6) { |
| 4216 |
carry_in = cpu_cc_dst; |
| 4217 |
end_op = CC_OP_ADCX; |
| 4218 |
} else {
|
| 4219 |
end_op = CC_OP_ADCOX; |
| 4220 |
} |
| 4221 |
break;
|
| 4222 |
case CC_OP_ADOX:
|
| 4223 |
if (b == 0x1f6) { |
| 4224 |
end_op = CC_OP_ADCOX; |
| 4225 |
} else {
|
| 4226 |
carry_in = cpu_cc_src2; |
| 4227 |
end_op = CC_OP_ADOX; |
| 4228 |
} |
| 4229 |
break;
|
| 4230 |
case CC_OP_ADCOX:
|
| 4231 |
end_op = CC_OP_ADCOX; |
| 4232 |
carry_in = carry_out; |
| 4233 |
break;
|
| 4234 |
default:
|
| 4235 |
end_op = (b == 0x1f6 ? CC_OP_ADCX : CC_OP_ADCOX);
|
| 4236 |
break;
|
| 4237 |
} |
| 4238 |
/* If we can't reuse carry-out, get it out of EFLAGS. */
|
| 4239 |
if (TCGV_IS_UNUSED(carry_in)) {
|
| 4240 |
if (s->cc_op != CC_OP_ADCX && s->cc_op != CC_OP_ADOX) {
|
| 4241 |
gen_compute_eflags(s); |
| 4242 |
} |
| 4243 |
carry_in = cpu_tmp0; |
| 4244 |
tcg_gen_shri_tl(carry_in, cpu_cc_src, |
| 4245 |
ctz32(b == 0x1f6 ? CC_C : CC_O));
|
| 4246 |
tcg_gen_andi_tl(carry_in, carry_in, 1);
|
| 4247 |
} |
| 4248 |
|
| 4249 |
switch (ot) {
|
| 4250 |
#ifdef TARGET_X86_64
|
| 4251 |
case OT_LONG:
|
| 4252 |
/* If we know TL is 64-bit, and we want a 32-bit
|
| 4253 |
result, just do everything in 64-bit arithmetic. */
|
| 4254 |
tcg_gen_ext32u_i64(cpu_regs[reg], cpu_regs[reg]); |
| 4255 |
tcg_gen_ext32u_i64(cpu_T[0], cpu_T[0]); |
| 4256 |
tcg_gen_add_i64(cpu_T[0], cpu_T[0], cpu_regs[reg]); |
| 4257 |
tcg_gen_add_i64(cpu_T[0], cpu_T[0], carry_in); |
| 4258 |
tcg_gen_ext32u_i64(cpu_regs[reg], cpu_T[0]);
|
| 4259 |
tcg_gen_shri_i64(carry_out, cpu_T[0], 32); |
| 4260 |
break;
|
| 4261 |
#endif
|
| 4262 |
default:
|
| 4263 |
/* Otherwise compute the carry-out in two steps. */
|
| 4264 |
tcg_gen_add_tl(cpu_T[0], cpu_T[0], cpu_regs[reg]); |
| 4265 |
tcg_gen_setcond_tl(TCG_COND_LTU, cpu_tmp4, |
| 4266 |
cpu_T[0], cpu_regs[reg]);
|
| 4267 |
tcg_gen_add_tl(cpu_regs[reg], cpu_T[0], carry_in);
|
| 4268 |
tcg_gen_setcond_tl(TCG_COND_LTU, carry_out, |
| 4269 |
cpu_regs[reg], cpu_T[0]);
|
| 4270 |
tcg_gen_or_tl(carry_out, carry_out, cpu_tmp4); |
| 4271 |
break;
|
| 4272 |
} |
| 4273 |
/* We began with all flags computed to CC_SRC, and we
|
| 4274 |
have now placed the carry-out in CC_DST. All that
|
| 4275 |
is left is to record the CC_OP. */
|
| 4276 |
set_cc_op(s, end_op); |
| 4277 |
} |
| 4278 |
break;
|
| 4279 |
|
| 4280 |
case 0x1f7: /* shlx Gy, Ey, By */ |
| 4281 |
case 0x2f7: /* sarx Gy, Ey, By */ |
| 4282 |
case 0x3f7: /* shrx Gy, Ey, By */ |
| 4283 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI2)
|
| 4284 |
|| !(s->prefix & PREFIX_VEX) |
| 4285 |
|| s->vex_l != 0) {
|
| 4286 |
goto illegal_op;
|
| 4287 |
} |
| 4288 |
ot = (s->dflag == 2 ? OT_QUAD : OT_LONG);
|
| 4289 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4290 |
if (ot == OT_QUAD) {
|
| 4291 |
tcg_gen_andi_tl(cpu_T[1], cpu_regs[s->vex_v], 63); |
| 4292 |
} else {
|
| 4293 |
tcg_gen_andi_tl(cpu_T[1], cpu_regs[s->vex_v], 31); |
| 4294 |
} |
| 4295 |
if (b == 0x1f7) { |
| 4296 |
tcg_gen_shl_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 4297 |
} else if (b == 0x2f7) { |
| 4298 |
if (ot != OT_QUAD) {
|
| 4299 |
tcg_gen_ext32s_tl(cpu_T[0], cpu_T[0]); |
| 4300 |
} |
| 4301 |
tcg_gen_sar_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 4302 |
} else {
|
| 4303 |
if (ot != OT_QUAD) {
|
| 4304 |
tcg_gen_ext32u_tl(cpu_T[0], cpu_T[0]); |
| 4305 |
} |
| 4306 |
tcg_gen_shr_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 4307 |
} |
| 4308 |
gen_op_mov_reg_T0(ot, reg); |
| 4309 |
break;
|
| 4310 |
|
| 4311 |
case 0x0f3: |
| 4312 |
case 0x1f3: |
| 4313 |
case 0x2f3: |
| 4314 |
case 0x3f3: /* Group 17 */ |
| 4315 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI1)
|
| 4316 |
|| !(s->prefix & PREFIX_VEX) |
| 4317 |
|| s->vex_l != 0) {
|
| 4318 |
goto illegal_op;
|
| 4319 |
} |
| 4320 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4321 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4322 |
|
| 4323 |
switch (reg & 7) { |
| 4324 |
case 1: /* blsr By,Ey */ |
| 4325 |
tcg_gen_neg_tl(cpu_T[1], cpu_T[0]); |
| 4326 |
tcg_gen_and_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 4327 |
gen_op_mov_reg_T0(ot, s->vex_v); |
| 4328 |
gen_op_update2_cc(); |
| 4329 |
set_cc_op(s, CC_OP_BMILGB + ot); |
| 4330 |
break;
|
| 4331 |
|
| 4332 |
case 2: /* blsmsk By,Ey */ |
| 4333 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[0]);
|
| 4334 |
tcg_gen_subi_tl(cpu_T[0], cpu_T[0], 1); |
| 4335 |
tcg_gen_xor_tl(cpu_T[0], cpu_T[0], cpu_cc_src); |
| 4336 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 4337 |
set_cc_op(s, CC_OP_BMILGB + ot); |
| 4338 |
break;
|
| 4339 |
|
| 4340 |
case 3: /* blsi By, Ey */ |
| 4341 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[0]);
|
| 4342 |
tcg_gen_subi_tl(cpu_T[0], cpu_T[0], 1); |
| 4343 |
tcg_gen_and_tl(cpu_T[0], cpu_T[0], cpu_cc_src); |
| 4344 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 4345 |
set_cc_op(s, CC_OP_BMILGB + ot); |
| 4346 |
break;
|
| 4347 |
|
| 4348 |
default:
|
| 4349 |
goto illegal_op;
|
| 4350 |
} |
| 4351 |
break;
|
| 4352 |
|
| 4353 |
default:
|
| 4354 |
goto illegal_op;
|
| 4355 |
} |
| 4356 |
break;
|
| 4357 |
|
| 4358 |
case 0x03a: |
| 4359 |
case 0x13a: |
| 4360 |
b = modrm; |
| 4361 |
modrm = cpu_ldub_code(env, s->pc++); |
| 4362 |
rm = modrm & 7;
|
| 4363 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 4364 |
mod = (modrm >> 6) & 3; |
| 4365 |
if (b1 >= 2) { |
| 4366 |
goto illegal_op;
|
| 4367 |
} |
| 4368 |
|
| 4369 |
sse_fn_eppi = sse_op_table7[b].op[b1]; |
| 4370 |
if (!sse_fn_eppi) {
|
| 4371 |
goto illegal_op;
|
| 4372 |
} |
| 4373 |
if (!(s->cpuid_ext_features & sse_op_table7[b].ext_mask))
|
| 4374 |
goto illegal_op;
|
| 4375 |
|
| 4376 |
if (sse_fn_eppi == SSE_SPECIAL) {
|
| 4377 |
ot = (s->dflag == 2) ? OT_QUAD : OT_LONG;
|
| 4378 |
rm = (modrm & 7) | REX_B(s);
|
| 4379 |
if (mod != 3) |
| 4380 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4381 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 4382 |
val = cpu_ldub_code(env, s->pc++); |
| 4383 |
switch (b) {
|
| 4384 |
case 0x14: /* pextrb */ |
| 4385 |
tcg_gen_ld8u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,
|
| 4386 |
xmm_regs[reg].XMM_B(val & 15)));
|
| 4387 |
if (mod == 3) |
| 4388 |
gen_op_mov_reg_T0(ot, rm); |
| 4389 |
else
|
| 4390 |
tcg_gen_qemu_st8(cpu_T[0], cpu_A0,
|
| 4391 |
(s->mem_index >> 2) - 1); |
| 4392 |
break;
|
| 4393 |
case 0x15: /* pextrw */ |
| 4394 |
tcg_gen_ld16u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,
|
| 4395 |
xmm_regs[reg].XMM_W(val & 7)));
|
| 4396 |
if (mod == 3) |
| 4397 |
gen_op_mov_reg_T0(ot, rm); |
| 4398 |
else
|
| 4399 |
tcg_gen_qemu_st16(cpu_T[0], cpu_A0,
|
| 4400 |
(s->mem_index >> 2) - 1); |
| 4401 |
break;
|
| 4402 |
case 0x16: |
| 4403 |
if (ot == OT_LONG) { /* pextrd */ |
| 4404 |
tcg_gen_ld_i32(cpu_tmp2_i32, cpu_env, |
| 4405 |
offsetof(CPUX86State, |
| 4406 |
xmm_regs[reg].XMM_L(val & 3)));
|
| 4407 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 4408 |
if (mod == 3) |
| 4409 |
gen_op_mov_reg_v(ot, rm, cpu_T[0]);
|
| 4410 |
else
|
| 4411 |
tcg_gen_qemu_st32(cpu_T[0], cpu_A0,
|
| 4412 |
(s->mem_index >> 2) - 1); |
| 4413 |
} else { /* pextrq */ |
| 4414 |
#ifdef TARGET_X86_64
|
| 4415 |
tcg_gen_ld_i64(cpu_tmp1_i64, cpu_env, |
| 4416 |
offsetof(CPUX86State, |
| 4417 |
xmm_regs[reg].XMM_Q(val & 1)));
|
| 4418 |
if (mod == 3) |
| 4419 |
gen_op_mov_reg_v(ot, rm, cpu_tmp1_i64); |
| 4420 |
else
|
| 4421 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_A0, |
| 4422 |
(s->mem_index >> 2) - 1); |
| 4423 |
#else
|
| 4424 |
goto illegal_op;
|
| 4425 |
#endif
|
| 4426 |
} |
| 4427 |
break;
|
| 4428 |
case 0x17: /* extractps */ |
| 4429 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,
|
| 4430 |
xmm_regs[reg].XMM_L(val & 3)));
|
| 4431 |
if (mod == 3) |
| 4432 |
gen_op_mov_reg_T0(ot, rm); |
| 4433 |
else
|
| 4434 |
tcg_gen_qemu_st32(cpu_T[0], cpu_A0,
|
| 4435 |
(s->mem_index >> 2) - 1); |
| 4436 |
break;
|
| 4437 |
case 0x20: /* pinsrb */ |
| 4438 |
if (mod == 3) |
| 4439 |
gen_op_mov_TN_reg(OT_LONG, 0, rm);
|
| 4440 |
else
|
| 4441 |
tcg_gen_qemu_ld8u(cpu_tmp0, cpu_A0, |
| 4442 |
(s->mem_index >> 2) - 1); |
| 4443 |
tcg_gen_st8_tl(cpu_tmp0, cpu_env, offsetof(CPUX86State, |
| 4444 |
xmm_regs[reg].XMM_B(val & 15)));
|
| 4445 |
break;
|
| 4446 |
case 0x21: /* insertps */ |
| 4447 |
if (mod == 3) { |
| 4448 |
tcg_gen_ld_i32(cpu_tmp2_i32, cpu_env, |
| 4449 |
offsetof(CPUX86State,xmm_regs[rm] |
| 4450 |
.XMM_L((val >> 6) & 3))); |
| 4451 |
} else {
|
| 4452 |
tcg_gen_qemu_ld32u(cpu_tmp0, cpu_A0, |
| 4453 |
(s->mem_index >> 2) - 1); |
| 4454 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_tmp0); |
| 4455 |
} |
| 4456 |
tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, |
| 4457 |
offsetof(CPUX86State,xmm_regs[reg] |
| 4458 |
.XMM_L((val >> 4) & 3))); |
| 4459 |
if ((val >> 0) & 1) |
| 4460 |
tcg_gen_st_i32(tcg_const_i32(0 /*float32_zero*/), |
| 4461 |
cpu_env, offsetof(CPUX86State, |
| 4462 |
xmm_regs[reg].XMM_L(0)));
|
| 4463 |
if ((val >> 1) & 1) |
| 4464 |
tcg_gen_st_i32(tcg_const_i32(0 /*float32_zero*/), |
| 4465 |
cpu_env, offsetof(CPUX86State, |
| 4466 |
xmm_regs[reg].XMM_L(1)));
|
| 4467 |
if ((val >> 2) & 1) |
| 4468 |
tcg_gen_st_i32(tcg_const_i32(0 /*float32_zero*/), |
| 4469 |
cpu_env, offsetof(CPUX86State, |
| 4470 |
xmm_regs[reg].XMM_L(2)));
|
| 4471 |
if ((val >> 3) & 1) |
| 4472 |
tcg_gen_st_i32(tcg_const_i32(0 /*float32_zero*/), |
| 4473 |
cpu_env, offsetof(CPUX86State, |
| 4474 |
xmm_regs[reg].XMM_L(3)));
|
| 4475 |
break;
|
| 4476 |
case 0x22: |
| 4477 |
if (ot == OT_LONG) { /* pinsrd */ |
| 4478 |
if (mod == 3) |
| 4479 |
gen_op_mov_v_reg(ot, cpu_tmp0, rm); |
| 4480 |
else
|
| 4481 |
tcg_gen_qemu_ld32u(cpu_tmp0, cpu_A0, |
| 4482 |
(s->mem_index >> 2) - 1); |
| 4483 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_tmp0); |
| 4484 |
tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, |
| 4485 |
offsetof(CPUX86State, |
| 4486 |
xmm_regs[reg].XMM_L(val & 3)));
|
| 4487 |
} else { /* pinsrq */ |
| 4488 |
#ifdef TARGET_X86_64
|
| 4489 |
if (mod == 3) |
| 4490 |
gen_op_mov_v_reg(ot, cpu_tmp1_i64, rm); |
| 4491 |
else
|
| 4492 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_A0, |
| 4493 |
(s->mem_index >> 2) - 1); |
| 4494 |
tcg_gen_st_i64(cpu_tmp1_i64, cpu_env, |
| 4495 |
offsetof(CPUX86State, |
| 4496 |
xmm_regs[reg].XMM_Q(val & 1)));
|
| 4497 |
#else
|
| 4498 |
goto illegal_op;
|
| 4499 |
#endif
|
| 4500 |
} |
| 4501 |
break;
|
| 4502 |
} |
| 4503 |
return;
|
| 4504 |
} |
| 4505 |
|
| 4506 |
if (b1) {
|
| 4507 |
op1_offset = offsetof(CPUX86State,xmm_regs[reg]); |
| 4508 |
if (mod == 3) { |
| 4509 |
op2_offset = offsetof(CPUX86State,xmm_regs[rm | REX_B(s)]); |
| 4510 |
} else {
|
| 4511 |
op2_offset = offsetof(CPUX86State,xmm_t0); |
| 4512 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4513 |
gen_ldo_env_A0(s->mem_index, op2_offset); |
| 4514 |
} |
| 4515 |
} else {
|
| 4516 |
op1_offset = offsetof(CPUX86State,fpregs[reg].mmx); |
| 4517 |
if (mod == 3) { |
| 4518 |
op2_offset = offsetof(CPUX86State,fpregs[rm].mmx); |
| 4519 |
} else {
|
| 4520 |
op2_offset = offsetof(CPUX86State,mmx_t0); |
| 4521 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4522 |
gen_ldq_env_A0(s->mem_index, op2_offset); |
| 4523 |
} |
| 4524 |
} |
| 4525 |
val = cpu_ldub_code(env, s->pc++); |
| 4526 |
|
| 4527 |
if ((b & 0xfc) == 0x60) { /* pcmpXstrX */ |
| 4528 |
set_cc_op(s, CC_OP_EFLAGS); |
| 4529 |
|
| 4530 |
if (s->dflag == 2) |
| 4531 |
/* The helper must use entire 64-bit gp registers */
|
| 4532 |
val |= 1 << 8; |
| 4533 |
} |
| 4534 |
|
| 4535 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 4536 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 4537 |
sse_fn_eppi(cpu_env, cpu_ptr0, cpu_ptr1, tcg_const_i32(val)); |
| 4538 |
break;
|
| 4539 |
|
| 4540 |
case 0x33a: |
| 4541 |
/* Various integer extensions at 0f 3a f[0-f]. */
|
| 4542 |
b = modrm | (b1 << 8);
|
| 4543 |
modrm = cpu_ldub_code(env, s->pc++); |
| 4544 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 4545 |
|
| 4546 |
switch (b) {
|
| 4547 |
case 0x3f0: /* rorx Gy,Ey, Ib */ |
| 4548 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI2)
|
| 4549 |
|| !(s->prefix & PREFIX_VEX) |
| 4550 |
|| s->vex_l != 0) {
|
| 4551 |
goto illegal_op;
|
| 4552 |
} |
| 4553 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 4554 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 4555 |
b = cpu_ldub_code(env, s->pc++); |
| 4556 |
if (ot == OT_QUAD) {
|
| 4557 |
tcg_gen_rotri_tl(cpu_T[0], cpu_T[0], b & 63); |
| 4558 |
} else {
|
| 4559 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 4560 |
tcg_gen_rotri_i32(cpu_tmp2_i32, cpu_tmp2_i32, b & 31);
|
| 4561 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 4562 |
} |
| 4563 |
gen_op_mov_reg_T0(ot, reg); |
| 4564 |
break;
|
| 4565 |
|
| 4566 |
default:
|
| 4567 |
goto illegal_op;
|
| 4568 |
} |
| 4569 |
break;
|
| 4570 |
|
| 4571 |
default:
|
| 4572 |
goto illegal_op;
|
| 4573 |
} |
| 4574 |
} else {
|
| 4575 |
/* generic MMX or SSE operation */
|
| 4576 |
switch(b) {
|
| 4577 |
case 0x70: /* pshufx insn */ |
| 4578 |
case 0xc6: /* pshufx insn */ |
| 4579 |
case 0xc2: /* compare insns */ |
| 4580 |
s->rip_offset = 1;
|
| 4581 |
break;
|
| 4582 |
default:
|
| 4583 |
break;
|
| 4584 |
} |
| 4585 |
if (is_xmm) {
|
| 4586 |
op1_offset = offsetof(CPUX86State,xmm_regs[reg]); |
| 4587 |
if (mod != 3) { |
| 4588 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4589 |
op2_offset = offsetof(CPUX86State,xmm_t0); |
| 4590 |
if (b1 >= 2 && ((b >= 0x50 && b <= 0x5f && b != 0x5b) || |
| 4591 |
b == 0xc2)) {
|
| 4592 |
/* specific case for SSE single instructions */
|
| 4593 |
if (b1 == 2) { |
| 4594 |
/* 32 bit access */
|
| 4595 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 4596 |
tcg_gen_st32_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,xmm_t0.XMM_L(0))); |
| 4597 |
} else {
|
| 4598 |
/* 64 bit access */
|
| 4599 |
gen_ldq_env_A0(s->mem_index, offsetof(CPUX86State,xmm_t0.XMM_D(0)));
|
| 4600 |
} |
| 4601 |
} else {
|
| 4602 |
gen_ldo_env_A0(s->mem_index, op2_offset); |
| 4603 |
} |
| 4604 |
} else {
|
| 4605 |
rm = (modrm & 7) | REX_B(s);
|
| 4606 |
op2_offset = offsetof(CPUX86State,xmm_regs[rm]); |
| 4607 |
} |
| 4608 |
} else {
|
| 4609 |
op1_offset = offsetof(CPUX86State,fpregs[reg].mmx); |
| 4610 |
if (mod != 3) { |
| 4611 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4612 |
op2_offset = offsetof(CPUX86State,mmx_t0); |
| 4613 |
gen_ldq_env_A0(s->mem_index, op2_offset); |
| 4614 |
} else {
|
| 4615 |
rm = (modrm & 7);
|
| 4616 |
op2_offset = offsetof(CPUX86State,fpregs[rm].mmx); |
| 4617 |
} |
| 4618 |
} |
| 4619 |
switch(b) {
|
| 4620 |
case 0x0f: /* 3DNow! data insns */ |
| 4621 |
if (!(s->cpuid_ext2_features & CPUID_EXT2_3DNOW))
|
| 4622 |
goto illegal_op;
|
| 4623 |
val = cpu_ldub_code(env, s->pc++); |
| 4624 |
sse_fn_epp = sse_op_table5[val]; |
| 4625 |
if (!sse_fn_epp) {
|
| 4626 |
goto illegal_op;
|
| 4627 |
} |
| 4628 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 4629 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 4630 |
sse_fn_epp(cpu_env, cpu_ptr0, cpu_ptr1); |
| 4631 |
break;
|
| 4632 |
case 0x70: /* pshufx insn */ |
| 4633 |
case 0xc6: /* pshufx insn */ |
| 4634 |
val = cpu_ldub_code(env, s->pc++); |
| 4635 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 4636 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 4637 |
/* XXX: introduce a new table? */
|
| 4638 |
sse_fn_ppi = (SSEFunc_0_ppi)sse_fn_epp; |
| 4639 |
sse_fn_ppi(cpu_ptr0, cpu_ptr1, tcg_const_i32(val)); |
| 4640 |
break;
|
| 4641 |
case 0xc2: |
| 4642 |
/* compare insns */
|
| 4643 |
val = cpu_ldub_code(env, s->pc++); |
| 4644 |
if (val >= 8) |
| 4645 |
goto illegal_op;
|
| 4646 |
sse_fn_epp = sse_op_table4[val][b1]; |
| 4647 |
|
| 4648 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 4649 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 4650 |
sse_fn_epp(cpu_env, cpu_ptr0, cpu_ptr1); |
| 4651 |
break;
|
| 4652 |
case 0xf7: |
| 4653 |
/* maskmov : we must prepare A0 */
|
| 4654 |
if (mod != 3) |
| 4655 |
goto illegal_op;
|
| 4656 |
#ifdef TARGET_X86_64
|
| 4657 |
if (s->aflag == 2) { |
| 4658 |
gen_op_movq_A0_reg(R_EDI); |
| 4659 |
} else
|
| 4660 |
#endif
|
| 4661 |
{
|
| 4662 |
gen_op_movl_A0_reg(R_EDI); |
| 4663 |
if (s->aflag == 0) |
| 4664 |
gen_op_andl_A0_ffff(); |
| 4665 |
} |
| 4666 |
gen_add_A0_ds_seg(s); |
| 4667 |
|
| 4668 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 4669 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 4670 |
/* XXX: introduce a new table? */
|
| 4671 |
sse_fn_eppt = (SSEFunc_0_eppt)sse_fn_epp; |
| 4672 |
sse_fn_eppt(cpu_env, cpu_ptr0, cpu_ptr1, cpu_A0); |
| 4673 |
break;
|
| 4674 |
default:
|
| 4675 |
tcg_gen_addi_ptr(cpu_ptr0, cpu_env, op1_offset); |
| 4676 |
tcg_gen_addi_ptr(cpu_ptr1, cpu_env, op2_offset); |
| 4677 |
sse_fn_epp(cpu_env, cpu_ptr0, cpu_ptr1); |
| 4678 |
break;
|
| 4679 |
} |
| 4680 |
if (b == 0x2e || b == 0x2f) { |
| 4681 |
set_cc_op(s, CC_OP_EFLAGS); |
| 4682 |
} |
| 4683 |
} |
| 4684 |
} |
| 4685 |
|
| 4686 |
/* convert one instruction. s->is_jmp is set if the translation must
|
| 4687 |
be stopped. Return the next pc value */
|
| 4688 |
static target_ulong disas_insn(CPUX86State *env, DisasContext *s,
|
| 4689 |
target_ulong pc_start) |
| 4690 |
{
|
| 4691 |
int b, prefixes, aflag, dflag;
|
| 4692 |
int shift, ot;
|
| 4693 |
int modrm, reg, rm, mod, reg_addr, op, opreg, offset_addr, val;
|
| 4694 |
target_ulong next_eip, tval; |
| 4695 |
int rex_w, rex_r;
|
| 4696 |
|
| 4697 |
if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_OP | CPU_LOG_TB_OP_OPT))) {
|
| 4698 |
tcg_gen_debug_insn_start(pc_start); |
| 4699 |
} |
| 4700 |
s->pc = pc_start; |
| 4701 |
prefixes = 0;
|
| 4702 |
aflag = s->code32; |
| 4703 |
dflag = s->code32; |
| 4704 |
s->override = -1;
|
| 4705 |
rex_w = -1;
|
| 4706 |
rex_r = 0;
|
| 4707 |
#ifdef TARGET_X86_64
|
| 4708 |
s->rex_x = 0;
|
| 4709 |
s->rex_b = 0;
|
| 4710 |
x86_64_hregs = 0;
|
| 4711 |
#endif
|
| 4712 |
s->rip_offset = 0; /* for relative ip address */ |
| 4713 |
s->vex_l = 0;
|
| 4714 |
s->vex_v = 0;
|
| 4715 |
next_byte:
|
| 4716 |
b = cpu_ldub_code(env, s->pc); |
| 4717 |
s->pc++; |
| 4718 |
/* Collect prefixes. */
|
| 4719 |
switch (b) {
|
| 4720 |
case 0xf3: |
| 4721 |
prefixes |= PREFIX_REPZ; |
| 4722 |
goto next_byte;
|
| 4723 |
case 0xf2: |
| 4724 |
prefixes |= PREFIX_REPNZ; |
| 4725 |
goto next_byte;
|
| 4726 |
case 0xf0: |
| 4727 |
prefixes |= PREFIX_LOCK; |
| 4728 |
goto next_byte;
|
| 4729 |
case 0x2e: |
| 4730 |
s->override = R_CS; |
| 4731 |
goto next_byte;
|
| 4732 |
case 0x36: |
| 4733 |
s->override = R_SS; |
| 4734 |
goto next_byte;
|
| 4735 |
case 0x3e: |
| 4736 |
s->override = R_DS; |
| 4737 |
goto next_byte;
|
| 4738 |
case 0x26: |
| 4739 |
s->override = R_ES; |
| 4740 |
goto next_byte;
|
| 4741 |
case 0x64: |
| 4742 |
s->override = R_FS; |
| 4743 |
goto next_byte;
|
| 4744 |
case 0x65: |
| 4745 |
s->override = R_GS; |
| 4746 |
goto next_byte;
|
| 4747 |
case 0x66: |
| 4748 |
prefixes |= PREFIX_DATA; |
| 4749 |
goto next_byte;
|
| 4750 |
case 0x67: |
| 4751 |
prefixes |= PREFIX_ADR; |
| 4752 |
goto next_byte;
|
| 4753 |
#ifdef TARGET_X86_64
|
| 4754 |
case 0x40 ... 0x4f: |
| 4755 |
if (CODE64(s)) {
|
| 4756 |
/* REX prefix */
|
| 4757 |
rex_w = (b >> 3) & 1; |
| 4758 |
rex_r = (b & 0x4) << 1; |
| 4759 |
s->rex_x = (b & 0x2) << 2; |
| 4760 |
REX_B(s) = (b & 0x1) << 3; |
| 4761 |
x86_64_hregs = 1; /* select uniform byte register addressing */ |
| 4762 |
goto next_byte;
|
| 4763 |
} |
| 4764 |
break;
|
| 4765 |
#endif
|
| 4766 |
case 0xc5: /* 2-byte VEX */ |
| 4767 |
case 0xc4: /* 3-byte VEX */ |
| 4768 |
/* VEX prefixes cannot be used except in 32-bit mode.
|
| 4769 |
Otherwise the instruction is LES or LDS. */
|
| 4770 |
if (s->code32 && !s->vm86) {
|
| 4771 |
static const int pp_prefix[4] = { |
| 4772 |
0, PREFIX_DATA, PREFIX_REPZ, PREFIX_REPNZ
|
| 4773 |
}; |
| 4774 |
int vex3, vex2 = cpu_ldub_code(env, s->pc);
|
| 4775 |
|
| 4776 |
if (!CODE64(s) && (vex2 & 0xc0) != 0xc0) { |
| 4777 |
/* 4.1.4.6: In 32-bit mode, bits [7:6] must be 11b,
|
| 4778 |
otherwise the instruction is LES or LDS. */
|
| 4779 |
break;
|
| 4780 |
} |
| 4781 |
s->pc++; |
| 4782 |
|
| 4783 |
/* 4.1.1-4.1.3: No preceeding lock, 66, f2, f3, or rex prefixes. */
|
| 4784 |
if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ
|
| 4785 |
| PREFIX_LOCK | PREFIX_DATA)) {
|
| 4786 |
goto illegal_op;
|
| 4787 |
} |
| 4788 |
#ifdef TARGET_X86_64
|
| 4789 |
if (x86_64_hregs) {
|
| 4790 |
goto illegal_op;
|
| 4791 |
} |
| 4792 |
#endif
|
| 4793 |
rex_r = (~vex2 >> 4) & 8; |
| 4794 |
if (b == 0xc5) { |
| 4795 |
vex3 = vex2; |
| 4796 |
b = cpu_ldub_code(env, s->pc++); |
| 4797 |
} else {
|
| 4798 |
#ifdef TARGET_X86_64
|
| 4799 |
s->rex_x = (~vex2 >> 3) & 8; |
| 4800 |
s->rex_b = (~vex2 >> 2) & 8; |
| 4801 |
#endif
|
| 4802 |
vex3 = cpu_ldub_code(env, s->pc++); |
| 4803 |
rex_w = (vex3 >> 7) & 1; |
| 4804 |
switch (vex2 & 0x1f) { |
| 4805 |
case 0x01: /* Implied 0f leading opcode bytes. */ |
| 4806 |
b = cpu_ldub_code(env, s->pc++) | 0x100;
|
| 4807 |
break;
|
| 4808 |
case 0x02: /* Implied 0f 38 leading opcode bytes. */ |
| 4809 |
b = 0x138;
|
| 4810 |
break;
|
| 4811 |
case 0x03: /* Implied 0f 3a leading opcode bytes. */ |
| 4812 |
b = 0x13a;
|
| 4813 |
break;
|
| 4814 |
default: /* Reserved for future use. */ |
| 4815 |
goto illegal_op;
|
| 4816 |
} |
| 4817 |
} |
| 4818 |
s->vex_v = (~vex3 >> 3) & 0xf; |
| 4819 |
s->vex_l = (vex3 >> 2) & 1; |
| 4820 |
prefixes |= pp_prefix[vex3 & 3] | PREFIX_VEX;
|
| 4821 |
} |
| 4822 |
break;
|
| 4823 |
} |
| 4824 |
|
| 4825 |
/* Post-process prefixes. */
|
| 4826 |
if (prefixes & PREFIX_DATA) {
|
| 4827 |
dflag ^= 1;
|
| 4828 |
} |
| 4829 |
if (prefixes & PREFIX_ADR) {
|
| 4830 |
aflag ^= 1;
|
| 4831 |
} |
| 4832 |
#ifdef TARGET_X86_64
|
| 4833 |
if (CODE64(s)) {
|
| 4834 |
if (rex_w == 1) { |
| 4835 |
/* 0x66 is ignored if rex.w is set */
|
| 4836 |
dflag = 2;
|
| 4837 |
} |
| 4838 |
if (!(prefixes & PREFIX_ADR)) {
|
| 4839 |
aflag = 2;
|
| 4840 |
} |
| 4841 |
} |
| 4842 |
#endif
|
| 4843 |
|
| 4844 |
s->prefix = prefixes; |
| 4845 |
s->aflag = aflag; |
| 4846 |
s->dflag = dflag; |
| 4847 |
|
| 4848 |
/* lock generation */
|
| 4849 |
if (prefixes & PREFIX_LOCK)
|
| 4850 |
gen_helper_lock(); |
| 4851 |
|
| 4852 |
/* now check op code */
|
| 4853 |
reswitch:
|
| 4854 |
switch(b) {
|
| 4855 |
case 0x0f: |
| 4856 |
/**************************/
|
| 4857 |
/* extended op code */
|
| 4858 |
b = cpu_ldub_code(env, s->pc++) | 0x100;
|
| 4859 |
goto reswitch;
|
| 4860 |
|
| 4861 |
/**************************/
|
| 4862 |
/* arith & logic */
|
| 4863 |
case 0x00 ... 0x05: |
| 4864 |
case 0x08 ... 0x0d: |
| 4865 |
case 0x10 ... 0x15: |
| 4866 |
case 0x18 ... 0x1d: |
| 4867 |
case 0x20 ... 0x25: |
| 4868 |
case 0x28 ... 0x2d: |
| 4869 |
case 0x30 ... 0x35: |
| 4870 |
case 0x38 ... 0x3d: |
| 4871 |
{
|
| 4872 |
int op, f, val;
|
| 4873 |
op = (b >> 3) & 7; |
| 4874 |
f = (b >> 1) & 3; |
| 4875 |
|
| 4876 |
if ((b & 1) == 0) |
| 4877 |
ot = OT_BYTE; |
| 4878 |
else
|
| 4879 |
ot = dflag + OT_WORD; |
| 4880 |
|
| 4881 |
switch(f) {
|
| 4882 |
case 0: /* OP Ev, Gv */ |
| 4883 |
modrm = cpu_ldub_code(env, s->pc++); |
| 4884 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 4885 |
mod = (modrm >> 6) & 3; |
| 4886 |
rm = (modrm & 7) | REX_B(s);
|
| 4887 |
if (mod != 3) { |
| 4888 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4889 |
opreg = OR_TMP0; |
| 4890 |
} else if (op == OP_XORL && rm == reg) { |
| 4891 |
xor_zero:
|
| 4892 |
/* xor reg, reg optimisation */
|
| 4893 |
gen_op_movl_T0_0(); |
| 4894 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 4895 |
gen_op_mov_reg_T0(ot, reg); |
| 4896 |
gen_op_update1_cc(); |
| 4897 |
break;
|
| 4898 |
} else {
|
| 4899 |
opreg = rm; |
| 4900 |
} |
| 4901 |
gen_op_mov_TN_reg(ot, 1, reg);
|
| 4902 |
gen_op(s, op, ot, opreg); |
| 4903 |
break;
|
| 4904 |
case 1: /* OP Gv, Ev */ |
| 4905 |
modrm = cpu_ldub_code(env, s->pc++); |
| 4906 |
mod = (modrm >> 6) & 3; |
| 4907 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 4908 |
rm = (modrm & 7) | REX_B(s);
|
| 4909 |
if (mod != 3) { |
| 4910 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4911 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 4912 |
} else if (op == OP_XORL && rm == reg) { |
| 4913 |
goto xor_zero;
|
| 4914 |
} else {
|
| 4915 |
gen_op_mov_TN_reg(ot, 1, rm);
|
| 4916 |
} |
| 4917 |
gen_op(s, op, ot, reg); |
| 4918 |
break;
|
| 4919 |
case 2: /* OP A, Iv */ |
| 4920 |
val = insn_get(env, s, ot); |
| 4921 |
gen_op_movl_T1_im(val); |
| 4922 |
gen_op(s, op, ot, OR_EAX); |
| 4923 |
break;
|
| 4924 |
} |
| 4925 |
} |
| 4926 |
break;
|
| 4927 |
|
| 4928 |
case 0x82: |
| 4929 |
if (CODE64(s))
|
| 4930 |
goto illegal_op;
|
| 4931 |
case 0x80: /* GRP1 */ |
| 4932 |
case 0x81: |
| 4933 |
case 0x83: |
| 4934 |
{
|
| 4935 |
int val;
|
| 4936 |
|
| 4937 |
if ((b & 1) == 0) |
| 4938 |
ot = OT_BYTE; |
| 4939 |
else
|
| 4940 |
ot = dflag + OT_WORD; |
| 4941 |
|
| 4942 |
modrm = cpu_ldub_code(env, s->pc++); |
| 4943 |
mod = (modrm >> 6) & 3; |
| 4944 |
rm = (modrm & 7) | REX_B(s);
|
| 4945 |
op = (modrm >> 3) & 7; |
| 4946 |
|
| 4947 |
if (mod != 3) { |
| 4948 |
if (b == 0x83) |
| 4949 |
s->rip_offset = 1;
|
| 4950 |
else
|
| 4951 |
s->rip_offset = insn_const_size(ot); |
| 4952 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4953 |
opreg = OR_TMP0; |
| 4954 |
} else {
|
| 4955 |
opreg = rm; |
| 4956 |
} |
| 4957 |
|
| 4958 |
switch(b) {
|
| 4959 |
default:
|
| 4960 |
case 0x80: |
| 4961 |
case 0x81: |
| 4962 |
case 0x82: |
| 4963 |
val = insn_get(env, s, ot); |
| 4964 |
break;
|
| 4965 |
case 0x83: |
| 4966 |
val = (int8_t)insn_get(env, s, OT_BYTE); |
| 4967 |
break;
|
| 4968 |
} |
| 4969 |
gen_op_movl_T1_im(val); |
| 4970 |
gen_op(s, op, ot, opreg); |
| 4971 |
} |
| 4972 |
break;
|
| 4973 |
|
| 4974 |
/**************************/
|
| 4975 |
/* inc, dec, and other misc arith */
|
| 4976 |
case 0x40 ... 0x47: /* inc Gv */ |
| 4977 |
ot = dflag ? OT_LONG : OT_WORD; |
| 4978 |
gen_inc(s, ot, OR_EAX + (b & 7), 1); |
| 4979 |
break;
|
| 4980 |
case 0x48 ... 0x4f: /* dec Gv */ |
| 4981 |
ot = dflag ? OT_LONG : OT_WORD; |
| 4982 |
gen_inc(s, ot, OR_EAX + (b & 7), -1); |
| 4983 |
break;
|
| 4984 |
case 0xf6: /* GRP3 */ |
| 4985 |
case 0xf7: |
| 4986 |
if ((b & 1) == 0) |
| 4987 |
ot = OT_BYTE; |
| 4988 |
else
|
| 4989 |
ot = dflag + OT_WORD; |
| 4990 |
|
| 4991 |
modrm = cpu_ldub_code(env, s->pc++); |
| 4992 |
mod = (modrm >> 6) & 3; |
| 4993 |
rm = (modrm & 7) | REX_B(s);
|
| 4994 |
op = (modrm >> 3) & 7; |
| 4995 |
if (mod != 3) { |
| 4996 |
if (op == 0) |
| 4997 |
s->rip_offset = insn_const_size(ot); |
| 4998 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 4999 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 5000 |
} else {
|
| 5001 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 5002 |
} |
| 5003 |
|
| 5004 |
switch(op) {
|
| 5005 |
case 0: /* test */ |
| 5006 |
val = insn_get(env, s, ot); |
| 5007 |
gen_op_movl_T1_im(val); |
| 5008 |
gen_op_testl_T0_T1_cc(); |
| 5009 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 5010 |
break;
|
| 5011 |
case 2: /* not */ |
| 5012 |
tcg_gen_not_tl(cpu_T[0], cpu_T[0]); |
| 5013 |
if (mod != 3) { |
| 5014 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 5015 |
} else {
|
| 5016 |
gen_op_mov_reg_T0(ot, rm); |
| 5017 |
} |
| 5018 |
break;
|
| 5019 |
case 3: /* neg */ |
| 5020 |
tcg_gen_neg_tl(cpu_T[0], cpu_T[0]); |
| 5021 |
if (mod != 3) { |
| 5022 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 5023 |
} else {
|
| 5024 |
gen_op_mov_reg_T0(ot, rm); |
| 5025 |
} |
| 5026 |
gen_op_update_neg_cc(); |
| 5027 |
set_cc_op(s, CC_OP_SUBB + ot); |
| 5028 |
break;
|
| 5029 |
case 4: /* mul */ |
| 5030 |
switch(ot) {
|
| 5031 |
case OT_BYTE:
|
| 5032 |
gen_op_mov_TN_reg(OT_BYTE, 1, R_EAX);
|
| 5033 |
tcg_gen_ext8u_tl(cpu_T[0], cpu_T[0]); |
| 5034 |
tcg_gen_ext8u_tl(cpu_T[1], cpu_T[1]); |
| 5035 |
/* XXX: use 32 bit mul which could be faster */
|
| 5036 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5037 |
gen_op_mov_reg_T0(OT_WORD, R_EAX); |
| 5038 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5039 |
tcg_gen_andi_tl(cpu_cc_src, cpu_T[0], 0xff00); |
| 5040 |
set_cc_op(s, CC_OP_MULB); |
| 5041 |
break;
|
| 5042 |
case OT_WORD:
|
| 5043 |
gen_op_mov_TN_reg(OT_WORD, 1, R_EAX);
|
| 5044 |
tcg_gen_ext16u_tl(cpu_T[0], cpu_T[0]); |
| 5045 |
tcg_gen_ext16u_tl(cpu_T[1], cpu_T[1]); |
| 5046 |
/* XXX: use 32 bit mul which could be faster */
|
| 5047 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5048 |
gen_op_mov_reg_T0(OT_WORD, R_EAX); |
| 5049 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5050 |
tcg_gen_shri_tl(cpu_T[0], cpu_T[0], 16); |
| 5051 |
gen_op_mov_reg_T0(OT_WORD, R_EDX); |
| 5052 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[0]);
|
| 5053 |
set_cc_op(s, CC_OP_MULW); |
| 5054 |
break;
|
| 5055 |
default:
|
| 5056 |
case OT_LONG:
|
| 5057 |
#ifdef TARGET_X86_64
|
| 5058 |
gen_op_mov_TN_reg(OT_LONG, 1, R_EAX);
|
| 5059 |
tcg_gen_ext32u_tl(cpu_T[0], cpu_T[0]); |
| 5060 |
tcg_gen_ext32u_tl(cpu_T[1], cpu_T[1]); |
| 5061 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5062 |
gen_op_mov_reg_T0(OT_LONG, R_EAX); |
| 5063 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5064 |
tcg_gen_shri_tl(cpu_T[0], cpu_T[0], 32); |
| 5065 |
gen_op_mov_reg_T0(OT_LONG, R_EDX); |
| 5066 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[0]);
|
| 5067 |
#else
|
| 5068 |
{
|
| 5069 |
TCGv_i64 t0, t1; |
| 5070 |
t0 = tcg_temp_new_i64(); |
| 5071 |
t1 = tcg_temp_new_i64(); |
| 5072 |
gen_op_mov_TN_reg(OT_LONG, 1, R_EAX);
|
| 5073 |
tcg_gen_extu_i32_i64(t0, cpu_T[0]);
|
| 5074 |
tcg_gen_extu_i32_i64(t1, cpu_T[1]);
|
| 5075 |
tcg_gen_mul_i64(t0, t0, t1); |
| 5076 |
tcg_gen_trunc_i64_i32(cpu_T[0], t0);
|
| 5077 |
gen_op_mov_reg_T0(OT_LONG, R_EAX); |
| 5078 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5079 |
tcg_gen_shri_i64(t0, t0, 32);
|
| 5080 |
tcg_gen_trunc_i64_i32(cpu_T[0], t0);
|
| 5081 |
gen_op_mov_reg_T0(OT_LONG, R_EDX); |
| 5082 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[0]);
|
| 5083 |
} |
| 5084 |
#endif
|
| 5085 |
set_cc_op(s, CC_OP_MULL); |
| 5086 |
break;
|
| 5087 |
#ifdef TARGET_X86_64
|
| 5088 |
case OT_QUAD:
|
| 5089 |
gen_helper_mulq_EAX_T0(cpu_env, cpu_T[0]);
|
| 5090 |
set_cc_op(s, CC_OP_MULQ); |
| 5091 |
break;
|
| 5092 |
#endif
|
| 5093 |
} |
| 5094 |
break;
|
| 5095 |
case 5: /* imul */ |
| 5096 |
switch(ot) {
|
| 5097 |
case OT_BYTE:
|
| 5098 |
gen_op_mov_TN_reg(OT_BYTE, 1, R_EAX);
|
| 5099 |
tcg_gen_ext8s_tl(cpu_T[0], cpu_T[0]); |
| 5100 |
tcg_gen_ext8s_tl(cpu_T[1], cpu_T[1]); |
| 5101 |
/* XXX: use 32 bit mul which could be faster */
|
| 5102 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5103 |
gen_op_mov_reg_T0(OT_WORD, R_EAX); |
| 5104 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5105 |
tcg_gen_ext8s_tl(cpu_tmp0, cpu_T[0]);
|
| 5106 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[0], cpu_tmp0);
|
| 5107 |
set_cc_op(s, CC_OP_MULB); |
| 5108 |
break;
|
| 5109 |
case OT_WORD:
|
| 5110 |
gen_op_mov_TN_reg(OT_WORD, 1, R_EAX);
|
| 5111 |
tcg_gen_ext16s_tl(cpu_T[0], cpu_T[0]); |
| 5112 |
tcg_gen_ext16s_tl(cpu_T[1], cpu_T[1]); |
| 5113 |
/* XXX: use 32 bit mul which could be faster */
|
| 5114 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5115 |
gen_op_mov_reg_T0(OT_WORD, R_EAX); |
| 5116 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5117 |
tcg_gen_ext16s_tl(cpu_tmp0, cpu_T[0]);
|
| 5118 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[0], cpu_tmp0);
|
| 5119 |
tcg_gen_shri_tl(cpu_T[0], cpu_T[0], 16); |
| 5120 |
gen_op_mov_reg_T0(OT_WORD, R_EDX); |
| 5121 |
set_cc_op(s, CC_OP_MULW); |
| 5122 |
break;
|
| 5123 |
default:
|
| 5124 |
case OT_LONG:
|
| 5125 |
#ifdef TARGET_X86_64
|
| 5126 |
gen_op_mov_TN_reg(OT_LONG, 1, R_EAX);
|
| 5127 |
tcg_gen_ext32s_tl(cpu_T[0], cpu_T[0]); |
| 5128 |
tcg_gen_ext32s_tl(cpu_T[1], cpu_T[1]); |
| 5129 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5130 |
gen_op_mov_reg_T0(OT_LONG, R_EAX); |
| 5131 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5132 |
tcg_gen_ext32s_tl(cpu_tmp0, cpu_T[0]);
|
| 5133 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[0], cpu_tmp0);
|
| 5134 |
tcg_gen_shri_tl(cpu_T[0], cpu_T[0], 32); |
| 5135 |
gen_op_mov_reg_T0(OT_LONG, R_EDX); |
| 5136 |
#else
|
| 5137 |
{
|
| 5138 |
TCGv_i64 t0, t1; |
| 5139 |
t0 = tcg_temp_new_i64(); |
| 5140 |
t1 = tcg_temp_new_i64(); |
| 5141 |
gen_op_mov_TN_reg(OT_LONG, 1, R_EAX);
|
| 5142 |
tcg_gen_ext_i32_i64(t0, cpu_T[0]);
|
| 5143 |
tcg_gen_ext_i32_i64(t1, cpu_T[1]);
|
| 5144 |
tcg_gen_mul_i64(t0, t0, t1); |
| 5145 |
tcg_gen_trunc_i64_i32(cpu_T[0], t0);
|
| 5146 |
gen_op_mov_reg_T0(OT_LONG, R_EAX); |
| 5147 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5148 |
tcg_gen_sari_tl(cpu_tmp0, cpu_T[0], 31); |
| 5149 |
tcg_gen_shri_i64(t0, t0, 32);
|
| 5150 |
tcg_gen_trunc_i64_i32(cpu_T[0], t0);
|
| 5151 |
gen_op_mov_reg_T0(OT_LONG, R_EDX); |
| 5152 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[0], cpu_tmp0);
|
| 5153 |
} |
| 5154 |
#endif
|
| 5155 |
set_cc_op(s, CC_OP_MULL); |
| 5156 |
break;
|
| 5157 |
#ifdef TARGET_X86_64
|
| 5158 |
case OT_QUAD:
|
| 5159 |
gen_helper_imulq_EAX_T0(cpu_env, cpu_T[0]);
|
| 5160 |
set_cc_op(s, CC_OP_MULQ); |
| 5161 |
break;
|
| 5162 |
#endif
|
| 5163 |
} |
| 5164 |
break;
|
| 5165 |
case 6: /* div */ |
| 5166 |
switch(ot) {
|
| 5167 |
case OT_BYTE:
|
| 5168 |
gen_jmp_im(pc_start - s->cs_base); |
| 5169 |
gen_helper_divb_AL(cpu_env, cpu_T[0]);
|
| 5170 |
break;
|
| 5171 |
case OT_WORD:
|
| 5172 |
gen_jmp_im(pc_start - s->cs_base); |
| 5173 |
gen_helper_divw_AX(cpu_env, cpu_T[0]);
|
| 5174 |
break;
|
| 5175 |
default:
|
| 5176 |
case OT_LONG:
|
| 5177 |
gen_jmp_im(pc_start - s->cs_base); |
| 5178 |
gen_helper_divl_EAX(cpu_env, cpu_T[0]);
|
| 5179 |
break;
|
| 5180 |
#ifdef TARGET_X86_64
|
| 5181 |
case OT_QUAD:
|
| 5182 |
gen_jmp_im(pc_start - s->cs_base); |
| 5183 |
gen_helper_divq_EAX(cpu_env, cpu_T[0]);
|
| 5184 |
break;
|
| 5185 |
#endif
|
| 5186 |
} |
| 5187 |
break;
|
| 5188 |
case 7: /* idiv */ |
| 5189 |
switch(ot) {
|
| 5190 |
case OT_BYTE:
|
| 5191 |
gen_jmp_im(pc_start - s->cs_base); |
| 5192 |
gen_helper_idivb_AL(cpu_env, cpu_T[0]);
|
| 5193 |
break;
|
| 5194 |
case OT_WORD:
|
| 5195 |
gen_jmp_im(pc_start - s->cs_base); |
| 5196 |
gen_helper_idivw_AX(cpu_env, cpu_T[0]);
|
| 5197 |
break;
|
| 5198 |
default:
|
| 5199 |
case OT_LONG:
|
| 5200 |
gen_jmp_im(pc_start - s->cs_base); |
| 5201 |
gen_helper_idivl_EAX(cpu_env, cpu_T[0]);
|
| 5202 |
break;
|
| 5203 |
#ifdef TARGET_X86_64
|
| 5204 |
case OT_QUAD:
|
| 5205 |
gen_jmp_im(pc_start - s->cs_base); |
| 5206 |
gen_helper_idivq_EAX(cpu_env, cpu_T[0]);
|
| 5207 |
break;
|
| 5208 |
#endif
|
| 5209 |
} |
| 5210 |
break;
|
| 5211 |
default:
|
| 5212 |
goto illegal_op;
|
| 5213 |
} |
| 5214 |
break;
|
| 5215 |
|
| 5216 |
case 0xfe: /* GRP4 */ |
| 5217 |
case 0xff: /* GRP5 */ |
| 5218 |
if ((b & 1) == 0) |
| 5219 |
ot = OT_BYTE; |
| 5220 |
else
|
| 5221 |
ot = dflag + OT_WORD; |
| 5222 |
|
| 5223 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5224 |
mod = (modrm >> 6) & 3; |
| 5225 |
rm = (modrm & 7) | REX_B(s);
|
| 5226 |
op = (modrm >> 3) & 7; |
| 5227 |
if (op >= 2 && b == 0xfe) { |
| 5228 |
goto illegal_op;
|
| 5229 |
} |
| 5230 |
if (CODE64(s)) {
|
| 5231 |
if (op == 2 || op == 4) { |
| 5232 |
/* operand size for jumps is 64 bit */
|
| 5233 |
ot = OT_QUAD; |
| 5234 |
} else if (op == 3 || op == 5) { |
| 5235 |
ot = dflag ? OT_LONG + (rex_w == 1) : OT_WORD;
|
| 5236 |
} else if (op == 6) { |
| 5237 |
/* default push size is 64 bit */
|
| 5238 |
ot = dflag ? OT_QUAD : OT_WORD; |
| 5239 |
} |
| 5240 |
} |
| 5241 |
if (mod != 3) { |
| 5242 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5243 |
if (op >= 2 && op != 3 && op != 5) |
| 5244 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 5245 |
} else {
|
| 5246 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 5247 |
} |
| 5248 |
|
| 5249 |
switch(op) {
|
| 5250 |
case 0: /* inc Ev */ |
| 5251 |
if (mod != 3) |
| 5252 |
opreg = OR_TMP0; |
| 5253 |
else
|
| 5254 |
opreg = rm; |
| 5255 |
gen_inc(s, ot, opreg, 1);
|
| 5256 |
break;
|
| 5257 |
case 1: /* dec Ev */ |
| 5258 |
if (mod != 3) |
| 5259 |
opreg = OR_TMP0; |
| 5260 |
else
|
| 5261 |
opreg = rm; |
| 5262 |
gen_inc(s, ot, opreg, -1);
|
| 5263 |
break;
|
| 5264 |
case 2: /* call Ev */ |
| 5265 |
/* XXX: optimize if memory (no 'and' is necessary) */
|
| 5266 |
if (s->dflag == 0) |
| 5267 |
gen_op_andl_T0_ffff(); |
| 5268 |
next_eip = s->pc - s->cs_base; |
| 5269 |
gen_movtl_T1_im(next_eip); |
| 5270 |
gen_push_T1(s); |
| 5271 |
gen_op_jmp_T0(); |
| 5272 |
gen_eob(s); |
| 5273 |
break;
|
| 5274 |
case 3: /* lcall Ev */ |
| 5275 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 5276 |
gen_add_A0_im(s, 1 << (ot - OT_WORD + 1)); |
| 5277 |
gen_op_ldu_T0_A0(OT_WORD + s->mem_index); |
| 5278 |
do_lcall:
|
| 5279 |
if (s->pe && !s->vm86) {
|
| 5280 |
gen_update_cc_op(s); |
| 5281 |
gen_jmp_im(pc_start - s->cs_base); |
| 5282 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 5283 |
gen_helper_lcall_protected(cpu_env, cpu_tmp2_i32, cpu_T[1],
|
| 5284 |
tcg_const_i32(dflag), |
| 5285 |
tcg_const_i32(s->pc - pc_start)); |
| 5286 |
} else {
|
| 5287 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 5288 |
gen_helper_lcall_real(cpu_env, cpu_tmp2_i32, cpu_T[1],
|
| 5289 |
tcg_const_i32(dflag), |
| 5290 |
tcg_const_i32(s->pc - s->cs_base)); |
| 5291 |
} |
| 5292 |
gen_eob(s); |
| 5293 |
break;
|
| 5294 |
case 4: /* jmp Ev */ |
| 5295 |
if (s->dflag == 0) |
| 5296 |
gen_op_andl_T0_ffff(); |
| 5297 |
gen_op_jmp_T0(); |
| 5298 |
gen_eob(s); |
| 5299 |
break;
|
| 5300 |
case 5: /* ljmp Ev */ |
| 5301 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 5302 |
gen_add_A0_im(s, 1 << (ot - OT_WORD + 1)); |
| 5303 |
gen_op_ldu_T0_A0(OT_WORD + s->mem_index); |
| 5304 |
do_ljmp:
|
| 5305 |
if (s->pe && !s->vm86) {
|
| 5306 |
gen_update_cc_op(s); |
| 5307 |
gen_jmp_im(pc_start - s->cs_base); |
| 5308 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 5309 |
gen_helper_ljmp_protected(cpu_env, cpu_tmp2_i32, cpu_T[1],
|
| 5310 |
tcg_const_i32(s->pc - pc_start)); |
| 5311 |
} else {
|
| 5312 |
gen_op_movl_seg_T0_vm(R_CS); |
| 5313 |
gen_op_movl_T0_T1(); |
| 5314 |
gen_op_jmp_T0(); |
| 5315 |
} |
| 5316 |
gen_eob(s); |
| 5317 |
break;
|
| 5318 |
case 6: /* push Ev */ |
| 5319 |
gen_push_T0(s); |
| 5320 |
break;
|
| 5321 |
default:
|
| 5322 |
goto illegal_op;
|
| 5323 |
} |
| 5324 |
break;
|
| 5325 |
|
| 5326 |
case 0x84: /* test Ev, Gv */ |
| 5327 |
case 0x85: |
| 5328 |
if ((b & 1) == 0) |
| 5329 |
ot = OT_BYTE; |
| 5330 |
else
|
| 5331 |
ot = dflag + OT_WORD; |
| 5332 |
|
| 5333 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5334 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5335 |
|
| 5336 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 5337 |
gen_op_mov_TN_reg(ot, 1, reg);
|
| 5338 |
gen_op_testl_T0_T1_cc(); |
| 5339 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 5340 |
break;
|
| 5341 |
|
| 5342 |
case 0xa8: /* test eAX, Iv */ |
| 5343 |
case 0xa9: |
| 5344 |
if ((b & 1) == 0) |
| 5345 |
ot = OT_BYTE; |
| 5346 |
else
|
| 5347 |
ot = dflag + OT_WORD; |
| 5348 |
val = insn_get(env, s, ot); |
| 5349 |
|
| 5350 |
gen_op_mov_TN_reg(ot, 0, OR_EAX);
|
| 5351 |
gen_op_movl_T1_im(val); |
| 5352 |
gen_op_testl_T0_T1_cc(); |
| 5353 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 5354 |
break;
|
| 5355 |
|
| 5356 |
case 0x98: /* CWDE/CBW */ |
| 5357 |
#ifdef TARGET_X86_64
|
| 5358 |
if (dflag == 2) { |
| 5359 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EAX);
|
| 5360 |
tcg_gen_ext32s_tl(cpu_T[0], cpu_T[0]); |
| 5361 |
gen_op_mov_reg_T0(OT_QUAD, R_EAX); |
| 5362 |
} else
|
| 5363 |
#endif
|
| 5364 |
if (dflag == 1) { |
| 5365 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EAX);
|
| 5366 |
tcg_gen_ext16s_tl(cpu_T[0], cpu_T[0]); |
| 5367 |
gen_op_mov_reg_T0(OT_LONG, R_EAX); |
| 5368 |
} else {
|
| 5369 |
gen_op_mov_TN_reg(OT_BYTE, 0, R_EAX);
|
| 5370 |
tcg_gen_ext8s_tl(cpu_T[0], cpu_T[0]); |
| 5371 |
gen_op_mov_reg_T0(OT_WORD, R_EAX); |
| 5372 |
} |
| 5373 |
break;
|
| 5374 |
case 0x99: /* CDQ/CWD */ |
| 5375 |
#ifdef TARGET_X86_64
|
| 5376 |
if (dflag == 2) { |
| 5377 |
gen_op_mov_TN_reg(OT_QUAD, 0, R_EAX);
|
| 5378 |
tcg_gen_sari_tl(cpu_T[0], cpu_T[0], 63); |
| 5379 |
gen_op_mov_reg_T0(OT_QUAD, R_EDX); |
| 5380 |
} else
|
| 5381 |
#endif
|
| 5382 |
if (dflag == 1) { |
| 5383 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EAX);
|
| 5384 |
tcg_gen_ext32s_tl(cpu_T[0], cpu_T[0]); |
| 5385 |
tcg_gen_sari_tl(cpu_T[0], cpu_T[0], 31); |
| 5386 |
gen_op_mov_reg_T0(OT_LONG, R_EDX); |
| 5387 |
} else {
|
| 5388 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EAX);
|
| 5389 |
tcg_gen_ext16s_tl(cpu_T[0], cpu_T[0]); |
| 5390 |
tcg_gen_sari_tl(cpu_T[0], cpu_T[0], 15); |
| 5391 |
gen_op_mov_reg_T0(OT_WORD, R_EDX); |
| 5392 |
} |
| 5393 |
break;
|
| 5394 |
case 0x1af: /* imul Gv, Ev */ |
| 5395 |
case 0x69: /* imul Gv, Ev, I */ |
| 5396 |
case 0x6b: |
| 5397 |
ot = dflag + OT_WORD; |
| 5398 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5399 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5400 |
if (b == 0x69) |
| 5401 |
s->rip_offset = insn_const_size(ot); |
| 5402 |
else if (b == 0x6b) |
| 5403 |
s->rip_offset = 1;
|
| 5404 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 5405 |
if (b == 0x69) { |
| 5406 |
val = insn_get(env, s, ot); |
| 5407 |
gen_op_movl_T1_im(val); |
| 5408 |
} else if (b == 0x6b) { |
| 5409 |
val = (int8_t)insn_get(env, s, OT_BYTE); |
| 5410 |
gen_op_movl_T1_im(val); |
| 5411 |
} else {
|
| 5412 |
gen_op_mov_TN_reg(ot, 1, reg);
|
| 5413 |
} |
| 5414 |
|
| 5415 |
#ifdef TARGET_X86_64
|
| 5416 |
if (ot == OT_QUAD) {
|
| 5417 |
gen_helper_imulq_T0_T1(cpu_T[0], cpu_env, cpu_T[0], cpu_T[1]); |
| 5418 |
} else
|
| 5419 |
#endif
|
| 5420 |
if (ot == OT_LONG) {
|
| 5421 |
#ifdef TARGET_X86_64
|
| 5422 |
tcg_gen_ext32s_tl(cpu_T[0], cpu_T[0]); |
| 5423 |
tcg_gen_ext32s_tl(cpu_T[1], cpu_T[1]); |
| 5424 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5425 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5426 |
tcg_gen_ext32s_tl(cpu_tmp0, cpu_T[0]);
|
| 5427 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[0], cpu_tmp0);
|
| 5428 |
#else
|
| 5429 |
{
|
| 5430 |
TCGv_i64 t0, t1; |
| 5431 |
t0 = tcg_temp_new_i64(); |
| 5432 |
t1 = tcg_temp_new_i64(); |
| 5433 |
tcg_gen_ext_i32_i64(t0, cpu_T[0]);
|
| 5434 |
tcg_gen_ext_i32_i64(t1, cpu_T[1]);
|
| 5435 |
tcg_gen_mul_i64(t0, t0, t1); |
| 5436 |
tcg_gen_trunc_i64_i32(cpu_T[0], t0);
|
| 5437 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5438 |
tcg_gen_sari_tl(cpu_tmp0, cpu_T[0], 31); |
| 5439 |
tcg_gen_shri_i64(t0, t0, 32);
|
| 5440 |
tcg_gen_trunc_i64_i32(cpu_T[1], t0);
|
| 5441 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[1], cpu_tmp0);
|
| 5442 |
} |
| 5443 |
#endif
|
| 5444 |
} else {
|
| 5445 |
tcg_gen_ext16s_tl(cpu_T[0], cpu_T[0]); |
| 5446 |
tcg_gen_ext16s_tl(cpu_T[1], cpu_T[1]); |
| 5447 |
/* XXX: use 32 bit mul which could be faster */
|
| 5448 |
tcg_gen_mul_tl(cpu_T[0], cpu_T[0], cpu_T[1]); |
| 5449 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 5450 |
tcg_gen_ext16s_tl(cpu_tmp0, cpu_T[0]);
|
| 5451 |
tcg_gen_sub_tl(cpu_cc_src, cpu_T[0], cpu_tmp0);
|
| 5452 |
} |
| 5453 |
gen_op_mov_reg_T0(ot, reg); |
| 5454 |
set_cc_op(s, CC_OP_MULB + ot); |
| 5455 |
break;
|
| 5456 |
case 0x1c0: |
| 5457 |
case 0x1c1: /* xadd Ev, Gv */ |
| 5458 |
if ((b & 1) == 0) |
| 5459 |
ot = OT_BYTE; |
| 5460 |
else
|
| 5461 |
ot = dflag + OT_WORD; |
| 5462 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5463 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5464 |
mod = (modrm >> 6) & 3; |
| 5465 |
if (mod == 3) { |
| 5466 |
rm = (modrm & 7) | REX_B(s);
|
| 5467 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 5468 |
gen_op_mov_TN_reg(ot, 1, rm);
|
| 5469 |
gen_op_addl_T0_T1(); |
| 5470 |
gen_op_mov_reg_T1(ot, reg); |
| 5471 |
gen_op_mov_reg_T0(ot, rm); |
| 5472 |
} else {
|
| 5473 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5474 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 5475 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 5476 |
gen_op_addl_T0_T1(); |
| 5477 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 5478 |
gen_op_mov_reg_T1(ot, reg); |
| 5479 |
} |
| 5480 |
gen_op_update2_cc(); |
| 5481 |
set_cc_op(s, CC_OP_ADDB + ot); |
| 5482 |
break;
|
| 5483 |
case 0x1b0: |
| 5484 |
case 0x1b1: /* cmpxchg Ev, Gv */ |
| 5485 |
{
|
| 5486 |
int label1, label2;
|
| 5487 |
TCGv t0, t1, t2, a0; |
| 5488 |
|
| 5489 |
if ((b & 1) == 0) |
| 5490 |
ot = OT_BYTE; |
| 5491 |
else
|
| 5492 |
ot = dflag + OT_WORD; |
| 5493 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5494 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5495 |
mod = (modrm >> 6) & 3; |
| 5496 |
t0 = tcg_temp_local_new(); |
| 5497 |
t1 = tcg_temp_local_new(); |
| 5498 |
t2 = tcg_temp_local_new(); |
| 5499 |
a0 = tcg_temp_local_new(); |
| 5500 |
gen_op_mov_v_reg(ot, t1, reg); |
| 5501 |
if (mod == 3) { |
| 5502 |
rm = (modrm & 7) | REX_B(s);
|
| 5503 |
gen_op_mov_v_reg(ot, t0, rm); |
| 5504 |
} else {
|
| 5505 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5506 |
tcg_gen_mov_tl(a0, cpu_A0); |
| 5507 |
gen_op_ld_v(ot + s->mem_index, t0, a0); |
| 5508 |
rm = 0; /* avoid warning */ |
| 5509 |
} |
| 5510 |
label1 = gen_new_label(); |
| 5511 |
tcg_gen_mov_tl(t2, cpu_regs[R_EAX]); |
| 5512 |
gen_extu(ot, t0); |
| 5513 |
gen_extu(ot, t2); |
| 5514 |
tcg_gen_brcond_tl(TCG_COND_EQ, t2, t0, label1); |
| 5515 |
label2 = gen_new_label(); |
| 5516 |
if (mod == 3) { |
| 5517 |
gen_op_mov_reg_v(ot, R_EAX, t0); |
| 5518 |
tcg_gen_br(label2); |
| 5519 |
gen_set_label(label1); |
| 5520 |
gen_op_mov_reg_v(ot, rm, t1); |
| 5521 |
} else {
|
| 5522 |
/* perform no-op store cycle like physical cpu; must be
|
| 5523 |
before changing accumulator to ensure idempotency if
|
| 5524 |
the store faults and the instruction is restarted */
|
| 5525 |
gen_op_st_v(ot + s->mem_index, t0, a0); |
| 5526 |
gen_op_mov_reg_v(ot, R_EAX, t0); |
| 5527 |
tcg_gen_br(label2); |
| 5528 |
gen_set_label(label1); |
| 5529 |
gen_op_st_v(ot + s->mem_index, t1, a0); |
| 5530 |
} |
| 5531 |
gen_set_label(label2); |
| 5532 |
tcg_gen_mov_tl(cpu_cc_src, t0); |
| 5533 |
tcg_gen_mov_tl(cpu_cc_srcT, t2); |
| 5534 |
tcg_gen_sub_tl(cpu_cc_dst, t2, t0); |
| 5535 |
set_cc_op(s, CC_OP_SUBB + ot); |
| 5536 |
tcg_temp_free(t0); |
| 5537 |
tcg_temp_free(t1); |
| 5538 |
tcg_temp_free(t2); |
| 5539 |
tcg_temp_free(a0); |
| 5540 |
} |
| 5541 |
break;
|
| 5542 |
case 0x1c7: /* cmpxchg8b */ |
| 5543 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5544 |
mod = (modrm >> 6) & 3; |
| 5545 |
if ((mod == 3) || ((modrm & 0x38) != 0x8)) |
| 5546 |
goto illegal_op;
|
| 5547 |
#ifdef TARGET_X86_64
|
| 5548 |
if (dflag == 2) { |
| 5549 |
if (!(s->cpuid_ext_features & CPUID_EXT_CX16))
|
| 5550 |
goto illegal_op;
|
| 5551 |
gen_jmp_im(pc_start - s->cs_base); |
| 5552 |
gen_update_cc_op(s); |
| 5553 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5554 |
gen_helper_cmpxchg16b(cpu_env, cpu_A0); |
| 5555 |
} else
|
| 5556 |
#endif
|
| 5557 |
{
|
| 5558 |
if (!(s->cpuid_features & CPUID_CX8))
|
| 5559 |
goto illegal_op;
|
| 5560 |
gen_jmp_im(pc_start - s->cs_base); |
| 5561 |
gen_update_cc_op(s); |
| 5562 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5563 |
gen_helper_cmpxchg8b(cpu_env, cpu_A0); |
| 5564 |
} |
| 5565 |
set_cc_op(s, CC_OP_EFLAGS); |
| 5566 |
break;
|
| 5567 |
|
| 5568 |
/**************************/
|
| 5569 |
/* push/pop */
|
| 5570 |
case 0x50 ... 0x57: /* push */ |
| 5571 |
gen_op_mov_TN_reg(OT_LONG, 0, (b & 7) | REX_B(s)); |
| 5572 |
gen_push_T0(s); |
| 5573 |
break;
|
| 5574 |
case 0x58 ... 0x5f: /* pop */ |
| 5575 |
if (CODE64(s)) {
|
| 5576 |
ot = dflag ? OT_QUAD : OT_WORD; |
| 5577 |
} else {
|
| 5578 |
ot = dflag + OT_WORD; |
| 5579 |
} |
| 5580 |
gen_pop_T0(s); |
| 5581 |
/* NOTE: order is important for pop %sp */
|
| 5582 |
gen_pop_update(s); |
| 5583 |
gen_op_mov_reg_T0(ot, (b & 7) | REX_B(s));
|
| 5584 |
break;
|
| 5585 |
case 0x60: /* pusha */ |
| 5586 |
if (CODE64(s))
|
| 5587 |
goto illegal_op;
|
| 5588 |
gen_pusha(s); |
| 5589 |
break;
|
| 5590 |
case 0x61: /* popa */ |
| 5591 |
if (CODE64(s))
|
| 5592 |
goto illegal_op;
|
| 5593 |
gen_popa(s); |
| 5594 |
break;
|
| 5595 |
case 0x68: /* push Iv */ |
| 5596 |
case 0x6a: |
| 5597 |
if (CODE64(s)) {
|
| 5598 |
ot = dflag ? OT_QUAD : OT_WORD; |
| 5599 |
} else {
|
| 5600 |
ot = dflag + OT_WORD; |
| 5601 |
} |
| 5602 |
if (b == 0x68) |
| 5603 |
val = insn_get(env, s, ot); |
| 5604 |
else
|
| 5605 |
val = (int8_t)insn_get(env, s, OT_BYTE); |
| 5606 |
gen_op_movl_T0_im(val); |
| 5607 |
gen_push_T0(s); |
| 5608 |
break;
|
| 5609 |
case 0x8f: /* pop Ev */ |
| 5610 |
if (CODE64(s)) {
|
| 5611 |
ot = dflag ? OT_QUAD : OT_WORD; |
| 5612 |
} else {
|
| 5613 |
ot = dflag + OT_WORD; |
| 5614 |
} |
| 5615 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5616 |
mod = (modrm >> 6) & 3; |
| 5617 |
gen_pop_T0(s); |
| 5618 |
if (mod == 3) { |
| 5619 |
/* NOTE: order is important for pop %sp */
|
| 5620 |
gen_pop_update(s); |
| 5621 |
rm = (modrm & 7) | REX_B(s);
|
| 5622 |
gen_op_mov_reg_T0(ot, rm); |
| 5623 |
} else {
|
| 5624 |
/* NOTE: order is important too for MMU exceptions */
|
| 5625 |
s->popl_esp_hack = 1 << ot;
|
| 5626 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 1);
|
| 5627 |
s->popl_esp_hack = 0;
|
| 5628 |
gen_pop_update(s); |
| 5629 |
} |
| 5630 |
break;
|
| 5631 |
case 0xc8: /* enter */ |
| 5632 |
{
|
| 5633 |
int level;
|
| 5634 |
val = cpu_lduw_code(env, s->pc); |
| 5635 |
s->pc += 2;
|
| 5636 |
level = cpu_ldub_code(env, s->pc++); |
| 5637 |
gen_enter(s, val, level); |
| 5638 |
} |
| 5639 |
break;
|
| 5640 |
case 0xc9: /* leave */ |
| 5641 |
/* XXX: exception not precise (ESP is updated before potential exception) */
|
| 5642 |
if (CODE64(s)) {
|
| 5643 |
gen_op_mov_TN_reg(OT_QUAD, 0, R_EBP);
|
| 5644 |
gen_op_mov_reg_T0(OT_QUAD, R_ESP); |
| 5645 |
} else if (s->ss32) { |
| 5646 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EBP);
|
| 5647 |
gen_op_mov_reg_T0(OT_LONG, R_ESP); |
| 5648 |
} else {
|
| 5649 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EBP);
|
| 5650 |
gen_op_mov_reg_T0(OT_WORD, R_ESP); |
| 5651 |
} |
| 5652 |
gen_pop_T0(s); |
| 5653 |
if (CODE64(s)) {
|
| 5654 |
ot = dflag ? OT_QUAD : OT_WORD; |
| 5655 |
} else {
|
| 5656 |
ot = dflag + OT_WORD; |
| 5657 |
} |
| 5658 |
gen_op_mov_reg_T0(ot, R_EBP); |
| 5659 |
gen_pop_update(s); |
| 5660 |
break;
|
| 5661 |
case 0x06: /* push es */ |
| 5662 |
case 0x0e: /* push cs */ |
| 5663 |
case 0x16: /* push ss */ |
| 5664 |
case 0x1e: /* push ds */ |
| 5665 |
if (CODE64(s))
|
| 5666 |
goto illegal_op;
|
| 5667 |
gen_op_movl_T0_seg(b >> 3);
|
| 5668 |
gen_push_T0(s); |
| 5669 |
break;
|
| 5670 |
case 0x1a0: /* push fs */ |
| 5671 |
case 0x1a8: /* push gs */ |
| 5672 |
gen_op_movl_T0_seg((b >> 3) & 7); |
| 5673 |
gen_push_T0(s); |
| 5674 |
break;
|
| 5675 |
case 0x07: /* pop es */ |
| 5676 |
case 0x17: /* pop ss */ |
| 5677 |
case 0x1f: /* pop ds */ |
| 5678 |
if (CODE64(s))
|
| 5679 |
goto illegal_op;
|
| 5680 |
reg = b >> 3;
|
| 5681 |
gen_pop_T0(s); |
| 5682 |
gen_movl_seg_T0(s, reg, pc_start - s->cs_base); |
| 5683 |
gen_pop_update(s); |
| 5684 |
if (reg == R_SS) {
|
| 5685 |
/* if reg == SS, inhibit interrupts/trace. */
|
| 5686 |
/* If several instructions disable interrupts, only the
|
| 5687 |
_first_ does it */
|
| 5688 |
if (!(s->tb->flags & HF_INHIBIT_IRQ_MASK))
|
| 5689 |
gen_helper_set_inhibit_irq(cpu_env); |
| 5690 |
s->tf = 0;
|
| 5691 |
} |
| 5692 |
if (s->is_jmp) {
|
| 5693 |
gen_jmp_im(s->pc - s->cs_base); |
| 5694 |
gen_eob(s); |
| 5695 |
} |
| 5696 |
break;
|
| 5697 |
case 0x1a1: /* pop fs */ |
| 5698 |
case 0x1a9: /* pop gs */ |
| 5699 |
gen_pop_T0(s); |
| 5700 |
gen_movl_seg_T0(s, (b >> 3) & 7, pc_start - s->cs_base); |
| 5701 |
gen_pop_update(s); |
| 5702 |
if (s->is_jmp) {
|
| 5703 |
gen_jmp_im(s->pc - s->cs_base); |
| 5704 |
gen_eob(s); |
| 5705 |
} |
| 5706 |
break;
|
| 5707 |
|
| 5708 |
/**************************/
|
| 5709 |
/* mov */
|
| 5710 |
case 0x88: |
| 5711 |
case 0x89: /* mov Gv, Ev */ |
| 5712 |
if ((b & 1) == 0) |
| 5713 |
ot = OT_BYTE; |
| 5714 |
else
|
| 5715 |
ot = dflag + OT_WORD; |
| 5716 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5717 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5718 |
|
| 5719 |
/* generate a generic store */
|
| 5720 |
gen_ldst_modrm(env, s, modrm, ot, reg, 1);
|
| 5721 |
break;
|
| 5722 |
case 0xc6: |
| 5723 |
case 0xc7: /* mov Ev, Iv */ |
| 5724 |
if ((b & 1) == 0) |
| 5725 |
ot = OT_BYTE; |
| 5726 |
else
|
| 5727 |
ot = dflag + OT_WORD; |
| 5728 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5729 |
mod = (modrm >> 6) & 3; |
| 5730 |
if (mod != 3) { |
| 5731 |
s->rip_offset = insn_const_size(ot); |
| 5732 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5733 |
} |
| 5734 |
val = insn_get(env, s, ot); |
| 5735 |
gen_op_movl_T0_im(val); |
| 5736 |
if (mod != 3) |
| 5737 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 5738 |
else
|
| 5739 |
gen_op_mov_reg_T0(ot, (modrm & 7) | REX_B(s));
|
| 5740 |
break;
|
| 5741 |
case 0x8a: |
| 5742 |
case 0x8b: /* mov Ev, Gv */ |
| 5743 |
if ((b & 1) == 0) |
| 5744 |
ot = OT_BYTE; |
| 5745 |
else
|
| 5746 |
ot = OT_WORD + dflag; |
| 5747 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5748 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5749 |
|
| 5750 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 5751 |
gen_op_mov_reg_T0(ot, reg); |
| 5752 |
break;
|
| 5753 |
case 0x8e: /* mov seg, Gv */ |
| 5754 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5755 |
reg = (modrm >> 3) & 7; |
| 5756 |
if (reg >= 6 || reg == R_CS) |
| 5757 |
goto illegal_op;
|
| 5758 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 5759 |
gen_movl_seg_T0(s, reg, pc_start - s->cs_base); |
| 5760 |
if (reg == R_SS) {
|
| 5761 |
/* if reg == SS, inhibit interrupts/trace */
|
| 5762 |
/* If several instructions disable interrupts, only the
|
| 5763 |
_first_ does it */
|
| 5764 |
if (!(s->tb->flags & HF_INHIBIT_IRQ_MASK))
|
| 5765 |
gen_helper_set_inhibit_irq(cpu_env); |
| 5766 |
s->tf = 0;
|
| 5767 |
} |
| 5768 |
if (s->is_jmp) {
|
| 5769 |
gen_jmp_im(s->pc - s->cs_base); |
| 5770 |
gen_eob(s); |
| 5771 |
} |
| 5772 |
break;
|
| 5773 |
case 0x8c: /* mov Gv, seg */ |
| 5774 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5775 |
reg = (modrm >> 3) & 7; |
| 5776 |
mod = (modrm >> 6) & 3; |
| 5777 |
if (reg >= 6) |
| 5778 |
goto illegal_op;
|
| 5779 |
gen_op_movl_T0_seg(reg); |
| 5780 |
if (mod == 3) |
| 5781 |
ot = OT_WORD + dflag; |
| 5782 |
else
|
| 5783 |
ot = OT_WORD; |
| 5784 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 1);
|
| 5785 |
break;
|
| 5786 |
|
| 5787 |
case 0x1b6: /* movzbS Gv, Eb */ |
| 5788 |
case 0x1b7: /* movzwS Gv, Eb */ |
| 5789 |
case 0x1be: /* movsbS Gv, Eb */ |
| 5790 |
case 0x1bf: /* movswS Gv, Eb */ |
| 5791 |
{
|
| 5792 |
int d_ot;
|
| 5793 |
/* d_ot is the size of destination */
|
| 5794 |
d_ot = dflag + OT_WORD; |
| 5795 |
/* ot is the size of source */
|
| 5796 |
ot = (b & 1) + OT_BYTE;
|
| 5797 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5798 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5799 |
mod = (modrm >> 6) & 3; |
| 5800 |
rm = (modrm & 7) | REX_B(s);
|
| 5801 |
|
| 5802 |
if (mod == 3) { |
| 5803 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 5804 |
switch(ot | (b & 8)) { |
| 5805 |
case OT_BYTE:
|
| 5806 |
tcg_gen_ext8u_tl(cpu_T[0], cpu_T[0]); |
| 5807 |
break;
|
| 5808 |
case OT_BYTE | 8: |
| 5809 |
tcg_gen_ext8s_tl(cpu_T[0], cpu_T[0]); |
| 5810 |
break;
|
| 5811 |
case OT_WORD:
|
| 5812 |
tcg_gen_ext16u_tl(cpu_T[0], cpu_T[0]); |
| 5813 |
break;
|
| 5814 |
default:
|
| 5815 |
case OT_WORD | 8: |
| 5816 |
tcg_gen_ext16s_tl(cpu_T[0], cpu_T[0]); |
| 5817 |
break;
|
| 5818 |
} |
| 5819 |
gen_op_mov_reg_T0(d_ot, reg); |
| 5820 |
} else {
|
| 5821 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5822 |
if (b & 8) { |
| 5823 |
gen_op_lds_T0_A0(ot + s->mem_index); |
| 5824 |
} else {
|
| 5825 |
gen_op_ldu_T0_A0(ot + s->mem_index); |
| 5826 |
} |
| 5827 |
gen_op_mov_reg_T0(d_ot, reg); |
| 5828 |
} |
| 5829 |
} |
| 5830 |
break;
|
| 5831 |
|
| 5832 |
case 0x8d: /* lea */ |
| 5833 |
ot = dflag + OT_WORD; |
| 5834 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5835 |
mod = (modrm >> 6) & 3; |
| 5836 |
if (mod == 3) |
| 5837 |
goto illegal_op;
|
| 5838 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5839 |
/* we must ensure that no segment is added */
|
| 5840 |
s->override = -1;
|
| 5841 |
val = s->addseg; |
| 5842 |
s->addseg = 0;
|
| 5843 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5844 |
s->addseg = val; |
| 5845 |
gen_op_mov_reg_A0(ot - OT_WORD, reg); |
| 5846 |
break;
|
| 5847 |
|
| 5848 |
case 0xa0: /* mov EAX, Ov */ |
| 5849 |
case 0xa1: |
| 5850 |
case 0xa2: /* mov Ov, EAX */ |
| 5851 |
case 0xa3: |
| 5852 |
{
|
| 5853 |
target_ulong offset_addr; |
| 5854 |
|
| 5855 |
if ((b & 1) == 0) |
| 5856 |
ot = OT_BYTE; |
| 5857 |
else
|
| 5858 |
ot = dflag + OT_WORD; |
| 5859 |
#ifdef TARGET_X86_64
|
| 5860 |
if (s->aflag == 2) { |
| 5861 |
offset_addr = cpu_ldq_code(env, s->pc); |
| 5862 |
s->pc += 8;
|
| 5863 |
gen_op_movq_A0_im(offset_addr); |
| 5864 |
} else
|
| 5865 |
#endif
|
| 5866 |
{
|
| 5867 |
if (s->aflag) {
|
| 5868 |
offset_addr = insn_get(env, s, OT_LONG); |
| 5869 |
} else {
|
| 5870 |
offset_addr = insn_get(env, s, OT_WORD); |
| 5871 |
} |
| 5872 |
gen_op_movl_A0_im(offset_addr); |
| 5873 |
} |
| 5874 |
gen_add_A0_ds_seg(s); |
| 5875 |
if ((b & 2) == 0) { |
| 5876 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 5877 |
gen_op_mov_reg_T0(ot, R_EAX); |
| 5878 |
} else {
|
| 5879 |
gen_op_mov_TN_reg(ot, 0, R_EAX);
|
| 5880 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 5881 |
} |
| 5882 |
} |
| 5883 |
break;
|
| 5884 |
case 0xd7: /* xlat */ |
| 5885 |
#ifdef TARGET_X86_64
|
| 5886 |
if (s->aflag == 2) { |
| 5887 |
gen_op_movq_A0_reg(R_EBX); |
| 5888 |
gen_op_mov_TN_reg(OT_QUAD, 0, R_EAX);
|
| 5889 |
tcg_gen_andi_tl(cpu_T[0], cpu_T[0], 0xff); |
| 5890 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_T[0]);
|
| 5891 |
} else
|
| 5892 |
#endif
|
| 5893 |
{
|
| 5894 |
gen_op_movl_A0_reg(R_EBX); |
| 5895 |
gen_op_mov_TN_reg(OT_LONG, 0, R_EAX);
|
| 5896 |
tcg_gen_andi_tl(cpu_T[0], cpu_T[0], 0xff); |
| 5897 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_T[0]);
|
| 5898 |
if (s->aflag == 0) |
| 5899 |
gen_op_andl_A0_ffff(); |
| 5900 |
else
|
| 5901 |
tcg_gen_andi_tl(cpu_A0, cpu_A0, 0xffffffff);
|
| 5902 |
} |
| 5903 |
gen_add_A0_ds_seg(s); |
| 5904 |
gen_op_ldu_T0_A0(OT_BYTE + s->mem_index); |
| 5905 |
gen_op_mov_reg_T0(OT_BYTE, R_EAX); |
| 5906 |
break;
|
| 5907 |
case 0xb0 ... 0xb7: /* mov R, Ib */ |
| 5908 |
val = insn_get(env, s, OT_BYTE); |
| 5909 |
gen_op_movl_T0_im(val); |
| 5910 |
gen_op_mov_reg_T0(OT_BYTE, (b & 7) | REX_B(s));
|
| 5911 |
break;
|
| 5912 |
case 0xb8 ... 0xbf: /* mov R, Iv */ |
| 5913 |
#ifdef TARGET_X86_64
|
| 5914 |
if (dflag == 2) { |
| 5915 |
uint64_t tmp; |
| 5916 |
/* 64 bit case */
|
| 5917 |
tmp = cpu_ldq_code(env, s->pc); |
| 5918 |
s->pc += 8;
|
| 5919 |
reg = (b & 7) | REX_B(s);
|
| 5920 |
gen_movtl_T0_im(tmp); |
| 5921 |
gen_op_mov_reg_T0(OT_QUAD, reg); |
| 5922 |
} else
|
| 5923 |
#endif
|
| 5924 |
{
|
| 5925 |
ot = dflag ? OT_LONG : OT_WORD; |
| 5926 |
val = insn_get(env, s, ot); |
| 5927 |
reg = (b & 7) | REX_B(s);
|
| 5928 |
gen_op_movl_T0_im(val); |
| 5929 |
gen_op_mov_reg_T0(ot, reg); |
| 5930 |
} |
| 5931 |
break;
|
| 5932 |
|
| 5933 |
case 0x91 ... 0x97: /* xchg R, EAX */ |
| 5934 |
do_xchg_reg_eax: |
| 5935 |
ot = dflag + OT_WORD; |
| 5936 |
reg = (b & 7) | REX_B(s);
|
| 5937 |
rm = R_EAX; |
| 5938 |
goto do_xchg_reg;
|
| 5939 |
case 0x86: |
| 5940 |
case 0x87: /* xchg Ev, Gv */ |
| 5941 |
if ((b & 1) == 0) |
| 5942 |
ot = OT_BYTE; |
| 5943 |
else
|
| 5944 |
ot = dflag + OT_WORD; |
| 5945 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5946 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5947 |
mod = (modrm >> 6) & 3; |
| 5948 |
if (mod == 3) { |
| 5949 |
rm = (modrm & 7) | REX_B(s);
|
| 5950 |
do_xchg_reg:
|
| 5951 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 5952 |
gen_op_mov_TN_reg(ot, 1, rm);
|
| 5953 |
gen_op_mov_reg_T0(ot, rm); |
| 5954 |
gen_op_mov_reg_T1(ot, reg); |
| 5955 |
} else {
|
| 5956 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5957 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 5958 |
/* for xchg, lock is implicit */
|
| 5959 |
if (!(prefixes & PREFIX_LOCK))
|
| 5960 |
gen_helper_lock(); |
| 5961 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 5962 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 5963 |
if (!(prefixes & PREFIX_LOCK))
|
| 5964 |
gen_helper_unlock(); |
| 5965 |
gen_op_mov_reg_T1(ot, reg); |
| 5966 |
} |
| 5967 |
break;
|
| 5968 |
case 0xc4: /* les Gv */ |
| 5969 |
/* In CODE64 this is VEX3; see above. */
|
| 5970 |
op = R_ES; |
| 5971 |
goto do_lxx;
|
| 5972 |
case 0xc5: /* lds Gv */ |
| 5973 |
/* In CODE64 this is VEX2; see above. */
|
| 5974 |
op = R_DS; |
| 5975 |
goto do_lxx;
|
| 5976 |
case 0x1b2: /* lss Gv */ |
| 5977 |
op = R_SS; |
| 5978 |
goto do_lxx;
|
| 5979 |
case 0x1b4: /* lfs Gv */ |
| 5980 |
op = R_FS; |
| 5981 |
goto do_lxx;
|
| 5982 |
case 0x1b5: /* lgs Gv */ |
| 5983 |
op = R_GS; |
| 5984 |
do_lxx:
|
| 5985 |
ot = dflag ? OT_LONG : OT_WORD; |
| 5986 |
modrm = cpu_ldub_code(env, s->pc++); |
| 5987 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 5988 |
mod = (modrm >> 6) & 3; |
| 5989 |
if (mod == 3) |
| 5990 |
goto illegal_op;
|
| 5991 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 5992 |
gen_op_ld_T1_A0(ot + s->mem_index); |
| 5993 |
gen_add_A0_im(s, 1 << (ot - OT_WORD + 1)); |
| 5994 |
/* load the segment first to handle exceptions properly */
|
| 5995 |
gen_op_ldu_T0_A0(OT_WORD + s->mem_index); |
| 5996 |
gen_movl_seg_T0(s, op, pc_start - s->cs_base); |
| 5997 |
/* then put the data */
|
| 5998 |
gen_op_mov_reg_T1(ot, reg); |
| 5999 |
if (s->is_jmp) {
|
| 6000 |
gen_jmp_im(s->pc - s->cs_base); |
| 6001 |
gen_eob(s); |
| 6002 |
} |
| 6003 |
break;
|
| 6004 |
|
| 6005 |
/************************/
|
| 6006 |
/* shifts */
|
| 6007 |
case 0xc0: |
| 6008 |
case 0xc1: |
| 6009 |
/* shift Ev,Ib */
|
| 6010 |
shift = 2;
|
| 6011 |
grp2:
|
| 6012 |
{
|
| 6013 |
if ((b & 1) == 0) |
| 6014 |
ot = OT_BYTE; |
| 6015 |
else
|
| 6016 |
ot = dflag + OT_WORD; |
| 6017 |
|
| 6018 |
modrm = cpu_ldub_code(env, s->pc++); |
| 6019 |
mod = (modrm >> 6) & 3; |
| 6020 |
op = (modrm >> 3) & 7; |
| 6021 |
|
| 6022 |
if (mod != 3) { |
| 6023 |
if (shift == 2) { |
| 6024 |
s->rip_offset = 1;
|
| 6025 |
} |
| 6026 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 6027 |
opreg = OR_TMP0; |
| 6028 |
} else {
|
| 6029 |
opreg = (modrm & 7) | REX_B(s);
|
| 6030 |
} |
| 6031 |
|
| 6032 |
/* simpler op */
|
| 6033 |
if (shift == 0) { |
| 6034 |
gen_shift(s, op, ot, opreg, OR_ECX); |
| 6035 |
} else {
|
| 6036 |
if (shift == 2) { |
| 6037 |
shift = cpu_ldub_code(env, s->pc++); |
| 6038 |
} |
| 6039 |
gen_shifti(s, op, ot, opreg, shift); |
| 6040 |
} |
| 6041 |
} |
| 6042 |
break;
|
| 6043 |
case 0xd0: |
| 6044 |
case 0xd1: |
| 6045 |
/* shift Ev,1 */
|
| 6046 |
shift = 1;
|
| 6047 |
goto grp2;
|
| 6048 |
case 0xd2: |
| 6049 |
case 0xd3: |
| 6050 |
/* shift Ev,cl */
|
| 6051 |
shift = 0;
|
| 6052 |
goto grp2;
|
| 6053 |
|
| 6054 |
case 0x1a4: /* shld imm */ |
| 6055 |
op = 0;
|
| 6056 |
shift = 1;
|
| 6057 |
goto do_shiftd;
|
| 6058 |
case 0x1a5: /* shld cl */ |
| 6059 |
op = 0;
|
| 6060 |
shift = 0;
|
| 6061 |
goto do_shiftd;
|
| 6062 |
case 0x1ac: /* shrd imm */ |
| 6063 |
op = 1;
|
| 6064 |
shift = 1;
|
| 6065 |
goto do_shiftd;
|
| 6066 |
case 0x1ad: /* shrd cl */ |
| 6067 |
op = 1;
|
| 6068 |
shift = 0;
|
| 6069 |
do_shiftd:
|
| 6070 |
ot = dflag + OT_WORD; |
| 6071 |
modrm = cpu_ldub_code(env, s->pc++); |
| 6072 |
mod = (modrm >> 6) & 3; |
| 6073 |
rm = (modrm & 7) | REX_B(s);
|
| 6074 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 6075 |
if (mod != 3) { |
| 6076 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 6077 |
opreg = OR_TMP0; |
| 6078 |
} else {
|
| 6079 |
opreg = rm; |
| 6080 |
} |
| 6081 |
gen_op_mov_TN_reg(ot, 1, reg);
|
| 6082 |
|
| 6083 |
if (shift) {
|
| 6084 |
TCGv imm = tcg_const_tl(cpu_ldub_code(env, s->pc++)); |
| 6085 |
gen_shiftd_rm_T1(s, ot, opreg, op, imm); |
| 6086 |
tcg_temp_free(imm); |
| 6087 |
} else {
|
| 6088 |
gen_shiftd_rm_T1(s, ot, opreg, op, cpu_regs[R_ECX]); |
| 6089 |
} |
| 6090 |
break;
|
| 6091 |
|
| 6092 |
/************************/
|
| 6093 |
/* floats */
|
| 6094 |
case 0xd8 ... 0xdf: |
| 6095 |
if (s->flags & (HF_EM_MASK | HF_TS_MASK)) {
|
| 6096 |
/* if CR0.EM or CR0.TS are set, generate an FPU exception */
|
| 6097 |
/* XXX: what to do if illegal op ? */
|
| 6098 |
gen_exception(s, EXCP07_PREX, pc_start - s->cs_base); |
| 6099 |
break;
|
| 6100 |
} |
| 6101 |
modrm = cpu_ldub_code(env, s->pc++); |
| 6102 |
mod = (modrm >> 6) & 3; |
| 6103 |
rm = modrm & 7;
|
| 6104 |
op = ((b & 7) << 3) | ((modrm >> 3) & 7); |
| 6105 |
if (mod != 3) { |
| 6106 |
/* memory op */
|
| 6107 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 6108 |
switch(op) {
|
| 6109 |
case 0x00 ... 0x07: /* fxxxs */ |
| 6110 |
case 0x10 ... 0x17: /* fixxxl */ |
| 6111 |
case 0x20 ... 0x27: /* fxxxl */ |
| 6112 |
case 0x30 ... 0x37: /* fixxx */ |
| 6113 |
{
|
| 6114 |
int op1;
|
| 6115 |
op1 = op & 7;
|
| 6116 |
|
| 6117 |
switch(op >> 4) { |
| 6118 |
case 0: |
| 6119 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 6120 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6121 |
gen_helper_flds_FT0(cpu_env, cpu_tmp2_i32); |
| 6122 |
break;
|
| 6123 |
case 1: |
| 6124 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 6125 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6126 |
gen_helper_fildl_FT0(cpu_env, cpu_tmp2_i32); |
| 6127 |
break;
|
| 6128 |
case 2: |
| 6129 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_A0, |
| 6130 |
(s->mem_index >> 2) - 1); |
| 6131 |
gen_helper_fldl_FT0(cpu_env, cpu_tmp1_i64); |
| 6132 |
break;
|
| 6133 |
case 3: |
| 6134 |
default:
|
| 6135 |
gen_op_lds_T0_A0(OT_WORD + s->mem_index); |
| 6136 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6137 |
gen_helper_fildl_FT0(cpu_env, cpu_tmp2_i32); |
| 6138 |
break;
|
| 6139 |
} |
| 6140 |
|
| 6141 |
gen_helper_fp_arith_ST0_FT0(op1); |
| 6142 |
if (op1 == 3) { |
| 6143 |
/* fcomp needs pop */
|
| 6144 |
gen_helper_fpop(cpu_env); |
| 6145 |
} |
| 6146 |
} |
| 6147 |
break;
|
| 6148 |
case 0x08: /* flds */ |
| 6149 |
case 0x0a: /* fsts */ |
| 6150 |
case 0x0b: /* fstps */ |
| 6151 |
case 0x18 ... 0x1b: /* fildl, fisttpl, fistl, fistpl */ |
| 6152 |
case 0x28 ... 0x2b: /* fldl, fisttpll, fstl, fstpl */ |
| 6153 |
case 0x38 ... 0x3b: /* filds, fisttps, fists, fistps */ |
| 6154 |
switch(op & 7) { |
| 6155 |
case 0: |
| 6156 |
switch(op >> 4) { |
| 6157 |
case 0: |
| 6158 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 6159 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6160 |
gen_helper_flds_ST0(cpu_env, cpu_tmp2_i32); |
| 6161 |
break;
|
| 6162 |
case 1: |
| 6163 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 6164 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6165 |
gen_helper_fildl_ST0(cpu_env, cpu_tmp2_i32); |
| 6166 |
break;
|
| 6167 |
case 2: |
| 6168 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_A0, |
| 6169 |
(s->mem_index >> 2) - 1); |
| 6170 |
gen_helper_fldl_ST0(cpu_env, cpu_tmp1_i64); |
| 6171 |
break;
|
| 6172 |
case 3: |
| 6173 |
default:
|
| 6174 |
gen_op_lds_T0_A0(OT_WORD + s->mem_index); |
| 6175 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6176 |
gen_helper_fildl_ST0(cpu_env, cpu_tmp2_i32); |
| 6177 |
break;
|
| 6178 |
} |
| 6179 |
break;
|
| 6180 |
case 1: |
| 6181 |
/* XXX: the corresponding CPUID bit must be tested ! */
|
| 6182 |
switch(op >> 4) { |
| 6183 |
case 1: |
| 6184 |
gen_helper_fisttl_ST0(cpu_tmp2_i32, cpu_env); |
| 6185 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6186 |
gen_op_st_T0_A0(OT_LONG + s->mem_index); |
| 6187 |
break;
|
| 6188 |
case 2: |
| 6189 |
gen_helper_fisttll_ST0(cpu_tmp1_i64, cpu_env); |
| 6190 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_A0, |
| 6191 |
(s->mem_index >> 2) - 1); |
| 6192 |
break;
|
| 6193 |
case 3: |
| 6194 |
default:
|
| 6195 |
gen_helper_fistt_ST0(cpu_tmp2_i32, cpu_env); |
| 6196 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6197 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 6198 |
break;
|
| 6199 |
} |
| 6200 |
gen_helper_fpop(cpu_env); |
| 6201 |
break;
|
| 6202 |
default:
|
| 6203 |
switch(op >> 4) { |
| 6204 |
case 0: |
| 6205 |
gen_helper_fsts_ST0(cpu_tmp2_i32, cpu_env); |
| 6206 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6207 |
gen_op_st_T0_A0(OT_LONG + s->mem_index); |
| 6208 |
break;
|
| 6209 |
case 1: |
| 6210 |
gen_helper_fistl_ST0(cpu_tmp2_i32, cpu_env); |
| 6211 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6212 |
gen_op_st_T0_A0(OT_LONG + s->mem_index); |
| 6213 |
break;
|
| 6214 |
case 2: |
| 6215 |
gen_helper_fstl_ST0(cpu_tmp1_i64, cpu_env); |
| 6216 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_A0, |
| 6217 |
(s->mem_index >> 2) - 1); |
| 6218 |
break;
|
| 6219 |
case 3: |
| 6220 |
default:
|
| 6221 |
gen_helper_fist_ST0(cpu_tmp2_i32, cpu_env); |
| 6222 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6223 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 6224 |
break;
|
| 6225 |
} |
| 6226 |
if ((op & 7) == 3) |
| 6227 |
gen_helper_fpop(cpu_env); |
| 6228 |
break;
|
| 6229 |
} |
| 6230 |
break;
|
| 6231 |
case 0x0c: /* fldenv mem */ |
| 6232 |
gen_update_cc_op(s); |
| 6233 |
gen_jmp_im(pc_start - s->cs_base); |
| 6234 |
gen_helper_fldenv(cpu_env, cpu_A0, tcg_const_i32(s->dflag)); |
| 6235 |
break;
|
| 6236 |
case 0x0d: /* fldcw mem */ |
| 6237 |
gen_op_ld_T0_A0(OT_WORD + s->mem_index); |
| 6238 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6239 |
gen_helper_fldcw(cpu_env, cpu_tmp2_i32); |
| 6240 |
break;
|
| 6241 |
case 0x0e: /* fnstenv mem */ |
| 6242 |
gen_update_cc_op(s); |
| 6243 |
gen_jmp_im(pc_start - s->cs_base); |
| 6244 |
gen_helper_fstenv(cpu_env, cpu_A0, tcg_const_i32(s->dflag)); |
| 6245 |
break;
|
| 6246 |
case 0x0f: /* fnstcw mem */ |
| 6247 |
gen_helper_fnstcw(cpu_tmp2_i32, cpu_env); |
| 6248 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6249 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 6250 |
break;
|
| 6251 |
case 0x1d: /* fldt mem */ |
| 6252 |
gen_update_cc_op(s); |
| 6253 |
gen_jmp_im(pc_start - s->cs_base); |
| 6254 |
gen_helper_fldt_ST0(cpu_env, cpu_A0); |
| 6255 |
break;
|
| 6256 |
case 0x1f: /* fstpt mem */ |
| 6257 |
gen_update_cc_op(s); |
| 6258 |
gen_jmp_im(pc_start - s->cs_base); |
| 6259 |
gen_helper_fstt_ST0(cpu_env, cpu_A0); |
| 6260 |
gen_helper_fpop(cpu_env); |
| 6261 |
break;
|
| 6262 |
case 0x2c: /* frstor mem */ |
| 6263 |
gen_update_cc_op(s); |
| 6264 |
gen_jmp_im(pc_start - s->cs_base); |
| 6265 |
gen_helper_frstor(cpu_env, cpu_A0, tcg_const_i32(s->dflag)); |
| 6266 |
break;
|
| 6267 |
case 0x2e: /* fnsave mem */ |
| 6268 |
gen_update_cc_op(s); |
| 6269 |
gen_jmp_im(pc_start - s->cs_base); |
| 6270 |
gen_helper_fsave(cpu_env, cpu_A0, tcg_const_i32(s->dflag)); |
| 6271 |
break;
|
| 6272 |
case 0x2f: /* fnstsw mem */ |
| 6273 |
gen_helper_fnstsw(cpu_tmp2_i32, cpu_env); |
| 6274 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6275 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 6276 |
break;
|
| 6277 |
case 0x3c: /* fbld */ |
| 6278 |
gen_update_cc_op(s); |
| 6279 |
gen_jmp_im(pc_start - s->cs_base); |
| 6280 |
gen_helper_fbld_ST0(cpu_env, cpu_A0); |
| 6281 |
break;
|
| 6282 |
case 0x3e: /* fbstp */ |
| 6283 |
gen_update_cc_op(s); |
| 6284 |
gen_jmp_im(pc_start - s->cs_base); |
| 6285 |
gen_helper_fbst_ST0(cpu_env, cpu_A0); |
| 6286 |
gen_helper_fpop(cpu_env); |
| 6287 |
break;
|
| 6288 |
case 0x3d: /* fildll */ |
| 6289 |
tcg_gen_qemu_ld64(cpu_tmp1_i64, cpu_A0, |
| 6290 |
(s->mem_index >> 2) - 1); |
| 6291 |
gen_helper_fildll_ST0(cpu_env, cpu_tmp1_i64); |
| 6292 |
break;
|
| 6293 |
case 0x3f: /* fistpll */ |
| 6294 |
gen_helper_fistll_ST0(cpu_tmp1_i64, cpu_env); |
| 6295 |
tcg_gen_qemu_st64(cpu_tmp1_i64, cpu_A0, |
| 6296 |
(s->mem_index >> 2) - 1); |
| 6297 |
gen_helper_fpop(cpu_env); |
| 6298 |
break;
|
| 6299 |
default:
|
| 6300 |
goto illegal_op;
|
| 6301 |
} |
| 6302 |
} else {
|
| 6303 |
/* register float ops */
|
| 6304 |
opreg = rm; |
| 6305 |
|
| 6306 |
switch(op) {
|
| 6307 |
case 0x08: /* fld sti */ |
| 6308 |
gen_helper_fpush(cpu_env); |
| 6309 |
gen_helper_fmov_ST0_STN(cpu_env, |
| 6310 |
tcg_const_i32((opreg + 1) & 7)); |
| 6311 |
break;
|
| 6312 |
case 0x09: /* fxchg sti */ |
| 6313 |
case 0x29: /* fxchg4 sti, undocumented op */ |
| 6314 |
case 0x39: /* fxchg7 sti, undocumented op */ |
| 6315 |
gen_helper_fxchg_ST0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6316 |
break;
|
| 6317 |
case 0x0a: /* grp d9/2 */ |
| 6318 |
switch(rm) {
|
| 6319 |
case 0: /* fnop */ |
| 6320 |
/* check exceptions (FreeBSD FPU probe) */
|
| 6321 |
gen_update_cc_op(s); |
| 6322 |
gen_jmp_im(pc_start - s->cs_base); |
| 6323 |
gen_helper_fwait(cpu_env); |
| 6324 |
break;
|
| 6325 |
default:
|
| 6326 |
goto illegal_op;
|
| 6327 |
} |
| 6328 |
break;
|
| 6329 |
case 0x0c: /* grp d9/4 */ |
| 6330 |
switch(rm) {
|
| 6331 |
case 0: /* fchs */ |
| 6332 |
gen_helper_fchs_ST0(cpu_env); |
| 6333 |
break;
|
| 6334 |
case 1: /* fabs */ |
| 6335 |
gen_helper_fabs_ST0(cpu_env); |
| 6336 |
break;
|
| 6337 |
case 4: /* ftst */ |
| 6338 |
gen_helper_fldz_FT0(cpu_env); |
| 6339 |
gen_helper_fcom_ST0_FT0(cpu_env); |
| 6340 |
break;
|
| 6341 |
case 5: /* fxam */ |
| 6342 |
gen_helper_fxam_ST0(cpu_env); |
| 6343 |
break;
|
| 6344 |
default:
|
| 6345 |
goto illegal_op;
|
| 6346 |
} |
| 6347 |
break;
|
| 6348 |
case 0x0d: /* grp d9/5 */ |
| 6349 |
{
|
| 6350 |
switch(rm) {
|
| 6351 |
case 0: |
| 6352 |
gen_helper_fpush(cpu_env); |
| 6353 |
gen_helper_fld1_ST0(cpu_env); |
| 6354 |
break;
|
| 6355 |
case 1: |
| 6356 |
gen_helper_fpush(cpu_env); |
| 6357 |
gen_helper_fldl2t_ST0(cpu_env); |
| 6358 |
break;
|
| 6359 |
case 2: |
| 6360 |
gen_helper_fpush(cpu_env); |
| 6361 |
gen_helper_fldl2e_ST0(cpu_env); |
| 6362 |
break;
|
| 6363 |
case 3: |
| 6364 |
gen_helper_fpush(cpu_env); |
| 6365 |
gen_helper_fldpi_ST0(cpu_env); |
| 6366 |
break;
|
| 6367 |
case 4: |
| 6368 |
gen_helper_fpush(cpu_env); |
| 6369 |
gen_helper_fldlg2_ST0(cpu_env); |
| 6370 |
break;
|
| 6371 |
case 5: |
| 6372 |
gen_helper_fpush(cpu_env); |
| 6373 |
gen_helper_fldln2_ST0(cpu_env); |
| 6374 |
break;
|
| 6375 |
case 6: |
| 6376 |
gen_helper_fpush(cpu_env); |
| 6377 |
gen_helper_fldz_ST0(cpu_env); |
| 6378 |
break;
|
| 6379 |
default:
|
| 6380 |
goto illegal_op;
|
| 6381 |
} |
| 6382 |
} |
| 6383 |
break;
|
| 6384 |
case 0x0e: /* grp d9/6 */ |
| 6385 |
switch(rm) {
|
| 6386 |
case 0: /* f2xm1 */ |
| 6387 |
gen_helper_f2xm1(cpu_env); |
| 6388 |
break;
|
| 6389 |
case 1: /* fyl2x */ |
| 6390 |
gen_helper_fyl2x(cpu_env); |
| 6391 |
break;
|
| 6392 |
case 2: /* fptan */ |
| 6393 |
gen_helper_fptan(cpu_env); |
| 6394 |
break;
|
| 6395 |
case 3: /* fpatan */ |
| 6396 |
gen_helper_fpatan(cpu_env); |
| 6397 |
break;
|
| 6398 |
case 4: /* fxtract */ |
| 6399 |
gen_helper_fxtract(cpu_env); |
| 6400 |
break;
|
| 6401 |
case 5: /* fprem1 */ |
| 6402 |
gen_helper_fprem1(cpu_env); |
| 6403 |
break;
|
| 6404 |
case 6: /* fdecstp */ |
| 6405 |
gen_helper_fdecstp(cpu_env); |
| 6406 |
break;
|
| 6407 |
default:
|
| 6408 |
case 7: /* fincstp */ |
| 6409 |
gen_helper_fincstp(cpu_env); |
| 6410 |
break;
|
| 6411 |
} |
| 6412 |
break;
|
| 6413 |
case 0x0f: /* grp d9/7 */ |
| 6414 |
switch(rm) {
|
| 6415 |
case 0: /* fprem */ |
| 6416 |
gen_helper_fprem(cpu_env); |
| 6417 |
break;
|
| 6418 |
case 1: /* fyl2xp1 */ |
| 6419 |
gen_helper_fyl2xp1(cpu_env); |
| 6420 |
break;
|
| 6421 |
case 2: /* fsqrt */ |
| 6422 |
gen_helper_fsqrt(cpu_env); |
| 6423 |
break;
|
| 6424 |
case 3: /* fsincos */ |
| 6425 |
gen_helper_fsincos(cpu_env); |
| 6426 |
break;
|
| 6427 |
case 5: /* fscale */ |
| 6428 |
gen_helper_fscale(cpu_env); |
| 6429 |
break;
|
| 6430 |
case 4: /* frndint */ |
| 6431 |
gen_helper_frndint(cpu_env); |
| 6432 |
break;
|
| 6433 |
case 6: /* fsin */ |
| 6434 |
gen_helper_fsin(cpu_env); |
| 6435 |
break;
|
| 6436 |
default:
|
| 6437 |
case 7: /* fcos */ |
| 6438 |
gen_helper_fcos(cpu_env); |
| 6439 |
break;
|
| 6440 |
} |
| 6441 |
break;
|
| 6442 |
case 0x00: case 0x01: case 0x04 ... 0x07: /* fxxx st, sti */ |
| 6443 |
case 0x20: case 0x21: case 0x24 ... 0x27: /* fxxx sti, st */ |
| 6444 |
case 0x30: case 0x31: case 0x34 ... 0x37: /* fxxxp sti, st */ |
| 6445 |
{
|
| 6446 |
int op1;
|
| 6447 |
|
| 6448 |
op1 = op & 7;
|
| 6449 |
if (op >= 0x20) { |
| 6450 |
gen_helper_fp_arith_STN_ST0(op1, opreg); |
| 6451 |
if (op >= 0x30) |
| 6452 |
gen_helper_fpop(cpu_env); |
| 6453 |
} else {
|
| 6454 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6455 |
gen_helper_fp_arith_ST0_FT0(op1); |
| 6456 |
} |
| 6457 |
} |
| 6458 |
break;
|
| 6459 |
case 0x02: /* fcom */ |
| 6460 |
case 0x22: /* fcom2, undocumented op */ |
| 6461 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6462 |
gen_helper_fcom_ST0_FT0(cpu_env); |
| 6463 |
break;
|
| 6464 |
case 0x03: /* fcomp */ |
| 6465 |
case 0x23: /* fcomp3, undocumented op */ |
| 6466 |
case 0x32: /* fcomp5, undocumented op */ |
| 6467 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6468 |
gen_helper_fcom_ST0_FT0(cpu_env); |
| 6469 |
gen_helper_fpop(cpu_env); |
| 6470 |
break;
|
| 6471 |
case 0x15: /* da/5 */ |
| 6472 |
switch(rm) {
|
| 6473 |
case 1: /* fucompp */ |
| 6474 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(1));
|
| 6475 |
gen_helper_fucom_ST0_FT0(cpu_env); |
| 6476 |
gen_helper_fpop(cpu_env); |
| 6477 |
gen_helper_fpop(cpu_env); |
| 6478 |
break;
|
| 6479 |
default:
|
| 6480 |
goto illegal_op;
|
| 6481 |
} |
| 6482 |
break;
|
| 6483 |
case 0x1c: |
| 6484 |
switch(rm) {
|
| 6485 |
case 0: /* feni (287 only, just do nop here) */ |
| 6486 |
break;
|
| 6487 |
case 1: /* fdisi (287 only, just do nop here) */ |
| 6488 |
break;
|
| 6489 |
case 2: /* fclex */ |
| 6490 |
gen_helper_fclex(cpu_env); |
| 6491 |
break;
|
| 6492 |
case 3: /* fninit */ |
| 6493 |
gen_helper_fninit(cpu_env); |
| 6494 |
break;
|
| 6495 |
case 4: /* fsetpm (287 only, just do nop here) */ |
| 6496 |
break;
|
| 6497 |
default:
|
| 6498 |
goto illegal_op;
|
| 6499 |
} |
| 6500 |
break;
|
| 6501 |
case 0x1d: /* fucomi */ |
| 6502 |
gen_update_cc_op(s); |
| 6503 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6504 |
gen_helper_fucomi_ST0_FT0(cpu_env); |
| 6505 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6506 |
break;
|
| 6507 |
case 0x1e: /* fcomi */ |
| 6508 |
gen_update_cc_op(s); |
| 6509 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6510 |
gen_helper_fcomi_ST0_FT0(cpu_env); |
| 6511 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6512 |
break;
|
| 6513 |
case 0x28: /* ffree sti */ |
| 6514 |
gen_helper_ffree_STN(cpu_env, tcg_const_i32(opreg)); |
| 6515 |
break;
|
| 6516 |
case 0x2a: /* fst sti */ |
| 6517 |
gen_helper_fmov_STN_ST0(cpu_env, tcg_const_i32(opreg)); |
| 6518 |
break;
|
| 6519 |
case 0x2b: /* fstp sti */ |
| 6520 |
case 0x0b: /* fstp1 sti, undocumented op */ |
| 6521 |
case 0x3a: /* fstp8 sti, undocumented op */ |
| 6522 |
case 0x3b: /* fstp9 sti, undocumented op */ |
| 6523 |
gen_helper_fmov_STN_ST0(cpu_env, tcg_const_i32(opreg)); |
| 6524 |
gen_helper_fpop(cpu_env); |
| 6525 |
break;
|
| 6526 |
case 0x2c: /* fucom st(i) */ |
| 6527 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6528 |
gen_helper_fucom_ST0_FT0(cpu_env); |
| 6529 |
break;
|
| 6530 |
case 0x2d: /* fucomp st(i) */ |
| 6531 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6532 |
gen_helper_fucom_ST0_FT0(cpu_env); |
| 6533 |
gen_helper_fpop(cpu_env); |
| 6534 |
break;
|
| 6535 |
case 0x33: /* de/3 */ |
| 6536 |
switch(rm) {
|
| 6537 |
case 1: /* fcompp */ |
| 6538 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(1));
|
| 6539 |
gen_helper_fcom_ST0_FT0(cpu_env); |
| 6540 |
gen_helper_fpop(cpu_env); |
| 6541 |
gen_helper_fpop(cpu_env); |
| 6542 |
break;
|
| 6543 |
default:
|
| 6544 |
goto illegal_op;
|
| 6545 |
} |
| 6546 |
break;
|
| 6547 |
case 0x38: /* ffreep sti, undocumented op */ |
| 6548 |
gen_helper_ffree_STN(cpu_env, tcg_const_i32(opreg)); |
| 6549 |
gen_helper_fpop(cpu_env); |
| 6550 |
break;
|
| 6551 |
case 0x3c: /* df/4 */ |
| 6552 |
switch(rm) {
|
| 6553 |
case 0: |
| 6554 |
gen_helper_fnstsw(cpu_tmp2_i32, cpu_env); |
| 6555 |
tcg_gen_extu_i32_tl(cpu_T[0], cpu_tmp2_i32);
|
| 6556 |
gen_op_mov_reg_T0(OT_WORD, R_EAX); |
| 6557 |
break;
|
| 6558 |
default:
|
| 6559 |
goto illegal_op;
|
| 6560 |
} |
| 6561 |
break;
|
| 6562 |
case 0x3d: /* fucomip */ |
| 6563 |
gen_update_cc_op(s); |
| 6564 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6565 |
gen_helper_fucomi_ST0_FT0(cpu_env); |
| 6566 |
gen_helper_fpop(cpu_env); |
| 6567 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6568 |
break;
|
| 6569 |
case 0x3e: /* fcomip */ |
| 6570 |
gen_update_cc_op(s); |
| 6571 |
gen_helper_fmov_FT0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6572 |
gen_helper_fcomi_ST0_FT0(cpu_env); |
| 6573 |
gen_helper_fpop(cpu_env); |
| 6574 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6575 |
break;
|
| 6576 |
case 0x10 ... 0x13: /* fcmovxx */ |
| 6577 |
case 0x18 ... 0x1b: |
| 6578 |
{
|
| 6579 |
int op1, l1;
|
| 6580 |
static const uint8_t fcmov_cc[8] = { |
| 6581 |
(JCC_B << 1),
|
| 6582 |
(JCC_Z << 1),
|
| 6583 |
(JCC_BE << 1),
|
| 6584 |
(JCC_P << 1),
|
| 6585 |
}; |
| 6586 |
op1 = fcmov_cc[op & 3] | (((op >> 3) & 1) ^ 1); |
| 6587 |
l1 = gen_new_label(); |
| 6588 |
gen_jcc1_noeob(s, op1, l1); |
| 6589 |
gen_helper_fmov_ST0_STN(cpu_env, tcg_const_i32(opreg)); |
| 6590 |
gen_set_label(l1); |
| 6591 |
} |
| 6592 |
break;
|
| 6593 |
default:
|
| 6594 |
goto illegal_op;
|
| 6595 |
} |
| 6596 |
} |
| 6597 |
break;
|
| 6598 |
/************************/
|
| 6599 |
/* string ops */
|
| 6600 |
|
| 6601 |
case 0xa4: /* movsS */ |
| 6602 |
case 0xa5: |
| 6603 |
if ((b & 1) == 0) |
| 6604 |
ot = OT_BYTE; |
| 6605 |
else
|
| 6606 |
ot = dflag + OT_WORD; |
| 6607 |
|
| 6608 |
if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) {
|
| 6609 |
gen_repz_movs(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); |
| 6610 |
} else {
|
| 6611 |
gen_movs(s, ot); |
| 6612 |
} |
| 6613 |
break;
|
| 6614 |
|
| 6615 |
case 0xaa: /* stosS */ |
| 6616 |
case 0xab: |
| 6617 |
if ((b & 1) == 0) |
| 6618 |
ot = OT_BYTE; |
| 6619 |
else
|
| 6620 |
ot = dflag + OT_WORD; |
| 6621 |
|
| 6622 |
if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) {
|
| 6623 |
gen_repz_stos(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); |
| 6624 |
} else {
|
| 6625 |
gen_stos(s, ot); |
| 6626 |
} |
| 6627 |
break;
|
| 6628 |
case 0xac: /* lodsS */ |
| 6629 |
case 0xad: |
| 6630 |
if ((b & 1) == 0) |
| 6631 |
ot = OT_BYTE; |
| 6632 |
else
|
| 6633 |
ot = dflag + OT_WORD; |
| 6634 |
if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) {
|
| 6635 |
gen_repz_lods(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); |
| 6636 |
} else {
|
| 6637 |
gen_lods(s, ot); |
| 6638 |
} |
| 6639 |
break;
|
| 6640 |
case 0xae: /* scasS */ |
| 6641 |
case 0xaf: |
| 6642 |
if ((b & 1) == 0) |
| 6643 |
ot = OT_BYTE; |
| 6644 |
else
|
| 6645 |
ot = dflag + OT_WORD; |
| 6646 |
if (prefixes & PREFIX_REPNZ) {
|
| 6647 |
gen_repz_scas(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 1);
|
| 6648 |
} else if (prefixes & PREFIX_REPZ) { |
| 6649 |
gen_repz_scas(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 0);
|
| 6650 |
} else {
|
| 6651 |
gen_scas(s, ot); |
| 6652 |
} |
| 6653 |
break;
|
| 6654 |
|
| 6655 |
case 0xa6: /* cmpsS */ |
| 6656 |
case 0xa7: |
| 6657 |
if ((b & 1) == 0) |
| 6658 |
ot = OT_BYTE; |
| 6659 |
else
|
| 6660 |
ot = dflag + OT_WORD; |
| 6661 |
if (prefixes & PREFIX_REPNZ) {
|
| 6662 |
gen_repz_cmps(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 1);
|
| 6663 |
} else if (prefixes & PREFIX_REPZ) { |
| 6664 |
gen_repz_cmps(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 0);
|
| 6665 |
} else {
|
| 6666 |
gen_cmps(s, ot); |
| 6667 |
} |
| 6668 |
break;
|
| 6669 |
case 0x6c: /* insS */ |
| 6670 |
case 0x6d: |
| 6671 |
if ((b & 1) == 0) |
| 6672 |
ot = OT_BYTE; |
| 6673 |
else
|
| 6674 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6675 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EDX);
|
| 6676 |
gen_op_andl_T0_ffff(); |
| 6677 |
gen_check_io(s, ot, pc_start - s->cs_base, |
| 6678 |
SVM_IOIO_TYPE_MASK | svm_is_rep(prefixes) | 4);
|
| 6679 |
if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) {
|
| 6680 |
gen_repz_ins(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); |
| 6681 |
} else {
|
| 6682 |
gen_ins(s, ot); |
| 6683 |
if (use_icount) {
|
| 6684 |
gen_jmp(s, s->pc - s->cs_base); |
| 6685 |
} |
| 6686 |
} |
| 6687 |
break;
|
| 6688 |
case 0x6e: /* outsS */ |
| 6689 |
case 0x6f: |
| 6690 |
if ((b & 1) == 0) |
| 6691 |
ot = OT_BYTE; |
| 6692 |
else
|
| 6693 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6694 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EDX);
|
| 6695 |
gen_op_andl_T0_ffff(); |
| 6696 |
gen_check_io(s, ot, pc_start - s->cs_base, |
| 6697 |
svm_is_rep(prefixes) | 4);
|
| 6698 |
if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) {
|
| 6699 |
gen_repz_outs(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); |
| 6700 |
} else {
|
| 6701 |
gen_outs(s, ot); |
| 6702 |
if (use_icount) {
|
| 6703 |
gen_jmp(s, s->pc - s->cs_base); |
| 6704 |
} |
| 6705 |
} |
| 6706 |
break;
|
| 6707 |
|
| 6708 |
/************************/
|
| 6709 |
/* port I/O */
|
| 6710 |
|
| 6711 |
case 0xe4: |
| 6712 |
case 0xe5: |
| 6713 |
if ((b & 1) == 0) |
| 6714 |
ot = OT_BYTE; |
| 6715 |
else
|
| 6716 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6717 |
val = cpu_ldub_code(env, s->pc++); |
| 6718 |
gen_op_movl_T0_im(val); |
| 6719 |
gen_check_io(s, ot, pc_start - s->cs_base, |
| 6720 |
SVM_IOIO_TYPE_MASK | svm_is_rep(prefixes)); |
| 6721 |
if (use_icount)
|
| 6722 |
gen_io_start(); |
| 6723 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6724 |
gen_helper_in_func(ot, cpu_T[1], cpu_tmp2_i32);
|
| 6725 |
gen_op_mov_reg_T1(ot, R_EAX); |
| 6726 |
if (use_icount) {
|
| 6727 |
gen_io_end(); |
| 6728 |
gen_jmp(s, s->pc - s->cs_base); |
| 6729 |
} |
| 6730 |
break;
|
| 6731 |
case 0xe6: |
| 6732 |
case 0xe7: |
| 6733 |
if ((b & 1) == 0) |
| 6734 |
ot = OT_BYTE; |
| 6735 |
else
|
| 6736 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6737 |
val = cpu_ldub_code(env, s->pc++); |
| 6738 |
gen_op_movl_T0_im(val); |
| 6739 |
gen_check_io(s, ot, pc_start - s->cs_base, |
| 6740 |
svm_is_rep(prefixes)); |
| 6741 |
gen_op_mov_TN_reg(ot, 1, R_EAX);
|
| 6742 |
|
| 6743 |
if (use_icount)
|
| 6744 |
gen_io_start(); |
| 6745 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6746 |
tcg_gen_trunc_tl_i32(cpu_tmp3_i32, cpu_T[1]);
|
| 6747 |
gen_helper_out_func(ot, cpu_tmp2_i32, cpu_tmp3_i32); |
| 6748 |
if (use_icount) {
|
| 6749 |
gen_io_end(); |
| 6750 |
gen_jmp(s, s->pc - s->cs_base); |
| 6751 |
} |
| 6752 |
break;
|
| 6753 |
case 0xec: |
| 6754 |
case 0xed: |
| 6755 |
if ((b & 1) == 0) |
| 6756 |
ot = OT_BYTE; |
| 6757 |
else
|
| 6758 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6759 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EDX);
|
| 6760 |
gen_op_andl_T0_ffff(); |
| 6761 |
gen_check_io(s, ot, pc_start - s->cs_base, |
| 6762 |
SVM_IOIO_TYPE_MASK | svm_is_rep(prefixes)); |
| 6763 |
if (use_icount)
|
| 6764 |
gen_io_start(); |
| 6765 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6766 |
gen_helper_in_func(ot, cpu_T[1], cpu_tmp2_i32);
|
| 6767 |
gen_op_mov_reg_T1(ot, R_EAX); |
| 6768 |
if (use_icount) {
|
| 6769 |
gen_io_end(); |
| 6770 |
gen_jmp(s, s->pc - s->cs_base); |
| 6771 |
} |
| 6772 |
break;
|
| 6773 |
case 0xee: |
| 6774 |
case 0xef: |
| 6775 |
if ((b & 1) == 0) |
| 6776 |
ot = OT_BYTE; |
| 6777 |
else
|
| 6778 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6779 |
gen_op_mov_TN_reg(OT_WORD, 0, R_EDX);
|
| 6780 |
gen_op_andl_T0_ffff(); |
| 6781 |
gen_check_io(s, ot, pc_start - s->cs_base, |
| 6782 |
svm_is_rep(prefixes)); |
| 6783 |
gen_op_mov_TN_reg(ot, 1, R_EAX);
|
| 6784 |
|
| 6785 |
if (use_icount)
|
| 6786 |
gen_io_start(); |
| 6787 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 6788 |
tcg_gen_trunc_tl_i32(cpu_tmp3_i32, cpu_T[1]);
|
| 6789 |
gen_helper_out_func(ot, cpu_tmp2_i32, cpu_tmp3_i32); |
| 6790 |
if (use_icount) {
|
| 6791 |
gen_io_end(); |
| 6792 |
gen_jmp(s, s->pc - s->cs_base); |
| 6793 |
} |
| 6794 |
break;
|
| 6795 |
|
| 6796 |
/************************/
|
| 6797 |
/* control */
|
| 6798 |
case 0xc2: /* ret im */ |
| 6799 |
val = cpu_ldsw_code(env, s->pc); |
| 6800 |
s->pc += 2;
|
| 6801 |
gen_pop_T0(s); |
| 6802 |
if (CODE64(s) && s->dflag)
|
| 6803 |
s->dflag = 2;
|
| 6804 |
gen_stack_update(s, val + (2 << s->dflag));
|
| 6805 |
if (s->dflag == 0) |
| 6806 |
gen_op_andl_T0_ffff(); |
| 6807 |
gen_op_jmp_T0(); |
| 6808 |
gen_eob(s); |
| 6809 |
break;
|
| 6810 |
case 0xc3: /* ret */ |
| 6811 |
gen_pop_T0(s); |
| 6812 |
gen_pop_update(s); |
| 6813 |
if (s->dflag == 0) |
| 6814 |
gen_op_andl_T0_ffff(); |
| 6815 |
gen_op_jmp_T0(); |
| 6816 |
gen_eob(s); |
| 6817 |
break;
|
| 6818 |
case 0xca: /* lret im */ |
| 6819 |
val = cpu_ldsw_code(env, s->pc); |
| 6820 |
s->pc += 2;
|
| 6821 |
do_lret:
|
| 6822 |
if (s->pe && !s->vm86) {
|
| 6823 |
gen_update_cc_op(s); |
| 6824 |
gen_jmp_im(pc_start - s->cs_base); |
| 6825 |
gen_helper_lret_protected(cpu_env, tcg_const_i32(s->dflag), |
| 6826 |
tcg_const_i32(val)); |
| 6827 |
} else {
|
| 6828 |
gen_stack_A0(s); |
| 6829 |
/* pop offset */
|
| 6830 |
gen_op_ld_T0_A0(1 + s->dflag + s->mem_index);
|
| 6831 |
if (s->dflag == 0) |
| 6832 |
gen_op_andl_T0_ffff(); |
| 6833 |
/* NOTE: keeping EIP updated is not a problem in case of
|
| 6834 |
exception */
|
| 6835 |
gen_op_jmp_T0(); |
| 6836 |
/* pop selector */
|
| 6837 |
gen_op_addl_A0_im(2 << s->dflag);
|
| 6838 |
gen_op_ld_T0_A0(1 + s->dflag + s->mem_index);
|
| 6839 |
gen_op_movl_seg_T0_vm(R_CS); |
| 6840 |
/* add stack offset */
|
| 6841 |
gen_stack_update(s, val + (4 << s->dflag));
|
| 6842 |
} |
| 6843 |
gen_eob(s); |
| 6844 |
break;
|
| 6845 |
case 0xcb: /* lret */ |
| 6846 |
val = 0;
|
| 6847 |
goto do_lret;
|
| 6848 |
case 0xcf: /* iret */ |
| 6849 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_IRET); |
| 6850 |
if (!s->pe) {
|
| 6851 |
/* real mode */
|
| 6852 |
gen_helper_iret_real(cpu_env, tcg_const_i32(s->dflag)); |
| 6853 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6854 |
} else if (s->vm86) { |
| 6855 |
if (s->iopl != 3) { |
| 6856 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 6857 |
} else {
|
| 6858 |
gen_helper_iret_real(cpu_env, tcg_const_i32(s->dflag)); |
| 6859 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6860 |
} |
| 6861 |
} else {
|
| 6862 |
gen_update_cc_op(s); |
| 6863 |
gen_jmp_im(pc_start - s->cs_base); |
| 6864 |
gen_helper_iret_protected(cpu_env, tcg_const_i32(s->dflag), |
| 6865 |
tcg_const_i32(s->pc - s->cs_base)); |
| 6866 |
set_cc_op(s, CC_OP_EFLAGS); |
| 6867 |
} |
| 6868 |
gen_eob(s); |
| 6869 |
break;
|
| 6870 |
case 0xe8: /* call im */ |
| 6871 |
{
|
| 6872 |
if (dflag)
|
| 6873 |
tval = (int32_t)insn_get(env, s, OT_LONG); |
| 6874 |
else
|
| 6875 |
tval = (int16_t)insn_get(env, s, OT_WORD); |
| 6876 |
next_eip = s->pc - s->cs_base; |
| 6877 |
tval += next_eip; |
| 6878 |
if (s->dflag == 0) |
| 6879 |
tval &= 0xffff;
|
| 6880 |
else if(!CODE64(s)) |
| 6881 |
tval &= 0xffffffff;
|
| 6882 |
gen_movtl_T0_im(next_eip); |
| 6883 |
gen_push_T0(s); |
| 6884 |
gen_jmp(s, tval); |
| 6885 |
} |
| 6886 |
break;
|
| 6887 |
case 0x9a: /* lcall im */ |
| 6888 |
{
|
| 6889 |
unsigned int selector, offset; |
| 6890 |
|
| 6891 |
if (CODE64(s))
|
| 6892 |
goto illegal_op;
|
| 6893 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6894 |
offset = insn_get(env, s, ot); |
| 6895 |
selector = insn_get(env, s, OT_WORD); |
| 6896 |
|
| 6897 |
gen_op_movl_T0_im(selector); |
| 6898 |
gen_op_movl_T1_imu(offset); |
| 6899 |
} |
| 6900 |
goto do_lcall;
|
| 6901 |
case 0xe9: /* jmp im */ |
| 6902 |
if (dflag)
|
| 6903 |
tval = (int32_t)insn_get(env, s, OT_LONG); |
| 6904 |
else
|
| 6905 |
tval = (int16_t)insn_get(env, s, OT_WORD); |
| 6906 |
tval += s->pc - s->cs_base; |
| 6907 |
if (s->dflag == 0) |
| 6908 |
tval &= 0xffff;
|
| 6909 |
else if(!CODE64(s)) |
| 6910 |
tval &= 0xffffffff;
|
| 6911 |
gen_jmp(s, tval); |
| 6912 |
break;
|
| 6913 |
case 0xea: /* ljmp im */ |
| 6914 |
{
|
| 6915 |
unsigned int selector, offset; |
| 6916 |
|
| 6917 |
if (CODE64(s))
|
| 6918 |
goto illegal_op;
|
| 6919 |
ot = dflag ? OT_LONG : OT_WORD; |
| 6920 |
offset = insn_get(env, s, ot); |
| 6921 |
selector = insn_get(env, s, OT_WORD); |
| 6922 |
|
| 6923 |
gen_op_movl_T0_im(selector); |
| 6924 |
gen_op_movl_T1_imu(offset); |
| 6925 |
} |
| 6926 |
goto do_ljmp;
|
| 6927 |
case 0xeb: /* jmp Jb */ |
| 6928 |
tval = (int8_t)insn_get(env, s, OT_BYTE); |
| 6929 |
tval += s->pc - s->cs_base; |
| 6930 |
if (s->dflag == 0) |
| 6931 |
tval &= 0xffff;
|
| 6932 |
gen_jmp(s, tval); |
| 6933 |
break;
|
| 6934 |
case 0x70 ... 0x7f: /* jcc Jb */ |
| 6935 |
tval = (int8_t)insn_get(env, s, OT_BYTE); |
| 6936 |
goto do_jcc;
|
| 6937 |
case 0x180 ... 0x18f: /* jcc Jv */ |
| 6938 |
if (dflag) {
|
| 6939 |
tval = (int32_t)insn_get(env, s, OT_LONG); |
| 6940 |
} else {
|
| 6941 |
tval = (int16_t)insn_get(env, s, OT_WORD); |
| 6942 |
} |
| 6943 |
do_jcc:
|
| 6944 |
next_eip = s->pc - s->cs_base; |
| 6945 |
tval += next_eip; |
| 6946 |
if (s->dflag == 0) |
| 6947 |
tval &= 0xffff;
|
| 6948 |
gen_jcc(s, b, tval, next_eip); |
| 6949 |
break;
|
| 6950 |
|
| 6951 |
case 0x190 ... 0x19f: /* setcc Gv */ |
| 6952 |
modrm = cpu_ldub_code(env, s->pc++); |
| 6953 |
gen_setcc1(s, b, cpu_T[0]);
|
| 6954 |
gen_ldst_modrm(env, s, modrm, OT_BYTE, OR_TMP0, 1);
|
| 6955 |
break;
|
| 6956 |
case 0x140 ... 0x14f: /* cmov Gv, Ev */ |
| 6957 |
ot = dflag + OT_WORD; |
| 6958 |
modrm = cpu_ldub_code(env, s->pc++); |
| 6959 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 6960 |
gen_cmovcc1(env, s, ot, b, modrm, reg); |
| 6961 |
break;
|
| 6962 |
|
| 6963 |
/************************/
|
| 6964 |
/* flags */
|
| 6965 |
case 0x9c: /* pushf */ |
| 6966 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_PUSHF); |
| 6967 |
if (s->vm86 && s->iopl != 3) { |
| 6968 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 6969 |
} else {
|
| 6970 |
gen_update_cc_op(s); |
| 6971 |
gen_helper_read_eflags(cpu_T[0], cpu_env);
|
| 6972 |
gen_push_T0(s); |
| 6973 |
} |
| 6974 |
break;
|
| 6975 |
case 0x9d: /* popf */ |
| 6976 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_POPF); |
| 6977 |
if (s->vm86 && s->iopl != 3) { |
| 6978 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 6979 |
} else {
|
| 6980 |
gen_pop_T0(s); |
| 6981 |
if (s->cpl == 0) { |
| 6982 |
if (s->dflag) {
|
| 6983 |
gen_helper_write_eflags(cpu_env, cpu_T[0],
|
| 6984 |
tcg_const_i32((TF_MASK | AC_MASK | |
| 6985 |
ID_MASK | NT_MASK | |
| 6986 |
IF_MASK | |
| 6987 |
IOPL_MASK))); |
| 6988 |
} else {
|
| 6989 |
gen_helper_write_eflags(cpu_env, cpu_T[0],
|
| 6990 |
tcg_const_i32((TF_MASK | AC_MASK | |
| 6991 |
ID_MASK | NT_MASK | |
| 6992 |
IF_MASK | IOPL_MASK) |
| 6993 |
& 0xffff));
|
| 6994 |
} |
| 6995 |
} else {
|
| 6996 |
if (s->cpl <= s->iopl) {
|
| 6997 |
if (s->dflag) {
|
| 6998 |
gen_helper_write_eflags(cpu_env, cpu_T[0],
|
| 6999 |
tcg_const_i32((TF_MASK | |
| 7000 |
AC_MASK | |
| 7001 |
ID_MASK | |
| 7002 |
NT_MASK | |
| 7003 |
IF_MASK))); |
| 7004 |
} else {
|
| 7005 |
gen_helper_write_eflags(cpu_env, cpu_T[0],
|
| 7006 |
tcg_const_i32((TF_MASK | |
| 7007 |
AC_MASK | |
| 7008 |
ID_MASK | |
| 7009 |
NT_MASK | |
| 7010 |
IF_MASK) |
| 7011 |
& 0xffff));
|
| 7012 |
} |
| 7013 |
} else {
|
| 7014 |
if (s->dflag) {
|
| 7015 |
gen_helper_write_eflags(cpu_env, cpu_T[0],
|
| 7016 |
tcg_const_i32((TF_MASK | AC_MASK | |
| 7017 |
ID_MASK | NT_MASK))); |
| 7018 |
} else {
|
| 7019 |
gen_helper_write_eflags(cpu_env, cpu_T[0],
|
| 7020 |
tcg_const_i32((TF_MASK | AC_MASK | |
| 7021 |
ID_MASK | NT_MASK) |
| 7022 |
& 0xffff));
|
| 7023 |
} |
| 7024 |
} |
| 7025 |
} |
| 7026 |
gen_pop_update(s); |
| 7027 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7028 |
/* abort translation because TF/AC flag may change */
|
| 7029 |
gen_jmp_im(s->pc - s->cs_base); |
| 7030 |
gen_eob(s); |
| 7031 |
} |
| 7032 |
break;
|
| 7033 |
case 0x9e: /* sahf */ |
| 7034 |
if (CODE64(s) && !(s->cpuid_ext3_features & CPUID_EXT3_LAHF_LM))
|
| 7035 |
goto illegal_op;
|
| 7036 |
gen_op_mov_TN_reg(OT_BYTE, 0, R_AH);
|
| 7037 |
gen_compute_eflags(s); |
| 7038 |
tcg_gen_andi_tl(cpu_cc_src, cpu_cc_src, CC_O); |
| 7039 |
tcg_gen_andi_tl(cpu_T[0], cpu_T[0], CC_S | CC_Z | CC_A | CC_P | CC_C); |
| 7040 |
tcg_gen_or_tl(cpu_cc_src, cpu_cc_src, cpu_T[0]);
|
| 7041 |
break;
|
| 7042 |
case 0x9f: /* lahf */ |
| 7043 |
if (CODE64(s) && !(s->cpuid_ext3_features & CPUID_EXT3_LAHF_LM))
|
| 7044 |
goto illegal_op;
|
| 7045 |
gen_compute_eflags(s); |
| 7046 |
/* Note: gen_compute_eflags() only gives the condition codes */
|
| 7047 |
tcg_gen_ori_tl(cpu_T[0], cpu_cc_src, 0x02); |
| 7048 |
gen_op_mov_reg_T0(OT_BYTE, R_AH); |
| 7049 |
break;
|
| 7050 |
case 0xf5: /* cmc */ |
| 7051 |
gen_compute_eflags(s); |
| 7052 |
tcg_gen_xori_tl(cpu_cc_src, cpu_cc_src, CC_C); |
| 7053 |
break;
|
| 7054 |
case 0xf8: /* clc */ |
| 7055 |
gen_compute_eflags(s); |
| 7056 |
tcg_gen_andi_tl(cpu_cc_src, cpu_cc_src, ~CC_C); |
| 7057 |
break;
|
| 7058 |
case 0xf9: /* stc */ |
| 7059 |
gen_compute_eflags(s); |
| 7060 |
tcg_gen_ori_tl(cpu_cc_src, cpu_cc_src, CC_C); |
| 7061 |
break;
|
| 7062 |
case 0xfc: /* cld */ |
| 7063 |
tcg_gen_movi_i32(cpu_tmp2_i32, 1);
|
| 7064 |
tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, offsetof(CPUX86State, df)); |
| 7065 |
break;
|
| 7066 |
case 0xfd: /* std */ |
| 7067 |
tcg_gen_movi_i32(cpu_tmp2_i32, -1);
|
| 7068 |
tcg_gen_st_i32(cpu_tmp2_i32, cpu_env, offsetof(CPUX86State, df)); |
| 7069 |
break;
|
| 7070 |
|
| 7071 |
/************************/
|
| 7072 |
/* bit operations */
|
| 7073 |
case 0x1ba: /* bt/bts/btr/btc Gv, im */ |
| 7074 |
ot = dflag + OT_WORD; |
| 7075 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7076 |
op = (modrm >> 3) & 7; |
| 7077 |
mod = (modrm >> 6) & 3; |
| 7078 |
rm = (modrm & 7) | REX_B(s);
|
| 7079 |
if (mod != 3) { |
| 7080 |
s->rip_offset = 1;
|
| 7081 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7082 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 7083 |
} else {
|
| 7084 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 7085 |
} |
| 7086 |
/* load shift */
|
| 7087 |
val = cpu_ldub_code(env, s->pc++); |
| 7088 |
gen_op_movl_T1_im(val); |
| 7089 |
if (op < 4) |
| 7090 |
goto illegal_op;
|
| 7091 |
op -= 4;
|
| 7092 |
goto bt_op;
|
| 7093 |
case 0x1a3: /* bt Gv, Ev */ |
| 7094 |
op = 0;
|
| 7095 |
goto do_btx;
|
| 7096 |
case 0x1ab: /* bts */ |
| 7097 |
op = 1;
|
| 7098 |
goto do_btx;
|
| 7099 |
case 0x1b3: /* btr */ |
| 7100 |
op = 2;
|
| 7101 |
goto do_btx;
|
| 7102 |
case 0x1bb: /* btc */ |
| 7103 |
op = 3;
|
| 7104 |
do_btx:
|
| 7105 |
ot = dflag + OT_WORD; |
| 7106 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7107 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 7108 |
mod = (modrm >> 6) & 3; |
| 7109 |
rm = (modrm & 7) | REX_B(s);
|
| 7110 |
gen_op_mov_TN_reg(OT_LONG, 1, reg);
|
| 7111 |
if (mod != 3) { |
| 7112 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7113 |
/* specific case: we need to add a displacement */
|
| 7114 |
gen_exts(ot, cpu_T[1]);
|
| 7115 |
tcg_gen_sari_tl(cpu_tmp0, cpu_T[1], 3 + ot); |
| 7116 |
tcg_gen_shli_tl(cpu_tmp0, cpu_tmp0, ot); |
| 7117 |
tcg_gen_add_tl(cpu_A0, cpu_A0, cpu_tmp0); |
| 7118 |
gen_op_ld_T0_A0(ot + s->mem_index); |
| 7119 |
} else {
|
| 7120 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 7121 |
} |
| 7122 |
bt_op:
|
| 7123 |
tcg_gen_andi_tl(cpu_T[1], cpu_T[1], (1 << (3 + ot)) - 1); |
| 7124 |
switch(op) {
|
| 7125 |
case 0: |
| 7126 |
tcg_gen_shr_tl(cpu_cc_src, cpu_T[0], cpu_T[1]); |
| 7127 |
tcg_gen_movi_tl(cpu_cc_dst, 0);
|
| 7128 |
break;
|
| 7129 |
case 1: |
| 7130 |
tcg_gen_shr_tl(cpu_tmp4, cpu_T[0], cpu_T[1]); |
| 7131 |
tcg_gen_movi_tl(cpu_tmp0, 1);
|
| 7132 |
tcg_gen_shl_tl(cpu_tmp0, cpu_tmp0, cpu_T[1]);
|
| 7133 |
tcg_gen_or_tl(cpu_T[0], cpu_T[0], cpu_tmp0); |
| 7134 |
break;
|
| 7135 |
case 2: |
| 7136 |
tcg_gen_shr_tl(cpu_tmp4, cpu_T[0], cpu_T[1]); |
| 7137 |
tcg_gen_movi_tl(cpu_tmp0, 1);
|
| 7138 |
tcg_gen_shl_tl(cpu_tmp0, cpu_tmp0, cpu_T[1]);
|
| 7139 |
tcg_gen_not_tl(cpu_tmp0, cpu_tmp0); |
| 7140 |
tcg_gen_and_tl(cpu_T[0], cpu_T[0], cpu_tmp0); |
| 7141 |
break;
|
| 7142 |
default:
|
| 7143 |
case 3: |
| 7144 |
tcg_gen_shr_tl(cpu_tmp4, cpu_T[0], cpu_T[1]); |
| 7145 |
tcg_gen_movi_tl(cpu_tmp0, 1);
|
| 7146 |
tcg_gen_shl_tl(cpu_tmp0, cpu_tmp0, cpu_T[1]);
|
| 7147 |
tcg_gen_xor_tl(cpu_T[0], cpu_T[0], cpu_tmp0); |
| 7148 |
break;
|
| 7149 |
} |
| 7150 |
set_cc_op(s, CC_OP_SARB + ot); |
| 7151 |
if (op != 0) { |
| 7152 |
if (mod != 3) |
| 7153 |
gen_op_st_T0_A0(ot + s->mem_index); |
| 7154 |
else
|
| 7155 |
gen_op_mov_reg_T0(ot, rm); |
| 7156 |
tcg_gen_mov_tl(cpu_cc_src, cpu_tmp4); |
| 7157 |
tcg_gen_movi_tl(cpu_cc_dst, 0);
|
| 7158 |
} |
| 7159 |
break;
|
| 7160 |
case 0x1bc: /* bsf / tzcnt */ |
| 7161 |
case 0x1bd: /* bsr / lzcnt */ |
| 7162 |
ot = dflag + OT_WORD; |
| 7163 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7164 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 7165 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 7166 |
gen_extu(ot, cpu_T[0]);
|
| 7167 |
|
| 7168 |
/* Note that lzcnt and tzcnt are in different extensions. */
|
| 7169 |
if ((prefixes & PREFIX_REPZ)
|
| 7170 |
&& (b & 1
|
| 7171 |
? s->cpuid_ext3_features & CPUID_EXT3_ABM |
| 7172 |
: s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_BMI1)) {
|
| 7173 |
int size = 8 << ot; |
| 7174 |
tcg_gen_mov_tl(cpu_cc_src, cpu_T[0]);
|
| 7175 |
if (b & 1) { |
| 7176 |
/* For lzcnt, reduce the target_ulong result by the
|
| 7177 |
number of zeros that we expect to find at the top. */
|
| 7178 |
gen_helper_clz(cpu_T[0], cpu_T[0]); |
| 7179 |
tcg_gen_subi_tl(cpu_T[0], cpu_T[0], TARGET_LONG_BITS - size); |
| 7180 |
} else {
|
| 7181 |
/* For tzcnt, a zero input must return the operand size:
|
| 7182 |
force all bits outside the operand size to 1. */
|
| 7183 |
target_ulong mask = (target_ulong)-2 << (size - 1); |
| 7184 |
tcg_gen_ori_tl(cpu_T[0], cpu_T[0], mask); |
| 7185 |
gen_helper_ctz(cpu_T[0], cpu_T[0]); |
| 7186 |
} |
| 7187 |
/* For lzcnt/tzcnt, C and Z bits are defined and are
|
| 7188 |
related to the result. */
|
| 7189 |
gen_op_update1_cc(); |
| 7190 |
set_cc_op(s, CC_OP_BMILGB + ot); |
| 7191 |
} else {
|
| 7192 |
/* For bsr/bsf, only the Z bit is defined and it is related
|
| 7193 |
to the input and not the result. */
|
| 7194 |
tcg_gen_mov_tl(cpu_cc_dst, cpu_T[0]);
|
| 7195 |
set_cc_op(s, CC_OP_LOGICB + ot); |
| 7196 |
if (b & 1) { |
| 7197 |
/* For bsr, return the bit index of the first 1 bit,
|
| 7198 |
not the count of leading zeros. */
|
| 7199 |
gen_helper_clz(cpu_T[0], cpu_T[0]); |
| 7200 |
tcg_gen_xori_tl(cpu_T[0], cpu_T[0], TARGET_LONG_BITS - 1); |
| 7201 |
} else {
|
| 7202 |
gen_helper_ctz(cpu_T[0], cpu_T[0]); |
| 7203 |
} |
| 7204 |
/* ??? The manual says that the output is undefined when the
|
| 7205 |
input is zero, but real hardware leaves it unchanged, and
|
| 7206 |
real programs appear to depend on that. */
|
| 7207 |
tcg_gen_movi_tl(cpu_tmp0, 0);
|
| 7208 |
tcg_gen_movcond_tl(TCG_COND_EQ, cpu_T[0], cpu_cc_dst, cpu_tmp0,
|
| 7209 |
cpu_regs[reg], cpu_T[0]);
|
| 7210 |
} |
| 7211 |
gen_op_mov_reg_T0(ot, reg); |
| 7212 |
break;
|
| 7213 |
/************************/
|
| 7214 |
/* bcd */
|
| 7215 |
case 0x27: /* daa */ |
| 7216 |
if (CODE64(s))
|
| 7217 |
goto illegal_op;
|
| 7218 |
gen_update_cc_op(s); |
| 7219 |
gen_helper_daa(cpu_env); |
| 7220 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7221 |
break;
|
| 7222 |
case 0x2f: /* das */ |
| 7223 |
if (CODE64(s))
|
| 7224 |
goto illegal_op;
|
| 7225 |
gen_update_cc_op(s); |
| 7226 |
gen_helper_das(cpu_env); |
| 7227 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7228 |
break;
|
| 7229 |
case 0x37: /* aaa */ |
| 7230 |
if (CODE64(s))
|
| 7231 |
goto illegal_op;
|
| 7232 |
gen_update_cc_op(s); |
| 7233 |
gen_helper_aaa(cpu_env); |
| 7234 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7235 |
break;
|
| 7236 |
case 0x3f: /* aas */ |
| 7237 |
if (CODE64(s))
|
| 7238 |
goto illegal_op;
|
| 7239 |
gen_update_cc_op(s); |
| 7240 |
gen_helper_aas(cpu_env); |
| 7241 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7242 |
break;
|
| 7243 |
case 0xd4: /* aam */ |
| 7244 |
if (CODE64(s))
|
| 7245 |
goto illegal_op;
|
| 7246 |
val = cpu_ldub_code(env, s->pc++); |
| 7247 |
if (val == 0) { |
| 7248 |
gen_exception(s, EXCP00_DIVZ, pc_start - s->cs_base); |
| 7249 |
} else {
|
| 7250 |
gen_helper_aam(cpu_env, tcg_const_i32(val)); |
| 7251 |
set_cc_op(s, CC_OP_LOGICB); |
| 7252 |
} |
| 7253 |
break;
|
| 7254 |
case 0xd5: /* aad */ |
| 7255 |
if (CODE64(s))
|
| 7256 |
goto illegal_op;
|
| 7257 |
val = cpu_ldub_code(env, s->pc++); |
| 7258 |
gen_helper_aad(cpu_env, tcg_const_i32(val)); |
| 7259 |
set_cc_op(s, CC_OP_LOGICB); |
| 7260 |
break;
|
| 7261 |
/************************/
|
| 7262 |
/* misc */
|
| 7263 |
case 0x90: /* nop */ |
| 7264 |
/* XXX: correct lock test for all insn */
|
| 7265 |
if (prefixes & PREFIX_LOCK) {
|
| 7266 |
goto illegal_op;
|
| 7267 |
} |
| 7268 |
/* If REX_B is set, then this is xchg eax, r8d, not a nop. */
|
| 7269 |
if (REX_B(s)) {
|
| 7270 |
goto do_xchg_reg_eax;
|
| 7271 |
} |
| 7272 |
if (prefixes & PREFIX_REPZ) {
|
| 7273 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_PAUSE); |
| 7274 |
} |
| 7275 |
break;
|
| 7276 |
case 0x9b: /* fwait */ |
| 7277 |
if ((s->flags & (HF_MP_MASK | HF_TS_MASK)) ==
|
| 7278 |
(HF_MP_MASK | HF_TS_MASK)) {
|
| 7279 |
gen_exception(s, EXCP07_PREX, pc_start - s->cs_base); |
| 7280 |
} else {
|
| 7281 |
gen_update_cc_op(s); |
| 7282 |
gen_jmp_im(pc_start - s->cs_base); |
| 7283 |
gen_helper_fwait(cpu_env); |
| 7284 |
} |
| 7285 |
break;
|
| 7286 |
case 0xcc: /* int3 */ |
| 7287 |
gen_interrupt(s, EXCP03_INT3, pc_start - s->cs_base, s->pc - s->cs_base); |
| 7288 |
break;
|
| 7289 |
case 0xcd: /* int N */ |
| 7290 |
val = cpu_ldub_code(env, s->pc++); |
| 7291 |
if (s->vm86 && s->iopl != 3) { |
| 7292 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7293 |
} else {
|
| 7294 |
gen_interrupt(s, val, pc_start - s->cs_base, s->pc - s->cs_base); |
| 7295 |
} |
| 7296 |
break;
|
| 7297 |
case 0xce: /* into */ |
| 7298 |
if (CODE64(s))
|
| 7299 |
goto illegal_op;
|
| 7300 |
gen_update_cc_op(s); |
| 7301 |
gen_jmp_im(pc_start - s->cs_base); |
| 7302 |
gen_helper_into(cpu_env, tcg_const_i32(s->pc - pc_start)); |
| 7303 |
break;
|
| 7304 |
#ifdef WANT_ICEBP
|
| 7305 |
case 0xf1: /* icebp (undocumented, exits to external debugger) */ |
| 7306 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_ICEBP); |
| 7307 |
#if 1 |
| 7308 |
gen_debug(s, pc_start - s->cs_base); |
| 7309 |
#else
|
| 7310 |
/* start debug */
|
| 7311 |
tb_flush(env); |
| 7312 |
qemu_set_log(CPU_LOG_INT | CPU_LOG_TB_IN_ASM); |
| 7313 |
#endif
|
| 7314 |
break;
|
| 7315 |
#endif
|
| 7316 |
case 0xfa: /* cli */ |
| 7317 |
if (!s->vm86) {
|
| 7318 |
if (s->cpl <= s->iopl) {
|
| 7319 |
gen_helper_cli(cpu_env); |
| 7320 |
} else {
|
| 7321 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7322 |
} |
| 7323 |
} else {
|
| 7324 |
if (s->iopl == 3) { |
| 7325 |
gen_helper_cli(cpu_env); |
| 7326 |
} else {
|
| 7327 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7328 |
} |
| 7329 |
} |
| 7330 |
break;
|
| 7331 |
case 0xfb: /* sti */ |
| 7332 |
if (!s->vm86) {
|
| 7333 |
if (s->cpl <= s->iopl) {
|
| 7334 |
gen_sti:
|
| 7335 |
gen_helper_sti(cpu_env); |
| 7336 |
/* interruptions are enabled only the first insn after sti */
|
| 7337 |
/* If several instructions disable interrupts, only the
|
| 7338 |
_first_ does it */
|
| 7339 |
if (!(s->tb->flags & HF_INHIBIT_IRQ_MASK))
|
| 7340 |
gen_helper_set_inhibit_irq(cpu_env); |
| 7341 |
/* give a chance to handle pending irqs */
|
| 7342 |
gen_jmp_im(s->pc - s->cs_base); |
| 7343 |
gen_eob(s); |
| 7344 |
} else {
|
| 7345 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7346 |
} |
| 7347 |
} else {
|
| 7348 |
if (s->iopl == 3) { |
| 7349 |
goto gen_sti;
|
| 7350 |
} else {
|
| 7351 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7352 |
} |
| 7353 |
} |
| 7354 |
break;
|
| 7355 |
case 0x62: /* bound */ |
| 7356 |
if (CODE64(s))
|
| 7357 |
goto illegal_op;
|
| 7358 |
ot = dflag ? OT_LONG : OT_WORD; |
| 7359 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7360 |
reg = (modrm >> 3) & 7; |
| 7361 |
mod = (modrm >> 6) & 3; |
| 7362 |
if (mod == 3) |
| 7363 |
goto illegal_op;
|
| 7364 |
gen_op_mov_TN_reg(ot, 0, reg);
|
| 7365 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7366 |
gen_jmp_im(pc_start - s->cs_base); |
| 7367 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 7368 |
if (ot == OT_WORD) {
|
| 7369 |
gen_helper_boundw(cpu_env, cpu_A0, cpu_tmp2_i32); |
| 7370 |
} else {
|
| 7371 |
gen_helper_boundl(cpu_env, cpu_A0, cpu_tmp2_i32); |
| 7372 |
} |
| 7373 |
break;
|
| 7374 |
case 0x1c8 ... 0x1cf: /* bswap reg */ |
| 7375 |
reg = (b & 7) | REX_B(s);
|
| 7376 |
#ifdef TARGET_X86_64
|
| 7377 |
if (dflag == 2) { |
| 7378 |
gen_op_mov_TN_reg(OT_QUAD, 0, reg);
|
| 7379 |
tcg_gen_bswap64_i64(cpu_T[0], cpu_T[0]); |
| 7380 |
gen_op_mov_reg_T0(OT_QUAD, reg); |
| 7381 |
} else
|
| 7382 |
#endif
|
| 7383 |
{
|
| 7384 |
gen_op_mov_TN_reg(OT_LONG, 0, reg);
|
| 7385 |
tcg_gen_ext32u_tl(cpu_T[0], cpu_T[0]); |
| 7386 |
tcg_gen_bswap32_tl(cpu_T[0], cpu_T[0]); |
| 7387 |
gen_op_mov_reg_T0(OT_LONG, reg); |
| 7388 |
} |
| 7389 |
break;
|
| 7390 |
case 0xd6: /* salc */ |
| 7391 |
if (CODE64(s))
|
| 7392 |
goto illegal_op;
|
| 7393 |
gen_compute_eflags_c(s, cpu_T[0]);
|
| 7394 |
tcg_gen_neg_tl(cpu_T[0], cpu_T[0]); |
| 7395 |
gen_op_mov_reg_T0(OT_BYTE, R_EAX); |
| 7396 |
break;
|
| 7397 |
case 0xe0: /* loopnz */ |
| 7398 |
case 0xe1: /* loopz */ |
| 7399 |
case 0xe2: /* loop */ |
| 7400 |
case 0xe3: /* jecxz */ |
| 7401 |
{
|
| 7402 |
int l1, l2, l3;
|
| 7403 |
|
| 7404 |
tval = (int8_t)insn_get(env, s, OT_BYTE); |
| 7405 |
next_eip = s->pc - s->cs_base; |
| 7406 |
tval += next_eip; |
| 7407 |
if (s->dflag == 0) |
| 7408 |
tval &= 0xffff;
|
| 7409 |
|
| 7410 |
l1 = gen_new_label(); |
| 7411 |
l2 = gen_new_label(); |
| 7412 |
l3 = gen_new_label(); |
| 7413 |
b &= 3;
|
| 7414 |
switch(b) {
|
| 7415 |
case 0: /* loopnz */ |
| 7416 |
case 1: /* loopz */ |
| 7417 |
gen_op_add_reg_im(s->aflag, R_ECX, -1);
|
| 7418 |
gen_op_jz_ecx(s->aflag, l3); |
| 7419 |
gen_jcc1(s, (JCC_Z << 1) | (b ^ 1), l1); |
| 7420 |
break;
|
| 7421 |
case 2: /* loop */ |
| 7422 |
gen_op_add_reg_im(s->aflag, R_ECX, -1);
|
| 7423 |
gen_op_jnz_ecx(s->aflag, l1); |
| 7424 |
break;
|
| 7425 |
default:
|
| 7426 |
case 3: /* jcxz */ |
| 7427 |
gen_op_jz_ecx(s->aflag, l1); |
| 7428 |
break;
|
| 7429 |
} |
| 7430 |
|
| 7431 |
gen_set_label(l3); |
| 7432 |
gen_jmp_im(next_eip); |
| 7433 |
tcg_gen_br(l2); |
| 7434 |
|
| 7435 |
gen_set_label(l1); |
| 7436 |
gen_jmp_im(tval); |
| 7437 |
gen_set_label(l2); |
| 7438 |
gen_eob(s); |
| 7439 |
} |
| 7440 |
break;
|
| 7441 |
case 0x130: /* wrmsr */ |
| 7442 |
case 0x132: /* rdmsr */ |
| 7443 |
if (s->cpl != 0) { |
| 7444 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7445 |
} else {
|
| 7446 |
gen_update_cc_op(s); |
| 7447 |
gen_jmp_im(pc_start - s->cs_base); |
| 7448 |
if (b & 2) { |
| 7449 |
gen_helper_rdmsr(cpu_env); |
| 7450 |
} else {
|
| 7451 |
gen_helper_wrmsr(cpu_env); |
| 7452 |
} |
| 7453 |
} |
| 7454 |
break;
|
| 7455 |
case 0x131: /* rdtsc */ |
| 7456 |
gen_update_cc_op(s); |
| 7457 |
gen_jmp_im(pc_start - s->cs_base); |
| 7458 |
if (use_icount)
|
| 7459 |
gen_io_start(); |
| 7460 |
gen_helper_rdtsc(cpu_env); |
| 7461 |
if (use_icount) {
|
| 7462 |
gen_io_end(); |
| 7463 |
gen_jmp(s, s->pc - s->cs_base); |
| 7464 |
} |
| 7465 |
break;
|
| 7466 |
case 0x133: /* rdpmc */ |
| 7467 |
gen_update_cc_op(s); |
| 7468 |
gen_jmp_im(pc_start - s->cs_base); |
| 7469 |
gen_helper_rdpmc(cpu_env); |
| 7470 |
break;
|
| 7471 |
case 0x134: /* sysenter */ |
| 7472 |
/* For Intel SYSENTER is valid on 64-bit */
|
| 7473 |
if (CODE64(s) && env->cpuid_vendor1 != CPUID_VENDOR_INTEL_1)
|
| 7474 |
goto illegal_op;
|
| 7475 |
if (!s->pe) {
|
| 7476 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7477 |
} else {
|
| 7478 |
gen_update_cc_op(s); |
| 7479 |
gen_jmp_im(pc_start - s->cs_base); |
| 7480 |
gen_helper_sysenter(cpu_env); |
| 7481 |
gen_eob(s); |
| 7482 |
} |
| 7483 |
break;
|
| 7484 |
case 0x135: /* sysexit */ |
| 7485 |
/* For Intel SYSEXIT is valid on 64-bit */
|
| 7486 |
if (CODE64(s) && env->cpuid_vendor1 != CPUID_VENDOR_INTEL_1)
|
| 7487 |
goto illegal_op;
|
| 7488 |
if (!s->pe) {
|
| 7489 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7490 |
} else {
|
| 7491 |
gen_update_cc_op(s); |
| 7492 |
gen_jmp_im(pc_start - s->cs_base); |
| 7493 |
gen_helper_sysexit(cpu_env, tcg_const_i32(dflag)); |
| 7494 |
gen_eob(s); |
| 7495 |
} |
| 7496 |
break;
|
| 7497 |
#ifdef TARGET_X86_64
|
| 7498 |
case 0x105: /* syscall */ |
| 7499 |
/* XXX: is it usable in real mode ? */
|
| 7500 |
gen_update_cc_op(s); |
| 7501 |
gen_jmp_im(pc_start - s->cs_base); |
| 7502 |
gen_helper_syscall(cpu_env, tcg_const_i32(s->pc - pc_start)); |
| 7503 |
gen_eob(s); |
| 7504 |
break;
|
| 7505 |
case 0x107: /* sysret */ |
| 7506 |
if (!s->pe) {
|
| 7507 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7508 |
} else {
|
| 7509 |
gen_update_cc_op(s); |
| 7510 |
gen_jmp_im(pc_start - s->cs_base); |
| 7511 |
gen_helper_sysret(cpu_env, tcg_const_i32(s->dflag)); |
| 7512 |
/* condition codes are modified only in long mode */
|
| 7513 |
if (s->lma) {
|
| 7514 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7515 |
} |
| 7516 |
gen_eob(s); |
| 7517 |
} |
| 7518 |
break;
|
| 7519 |
#endif
|
| 7520 |
case 0x1a2: /* cpuid */ |
| 7521 |
gen_update_cc_op(s); |
| 7522 |
gen_jmp_im(pc_start - s->cs_base); |
| 7523 |
gen_helper_cpuid(cpu_env); |
| 7524 |
break;
|
| 7525 |
case 0xf4: /* hlt */ |
| 7526 |
if (s->cpl != 0) { |
| 7527 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7528 |
} else {
|
| 7529 |
gen_update_cc_op(s); |
| 7530 |
gen_jmp_im(pc_start - s->cs_base); |
| 7531 |
gen_helper_hlt(cpu_env, tcg_const_i32(s->pc - pc_start)); |
| 7532 |
s->is_jmp = DISAS_TB_JUMP; |
| 7533 |
} |
| 7534 |
break;
|
| 7535 |
case 0x100: |
| 7536 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7537 |
mod = (modrm >> 6) & 3; |
| 7538 |
op = (modrm >> 3) & 7; |
| 7539 |
switch(op) {
|
| 7540 |
case 0: /* sldt */ |
| 7541 |
if (!s->pe || s->vm86)
|
| 7542 |
goto illegal_op;
|
| 7543 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_LDTR_READ); |
| 7544 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,ldt.selector));
|
| 7545 |
ot = OT_WORD; |
| 7546 |
if (mod == 3) |
| 7547 |
ot += s->dflag; |
| 7548 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 1);
|
| 7549 |
break;
|
| 7550 |
case 2: /* lldt */ |
| 7551 |
if (!s->pe || s->vm86)
|
| 7552 |
goto illegal_op;
|
| 7553 |
if (s->cpl != 0) { |
| 7554 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7555 |
} else {
|
| 7556 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_LDTR_WRITE); |
| 7557 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 7558 |
gen_jmp_im(pc_start - s->cs_base); |
| 7559 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 7560 |
gen_helper_lldt(cpu_env, cpu_tmp2_i32); |
| 7561 |
} |
| 7562 |
break;
|
| 7563 |
case 1: /* str */ |
| 7564 |
if (!s->pe || s->vm86)
|
| 7565 |
goto illegal_op;
|
| 7566 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_TR_READ); |
| 7567 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,tr.selector));
|
| 7568 |
ot = OT_WORD; |
| 7569 |
if (mod == 3) |
| 7570 |
ot += s->dflag; |
| 7571 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 1);
|
| 7572 |
break;
|
| 7573 |
case 3: /* ltr */ |
| 7574 |
if (!s->pe || s->vm86)
|
| 7575 |
goto illegal_op;
|
| 7576 |
if (s->cpl != 0) { |
| 7577 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7578 |
} else {
|
| 7579 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_TR_WRITE); |
| 7580 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 7581 |
gen_jmp_im(pc_start - s->cs_base); |
| 7582 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 7583 |
gen_helper_ltr(cpu_env, cpu_tmp2_i32); |
| 7584 |
} |
| 7585 |
break;
|
| 7586 |
case 4: /* verr */ |
| 7587 |
case 5: /* verw */ |
| 7588 |
if (!s->pe || s->vm86)
|
| 7589 |
goto illegal_op;
|
| 7590 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 7591 |
gen_update_cc_op(s); |
| 7592 |
if (op == 4) { |
| 7593 |
gen_helper_verr(cpu_env, cpu_T[0]);
|
| 7594 |
} else {
|
| 7595 |
gen_helper_verw(cpu_env, cpu_T[0]);
|
| 7596 |
} |
| 7597 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7598 |
break;
|
| 7599 |
default:
|
| 7600 |
goto illegal_op;
|
| 7601 |
} |
| 7602 |
break;
|
| 7603 |
case 0x101: |
| 7604 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7605 |
mod = (modrm >> 6) & 3; |
| 7606 |
op = (modrm >> 3) & 7; |
| 7607 |
rm = modrm & 7;
|
| 7608 |
switch(op) {
|
| 7609 |
case 0: /* sgdt */ |
| 7610 |
if (mod == 3) |
| 7611 |
goto illegal_op;
|
| 7612 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_GDTR_READ); |
| 7613 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7614 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, gdt.limit));
|
| 7615 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 7616 |
gen_add_A0_im(s, 2);
|
| 7617 |
tcg_gen_ld_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, gdt.base));
|
| 7618 |
if (!s->dflag)
|
| 7619 |
gen_op_andl_T0_im(0xffffff);
|
| 7620 |
gen_op_st_T0_A0(CODE64(s) + OT_LONG + s->mem_index); |
| 7621 |
break;
|
| 7622 |
case 1: |
| 7623 |
if (mod == 3) { |
| 7624 |
switch (rm) {
|
| 7625 |
case 0: /* monitor */ |
| 7626 |
if (!(s->cpuid_ext_features & CPUID_EXT_MONITOR) ||
|
| 7627 |
s->cpl != 0)
|
| 7628 |
goto illegal_op;
|
| 7629 |
gen_update_cc_op(s); |
| 7630 |
gen_jmp_im(pc_start - s->cs_base); |
| 7631 |
#ifdef TARGET_X86_64
|
| 7632 |
if (s->aflag == 2) { |
| 7633 |
gen_op_movq_A0_reg(R_EAX); |
| 7634 |
} else
|
| 7635 |
#endif
|
| 7636 |
{
|
| 7637 |
gen_op_movl_A0_reg(R_EAX); |
| 7638 |
if (s->aflag == 0) |
| 7639 |
gen_op_andl_A0_ffff(); |
| 7640 |
} |
| 7641 |
gen_add_A0_ds_seg(s); |
| 7642 |
gen_helper_monitor(cpu_env, cpu_A0); |
| 7643 |
break;
|
| 7644 |
case 1: /* mwait */ |
| 7645 |
if (!(s->cpuid_ext_features & CPUID_EXT_MONITOR) ||
|
| 7646 |
s->cpl != 0)
|
| 7647 |
goto illegal_op;
|
| 7648 |
gen_update_cc_op(s); |
| 7649 |
gen_jmp_im(pc_start - s->cs_base); |
| 7650 |
gen_helper_mwait(cpu_env, tcg_const_i32(s->pc - pc_start)); |
| 7651 |
gen_eob(s); |
| 7652 |
break;
|
| 7653 |
case 2: /* clac */ |
| 7654 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_SMAP) ||
|
| 7655 |
s->cpl != 0) {
|
| 7656 |
goto illegal_op;
|
| 7657 |
} |
| 7658 |
gen_helper_clac(cpu_env); |
| 7659 |
gen_jmp_im(s->pc - s->cs_base); |
| 7660 |
gen_eob(s); |
| 7661 |
break;
|
| 7662 |
case 3: /* stac */ |
| 7663 |
if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_SMAP) ||
|
| 7664 |
s->cpl != 0) {
|
| 7665 |
goto illegal_op;
|
| 7666 |
} |
| 7667 |
gen_helper_stac(cpu_env); |
| 7668 |
gen_jmp_im(s->pc - s->cs_base); |
| 7669 |
gen_eob(s); |
| 7670 |
break;
|
| 7671 |
default:
|
| 7672 |
goto illegal_op;
|
| 7673 |
} |
| 7674 |
} else { /* sidt */ |
| 7675 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_IDTR_READ); |
| 7676 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7677 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, idt.limit));
|
| 7678 |
gen_op_st_T0_A0(OT_WORD + s->mem_index); |
| 7679 |
gen_add_A0_im(s, 2);
|
| 7680 |
tcg_gen_ld_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, idt.base));
|
| 7681 |
if (!s->dflag)
|
| 7682 |
gen_op_andl_T0_im(0xffffff);
|
| 7683 |
gen_op_st_T0_A0(CODE64(s) + OT_LONG + s->mem_index); |
| 7684 |
} |
| 7685 |
break;
|
| 7686 |
case 2: /* lgdt */ |
| 7687 |
case 3: /* lidt */ |
| 7688 |
if (mod == 3) { |
| 7689 |
gen_update_cc_op(s); |
| 7690 |
gen_jmp_im(pc_start - s->cs_base); |
| 7691 |
switch(rm) {
|
| 7692 |
case 0: /* VMRUN */ |
| 7693 |
if (!(s->flags & HF_SVME_MASK) || !s->pe)
|
| 7694 |
goto illegal_op;
|
| 7695 |
if (s->cpl != 0) { |
| 7696 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7697 |
break;
|
| 7698 |
} else {
|
| 7699 |
gen_helper_vmrun(cpu_env, tcg_const_i32(s->aflag), |
| 7700 |
tcg_const_i32(s->pc - pc_start)); |
| 7701 |
tcg_gen_exit_tb(0);
|
| 7702 |
s->is_jmp = DISAS_TB_JUMP; |
| 7703 |
} |
| 7704 |
break;
|
| 7705 |
case 1: /* VMMCALL */ |
| 7706 |
if (!(s->flags & HF_SVME_MASK))
|
| 7707 |
goto illegal_op;
|
| 7708 |
gen_helper_vmmcall(cpu_env); |
| 7709 |
break;
|
| 7710 |
case 2: /* VMLOAD */ |
| 7711 |
if (!(s->flags & HF_SVME_MASK) || !s->pe)
|
| 7712 |
goto illegal_op;
|
| 7713 |
if (s->cpl != 0) { |
| 7714 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7715 |
break;
|
| 7716 |
} else {
|
| 7717 |
gen_helper_vmload(cpu_env, tcg_const_i32(s->aflag)); |
| 7718 |
} |
| 7719 |
break;
|
| 7720 |
case 3: /* VMSAVE */ |
| 7721 |
if (!(s->flags & HF_SVME_MASK) || !s->pe)
|
| 7722 |
goto illegal_op;
|
| 7723 |
if (s->cpl != 0) { |
| 7724 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7725 |
break;
|
| 7726 |
} else {
|
| 7727 |
gen_helper_vmsave(cpu_env, tcg_const_i32(s->aflag)); |
| 7728 |
} |
| 7729 |
break;
|
| 7730 |
case 4: /* STGI */ |
| 7731 |
if ((!(s->flags & HF_SVME_MASK) &&
|
| 7732 |
!(s->cpuid_ext3_features & CPUID_EXT3_SKINIT)) || |
| 7733 |
!s->pe) |
| 7734 |
goto illegal_op;
|
| 7735 |
if (s->cpl != 0) { |
| 7736 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7737 |
break;
|
| 7738 |
} else {
|
| 7739 |
gen_helper_stgi(cpu_env); |
| 7740 |
} |
| 7741 |
break;
|
| 7742 |
case 5: /* CLGI */ |
| 7743 |
if (!(s->flags & HF_SVME_MASK) || !s->pe)
|
| 7744 |
goto illegal_op;
|
| 7745 |
if (s->cpl != 0) { |
| 7746 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7747 |
break;
|
| 7748 |
} else {
|
| 7749 |
gen_helper_clgi(cpu_env); |
| 7750 |
} |
| 7751 |
break;
|
| 7752 |
case 6: /* SKINIT */ |
| 7753 |
if ((!(s->flags & HF_SVME_MASK) &&
|
| 7754 |
!(s->cpuid_ext3_features & CPUID_EXT3_SKINIT)) || |
| 7755 |
!s->pe) |
| 7756 |
goto illegal_op;
|
| 7757 |
gen_helper_skinit(cpu_env); |
| 7758 |
break;
|
| 7759 |
case 7: /* INVLPGA */ |
| 7760 |
if (!(s->flags & HF_SVME_MASK) || !s->pe)
|
| 7761 |
goto illegal_op;
|
| 7762 |
if (s->cpl != 0) { |
| 7763 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7764 |
break;
|
| 7765 |
} else {
|
| 7766 |
gen_helper_invlpga(cpu_env, tcg_const_i32(s->aflag)); |
| 7767 |
} |
| 7768 |
break;
|
| 7769 |
default:
|
| 7770 |
goto illegal_op;
|
| 7771 |
} |
| 7772 |
} else if (s->cpl != 0) { |
| 7773 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7774 |
} else {
|
| 7775 |
gen_svm_check_intercept(s, pc_start, |
| 7776 |
op==2 ? SVM_EXIT_GDTR_WRITE : SVM_EXIT_IDTR_WRITE);
|
| 7777 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7778 |
gen_op_ld_T1_A0(OT_WORD + s->mem_index); |
| 7779 |
gen_add_A0_im(s, 2);
|
| 7780 |
gen_op_ld_T0_A0(CODE64(s) + OT_LONG + s->mem_index); |
| 7781 |
if (!s->dflag)
|
| 7782 |
gen_op_andl_T0_im(0xffffff);
|
| 7783 |
if (op == 2) { |
| 7784 |
tcg_gen_st_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,gdt.base));
|
| 7785 |
tcg_gen_st32_tl(cpu_T[1], cpu_env, offsetof(CPUX86State,gdt.limit));
|
| 7786 |
} else {
|
| 7787 |
tcg_gen_st_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,idt.base));
|
| 7788 |
tcg_gen_st32_tl(cpu_T[1], cpu_env, offsetof(CPUX86State,idt.limit));
|
| 7789 |
} |
| 7790 |
} |
| 7791 |
break;
|
| 7792 |
case 4: /* smsw */ |
| 7793 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_READ_CR0); |
| 7794 |
#if defined TARGET_X86_64 && defined HOST_WORDS_BIGENDIAN
|
| 7795 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,cr[0]) + 4); |
| 7796 |
#else
|
| 7797 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,cr[0])); |
| 7798 |
#endif
|
| 7799 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 1);
|
| 7800 |
break;
|
| 7801 |
case 6: /* lmsw */ |
| 7802 |
if (s->cpl != 0) { |
| 7803 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7804 |
} else {
|
| 7805 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_WRITE_CR0); |
| 7806 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 7807 |
gen_helper_lmsw(cpu_env, cpu_T[0]);
|
| 7808 |
gen_jmp_im(s->pc - s->cs_base); |
| 7809 |
gen_eob(s); |
| 7810 |
} |
| 7811 |
break;
|
| 7812 |
case 7: |
| 7813 |
if (mod != 3) { /* invlpg */ |
| 7814 |
if (s->cpl != 0) { |
| 7815 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7816 |
} else {
|
| 7817 |
gen_update_cc_op(s); |
| 7818 |
gen_jmp_im(pc_start - s->cs_base); |
| 7819 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7820 |
gen_helper_invlpg(cpu_env, cpu_A0); |
| 7821 |
gen_jmp_im(s->pc - s->cs_base); |
| 7822 |
gen_eob(s); |
| 7823 |
} |
| 7824 |
} else {
|
| 7825 |
switch (rm) {
|
| 7826 |
case 0: /* swapgs */ |
| 7827 |
#ifdef TARGET_X86_64
|
| 7828 |
if (CODE64(s)) {
|
| 7829 |
if (s->cpl != 0) { |
| 7830 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7831 |
} else {
|
| 7832 |
tcg_gen_ld_tl(cpu_T[0], cpu_env,
|
| 7833 |
offsetof(CPUX86State,segs[R_GS].base)); |
| 7834 |
tcg_gen_ld_tl(cpu_T[1], cpu_env,
|
| 7835 |
offsetof(CPUX86State,kernelgsbase)); |
| 7836 |
tcg_gen_st_tl(cpu_T[1], cpu_env,
|
| 7837 |
offsetof(CPUX86State,segs[R_GS].base)); |
| 7838 |
tcg_gen_st_tl(cpu_T[0], cpu_env,
|
| 7839 |
offsetof(CPUX86State,kernelgsbase)); |
| 7840 |
} |
| 7841 |
} else
|
| 7842 |
#endif
|
| 7843 |
{
|
| 7844 |
goto illegal_op;
|
| 7845 |
} |
| 7846 |
break;
|
| 7847 |
case 1: /* rdtscp */ |
| 7848 |
if (!(s->cpuid_ext2_features & CPUID_EXT2_RDTSCP))
|
| 7849 |
goto illegal_op;
|
| 7850 |
gen_update_cc_op(s); |
| 7851 |
gen_jmp_im(pc_start - s->cs_base); |
| 7852 |
if (use_icount)
|
| 7853 |
gen_io_start(); |
| 7854 |
gen_helper_rdtscp(cpu_env); |
| 7855 |
if (use_icount) {
|
| 7856 |
gen_io_end(); |
| 7857 |
gen_jmp(s, s->pc - s->cs_base); |
| 7858 |
} |
| 7859 |
break;
|
| 7860 |
default:
|
| 7861 |
goto illegal_op;
|
| 7862 |
} |
| 7863 |
} |
| 7864 |
break;
|
| 7865 |
default:
|
| 7866 |
goto illegal_op;
|
| 7867 |
} |
| 7868 |
break;
|
| 7869 |
case 0x108: /* invd */ |
| 7870 |
case 0x109: /* wbinvd */ |
| 7871 |
if (s->cpl != 0) { |
| 7872 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 7873 |
} else {
|
| 7874 |
gen_svm_check_intercept(s, pc_start, (b & 2) ? SVM_EXIT_INVD : SVM_EXIT_WBINVD);
|
| 7875 |
/* nothing to do */
|
| 7876 |
} |
| 7877 |
break;
|
| 7878 |
case 0x63: /* arpl or movslS (x86_64) */ |
| 7879 |
#ifdef TARGET_X86_64
|
| 7880 |
if (CODE64(s)) {
|
| 7881 |
int d_ot;
|
| 7882 |
/* d_ot is the size of destination */
|
| 7883 |
d_ot = dflag + OT_WORD; |
| 7884 |
|
| 7885 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7886 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 7887 |
mod = (modrm >> 6) & 3; |
| 7888 |
rm = (modrm & 7) | REX_B(s);
|
| 7889 |
|
| 7890 |
if (mod == 3) { |
| 7891 |
gen_op_mov_TN_reg(OT_LONG, 0, rm);
|
| 7892 |
/* sign extend */
|
| 7893 |
if (d_ot == OT_QUAD)
|
| 7894 |
tcg_gen_ext32s_tl(cpu_T[0], cpu_T[0]); |
| 7895 |
gen_op_mov_reg_T0(d_ot, reg); |
| 7896 |
} else {
|
| 7897 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7898 |
if (d_ot == OT_QUAD) {
|
| 7899 |
gen_op_lds_T0_A0(OT_LONG + s->mem_index); |
| 7900 |
} else {
|
| 7901 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 7902 |
} |
| 7903 |
gen_op_mov_reg_T0(d_ot, reg); |
| 7904 |
} |
| 7905 |
} else
|
| 7906 |
#endif
|
| 7907 |
{
|
| 7908 |
int label1;
|
| 7909 |
TCGv t0, t1, t2, a0; |
| 7910 |
|
| 7911 |
if (!s->pe || s->vm86)
|
| 7912 |
goto illegal_op;
|
| 7913 |
t0 = tcg_temp_local_new(); |
| 7914 |
t1 = tcg_temp_local_new(); |
| 7915 |
t2 = tcg_temp_local_new(); |
| 7916 |
ot = OT_WORD; |
| 7917 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7918 |
reg = (modrm >> 3) & 7; |
| 7919 |
mod = (modrm >> 6) & 3; |
| 7920 |
rm = modrm & 7;
|
| 7921 |
if (mod != 3) { |
| 7922 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7923 |
gen_op_ld_v(ot + s->mem_index, t0, cpu_A0); |
| 7924 |
a0 = tcg_temp_local_new(); |
| 7925 |
tcg_gen_mov_tl(a0, cpu_A0); |
| 7926 |
} else {
|
| 7927 |
gen_op_mov_v_reg(ot, t0, rm); |
| 7928 |
TCGV_UNUSED(a0); |
| 7929 |
} |
| 7930 |
gen_op_mov_v_reg(ot, t1, reg); |
| 7931 |
tcg_gen_andi_tl(cpu_tmp0, t0, 3);
|
| 7932 |
tcg_gen_andi_tl(t1, t1, 3);
|
| 7933 |
tcg_gen_movi_tl(t2, 0);
|
| 7934 |
label1 = gen_new_label(); |
| 7935 |
tcg_gen_brcond_tl(TCG_COND_GE, cpu_tmp0, t1, label1); |
| 7936 |
tcg_gen_andi_tl(t0, t0, ~3);
|
| 7937 |
tcg_gen_or_tl(t0, t0, t1); |
| 7938 |
tcg_gen_movi_tl(t2, CC_Z); |
| 7939 |
gen_set_label(label1); |
| 7940 |
if (mod != 3) { |
| 7941 |
gen_op_st_v(ot + s->mem_index, t0, a0); |
| 7942 |
tcg_temp_free(a0); |
| 7943 |
} else {
|
| 7944 |
gen_op_mov_reg_v(ot, rm, t0); |
| 7945 |
} |
| 7946 |
gen_compute_eflags(s); |
| 7947 |
tcg_gen_andi_tl(cpu_cc_src, cpu_cc_src, ~CC_Z); |
| 7948 |
tcg_gen_or_tl(cpu_cc_src, cpu_cc_src, t2); |
| 7949 |
tcg_temp_free(t0); |
| 7950 |
tcg_temp_free(t1); |
| 7951 |
tcg_temp_free(t2); |
| 7952 |
} |
| 7953 |
break;
|
| 7954 |
case 0x102: /* lar */ |
| 7955 |
case 0x103: /* lsl */ |
| 7956 |
{
|
| 7957 |
int label1;
|
| 7958 |
TCGv t0; |
| 7959 |
if (!s->pe || s->vm86)
|
| 7960 |
goto illegal_op;
|
| 7961 |
ot = dflag ? OT_LONG : OT_WORD; |
| 7962 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7963 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 7964 |
gen_ldst_modrm(env, s, modrm, OT_WORD, OR_TMP0, 0);
|
| 7965 |
t0 = tcg_temp_local_new(); |
| 7966 |
gen_update_cc_op(s); |
| 7967 |
if (b == 0x102) { |
| 7968 |
gen_helper_lar(t0, cpu_env, cpu_T[0]);
|
| 7969 |
} else {
|
| 7970 |
gen_helper_lsl(t0, cpu_env, cpu_T[0]);
|
| 7971 |
} |
| 7972 |
tcg_gen_andi_tl(cpu_tmp0, cpu_cc_src, CC_Z); |
| 7973 |
label1 = gen_new_label(); |
| 7974 |
tcg_gen_brcondi_tl(TCG_COND_EQ, cpu_tmp0, 0, label1);
|
| 7975 |
gen_op_mov_reg_v(ot, reg, t0); |
| 7976 |
gen_set_label(label1); |
| 7977 |
set_cc_op(s, CC_OP_EFLAGS); |
| 7978 |
tcg_temp_free(t0); |
| 7979 |
} |
| 7980 |
break;
|
| 7981 |
case 0x118: |
| 7982 |
modrm = cpu_ldub_code(env, s->pc++); |
| 7983 |
mod = (modrm >> 6) & 3; |
| 7984 |
op = (modrm >> 3) & 7; |
| 7985 |
switch(op) {
|
| 7986 |
case 0: /* prefetchnta */ |
| 7987 |
case 1: /* prefetchnt0 */ |
| 7988 |
case 2: /* prefetchnt0 */ |
| 7989 |
case 3: /* prefetchnt0 */ |
| 7990 |
if (mod == 3) |
| 7991 |
goto illegal_op;
|
| 7992 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 7993 |
/* nothing more to do */
|
| 7994 |
break;
|
| 7995 |
default: /* nop (multi byte) */ |
| 7996 |
gen_nop_modrm(env, s, modrm); |
| 7997 |
break;
|
| 7998 |
} |
| 7999 |
break;
|
| 8000 |
case 0x119 ... 0x11f: /* nop (multi byte) */ |
| 8001 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8002 |
gen_nop_modrm(env, s, modrm); |
| 8003 |
break;
|
| 8004 |
case 0x120: /* mov reg, crN */ |
| 8005 |
case 0x122: /* mov crN, reg */ |
| 8006 |
if (s->cpl != 0) { |
| 8007 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 8008 |
} else {
|
| 8009 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8010 |
/* Ignore the mod bits (assume (modrm&0xc0)==0xc0).
|
| 8011 |
* AMD documentation (24594.pdf) and testing of
|
| 8012 |
* intel 386 and 486 processors all show that the mod bits
|
| 8013 |
* are assumed to be 1's, regardless of actual values.
|
| 8014 |
*/
|
| 8015 |
rm = (modrm & 7) | REX_B(s);
|
| 8016 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 8017 |
if (CODE64(s))
|
| 8018 |
ot = OT_QUAD; |
| 8019 |
else
|
| 8020 |
ot = OT_LONG; |
| 8021 |
if ((prefixes & PREFIX_LOCK) && (reg == 0) && |
| 8022 |
(s->cpuid_ext3_features & CPUID_EXT3_CR8LEG)) {
|
| 8023 |
reg = 8;
|
| 8024 |
} |
| 8025 |
switch(reg) {
|
| 8026 |
case 0: |
| 8027 |
case 2: |
| 8028 |
case 3: |
| 8029 |
case 4: |
| 8030 |
case 8: |
| 8031 |
gen_update_cc_op(s); |
| 8032 |
gen_jmp_im(pc_start - s->cs_base); |
| 8033 |
if (b & 2) { |
| 8034 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 8035 |
gen_helper_write_crN(cpu_env, tcg_const_i32(reg), |
| 8036 |
cpu_T[0]);
|
| 8037 |
gen_jmp_im(s->pc - s->cs_base); |
| 8038 |
gen_eob(s); |
| 8039 |
} else {
|
| 8040 |
gen_helper_read_crN(cpu_T[0], cpu_env, tcg_const_i32(reg));
|
| 8041 |
gen_op_mov_reg_T0(ot, rm); |
| 8042 |
} |
| 8043 |
break;
|
| 8044 |
default:
|
| 8045 |
goto illegal_op;
|
| 8046 |
} |
| 8047 |
} |
| 8048 |
break;
|
| 8049 |
case 0x121: /* mov reg, drN */ |
| 8050 |
case 0x123: /* mov drN, reg */ |
| 8051 |
if (s->cpl != 0) { |
| 8052 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 8053 |
} else {
|
| 8054 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8055 |
/* Ignore the mod bits (assume (modrm&0xc0)==0xc0).
|
| 8056 |
* AMD documentation (24594.pdf) and testing of
|
| 8057 |
* intel 386 and 486 processors all show that the mod bits
|
| 8058 |
* are assumed to be 1's, regardless of actual values.
|
| 8059 |
*/
|
| 8060 |
rm = (modrm & 7) | REX_B(s);
|
| 8061 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 8062 |
if (CODE64(s))
|
| 8063 |
ot = OT_QUAD; |
| 8064 |
else
|
| 8065 |
ot = OT_LONG; |
| 8066 |
/* XXX: do it dynamically with CR4.DE bit */
|
| 8067 |
if (reg == 4 || reg == 5 || reg >= 8) |
| 8068 |
goto illegal_op;
|
| 8069 |
if (b & 2) { |
| 8070 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_WRITE_DR0 + reg); |
| 8071 |
gen_op_mov_TN_reg(ot, 0, rm);
|
| 8072 |
gen_helper_movl_drN_T0(cpu_env, tcg_const_i32(reg), cpu_T[0]);
|
| 8073 |
gen_jmp_im(s->pc - s->cs_base); |
| 8074 |
gen_eob(s); |
| 8075 |
} else {
|
| 8076 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_READ_DR0 + reg); |
| 8077 |
tcg_gen_ld_tl(cpu_T[0], cpu_env, offsetof(CPUX86State,dr[reg]));
|
| 8078 |
gen_op_mov_reg_T0(ot, rm); |
| 8079 |
} |
| 8080 |
} |
| 8081 |
break;
|
| 8082 |
case 0x106: /* clts */ |
| 8083 |
if (s->cpl != 0) { |
| 8084 |
gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); |
| 8085 |
} else {
|
| 8086 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_WRITE_CR0); |
| 8087 |
gen_helper_clts(cpu_env); |
| 8088 |
/* abort block because static cpu state changed */
|
| 8089 |
gen_jmp_im(s->pc - s->cs_base); |
| 8090 |
gen_eob(s); |
| 8091 |
} |
| 8092 |
break;
|
| 8093 |
/* MMX/3DNow!/SSE/SSE2/SSE3/SSSE3/SSE4 support */
|
| 8094 |
case 0x1c3: /* MOVNTI reg, mem */ |
| 8095 |
if (!(s->cpuid_features & CPUID_SSE2))
|
| 8096 |
goto illegal_op;
|
| 8097 |
ot = s->dflag == 2 ? OT_QUAD : OT_LONG;
|
| 8098 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8099 |
mod = (modrm >> 6) & 3; |
| 8100 |
if (mod == 3) |
| 8101 |
goto illegal_op;
|
| 8102 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 8103 |
/* generate a generic store */
|
| 8104 |
gen_ldst_modrm(env, s, modrm, ot, reg, 1);
|
| 8105 |
break;
|
| 8106 |
case 0x1ae: |
| 8107 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8108 |
mod = (modrm >> 6) & 3; |
| 8109 |
op = (modrm >> 3) & 7; |
| 8110 |
switch(op) {
|
| 8111 |
case 0: /* fxsave */ |
| 8112 |
if (mod == 3 || !(s->cpuid_features & CPUID_FXSR) || |
| 8113 |
(s->prefix & PREFIX_LOCK)) |
| 8114 |
goto illegal_op;
|
| 8115 |
if ((s->flags & HF_EM_MASK) || (s->flags & HF_TS_MASK)) {
|
| 8116 |
gen_exception(s, EXCP07_PREX, pc_start - s->cs_base); |
| 8117 |
break;
|
| 8118 |
} |
| 8119 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 8120 |
gen_update_cc_op(s); |
| 8121 |
gen_jmp_im(pc_start - s->cs_base); |
| 8122 |
gen_helper_fxsave(cpu_env, cpu_A0, tcg_const_i32((s->dflag == 2)));
|
| 8123 |
break;
|
| 8124 |
case 1: /* fxrstor */ |
| 8125 |
if (mod == 3 || !(s->cpuid_features & CPUID_FXSR) || |
| 8126 |
(s->prefix & PREFIX_LOCK)) |
| 8127 |
goto illegal_op;
|
| 8128 |
if ((s->flags & HF_EM_MASK) || (s->flags & HF_TS_MASK)) {
|
| 8129 |
gen_exception(s, EXCP07_PREX, pc_start - s->cs_base); |
| 8130 |
break;
|
| 8131 |
} |
| 8132 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 8133 |
gen_update_cc_op(s); |
| 8134 |
gen_jmp_im(pc_start - s->cs_base); |
| 8135 |
gen_helper_fxrstor(cpu_env, cpu_A0, |
| 8136 |
tcg_const_i32((s->dflag == 2)));
|
| 8137 |
break;
|
| 8138 |
case 2: /* ldmxcsr */ |
| 8139 |
case 3: /* stmxcsr */ |
| 8140 |
if (s->flags & HF_TS_MASK) {
|
| 8141 |
gen_exception(s, EXCP07_PREX, pc_start - s->cs_base); |
| 8142 |
break;
|
| 8143 |
} |
| 8144 |
if ((s->flags & HF_EM_MASK) || !(s->flags & HF_OSFXSR_MASK) ||
|
| 8145 |
mod == 3)
|
| 8146 |
goto illegal_op;
|
| 8147 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 8148 |
if (op == 2) { |
| 8149 |
gen_op_ld_T0_A0(OT_LONG + s->mem_index); |
| 8150 |
tcg_gen_trunc_tl_i32(cpu_tmp2_i32, cpu_T[0]);
|
| 8151 |
gen_helper_ldmxcsr(cpu_env, cpu_tmp2_i32); |
| 8152 |
} else {
|
| 8153 |
tcg_gen_ld32u_tl(cpu_T[0], cpu_env, offsetof(CPUX86State, mxcsr));
|
| 8154 |
gen_op_st_T0_A0(OT_LONG + s->mem_index); |
| 8155 |
} |
| 8156 |
break;
|
| 8157 |
case 5: /* lfence */ |
| 8158 |
case 6: /* mfence */ |
| 8159 |
if ((modrm & 0xc7) != 0xc0 || !(s->cpuid_features & CPUID_SSE2)) |
| 8160 |
goto illegal_op;
|
| 8161 |
break;
|
| 8162 |
case 7: /* sfence / clflush */ |
| 8163 |
if ((modrm & 0xc7) == 0xc0) { |
| 8164 |
/* sfence */
|
| 8165 |
/* XXX: also check for cpuid_ext2_features & CPUID_EXT2_EMMX */
|
| 8166 |
if (!(s->cpuid_features & CPUID_SSE))
|
| 8167 |
goto illegal_op;
|
| 8168 |
} else {
|
| 8169 |
/* clflush */
|
| 8170 |
if (!(s->cpuid_features & CPUID_CLFLUSH))
|
| 8171 |
goto illegal_op;
|
| 8172 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 8173 |
} |
| 8174 |
break;
|
| 8175 |
default:
|
| 8176 |
goto illegal_op;
|
| 8177 |
} |
| 8178 |
break;
|
| 8179 |
case 0x10d: /* 3DNow! prefetch(w) */ |
| 8180 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8181 |
mod = (modrm >> 6) & 3; |
| 8182 |
if (mod == 3) |
| 8183 |
goto illegal_op;
|
| 8184 |
gen_lea_modrm(env, s, modrm, ®_addr, &offset_addr); |
| 8185 |
/* ignore for now */
|
| 8186 |
break;
|
| 8187 |
case 0x1aa: /* rsm */ |
| 8188 |
gen_svm_check_intercept(s, pc_start, SVM_EXIT_RSM); |
| 8189 |
if (!(s->flags & HF_SMM_MASK))
|
| 8190 |
goto illegal_op;
|
| 8191 |
gen_update_cc_op(s); |
| 8192 |
gen_jmp_im(s->pc - s->cs_base); |
| 8193 |
gen_helper_rsm(cpu_env); |
| 8194 |
gen_eob(s); |
| 8195 |
break;
|
| 8196 |
case 0x1b8: /* SSE4.2 popcnt */ |
| 8197 |
if ((prefixes & (PREFIX_REPZ | PREFIX_LOCK | PREFIX_REPNZ)) !=
|
| 8198 |
PREFIX_REPZ) |
| 8199 |
goto illegal_op;
|
| 8200 |
if (!(s->cpuid_ext_features & CPUID_EXT_POPCNT))
|
| 8201 |
goto illegal_op;
|
| 8202 |
|
| 8203 |
modrm = cpu_ldub_code(env, s->pc++); |
| 8204 |
reg = ((modrm >> 3) & 7) | rex_r; |
| 8205 |
|
| 8206 |
if (s->prefix & PREFIX_DATA)
|
| 8207 |
ot = OT_WORD; |
| 8208 |
else if (s->dflag != 2) |
| 8209 |
ot = OT_LONG; |
| 8210 |
else
|
| 8211 |
ot = OT_QUAD; |
| 8212 |
|
| 8213 |
gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
|
| 8214 |
gen_helper_popcnt(cpu_T[0], cpu_env, cpu_T[0], tcg_const_i32(ot)); |
| 8215 |
gen_op_mov_reg_T0(ot, reg); |
| 8216 |
|
| 8217 |
set_cc_op(s, CC_OP_EFLAGS); |
| 8218 |
break;
|
| 8219 |
case 0x10e ... 0x10f: |
| 8220 |
/* 3DNow! instructions, ignore prefixes */
|
| 8221 |
s->prefix &= ~(PREFIX_REPZ | PREFIX_REPNZ | PREFIX_DATA); |
| 8222 |
case 0x110 ... 0x117: |
| 8223 |
case 0x128 ... 0x12f: |
| 8224 |
case 0x138 ... 0x13a: |
| 8225 |
case 0x150 ... 0x179: |
| 8226 |
case 0x17c ... 0x17f: |
| 8227 |
case 0x1c2: |
| 8228 |
case 0x1c4 ... 0x1c6: |
| 8229 |
case 0x1d0 ... 0x1fe: |
| 8230 |
gen_sse(env, s, b, pc_start, rex_r); |
| 8231 |
break;
|
| 8232 |
default:
|
| 8233 |
goto illegal_op;
|
| 8234 |
} |
| 8235 |
/* lock generation */
|
| 8236 |
if (s->prefix & PREFIX_LOCK)
|
| 8237 |
gen_helper_unlock(); |
| 8238 |
return s->pc;
|
| 8239 |
illegal_op:
|
| 8240 |
if (s->prefix & PREFIX_LOCK)
|
| 8241 |
gen_helper_unlock(); |
| 8242 |
/* XXX: ensure that no lock was generated */
|
| 8243 |
gen_exception(s, EXCP06_ILLOP, pc_start - s->cs_base); |
| 8244 |
return s->pc;
|
| 8245 |
} |
| 8246 |
|
| 8247 |
void optimize_flags_init(void) |
| 8248 |
{
|
| 8249 |
cpu_env = tcg_global_reg_new_ptr(TCG_AREG0, "env");
|
| 8250 |
cpu_cc_op = tcg_global_mem_new_i32(TCG_AREG0, |
| 8251 |
offsetof(CPUX86State, cc_op), "cc_op");
|
| 8252 |
cpu_cc_dst = tcg_global_mem_new(TCG_AREG0, offsetof(CPUX86State, cc_dst), |
| 8253 |
"cc_dst");
|
| 8254 |
cpu_cc_src = tcg_global_mem_new(TCG_AREG0, offsetof(CPUX86State, cc_src), |
| 8255 |
"cc_src");
|
| 8256 |
cpu_cc_src2 = tcg_global_mem_new(TCG_AREG0, offsetof(CPUX86State, cc_src2), |
| 8257 |
"cc_src2");
|
| 8258 |
|
| 8259 |
#ifdef TARGET_X86_64
|
| 8260 |
cpu_regs[R_EAX] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8261 |
offsetof(CPUX86State, regs[R_EAX]), "rax");
|
| 8262 |
cpu_regs[R_ECX] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8263 |
offsetof(CPUX86State, regs[R_ECX]), "rcx");
|
| 8264 |
cpu_regs[R_EDX] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8265 |
offsetof(CPUX86State, regs[R_EDX]), "rdx");
|
| 8266 |
cpu_regs[R_EBX] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8267 |
offsetof(CPUX86State, regs[R_EBX]), "rbx");
|
| 8268 |
cpu_regs[R_ESP] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8269 |
offsetof(CPUX86State, regs[R_ESP]), "rsp");
|
| 8270 |
cpu_regs[R_EBP] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8271 |
offsetof(CPUX86State, regs[R_EBP]), "rbp");
|
| 8272 |
cpu_regs[R_ESI] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8273 |
offsetof(CPUX86State, regs[R_ESI]), "rsi");
|
| 8274 |
cpu_regs[R_EDI] = tcg_global_mem_new_i64(TCG_AREG0, |
| 8275 |
offsetof(CPUX86State, regs[R_EDI]), "rdi");
|
| 8276 |
cpu_regs[8] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8277 |
offsetof(CPUX86State, regs[8]), "r8"); |
| 8278 |
cpu_regs[9] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8279 |
offsetof(CPUX86State, regs[9]), "r9"); |
| 8280 |
cpu_regs[10] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8281 |
offsetof(CPUX86State, regs[10]), "r10"); |
| 8282 |
cpu_regs[11] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8283 |
offsetof(CPUX86State, regs[11]), "r11"); |
| 8284 |
cpu_regs[12] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8285 |
offsetof(CPUX86State, regs[12]), "r12"); |
| 8286 |
cpu_regs[13] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8287 |
offsetof(CPUX86State, regs[13]), "r13"); |
| 8288 |
cpu_regs[14] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8289 |
offsetof(CPUX86State, regs[14]), "r14"); |
| 8290 |
cpu_regs[15] = tcg_global_mem_new_i64(TCG_AREG0,
|
| 8291 |
offsetof(CPUX86State, regs[15]), "r15"); |
| 8292 |
#else
|
| 8293 |
cpu_regs[R_EAX] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8294 |
offsetof(CPUX86State, regs[R_EAX]), "eax");
|
| 8295 |
cpu_regs[R_ECX] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8296 |
offsetof(CPUX86State, regs[R_ECX]), "ecx");
|
| 8297 |
cpu_regs[R_EDX] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8298 |
offsetof(CPUX86State, regs[R_EDX]), "edx");
|
| 8299 |
cpu_regs[R_EBX] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8300 |
offsetof(CPUX86State, regs[R_EBX]), "ebx");
|
| 8301 |
cpu_regs[R_ESP] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8302 |
offsetof(CPUX86State, regs[R_ESP]), "esp");
|
| 8303 |
cpu_regs[R_EBP] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8304 |
offsetof(CPUX86State, regs[R_EBP]), "ebp");
|
| 8305 |
cpu_regs[R_ESI] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8306 |
offsetof(CPUX86State, regs[R_ESI]), "esi");
|
| 8307 |
cpu_regs[R_EDI] = tcg_global_mem_new_i32(TCG_AREG0, |
| 8308 |
offsetof(CPUX86State, regs[R_EDI]), "edi");
|
| 8309 |
#endif
|
| 8310 |
|
| 8311 |
/* register helpers */
|
| 8312 |
#define GEN_HELPER 2 |
| 8313 |
#include "helper.h" |
| 8314 |
} |
| 8315 |
|
| 8316 |
/* generate intermediate code in gen_opc_buf and gen_opparam_buf for
|
| 8317 |
basic block 'tb'. If search_pc is TRUE, also generate PC
|
| 8318 |
information for each intermediate instruction. */
|
| 8319 |
static inline void gen_intermediate_code_internal(CPUX86State *env, |
| 8320 |
TranslationBlock *tb, |
| 8321 |
int search_pc)
|
| 8322 |
{
|
| 8323 |
DisasContext dc1, *dc = &dc1; |
| 8324 |
target_ulong pc_ptr; |
| 8325 |
uint16_t *gen_opc_end; |
| 8326 |
CPUBreakpoint *bp; |
| 8327 |
int j, lj;
|
| 8328 |
uint64_t flags; |
| 8329 |
target_ulong pc_start; |
| 8330 |
target_ulong cs_base; |
| 8331 |
int num_insns;
|
| 8332 |
int max_insns;
|
| 8333 |
|
| 8334 |
/* generate intermediate code */
|
| 8335 |
pc_start = tb->pc; |
| 8336 |
cs_base = tb->cs_base; |
| 8337 |
flags = tb->flags; |
| 8338 |
|
| 8339 |
dc->pe = (flags >> HF_PE_SHIFT) & 1;
|
| 8340 |
dc->code32 = (flags >> HF_CS32_SHIFT) & 1;
|
| 8341 |
dc->ss32 = (flags >> HF_SS32_SHIFT) & 1;
|
| 8342 |
dc->addseg = (flags >> HF_ADDSEG_SHIFT) & 1;
|
| 8343 |
dc->f_st = 0;
|
| 8344 |
dc->vm86 = (flags >> VM_SHIFT) & 1;
|
| 8345 |
dc->cpl = (flags >> HF_CPL_SHIFT) & 3;
|
| 8346 |
dc->iopl = (flags >> IOPL_SHIFT) & 3;
|
| 8347 |
dc->tf = (flags >> TF_SHIFT) & 1;
|
| 8348 |
dc->singlestep_enabled = env->singlestep_enabled; |
| 8349 |
dc->cc_op = CC_OP_DYNAMIC; |
| 8350 |
dc->cc_op_dirty = false;
|
| 8351 |
dc->cs_base = cs_base; |
| 8352 |
dc->tb = tb; |
| 8353 |
dc->popl_esp_hack = 0;
|
| 8354 |
/* select memory access functions */
|
| 8355 |
dc->mem_index = 0;
|
| 8356 |
if (flags & HF_SOFTMMU_MASK) {
|
| 8357 |
dc->mem_index = (cpu_mmu_index(env) + 1) << 2; |
| 8358 |
} |
| 8359 |
dc->cpuid_features = env->cpuid_features; |
| 8360 |
dc->cpuid_ext_features = env->cpuid_ext_features; |
| 8361 |
dc->cpuid_ext2_features = env->cpuid_ext2_features; |
| 8362 |
dc->cpuid_ext3_features = env->cpuid_ext3_features; |
| 8363 |
dc->cpuid_7_0_ebx_features = env->cpuid_7_0_ebx_features; |
| 8364 |
#ifdef TARGET_X86_64
|
| 8365 |
dc->lma = (flags >> HF_LMA_SHIFT) & 1;
|
| 8366 |
dc->code64 = (flags >> HF_CS64_SHIFT) & 1;
|
| 8367 |
#endif
|
| 8368 |
dc->flags = flags; |
| 8369 |
dc->jmp_opt = !(dc->tf || env->singlestep_enabled || |
| 8370 |
(flags & HF_INHIBIT_IRQ_MASK) |
| 8371 |
#ifndef CONFIG_SOFTMMU
|
| 8372 |
|| (flags & HF_SOFTMMU_MASK) |
| 8373 |
#endif
|
| 8374 |
); |
| 8375 |
#if 0
|
| 8376 |
/* check addseg logic */
|
| 8377 |
if (!dc->addseg && (dc->vm86 || !dc->pe || !dc->code32))
|
| 8378 |
printf("ERROR addseg\n");
|
| 8379 |
#endif
|
| 8380 |
|
| 8381 |
cpu_T[0] = tcg_temp_new();
|
| 8382 |
cpu_T[1] = tcg_temp_new();
|
| 8383 |
cpu_A0 = tcg_temp_new(); |
| 8384 |
|
| 8385 |
cpu_tmp0 = tcg_temp_new(); |
| 8386 |
cpu_tmp1_i64 = tcg_temp_new_i64(); |
| 8387 |
cpu_tmp2_i32 = tcg_temp_new_i32(); |
| 8388 |
cpu_tmp3_i32 = tcg_temp_new_i32(); |
| 8389 |
cpu_tmp4 = tcg_temp_new(); |
| 8390 |
cpu_tmp5 = tcg_temp_new(); |
| 8391 |
cpu_ptr0 = tcg_temp_new_ptr(); |
| 8392 |
cpu_ptr1 = tcg_temp_new_ptr(); |
| 8393 |
cpu_cc_srcT = tcg_temp_local_new(); |
| 8394 |
|
| 8395 |
gen_opc_end = tcg_ctx.gen_opc_buf + OPC_MAX_SIZE; |
| 8396 |
|
| 8397 |
dc->is_jmp = DISAS_NEXT; |
| 8398 |
pc_ptr = pc_start; |
| 8399 |
lj = -1;
|
| 8400 |
num_insns = 0;
|
| 8401 |
max_insns = tb->cflags & CF_COUNT_MASK; |
| 8402 |
if (max_insns == 0) |
| 8403 |
max_insns = CF_COUNT_MASK; |
| 8404 |
|
| 8405 |
gen_icount_start(); |
| 8406 |
for(;;) {
|
| 8407 |
if (unlikely(!QTAILQ_EMPTY(&env->breakpoints))) {
|
| 8408 |
QTAILQ_FOREACH(bp, &env->breakpoints, entry) {
|
| 8409 |
if (bp->pc == pc_ptr &&
|
| 8410 |
!((bp->flags & BP_CPU) && (tb->flags & HF_RF_MASK))) {
|
| 8411 |
gen_debug(dc, pc_ptr - dc->cs_base); |
| 8412 |
break;
|
| 8413 |
} |
| 8414 |
} |
| 8415 |
} |
| 8416 |
if (search_pc) {
|
| 8417 |
j = tcg_ctx.gen_opc_ptr - tcg_ctx.gen_opc_buf; |
| 8418 |
if (lj < j) {
|
| 8419 |
lj++; |
| 8420 |
while (lj < j)
|
| 8421 |
tcg_ctx.gen_opc_instr_start[lj++] = 0;
|
| 8422 |
} |
| 8423 |
tcg_ctx.gen_opc_pc[lj] = pc_ptr; |
| 8424 |
gen_opc_cc_op[lj] = dc->cc_op; |
| 8425 |
tcg_ctx.gen_opc_instr_start[lj] = 1;
|
| 8426 |
tcg_ctx.gen_opc_icount[lj] = num_insns; |
| 8427 |
} |
| 8428 |
if (num_insns + 1 == max_insns && (tb->cflags & CF_LAST_IO)) |
| 8429 |
gen_io_start(); |
| 8430 |
|
| 8431 |
pc_ptr = disas_insn(env, dc, pc_ptr); |
| 8432 |
num_insns++; |
| 8433 |
/* stop translation if indicated */
|
| 8434 |
if (dc->is_jmp)
|
| 8435 |
break;
|
| 8436 |
/* if single step mode, we generate only one instruction and
|
| 8437 |
generate an exception */
|
| 8438 |
/* if irq were inhibited with HF_INHIBIT_IRQ_MASK, we clear
|
| 8439 |
the flag and abort the translation to give the irqs a
|
| 8440 |
change to be happen */
|
| 8441 |
if (dc->tf || dc->singlestep_enabled ||
|
| 8442 |
(flags & HF_INHIBIT_IRQ_MASK)) {
|
| 8443 |
gen_jmp_im(pc_ptr - dc->cs_base); |
| 8444 |
gen_eob(dc); |
| 8445 |
break;
|
| 8446 |
} |
| 8447 |
/* if too long translation, stop generation too */
|
| 8448 |
if (tcg_ctx.gen_opc_ptr >= gen_opc_end ||
|
| 8449 |
(pc_ptr - pc_start) >= (TARGET_PAGE_SIZE - 32) ||
|
| 8450 |
num_insns >= max_insns) {
|
| 8451 |
gen_jmp_im(pc_ptr - dc->cs_base); |
| 8452 |
gen_eob(dc); |
| 8453 |
break;
|
| 8454 |
} |
| 8455 |
if (singlestep) {
|
| 8456 |
gen_jmp_im(pc_ptr - dc->cs_base); |
| 8457 |
gen_eob(dc); |
| 8458 |
break;
|
| 8459 |
} |
| 8460 |
} |
| 8461 |
if (tb->cflags & CF_LAST_IO)
|
| 8462 |
gen_io_end(); |
| 8463 |
gen_icount_end(tb, num_insns); |
| 8464 |
*tcg_ctx.gen_opc_ptr = INDEX_op_end; |
| 8465 |
/* we don't forget to fill the last values */
|
| 8466 |
if (search_pc) {
|
| 8467 |
j = tcg_ctx.gen_opc_ptr - tcg_ctx.gen_opc_buf; |
| 8468 |
lj++; |
| 8469 |
while (lj <= j)
|
| 8470 |
tcg_ctx.gen_opc_instr_start[lj++] = 0;
|
| 8471 |
} |
| 8472 |
|
| 8473 |
#ifdef DEBUG_DISAS
|
| 8474 |
if (qemu_loglevel_mask(CPU_LOG_TB_IN_ASM)) {
|
| 8475 |
int disas_flags;
|
| 8476 |
qemu_log("----------------\n");
|
| 8477 |
qemu_log("IN: %s\n", lookup_symbol(pc_start));
|
| 8478 |
#ifdef TARGET_X86_64
|
| 8479 |
if (dc->code64)
|
| 8480 |
disas_flags = 2;
|
| 8481 |
else
|
| 8482 |
#endif
|
| 8483 |
disas_flags = !dc->code32; |
| 8484 |
log_target_disas(env, pc_start, pc_ptr - pc_start, disas_flags); |
| 8485 |
qemu_log("\n");
|
| 8486 |
} |
| 8487 |
#endif
|
| 8488 |
|
| 8489 |
if (!search_pc) {
|
| 8490 |
tb->size = pc_ptr - pc_start; |
| 8491 |
tb->icount = num_insns; |
| 8492 |
} |
| 8493 |
} |
| 8494 |
|
| 8495 |
void gen_intermediate_code(CPUX86State *env, TranslationBlock *tb)
|
| 8496 |
{
|
| 8497 |
gen_intermediate_code_internal(env, tb, 0);
|
| 8498 |
} |
| 8499 |
|
| 8500 |
void gen_intermediate_code_pc(CPUX86State *env, TranslationBlock *tb)
|
| 8501 |
{
|
| 8502 |
gen_intermediate_code_internal(env, tb, 1);
|
| 8503 |
} |
| 8504 |
|
| 8505 |
void restore_state_to_opc(CPUX86State *env, TranslationBlock *tb, int pc_pos) |
| 8506 |
{
|
| 8507 |
int cc_op;
|
| 8508 |
#ifdef DEBUG_DISAS
|
| 8509 |
if (qemu_loglevel_mask(CPU_LOG_TB_OP)) {
|
| 8510 |
int i;
|
| 8511 |
qemu_log("RESTORE:\n");
|
| 8512 |
for(i = 0;i <= pc_pos; i++) { |
| 8513 |
if (tcg_ctx.gen_opc_instr_start[i]) {
|
| 8514 |
qemu_log("0x%04x: " TARGET_FMT_lx "\n", i, |
| 8515 |
tcg_ctx.gen_opc_pc[i]); |
| 8516 |
} |
| 8517 |
} |
| 8518 |
qemu_log("pc_pos=0x%x eip=" TARGET_FMT_lx " cs_base=%x\n", |
| 8519 |
pc_pos, tcg_ctx.gen_opc_pc[pc_pos] - tb->cs_base, |
| 8520 |
(uint32_t)tb->cs_base); |
| 8521 |
} |
| 8522 |
#endif
|
| 8523 |
env->eip = tcg_ctx.gen_opc_pc[pc_pos] - tb->cs_base; |
| 8524 |
cc_op = gen_opc_cc_op[pc_pos]; |
| 8525 |
if (cc_op != CC_OP_DYNAMIC)
|
| 8526 |
env->cc_op = cc_op; |
| 8527 |
} |