projects / qemu.git / blame
| Commit | Line | Data |
|---|---|---|
| 64654ded BS |
1 | /* |
| 2 | * PowerPC integer and vector emulation helpers for QEMU. | |
| 3 | * | |
| 4 | * Copyright (c) 2003-2007 Jocelyn Mayer | |
| 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 "cpu.h" | |
| 1de7afc9 | 20 | #include "qemu/host-utils.h" |
| 64654ded BS |
21 | #include "helper.h" |
| 22 | ||
| 23 | #include "helper_regs.h" | |
| 24 | /*****************************************************************************/ | |
| 25 | /* Fixed point operations helpers */ | |
| 26 | #if defined(TARGET_PPC64) | |
| 27 | ||
| d15f74fb | 28 | uint64_t helper_mulldo(CPUPPCState *env, uint64_t arg1, uint64_t arg2) |
| 64654ded BS |
29 | { |
| 30 | int64_t th; | |
| 31 | uint64_t tl; | |
| 32 | ||
| 33 | muls64(&tl, (uint64_t *)&th, arg1, arg2); | |
| 34 | /* If th != 0 && th != -1, then we had an overflow */ | |
| 35 | if (likely((uint64_t)(th + 1) <= 1)) { | |
| da91a00f | 36 | env->ov = 0; |
| 64654ded | 37 | } else { |
| da91a00f | 38 | env->so = env->ov = 1; |
| 64654ded BS |
39 | } |
| 40 | return (int64_t)tl; | |
| 41 | } | |
| 42 | #endif | |
| 43 | ||
| 6a4fda33 TM |
44 | target_ulong helper_divweu(CPUPPCState *env, target_ulong ra, target_ulong rb, |
| 45 | uint32_t oe) | |
| 46 | { | |
| 47 | uint64_t rt = 0; | |
| 48 | int overflow = 0; | |
| 49 | ||
| 50 | uint64_t dividend = (uint64_t)ra << 32; | |
| 51 | uint64_t divisor = (uint32_t)rb; | |
| 52 | ||
| 53 | if (unlikely(divisor == 0)) { | |
| 54 | overflow = 1; | |
| 55 | } else { | |
| 56 | rt = dividend / divisor; | |
| 57 | overflow = rt > UINT32_MAX; | |
| 58 | } | |
| 59 | ||
| 60 | if (unlikely(overflow)) { | |
| 61 | rt = 0; /* Undefined */ | |
| 62 | } | |
| 63 | ||
| 64 | if (oe) { | |
| 65 | if (unlikely(overflow)) { | |
| 66 | env->so = env->ov = 1; | |
| 67 | } else { | |
| 68 | env->ov = 0; | |
| 69 | } | |
| 70 | } | |
| 71 | ||
| 72 | return (target_ulong)rt; | |
| 73 | } | |
| 74 | ||
| a98eb9e9 TM |
75 | target_ulong helper_divwe(CPUPPCState *env, target_ulong ra, target_ulong rb, |
| 76 | uint32_t oe) | |
| 77 | { | |
| 78 | int64_t rt = 0; | |
| 79 | int overflow = 0; | |
| 80 | ||
| 81 | int64_t dividend = (int64_t)ra << 32; | |
| 82 | int64_t divisor = (int64_t)((int32_t)rb); | |
| 83 | ||
| 84 | if (unlikely((divisor == 0) || | |
| 85 | ((divisor == -1ull) && (dividend == INT64_MIN)))) { | |
| 86 | overflow = 1; | |
| 87 | } else { | |
| 88 | rt = dividend / divisor; | |
| 89 | overflow = rt != (int32_t)rt; | |
| 90 | } | |
| 91 | ||
| 92 | if (unlikely(overflow)) { | |
| 93 | rt = 0; /* Undefined */ | |
| 94 | } | |
| 95 | ||
| 96 | if (oe) { | |
| 97 | if (unlikely(overflow)) { | |
| 98 | env->so = env->ov = 1; | |
| 99 | } else { | |
| 100 | env->ov = 0; | |
| 101 | } | |
| 102 | } | |
| 103 | ||
| 104 | return (target_ulong)rt; | |
| 105 | } | |
| 106 | ||
| 98d1eb27 TM |
107 | #if defined(TARGET_PPC64) |
| 108 | ||
| 109 | uint64_t helper_divdeu(CPUPPCState *env, uint64_t ra, uint64_t rb, uint32_t oe) | |
| 110 | { | |
| 111 | uint64_t rt = 0; | |
| 112 | int overflow = 0; | |
| 113 | ||
| 114 | overflow = divu128(&rt, &ra, rb); | |
| 115 | ||
| 116 | if (unlikely(overflow)) { | |
| 117 | rt = 0; /* Undefined */ | |
| 118 | } | |
| 119 | ||
| 120 | if (oe) { | |
| 121 | if (unlikely(overflow)) { | |
| 122 | env->so = env->ov = 1; | |
| 123 | } else { | |
| 124 | env->ov = 0; | |
| 125 | } | |
| 126 | } | |
| 127 | ||
| 128 | return rt; | |
| 129 | } | |
| 130 | ||
| e44259b6 TM |
131 | uint64_t helper_divde(CPUPPCState *env, uint64_t rau, uint64_t rbu, uint32_t oe) |
| 132 | { | |
| 133 | int64_t rt = 0; | |
| 134 | int64_t ra = (int64_t)rau; | |
| 135 | int64_t rb = (int64_t)rbu; | |
| 136 | int overflow = divs128(&rt, &ra, rb); | |
| 137 | ||
| 138 | if (unlikely(overflow)) { | |
| 139 | rt = 0; /* Undefined */ | |
| 140 | } | |
| 141 | ||
| 142 | if (oe) { | |
| 143 | ||
| 144 | if (unlikely(overflow)) { | |
| 145 | env->so = env->ov = 1; | |
| 146 | } else { | |
| 147 | env->ov = 0; | |
| 148 | } | |
| 149 | } | |
| 150 | ||
| 151 | return rt; | |
| 152 | } | |
| 153 | ||
| 98d1eb27 TM |
154 | #endif |
| 155 | ||
| 156 | ||
| 64654ded BS |
157 | target_ulong helper_cntlzw(target_ulong t) |
| 158 | { | |
| 159 | return clz32(t); | |
| 160 | } | |
| 161 | ||
| 162 | #if defined(TARGET_PPC64) | |
| 163 | target_ulong helper_cntlzd(target_ulong t) | |
| 164 | { | |
| 165 | return clz64(t); | |
| 166 | } | |
| 167 | #endif | |
| 168 | ||
| 86ba37ed TM |
169 | #if defined(TARGET_PPC64) |
| 170 | ||
| 171 | uint64_t helper_bpermd(uint64_t rs, uint64_t rb) | |
| 172 | { | |
| 173 | int i; | |
| 174 | uint64_t ra = 0; | |
| 175 | ||
| 176 | for (i = 0; i < 8; i++) { | |
| 177 | int index = (rs >> (i*8)) & 0xFF; | |
| 178 | if (index < 64) { | |
| 179 | if (rb & (1ull << (63-index))) { | |
| 180 | ra |= 1 << i; | |
| 181 | } | |
| 182 | } | |
| 183 | } | |
| 184 | return ra; | |
| 185 | } | |
| 186 | ||
| 187 | #endif | |
| 188 | ||
| fcfda20f AJ |
189 | target_ulong helper_cmpb(target_ulong rs, target_ulong rb) |
| 190 | { | |
| 191 | target_ulong mask = 0xff; | |
| 192 | target_ulong ra = 0; | |
| 193 | int i; | |
| 194 | ||
| 195 | for (i = 0; i < sizeof(target_ulong); i++) { | |
| 196 | if ((rs & mask) == (rb & mask)) { | |
| 197 | ra |= mask; | |
| 198 | } | |
| 199 | mask <<= 8; | |
| 200 | } | |
| 201 | return ra; | |
| 202 | } | |
| 203 | ||
| 64654ded | 204 | /* shift right arithmetic helper */ |
| d15f74fb BS |
205 | target_ulong helper_sraw(CPUPPCState *env, target_ulong value, |
| 206 | target_ulong shift) | |
| 64654ded BS |
207 | { |
| 208 | int32_t ret; | |
| 209 | ||
| 210 | if (likely(!(shift & 0x20))) { | |
| 211 | if (likely((uint32_t)shift != 0)) { | |
| 212 | shift &= 0x1f; | |
| 213 | ret = (int32_t)value >> shift; | |
| 214 | if (likely(ret >= 0 || (value & ((1 << shift) - 1)) == 0)) { | |
| da91a00f | 215 | env->ca = 0; |
| 64654ded | 216 | } else { |
| da91a00f | 217 | env->ca = 1; |
| 64654ded BS |
218 | } |
| 219 | } else { | |
| 220 | ret = (int32_t)value; | |
| da91a00f | 221 | env->ca = 0; |
| 64654ded BS |
222 | } |
| 223 | } else { | |
| 224 | ret = (int32_t)value >> 31; | |
| da91a00f | 225 | env->ca = (ret != 0); |
| 64654ded BS |
226 | } |
| 227 | return (target_long)ret; | |
| 228 | } | |
| 229 | ||
| 230 | #if defined(TARGET_PPC64) | |
| d15f74fb BS |
231 | target_ulong helper_srad(CPUPPCState *env, target_ulong value, |
| 232 | target_ulong shift) | |
| 64654ded BS |
233 | { |
| 234 | int64_t ret; | |
| 235 | ||
| 236 | if (likely(!(shift & 0x40))) { | |
| 237 | if (likely((uint64_t)shift != 0)) { | |
| 238 | shift &= 0x3f; | |
| 239 | ret = (int64_t)value >> shift; | |
| 240 | if (likely(ret >= 0 || (value & ((1 << shift) - 1)) == 0)) { | |
| da91a00f | 241 | env->ca = 0; |
| 64654ded | 242 | } else { |
| da91a00f | 243 | env->ca = 1; |
| 64654ded BS |
244 | } |
| 245 | } else { | |
| 246 | ret = (int64_t)value; | |
| da91a00f | 247 | env->ca = 0; |
| 64654ded BS |
248 | } |
| 249 | } else { | |
| 250 | ret = (int64_t)value >> 63; | |
| da91a00f | 251 | env->ca = (ret != 0); |
| 64654ded BS |
252 | } |
| 253 | return ret; | |
| 254 | } | |
| 255 | #endif | |
| 256 | ||
| 257 | #if defined(TARGET_PPC64) | |
| 258 | target_ulong helper_popcntb(target_ulong val) | |
| 259 | { | |
| 260 | val = (val & 0x5555555555555555ULL) + ((val >> 1) & | |
| 261 | 0x5555555555555555ULL); | |
| 262 | val = (val & 0x3333333333333333ULL) + ((val >> 2) & | |
| 263 | 0x3333333333333333ULL); | |
| 264 | val = (val & 0x0f0f0f0f0f0f0f0fULL) + ((val >> 4) & | |
| 265 | 0x0f0f0f0f0f0f0f0fULL); | |
| 266 | return val; | |
| 267 | } | |
| 268 | ||
| 269 | target_ulong helper_popcntw(target_ulong val) | |
| 270 | { | |
| 271 | val = (val & 0x5555555555555555ULL) + ((val >> 1) & | |
| 272 | 0x5555555555555555ULL); | |
| 273 | val = (val & 0x3333333333333333ULL) + ((val >> 2) & | |
| 274 | 0x3333333333333333ULL); | |
| 275 | val = (val & 0x0f0f0f0f0f0f0f0fULL) + ((val >> 4) & | |
| 276 | 0x0f0f0f0f0f0f0f0fULL); | |
| 277 | val = (val & 0x00ff00ff00ff00ffULL) + ((val >> 8) & | |
| 278 | 0x00ff00ff00ff00ffULL); | |
| 279 | val = (val & 0x0000ffff0000ffffULL) + ((val >> 16) & | |
| 280 | 0x0000ffff0000ffffULL); | |
| 281 | return val; | |
| 282 | } | |
| 283 | ||
| 284 | target_ulong helper_popcntd(target_ulong val) | |
| 285 | { | |
| 286 | return ctpop64(val); | |
| 287 | } | |
| 288 | #else | |
| 289 | target_ulong helper_popcntb(target_ulong val) | |
| 290 | { | |
| 291 | val = (val & 0x55555555) + ((val >> 1) & 0x55555555); | |
| 292 | val = (val & 0x33333333) + ((val >> 2) & 0x33333333); | |
| 293 | val = (val & 0x0f0f0f0f) + ((val >> 4) & 0x0f0f0f0f); | |
| 294 | return val; | |
| 295 | } | |
| 296 | ||
| 297 | target_ulong helper_popcntw(target_ulong val) | |
| 298 | { | |
| 299 | val = (val & 0x55555555) + ((val >> 1) & 0x55555555); | |
| 300 | val = (val & 0x33333333) + ((val >> 2) & 0x33333333); | |
| 301 | val = (val & 0x0f0f0f0f) + ((val >> 4) & 0x0f0f0f0f); | |
| 302 | val = (val & 0x00ff00ff) + ((val >> 8) & 0x00ff00ff); | |
| 303 | val = (val & 0x0000ffff) + ((val >> 16) & 0x0000ffff); | |
| 304 | return val; | |
| 305 | } | |
| 306 | #endif | |
| 307 | ||
| 308 | /*****************************************************************************/ | |
| 309 | /* PowerPC 601 specific instructions (POWER bridge) */ | |
| d15f74fb | 310 | target_ulong helper_div(CPUPPCState *env, target_ulong arg1, target_ulong arg2) |
| 64654ded BS |
311 | { |
| 312 | uint64_t tmp = (uint64_t)arg1 << 32 | env->spr[SPR_MQ]; | |
| 313 | ||
| 314 | if (((int32_t)tmp == INT32_MIN && (int32_t)arg2 == (int32_t)-1) || | |
| 315 | (int32_t)arg2 == 0) { | |
| 316 | env->spr[SPR_MQ] = 0; | |
| 317 | return INT32_MIN; | |
| 318 | } else { | |
| 319 | env->spr[SPR_MQ] = tmp % arg2; | |
| 320 | return tmp / (int32_t)arg2; | |
| 321 | } | |
| 322 | } | |
| 323 | ||
| d15f74fb BS |
324 | target_ulong helper_divo(CPUPPCState *env, target_ulong arg1, |
| 325 | target_ulong arg2) | |
| 64654ded BS |
326 | { |
| 327 | uint64_t tmp = (uint64_t)arg1 << 32 | env->spr[SPR_MQ]; | |
| 328 | ||
| 329 | if (((int32_t)tmp == INT32_MIN && (int32_t)arg2 == (int32_t)-1) || | |
| 330 | (int32_t)arg2 == 0) { | |
| da91a00f | 331 | env->so = env->ov = 1; |
| 64654ded BS |
332 | env->spr[SPR_MQ] = 0; |
| 333 | return INT32_MIN; | |
| 334 | } else { | |
| 335 | env->spr[SPR_MQ] = tmp % arg2; | |
| 336 | tmp /= (int32_t)arg2; | |
| 337 | if ((int32_t)tmp != tmp) { | |
| da91a00f | 338 | env->so = env->ov = 1; |
| 64654ded | 339 | } else { |
| da91a00f | 340 | env->ov = 0; |
| 64654ded BS |
341 | } |
| 342 | return tmp; | |
| 343 | } | |
| 344 | } | |
| 345 | ||
| d15f74fb BS |
346 | target_ulong helper_divs(CPUPPCState *env, target_ulong arg1, |
| 347 | target_ulong arg2) | |
| 64654ded BS |
348 | { |
| 349 | if (((int32_t)arg1 == INT32_MIN && (int32_t)arg2 == (int32_t)-1) || | |
| 350 | (int32_t)arg2 == 0) { | |
| 351 | env->spr[SPR_MQ] = 0; | |
| 352 | return INT32_MIN; | |
| 353 | } else { | |
| 354 | env->spr[SPR_MQ] = (int32_t)arg1 % (int32_t)arg2; | |
| 355 | return (int32_t)arg1 / (int32_t)arg2; | |
| 356 | } | |
| 357 | } | |
| 358 | ||
| d15f74fb BS |
359 | target_ulong helper_divso(CPUPPCState *env, target_ulong arg1, |
| 360 | target_ulong arg2) | |
| 64654ded BS |
361 | { |
| 362 | if (((int32_t)arg1 == INT32_MIN && (int32_t)arg2 == (int32_t)-1) || | |
| 363 | (int32_t)arg2 == 0) { | |
| da91a00f | 364 | env->so = env->ov = 1; |
| 64654ded BS |
365 | env->spr[SPR_MQ] = 0; |
| 366 | return INT32_MIN; | |
| 367 | } else { | |
| da91a00f | 368 | env->ov = 0; |
| 64654ded BS |
369 | env->spr[SPR_MQ] = (int32_t)arg1 % (int32_t)arg2; |
| 370 | return (int32_t)arg1 / (int32_t)arg2; | |
| 371 | } | |
| 372 | } | |
| 373 | ||
| 374 | /*****************************************************************************/ | |
| 375 | /* 602 specific instructions */ | |
| 376 | /* mfrom is the most crazy instruction ever seen, imho ! */ | |
| 377 | /* Real implementation uses a ROM table. Do the same */ | |
| 378 | /* Extremely decomposed: | |
| 379 | * -arg / 256 | |
| 380 | * return 256 * log10(10 + 1.0) + 0.5 | |
| 381 | */ | |
| 382 | #if !defined(CONFIG_USER_ONLY) | |
| 383 | target_ulong helper_602_mfrom(target_ulong arg) | |
| 384 | { | |
| 385 | if (likely(arg < 602)) { | |
| 386 | #include "mfrom_table.c" | |
| 387 | return mfrom_ROM_table[arg]; | |
| 388 | } else { | |
| 389 | return 0; | |
| 390 | } | |
| 391 | } | |
| 392 | #endif | |
| 393 | ||
| 394 | /*****************************************************************************/ | |
| 395 | /* Altivec extension helpers */ | |
| 396 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 397 | #define HI_IDX 0 | |
| 398 | #define LO_IDX 1 | |
| 399 | #else | |
| 400 | #define HI_IDX 1 | |
| 401 | #define LO_IDX 0 | |
| 402 | #endif | |
| 403 | ||
| 404 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 405 | #define VECTOR_FOR_INORDER_I(index, element) \ | |
| 406 | for (index = 0; index < ARRAY_SIZE(r->element); index++) | |
| 407 | #else | |
| 408 | #define VECTOR_FOR_INORDER_I(index, element) \ | |
| 409 | for (index = ARRAY_SIZE(r->element)-1; index >= 0; index--) | |
| 410 | #endif | |
| 411 | ||
| 64654ded BS |
412 | /* Saturating arithmetic helpers. */ |
| 413 | #define SATCVT(from, to, from_type, to_type, min, max) \ | |
| 414 | static inline to_type cvt##from##to(from_type x, int *sat) \ | |
| 415 | { \ | |
| 416 | to_type r; \ | |
| 417 | \ | |
| 418 | if (x < (from_type)min) { \ | |
| 419 | r = min; \ | |
| 420 | *sat = 1; \ | |
| 421 | } else if (x > (from_type)max) { \ | |
| 422 | r = max; \ | |
| 423 | *sat = 1; \ | |
| 424 | } else { \ | |
| 425 | r = x; \ | |
| 426 | } \ | |
| 427 | return r; \ | |
| 428 | } | |
| 429 | #define SATCVTU(from, to, from_type, to_type, min, max) \ | |
| 430 | static inline to_type cvt##from##to(from_type x, int *sat) \ | |
| 431 | { \ | |
| 432 | to_type r; \ | |
| 433 | \ | |
| 434 | if (x > (from_type)max) { \ | |
| 435 | r = max; \ | |
| 436 | *sat = 1; \ | |
| 437 | } else { \ | |
| 438 | r = x; \ | |
| 439 | } \ | |
| 440 | return r; \ | |
| 441 | } | |
| 442 | SATCVT(sh, sb, int16_t, int8_t, INT8_MIN, INT8_MAX) | |
| 443 | SATCVT(sw, sh, int32_t, int16_t, INT16_MIN, INT16_MAX) | |
| 444 | SATCVT(sd, sw, int64_t, int32_t, INT32_MIN, INT32_MAX) | |
| 445 | ||
| 446 | SATCVTU(uh, ub, uint16_t, uint8_t, 0, UINT8_MAX) | |
| 447 | SATCVTU(uw, uh, uint32_t, uint16_t, 0, UINT16_MAX) | |
| 448 | SATCVTU(ud, uw, uint64_t, uint32_t, 0, UINT32_MAX) | |
| 449 | SATCVT(sh, ub, int16_t, uint8_t, 0, UINT8_MAX) | |
| 450 | SATCVT(sw, uh, int32_t, uint16_t, 0, UINT16_MAX) | |
| 451 | SATCVT(sd, uw, int64_t, uint32_t, 0, UINT32_MAX) | |
| 452 | #undef SATCVT | |
| 453 | #undef SATCVTU | |
| 454 | ||
| 455 | void helper_lvsl(ppc_avr_t *r, target_ulong sh) | |
| 456 | { | |
| 457 | int i, j = (sh & 0xf); | |
| 458 | ||
| 459 | VECTOR_FOR_INORDER_I(i, u8) { | |
| 460 | r->u8[i] = j++; | |
| 461 | } | |
| 462 | } | |
| 463 | ||
| 464 | void helper_lvsr(ppc_avr_t *r, target_ulong sh) | |
| 465 | { | |
| 466 | int i, j = 0x10 - (sh & 0xf); | |
| 467 | ||
| 468 | VECTOR_FOR_INORDER_I(i, u8) { | |
| 469 | r->u8[i] = j++; | |
| 470 | } | |
| 471 | } | |
| 472 | ||
| d15f74fb | 473 | void helper_mtvscr(CPUPPCState *env, ppc_avr_t *r) |
| 64654ded BS |
474 | { |
| 475 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 476 | env->vscr = r->u32[3]; | |
| 477 | #else | |
| 478 | env->vscr = r->u32[0]; | |
| 479 | #endif | |
| 480 | set_flush_to_zero(vscr_nj, &env->vec_status); | |
| 481 | } | |
| 482 | ||
| 483 | void helper_vaddcuw(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 484 | { | |
| 485 | int i; | |
| 486 | ||
| 487 | for (i = 0; i < ARRAY_SIZE(r->u32); i++) { | |
| 488 | r->u32[i] = ~a->u32[i] < b->u32[i]; | |
| 489 | } | |
| 490 | } | |
| 491 | ||
| 492 | #define VARITH_DO(name, op, element) \ | |
| 493 | void helper_v##name(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ | |
| 494 | { \ | |
| 495 | int i; \ | |
| 496 | \ | |
| 497 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 498 | r->element[i] = a->element[i] op b->element[i]; \ | |
| 499 | } \ | |
| 500 | } | |
| 501 | #define VARITH(suffix, element) \ | |
| 502 | VARITH_DO(add##suffix, +, element) \ | |
| 503 | VARITH_DO(sub##suffix, -, element) | |
| 504 | VARITH(ubm, u8) | |
| 505 | VARITH(uhm, u16) | |
| 506 | VARITH(uwm, u32) | |
| 56eabc75 | 507 | VARITH(udm, u64) |
| 953f0f58 | 508 | VARITH_DO(muluwm, *, u32) |
| 64654ded BS |
509 | #undef VARITH_DO |
| 510 | #undef VARITH | |
| 511 | ||
| 512 | #define VARITHFP(suffix, func) \ | |
| d15f74fb BS |
513 | void helper_v##suffix(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, \ |
| 514 | ppc_avr_t *b) \ | |
| 64654ded BS |
515 | { \ |
| 516 | int i; \ | |
| 517 | \ | |
| 518 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { \ | |
| ef9bd150 | 519 | r->f[i] = func(a->f[i], b->f[i], &env->vec_status); \ |
| 64654ded BS |
520 | } \ |
| 521 | } | |
| 522 | VARITHFP(addfp, float32_add) | |
| 523 | VARITHFP(subfp, float32_sub) | |
| db1babb8 AJ |
524 | VARITHFP(minfp, float32_min) |
| 525 | VARITHFP(maxfp, float32_max) | |
| 64654ded BS |
526 | #undef VARITHFP |
| 527 | ||
| 2f93c23f AJ |
528 | #define VARITHFPFMA(suffix, type) \ |
| 529 | void helper_v##suffix(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, \ | |
| 530 | ppc_avr_t *b, ppc_avr_t *c) \ | |
| 531 | { \ | |
| 532 | int i; \ | |
| 533 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { \ | |
| 534 | r->f[i] = float32_muladd(a->f[i], c->f[i], b->f[i], \ | |
| 535 | type, &env->vec_status); \ | |
| 536 | } \ | |
| 537 | } | |
| 538 | VARITHFPFMA(maddfp, 0); | |
| 539 | VARITHFPFMA(nmsubfp, float_muladd_negate_result | float_muladd_negate_c); | |
| 540 | #undef VARITHFPFMA | |
| 541 | ||
| 64654ded BS |
542 | #define VARITHSAT_CASE(type, op, cvt, element) \ |
| 543 | { \ | |
| 544 | type result = (type)a->element[i] op (type)b->element[i]; \ | |
| 545 | r->element[i] = cvt(result, &sat); \ | |
| 546 | } | |
| 547 | ||
| 548 | #define VARITHSAT_DO(name, op, optype, cvt, element) \ | |
| d15f74fb BS |
549 | void helper_v##name(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, \ |
| 550 | ppc_avr_t *b) \ | |
| 64654ded BS |
551 | { \ |
| 552 | int sat = 0; \ | |
| 553 | int i; \ | |
| 554 | \ | |
| 555 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 556 | switch (sizeof(r->element[0])) { \ | |
| 557 | case 1: \ | |
| 558 | VARITHSAT_CASE(optype, op, cvt, element); \ | |
| 559 | break; \ | |
| 560 | case 2: \ | |
| 561 | VARITHSAT_CASE(optype, op, cvt, element); \ | |
| 562 | break; \ | |
| 563 | case 4: \ | |
| 564 | VARITHSAT_CASE(optype, op, cvt, element); \ | |
| 565 | break; \ | |
| 566 | } \ | |
| 567 | } \ | |
| 568 | if (sat) { \ | |
| 569 | env->vscr |= (1 << VSCR_SAT); \ | |
| 570 | } \ | |
| 571 | } | |
| 572 | #define VARITHSAT_SIGNED(suffix, element, optype, cvt) \ | |
| 573 | VARITHSAT_DO(adds##suffix##s, +, optype, cvt, element) \ | |
| 574 | VARITHSAT_DO(subs##suffix##s, -, optype, cvt, element) | |
| 575 | #define VARITHSAT_UNSIGNED(suffix, element, optype, cvt) \ | |
| 576 | VARITHSAT_DO(addu##suffix##s, +, optype, cvt, element) \ | |
| 577 | VARITHSAT_DO(subu##suffix##s, -, optype, cvt, element) | |
| 578 | VARITHSAT_SIGNED(b, s8, int16_t, cvtshsb) | |
| 579 | VARITHSAT_SIGNED(h, s16, int32_t, cvtswsh) | |
| 580 | VARITHSAT_SIGNED(w, s32, int64_t, cvtsdsw) | |
| 581 | VARITHSAT_UNSIGNED(b, u8, uint16_t, cvtshub) | |
| 582 | VARITHSAT_UNSIGNED(h, u16, uint32_t, cvtswuh) | |
| 583 | VARITHSAT_UNSIGNED(w, u32, uint64_t, cvtsduw) | |
| 584 | #undef VARITHSAT_CASE | |
| 585 | #undef VARITHSAT_DO | |
| 586 | #undef VARITHSAT_SIGNED | |
| 587 | #undef VARITHSAT_UNSIGNED | |
| 588 | ||
| 589 | #define VAVG_DO(name, element, etype) \ | |
| 590 | void helper_v##name(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ | |
| 591 | { \ | |
| 592 | int i; \ | |
| 593 | \ | |
| 594 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 595 | etype x = (etype)a->element[i] + (etype)b->element[i] + 1; \ | |
| 596 | r->element[i] = x >> 1; \ | |
| 597 | } \ | |
| 598 | } | |
| 599 | ||
| 600 | #define VAVG(type, signed_element, signed_type, unsigned_element, \ | |
| 601 | unsigned_type) \ | |
| 602 | VAVG_DO(avgs##type, signed_element, signed_type) \ | |
| 603 | VAVG_DO(avgu##type, unsigned_element, unsigned_type) | |
| 604 | VAVG(b, s8, int16_t, u8, uint16_t) | |
| 605 | VAVG(h, s16, int32_t, u16, uint32_t) | |
| 606 | VAVG(w, s32, int64_t, u32, uint64_t) | |
| 607 | #undef VAVG_DO | |
| 608 | #undef VAVG | |
| 609 | ||
| 610 | #define VCF(suffix, cvt, element) \ | |
| d15f74fb BS |
611 | void helper_vcf##suffix(CPUPPCState *env, ppc_avr_t *r, \ |
| 612 | ppc_avr_t *b, uint32_t uim) \ | |
| 64654ded BS |
613 | { \ |
| 614 | int i; \ | |
| 615 | \ | |
| 616 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { \ | |
| 617 | float32 t = cvt(b->element[i], &env->vec_status); \ | |
| 618 | r->f[i] = float32_scalbn(t, -uim, &env->vec_status); \ | |
| 619 | } \ | |
| 620 | } | |
| 621 | VCF(ux, uint32_to_float32, u32) | |
| 622 | VCF(sx, int32_to_float32, s32) | |
| 623 | #undef VCF | |
| 624 | ||
| 625 | #define VCMP_DO(suffix, compare, element, record) \ | |
| d15f74fb BS |
626 | void helper_vcmp##suffix(CPUPPCState *env, ppc_avr_t *r, \ |
| 627 | ppc_avr_t *a, ppc_avr_t *b) \ | |
| 64654ded | 628 | { \ |
| 6f3dab41 TM |
629 | uint64_t ones = (uint64_t)-1; \ |
| 630 | uint64_t all = ones; \ | |
| 631 | uint64_t none = 0; \ | |
| 64654ded BS |
632 | int i; \ |
| 633 | \ | |
| 634 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 6f3dab41 | 635 | uint64_t result = (a->element[i] compare b->element[i] ? \ |
| 64654ded BS |
636 | ones : 0x0); \ |
| 637 | switch (sizeof(a->element[0])) { \ | |
| 6f3dab41 TM |
638 | case 8: \ |
| 639 | r->u64[i] = result; \ | |
| 640 | break; \ | |
| 64654ded BS |
641 | case 4: \ |
| 642 | r->u32[i] = result; \ | |
| 643 | break; \ | |
| 644 | case 2: \ | |
| 645 | r->u16[i] = result; \ | |
| 646 | break; \ | |
| 647 | case 1: \ | |
| 648 | r->u8[i] = result; \ | |
| 649 | break; \ | |
| 650 | } \ | |
| 651 | all &= result; \ | |
| 652 | none |= result; \ | |
| 653 | } \ | |
| 654 | if (record) { \ | |
| 655 | env->crf[6] = ((all != 0) << 3) | ((none == 0) << 1); \ | |
| 656 | } \ | |
| 657 | } | |
| 658 | #define VCMP(suffix, compare, element) \ | |
| 659 | VCMP_DO(suffix, compare, element, 0) \ | |
| 660 | VCMP_DO(suffix##_dot, compare, element, 1) | |
| 661 | VCMP(equb, ==, u8) | |
| 662 | VCMP(equh, ==, u16) | |
| 663 | VCMP(equw, ==, u32) | |
| 6f3dab41 | 664 | VCMP(equd, ==, u64) |
| 64654ded BS |
665 | VCMP(gtub, >, u8) |
| 666 | VCMP(gtuh, >, u16) | |
| 667 | VCMP(gtuw, >, u32) | |
| 6f3dab41 | 668 | VCMP(gtud, >, u64) |
| 64654ded BS |
669 | VCMP(gtsb, >, s8) |
| 670 | VCMP(gtsh, >, s16) | |
| 671 | VCMP(gtsw, >, s32) | |
| 6f3dab41 | 672 | VCMP(gtsd, >, s64) |
| 64654ded BS |
673 | #undef VCMP_DO |
| 674 | #undef VCMP | |
| 675 | ||
| 676 | #define VCMPFP_DO(suffix, compare, order, record) \ | |
| d15f74fb BS |
677 | void helper_vcmp##suffix(CPUPPCState *env, ppc_avr_t *r, \ |
| 678 | ppc_avr_t *a, ppc_avr_t *b) \ | |
| 64654ded BS |
679 | { \ |
| 680 | uint32_t ones = (uint32_t)-1; \ | |
| 681 | uint32_t all = ones; \ | |
| 682 | uint32_t none = 0; \ | |
| 683 | int i; \ | |
| 684 | \ | |
| 685 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { \ | |
| 686 | uint32_t result; \ | |
| 687 | int rel = float32_compare_quiet(a->f[i], b->f[i], \ | |
| 688 | &env->vec_status); \ | |
| 689 | if (rel == float_relation_unordered) { \ | |
| 690 | result = 0; \ | |
| 691 | } else if (rel compare order) { \ | |
| 692 | result = ones; \ | |
| 693 | } else { \ | |
| 694 | result = 0; \ | |
| 695 | } \ | |
| 696 | r->u32[i] = result; \ | |
| 697 | all &= result; \ | |
| 698 | none |= result; \ | |
| 699 | } \ | |
| 700 | if (record) { \ | |
| 701 | env->crf[6] = ((all != 0) << 3) | ((none == 0) << 1); \ | |
| 702 | } \ | |
| 703 | } | |
| 704 | #define VCMPFP(suffix, compare, order) \ | |
| 705 | VCMPFP_DO(suffix, compare, order, 0) \ | |
| 706 | VCMPFP_DO(suffix##_dot, compare, order, 1) | |
| 707 | VCMPFP(eqfp, ==, float_relation_equal) | |
| 708 | VCMPFP(gefp, !=, float_relation_less) | |
| 709 | VCMPFP(gtfp, ==, float_relation_greater) | |
| 710 | #undef VCMPFP_DO | |
| 711 | #undef VCMPFP | |
| 712 | ||
| d15f74fb BS |
713 | static inline void vcmpbfp_internal(CPUPPCState *env, ppc_avr_t *r, |
| 714 | ppc_avr_t *a, ppc_avr_t *b, int record) | |
| 64654ded BS |
715 | { |
| 716 | int i; | |
| 717 | int all_in = 0; | |
| 718 | ||
| 719 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { | |
| 720 | int le_rel = float32_compare_quiet(a->f[i], b->f[i], &env->vec_status); | |
| 721 | if (le_rel == float_relation_unordered) { | |
| 722 | r->u32[i] = 0xc0000000; | |
| 723 | /* ALL_IN does not need to be updated here. */ | |
| 724 | } else { | |
| 725 | float32 bneg = float32_chs(b->f[i]); | |
| 726 | int ge_rel = float32_compare_quiet(a->f[i], bneg, &env->vec_status); | |
| 727 | int le = le_rel != float_relation_greater; | |
| 728 | int ge = ge_rel != float_relation_less; | |
| 729 | ||
| 730 | r->u32[i] = ((!le) << 31) | ((!ge) << 30); | |
| 731 | all_in |= (!le | !ge); | |
| 732 | } | |
| 733 | } | |
| 734 | if (record) { | |
| 735 | env->crf[6] = (all_in == 0) << 1; | |
| 736 | } | |
| 737 | } | |
| 738 | ||
| d15f74fb | 739 | void helper_vcmpbfp(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) |
| 64654ded | 740 | { |
| d15f74fb | 741 | vcmpbfp_internal(env, r, a, b, 0); |
| 64654ded BS |
742 | } |
| 743 | ||
| d15f74fb BS |
744 | void helper_vcmpbfp_dot(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 745 | ppc_avr_t *b) | |
| 64654ded | 746 | { |
| d15f74fb | 747 | vcmpbfp_internal(env, r, a, b, 1); |
| 64654ded BS |
748 | } |
| 749 | ||
| 750 | #define VCT(suffix, satcvt, element) \ | |
| d15f74fb BS |
751 | void helper_vct##suffix(CPUPPCState *env, ppc_avr_t *r, \ |
| 752 | ppc_avr_t *b, uint32_t uim) \ | |
| 64654ded BS |
753 | { \ |
| 754 | int i; \ | |
| 755 | int sat = 0; \ | |
| 756 | float_status s = env->vec_status; \ | |
| 757 | \ | |
| 758 | set_float_rounding_mode(float_round_to_zero, &s); \ | |
| 759 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { \ | |
| 760 | if (float32_is_any_nan(b->f[i])) { \ | |
| 761 | r->element[i] = 0; \ | |
| 762 | } else { \ | |
| 763 | float64 t = float32_to_float64(b->f[i], &s); \ | |
| 764 | int64_t j; \ | |
| 765 | \ | |
| 766 | t = float64_scalbn(t, uim, &s); \ | |
| 767 | j = float64_to_int64(t, &s); \ | |
| 768 | r->element[i] = satcvt(j, &sat); \ | |
| 769 | } \ | |
| 770 | } \ | |
| 771 | if (sat) { \ | |
| 772 | env->vscr |= (1 << VSCR_SAT); \ | |
| 773 | } \ | |
| 774 | } | |
| 775 | VCT(uxs, cvtsduw, u32) | |
| 776 | VCT(sxs, cvtsdsw, s32) | |
| 777 | #undef VCT | |
| 778 | ||
| d15f74fb BS |
779 | void helper_vmhaddshs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 780 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
781 | { |
| 782 | int sat = 0; | |
| 783 | int i; | |
| 784 | ||
| 785 | for (i = 0; i < ARRAY_SIZE(r->s16); i++) { | |
| 786 | int32_t prod = a->s16[i] * b->s16[i]; | |
| 787 | int32_t t = (int32_t)c->s16[i] + (prod >> 15); | |
| 788 | ||
| 789 | r->s16[i] = cvtswsh(t, &sat); | |
| 790 | } | |
| 791 | ||
| 792 | if (sat) { | |
| 793 | env->vscr |= (1 << VSCR_SAT); | |
| 794 | } | |
| 795 | } | |
| 796 | ||
| d15f74fb BS |
797 | void helper_vmhraddshs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 798 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
799 | { |
| 800 | int sat = 0; | |
| 801 | int i; | |
| 802 | ||
| 803 | for (i = 0; i < ARRAY_SIZE(r->s16); i++) { | |
| 804 | int32_t prod = a->s16[i] * b->s16[i] + 0x00004000; | |
| 805 | int32_t t = (int32_t)c->s16[i] + (prod >> 15); | |
| 806 | r->s16[i] = cvtswsh(t, &sat); | |
| 807 | } | |
| 808 | ||
| 809 | if (sat) { | |
| 810 | env->vscr |= (1 << VSCR_SAT); | |
| 811 | } | |
| 812 | } | |
| 813 | ||
| 814 | #define VMINMAX_DO(name, compare, element) \ | |
| 815 | void helper_v##name(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ | |
| 816 | { \ | |
| 817 | int i; \ | |
| 818 | \ | |
| 819 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 820 | if (a->element[i] compare b->element[i]) { \ | |
| 821 | r->element[i] = b->element[i]; \ | |
| 822 | } else { \ | |
| 823 | r->element[i] = a->element[i]; \ | |
| 824 | } \ | |
| 825 | } \ | |
| 826 | } | |
| 827 | #define VMINMAX(suffix, element) \ | |
| 828 | VMINMAX_DO(min##suffix, >, element) \ | |
| 829 | VMINMAX_DO(max##suffix, <, element) | |
| 830 | VMINMAX(sb, s8) | |
| 831 | VMINMAX(sh, s16) | |
| 832 | VMINMAX(sw, s32) | |
| 8203e31b | 833 | VMINMAX(sd, s64) |
| 64654ded BS |
834 | VMINMAX(ub, u8) |
| 835 | VMINMAX(uh, u16) | |
| 836 | VMINMAX(uw, u32) | |
| 8203e31b | 837 | VMINMAX(ud, u64) |
| 64654ded BS |
838 | #undef VMINMAX_DO |
| 839 | #undef VMINMAX | |
| 840 | ||
| 64654ded BS |
841 | void helper_vmladduhm(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c) |
| 842 | { | |
| 843 | int i; | |
| 844 | ||
| 845 | for (i = 0; i < ARRAY_SIZE(r->s16); i++) { | |
| 846 | int32_t prod = a->s16[i] * b->s16[i]; | |
| 847 | r->s16[i] = (int16_t) (prod + c->s16[i]); | |
| 848 | } | |
| 849 | } | |
| 850 | ||
| 851 | #define VMRG_DO(name, element, highp) \ | |
| 852 | void helper_v##name(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ | |
| 853 | { \ | |
| 854 | ppc_avr_t result; \ | |
| 855 | int i; \ | |
| 856 | size_t n_elems = ARRAY_SIZE(r->element); \ | |
| 857 | \ | |
| 858 | for (i = 0; i < n_elems / 2; i++) { \ | |
| 859 | if (highp) { \ | |
| 860 | result.element[i*2+HI_IDX] = a->element[i]; \ | |
| 861 | result.element[i*2+LO_IDX] = b->element[i]; \ | |
| 862 | } else { \ | |
| 863 | result.element[n_elems - i * 2 - (1 + HI_IDX)] = \ | |
| 864 | b->element[n_elems - i - 1]; \ | |
| 865 | result.element[n_elems - i * 2 - (1 + LO_IDX)] = \ | |
| 866 | a->element[n_elems - i - 1]; \ | |
| 867 | } \ | |
| 868 | } \ | |
| 869 | *r = result; \ | |
| 870 | } | |
| 871 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 872 | #define MRGHI 0 | |
| 873 | #define MRGLO 1 | |
| 874 | #else | |
| 875 | #define MRGHI 1 | |
| 876 | #define MRGLO 0 | |
| 877 | #endif | |
| 878 | #define VMRG(suffix, element) \ | |
| 879 | VMRG_DO(mrgl##suffix, element, MRGHI) \ | |
| 880 | VMRG_DO(mrgh##suffix, element, MRGLO) | |
| 881 | VMRG(b, u8) | |
| 882 | VMRG(h, u16) | |
| 883 | VMRG(w, u32) | |
| 884 | #undef VMRG_DO | |
| 885 | #undef VMRG | |
| 886 | #undef MRGHI | |
| 887 | #undef MRGLO | |
| 888 | ||
| d15f74fb BS |
889 | void helper_vmsummbm(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 890 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
891 | { |
| 892 | int32_t prod[16]; | |
| 893 | int i; | |
| 894 | ||
| 895 | for (i = 0; i < ARRAY_SIZE(r->s8); i++) { | |
| 896 | prod[i] = (int32_t)a->s8[i] * b->u8[i]; | |
| 897 | } | |
| 898 | ||
| 899 | VECTOR_FOR_INORDER_I(i, s32) { | |
| 900 | r->s32[i] = c->s32[i] + prod[4 * i] + prod[4 * i + 1] + | |
| 901 | prod[4 * i + 2] + prod[4 * i + 3]; | |
| 902 | } | |
| 903 | } | |
| 904 | ||
| d15f74fb BS |
905 | void helper_vmsumshm(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 906 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
907 | { |
| 908 | int32_t prod[8]; | |
| 909 | int i; | |
| 910 | ||
| 911 | for (i = 0; i < ARRAY_SIZE(r->s16); i++) { | |
| 912 | prod[i] = a->s16[i] * b->s16[i]; | |
| 913 | } | |
| 914 | ||
| 915 | VECTOR_FOR_INORDER_I(i, s32) { | |
| 916 | r->s32[i] = c->s32[i] + prod[2 * i] + prod[2 * i + 1]; | |
| 917 | } | |
| 918 | } | |
| 919 | ||
| d15f74fb BS |
920 | void helper_vmsumshs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 921 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
922 | { |
| 923 | int32_t prod[8]; | |
| 924 | int i; | |
| 925 | int sat = 0; | |
| 926 | ||
| 927 | for (i = 0; i < ARRAY_SIZE(r->s16); i++) { | |
| 928 | prod[i] = (int32_t)a->s16[i] * b->s16[i]; | |
| 929 | } | |
| 930 | ||
| 931 | VECTOR_FOR_INORDER_I(i, s32) { | |
| 932 | int64_t t = (int64_t)c->s32[i] + prod[2 * i] + prod[2 * i + 1]; | |
| 933 | ||
| 934 | r->u32[i] = cvtsdsw(t, &sat); | |
| 935 | } | |
| 936 | ||
| 937 | if (sat) { | |
| 938 | env->vscr |= (1 << VSCR_SAT); | |
| 939 | } | |
| 940 | } | |
| 941 | ||
| d15f74fb BS |
942 | void helper_vmsumubm(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 943 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
944 | { |
| 945 | uint16_t prod[16]; | |
| 946 | int i; | |
| 947 | ||
| 948 | for (i = 0; i < ARRAY_SIZE(r->u8); i++) { | |
| 949 | prod[i] = a->u8[i] * b->u8[i]; | |
| 950 | } | |
| 951 | ||
| 952 | VECTOR_FOR_INORDER_I(i, u32) { | |
| 953 | r->u32[i] = c->u32[i] + prod[4 * i] + prod[4 * i + 1] + | |
| 954 | prod[4 * i + 2] + prod[4 * i + 3]; | |
| 955 | } | |
| 956 | } | |
| 957 | ||
| d15f74fb BS |
958 | void helper_vmsumuhm(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 959 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
960 | { |
| 961 | uint32_t prod[8]; | |
| 962 | int i; | |
| 963 | ||
| 964 | for (i = 0; i < ARRAY_SIZE(r->u16); i++) { | |
| 965 | prod[i] = a->u16[i] * b->u16[i]; | |
| 966 | } | |
| 967 | ||
| 968 | VECTOR_FOR_INORDER_I(i, u32) { | |
| 969 | r->u32[i] = c->u32[i] + prod[2 * i] + prod[2 * i + 1]; | |
| 970 | } | |
| 971 | } | |
| 972 | ||
| d15f74fb BS |
973 | void helper_vmsumuhs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, |
| 974 | ppc_avr_t *b, ppc_avr_t *c) | |
| 64654ded BS |
975 | { |
| 976 | uint32_t prod[8]; | |
| 977 | int i; | |
| 978 | int sat = 0; | |
| 979 | ||
| 980 | for (i = 0; i < ARRAY_SIZE(r->u16); i++) { | |
| 981 | prod[i] = a->u16[i] * b->u16[i]; | |
| 982 | } | |
| 983 | ||
| 984 | VECTOR_FOR_INORDER_I(i, s32) { | |
| 985 | uint64_t t = (uint64_t)c->u32[i] + prod[2 * i] + prod[2 * i + 1]; | |
| 986 | ||
| 987 | r->u32[i] = cvtuduw(t, &sat); | |
| 988 | } | |
| 989 | ||
| 990 | if (sat) { | |
| 991 | env->vscr |= (1 << VSCR_SAT); | |
| 992 | } | |
| 993 | } | |
| 994 | ||
| aa9e930c | 995 | #define VMUL_DO(name, mul_element, prod_element, cast, evenp) \ |
| 64654ded BS |
996 | void helper_v##name(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ |
| 997 | { \ | |
| 998 | int i; \ | |
| 999 | \ | |
| 1000 | VECTOR_FOR_INORDER_I(i, prod_element) { \ | |
| 1001 | if (evenp) { \ | |
| aa9e930c TM |
1002 | r->prod_element[i] = \ |
| 1003 | (cast)a->mul_element[i * 2 + HI_IDX] * \ | |
| 1004 | (cast)b->mul_element[i * 2 + HI_IDX]; \ | |
| 64654ded | 1005 | } else { \ |
| aa9e930c TM |
1006 | r->prod_element[i] = \ |
| 1007 | (cast)a->mul_element[i * 2 + LO_IDX] * \ | |
| 1008 | (cast)b->mul_element[i * 2 + LO_IDX]; \ | |
| 64654ded BS |
1009 | } \ |
| 1010 | } \ | |
| 1011 | } | |
| aa9e930c TM |
1012 | #define VMUL(suffix, mul_element, prod_element, cast) \ |
| 1013 | VMUL_DO(mule##suffix, mul_element, prod_element, cast, 1) \ | |
| 1014 | VMUL_DO(mulo##suffix, mul_element, prod_element, cast, 0) | |
| 1015 | VMUL(sb, s8, s16, int16_t) | |
| 1016 | VMUL(sh, s16, s32, int32_t) | |
| 63be0936 | 1017 | VMUL(sw, s32, s64, int64_t) |
| aa9e930c TM |
1018 | VMUL(ub, u8, u16, uint16_t) |
| 1019 | VMUL(uh, u16, u32, uint32_t) | |
| 63be0936 | 1020 | VMUL(uw, u32, u64, uint64_t) |
| 64654ded BS |
1021 | #undef VMUL_DO |
| 1022 | #undef VMUL | |
| 1023 | ||
| d15f74fb BS |
1024 | void helper_vperm(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, |
| 1025 | ppc_avr_t *c) | |
| 64654ded BS |
1026 | { |
| 1027 | ppc_avr_t result; | |
| 1028 | int i; | |
| 1029 | ||
| 1030 | VECTOR_FOR_INORDER_I(i, u8) { | |
| 1031 | int s = c->u8[i] & 0x1f; | |
| 1032 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1033 | int index = s & 0xf; | |
| 1034 | #else | |
| 1035 | int index = 15 - (s & 0xf); | |
| 1036 | #endif | |
| 1037 | ||
| 1038 | if (s & 0x10) { | |
| 1039 | result.u8[i] = b->u8[index]; | |
| 1040 | } else { | |
| 1041 | result.u8[i] = a->u8[index]; | |
| 1042 | } | |
| 1043 | } | |
| 1044 | *r = result; | |
| 1045 | } | |
| 1046 | ||
| 4d82038e TM |
1047 | #if defined(HOST_WORDS_BIGENDIAN) |
| 1048 | #define VBPERMQ_INDEX(avr, i) ((avr)->u8[(i)]) | |
| 1049 | #define VBPERMQ_DW(index) (((index) & 0x40) != 0) | |
| 1050 | #else | |
| 1051 | #define VBPERMQ_INDEX(avr, i) ((avr)->u8[15-(i)]) | |
| 1052 | #define VBPERMQ_DW(index) (((index) & 0x40) == 0) | |
| 1053 | #endif | |
| 1054 | ||
| 1055 | void helper_vbpermq(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1056 | { | |
| 1057 | int i; | |
| 1058 | uint64_t perm = 0; | |
| 1059 | ||
| 1060 | VECTOR_FOR_INORDER_I(i, u8) { | |
| 1061 | int index = VBPERMQ_INDEX(b, i); | |
| 1062 | ||
| 1063 | if (index < 128) { | |
| 1064 | uint64_t mask = (1ull << (63-(index & 0x3F))); | |
| 1065 | if (a->u64[VBPERMQ_DW(index)] & mask) { | |
| 1066 | perm |= (0x8000 >> i); | |
| 1067 | } | |
| 1068 | } | |
| 1069 | } | |
| 1070 | ||
| 1071 | r->u64[HI_IDX] = perm; | |
| 1072 | r->u64[LO_IDX] = 0; | |
| 1073 | } | |
| 1074 | ||
| 1075 | #undef VBPERMQ_INDEX | |
| 1076 | #undef VBPERMQ_DW | |
| 1077 | ||
| f1064f61 TM |
1078 | uint64_t VGBBD_MASKS[256] = { |
| 1079 | 0x0000000000000000ull, /* 00 */ | |
| 1080 | 0x0000000000000080ull, /* 01 */ | |
| 1081 | 0x0000000000008000ull, /* 02 */ | |
| 1082 | 0x0000000000008080ull, /* 03 */ | |
| 1083 | 0x0000000000800000ull, /* 04 */ | |
| 1084 | 0x0000000000800080ull, /* 05 */ | |
| 1085 | 0x0000000000808000ull, /* 06 */ | |
| 1086 | 0x0000000000808080ull, /* 07 */ | |
| 1087 | 0x0000000080000000ull, /* 08 */ | |
| 1088 | 0x0000000080000080ull, /* 09 */ | |
| 1089 | 0x0000000080008000ull, /* 0A */ | |
| 1090 | 0x0000000080008080ull, /* 0B */ | |
| 1091 | 0x0000000080800000ull, /* 0C */ | |
| 1092 | 0x0000000080800080ull, /* 0D */ | |
| 1093 | 0x0000000080808000ull, /* 0E */ | |
| 1094 | 0x0000000080808080ull, /* 0F */ | |
| 1095 | 0x0000008000000000ull, /* 10 */ | |
| 1096 | 0x0000008000000080ull, /* 11 */ | |
| 1097 | 0x0000008000008000ull, /* 12 */ | |
| 1098 | 0x0000008000008080ull, /* 13 */ | |
| 1099 | 0x0000008000800000ull, /* 14 */ | |
| 1100 | 0x0000008000800080ull, /* 15 */ | |
| 1101 | 0x0000008000808000ull, /* 16 */ | |
| 1102 | 0x0000008000808080ull, /* 17 */ | |
| 1103 | 0x0000008080000000ull, /* 18 */ | |
| 1104 | 0x0000008080000080ull, /* 19 */ | |
| 1105 | 0x0000008080008000ull, /* 1A */ | |
| 1106 | 0x0000008080008080ull, /* 1B */ | |
| 1107 | 0x0000008080800000ull, /* 1C */ | |
| 1108 | 0x0000008080800080ull, /* 1D */ | |
| 1109 | 0x0000008080808000ull, /* 1E */ | |
| 1110 | 0x0000008080808080ull, /* 1F */ | |
| 1111 | 0x0000800000000000ull, /* 20 */ | |
| 1112 | 0x0000800000000080ull, /* 21 */ | |
| 1113 | 0x0000800000008000ull, /* 22 */ | |
| 1114 | 0x0000800000008080ull, /* 23 */ | |
| 1115 | 0x0000800000800000ull, /* 24 */ | |
| 1116 | 0x0000800000800080ull, /* 25 */ | |
| 1117 | 0x0000800000808000ull, /* 26 */ | |
| 1118 | 0x0000800000808080ull, /* 27 */ | |
| 1119 | 0x0000800080000000ull, /* 28 */ | |
| 1120 | 0x0000800080000080ull, /* 29 */ | |
| 1121 | 0x0000800080008000ull, /* 2A */ | |
| 1122 | 0x0000800080008080ull, /* 2B */ | |
| 1123 | 0x0000800080800000ull, /* 2C */ | |
| 1124 | 0x0000800080800080ull, /* 2D */ | |
| 1125 | 0x0000800080808000ull, /* 2E */ | |
| 1126 | 0x0000800080808080ull, /* 2F */ | |
| 1127 | 0x0000808000000000ull, /* 30 */ | |
| 1128 | 0x0000808000000080ull, /* 31 */ | |
| 1129 | 0x0000808000008000ull, /* 32 */ | |
| 1130 | 0x0000808000008080ull, /* 33 */ | |
| 1131 | 0x0000808000800000ull, /* 34 */ | |
| 1132 | 0x0000808000800080ull, /* 35 */ | |
| 1133 | 0x0000808000808000ull, /* 36 */ | |
| 1134 | 0x0000808000808080ull, /* 37 */ | |
| 1135 | 0x0000808080000000ull, /* 38 */ | |
| 1136 | 0x0000808080000080ull, /* 39 */ | |
| 1137 | 0x0000808080008000ull, /* 3A */ | |
| 1138 | 0x0000808080008080ull, /* 3B */ | |
| 1139 | 0x0000808080800000ull, /* 3C */ | |
| 1140 | 0x0000808080800080ull, /* 3D */ | |
| 1141 | 0x0000808080808000ull, /* 3E */ | |
| 1142 | 0x0000808080808080ull, /* 3F */ | |
| 1143 | 0x0080000000000000ull, /* 40 */ | |
| 1144 | 0x0080000000000080ull, /* 41 */ | |
| 1145 | 0x0080000000008000ull, /* 42 */ | |
| 1146 | 0x0080000000008080ull, /* 43 */ | |
| 1147 | 0x0080000000800000ull, /* 44 */ | |
| 1148 | 0x0080000000800080ull, /* 45 */ | |
| 1149 | 0x0080000000808000ull, /* 46 */ | |
| 1150 | 0x0080000000808080ull, /* 47 */ | |
| 1151 | 0x0080000080000000ull, /* 48 */ | |
| 1152 | 0x0080000080000080ull, /* 49 */ | |
| 1153 | 0x0080000080008000ull, /* 4A */ | |
| 1154 | 0x0080000080008080ull, /* 4B */ | |
| 1155 | 0x0080000080800000ull, /* 4C */ | |
| 1156 | 0x0080000080800080ull, /* 4D */ | |
| 1157 | 0x0080000080808000ull, /* 4E */ | |
| 1158 | 0x0080000080808080ull, /* 4F */ | |
| 1159 | 0x0080008000000000ull, /* 50 */ | |
| 1160 | 0x0080008000000080ull, /* 51 */ | |
| 1161 | 0x0080008000008000ull, /* 52 */ | |
| 1162 | 0x0080008000008080ull, /* 53 */ | |
| 1163 | 0x0080008000800000ull, /* 54 */ | |
| 1164 | 0x0080008000800080ull, /* 55 */ | |
| 1165 | 0x0080008000808000ull, /* 56 */ | |
| 1166 | 0x0080008000808080ull, /* 57 */ | |
| 1167 | 0x0080008080000000ull, /* 58 */ | |
| 1168 | 0x0080008080000080ull, /* 59 */ | |
| 1169 | 0x0080008080008000ull, /* 5A */ | |
| 1170 | 0x0080008080008080ull, /* 5B */ | |
| 1171 | 0x0080008080800000ull, /* 5C */ | |
| 1172 | 0x0080008080800080ull, /* 5D */ | |
| 1173 | 0x0080008080808000ull, /* 5E */ | |
| 1174 | 0x0080008080808080ull, /* 5F */ | |
| 1175 | 0x0080800000000000ull, /* 60 */ | |
| 1176 | 0x0080800000000080ull, /* 61 */ | |
| 1177 | 0x0080800000008000ull, /* 62 */ | |
| 1178 | 0x0080800000008080ull, /* 63 */ | |
| 1179 | 0x0080800000800000ull, /* 64 */ | |
| 1180 | 0x0080800000800080ull, /* 65 */ | |
| 1181 | 0x0080800000808000ull, /* 66 */ | |
| 1182 | 0x0080800000808080ull, /* 67 */ | |
| 1183 | 0x0080800080000000ull, /* 68 */ | |
| 1184 | 0x0080800080000080ull, /* 69 */ | |
| 1185 | 0x0080800080008000ull, /* 6A */ | |
| 1186 | 0x0080800080008080ull, /* 6B */ | |
| 1187 | 0x0080800080800000ull, /* 6C */ | |
| 1188 | 0x0080800080800080ull, /* 6D */ | |
| 1189 | 0x0080800080808000ull, /* 6E */ | |
| 1190 | 0x0080800080808080ull, /* 6F */ | |
| 1191 | 0x0080808000000000ull, /* 70 */ | |
| 1192 | 0x0080808000000080ull, /* 71 */ | |
| 1193 | 0x0080808000008000ull, /* 72 */ | |
| 1194 | 0x0080808000008080ull, /* 73 */ | |
| 1195 | 0x0080808000800000ull, /* 74 */ | |
| 1196 | 0x0080808000800080ull, /* 75 */ | |
| 1197 | 0x0080808000808000ull, /* 76 */ | |
| 1198 | 0x0080808000808080ull, /* 77 */ | |
| 1199 | 0x0080808080000000ull, /* 78 */ | |
| 1200 | 0x0080808080000080ull, /* 79 */ | |
| 1201 | 0x0080808080008000ull, /* 7A */ | |
| 1202 | 0x0080808080008080ull, /* 7B */ | |
| 1203 | 0x0080808080800000ull, /* 7C */ | |
| 1204 | 0x0080808080800080ull, /* 7D */ | |
| 1205 | 0x0080808080808000ull, /* 7E */ | |
| 1206 | 0x0080808080808080ull, /* 7F */ | |
| 1207 | 0x8000000000000000ull, /* 80 */ | |
| 1208 | 0x8000000000000080ull, /* 81 */ | |
| 1209 | 0x8000000000008000ull, /* 82 */ | |
| 1210 | 0x8000000000008080ull, /* 83 */ | |
| 1211 | 0x8000000000800000ull, /* 84 */ | |
| 1212 | 0x8000000000800080ull, /* 85 */ | |
| 1213 | 0x8000000000808000ull, /* 86 */ | |
| 1214 | 0x8000000000808080ull, /* 87 */ | |
| 1215 | 0x8000000080000000ull, /* 88 */ | |
| 1216 | 0x8000000080000080ull, /* 89 */ | |
| 1217 | 0x8000000080008000ull, /* 8A */ | |
| 1218 | 0x8000000080008080ull, /* 8B */ | |
| 1219 | 0x8000000080800000ull, /* 8C */ | |
| 1220 | 0x8000000080800080ull, /* 8D */ | |
| 1221 | 0x8000000080808000ull, /* 8E */ | |
| 1222 | 0x8000000080808080ull, /* 8F */ | |
| 1223 | 0x8000008000000000ull, /* 90 */ | |
| 1224 | 0x8000008000000080ull, /* 91 */ | |
| 1225 | 0x8000008000008000ull, /* 92 */ | |
| 1226 | 0x8000008000008080ull, /* 93 */ | |
| 1227 | 0x8000008000800000ull, /* 94 */ | |
| 1228 | 0x8000008000800080ull, /* 95 */ | |
| 1229 | 0x8000008000808000ull, /* 96 */ | |
| 1230 | 0x8000008000808080ull, /* 97 */ | |
| 1231 | 0x8000008080000000ull, /* 98 */ | |
| 1232 | 0x8000008080000080ull, /* 99 */ | |
| 1233 | 0x8000008080008000ull, /* 9A */ | |
| 1234 | 0x8000008080008080ull, /* 9B */ | |
| 1235 | 0x8000008080800000ull, /* 9C */ | |
| 1236 | 0x8000008080800080ull, /* 9D */ | |
| 1237 | 0x8000008080808000ull, /* 9E */ | |
| 1238 | 0x8000008080808080ull, /* 9F */ | |
| 1239 | 0x8000800000000000ull, /* A0 */ | |
| 1240 | 0x8000800000000080ull, /* A1 */ | |
| 1241 | 0x8000800000008000ull, /* A2 */ | |
| 1242 | 0x8000800000008080ull, /* A3 */ | |
| 1243 | 0x8000800000800000ull, /* A4 */ | |
| 1244 | 0x8000800000800080ull, /* A5 */ | |
| 1245 | 0x8000800000808000ull, /* A6 */ | |
| 1246 | 0x8000800000808080ull, /* A7 */ | |
| 1247 | 0x8000800080000000ull, /* A8 */ | |
| 1248 | 0x8000800080000080ull, /* A9 */ | |
| 1249 | 0x8000800080008000ull, /* AA */ | |
| 1250 | 0x8000800080008080ull, /* AB */ | |
| 1251 | 0x8000800080800000ull, /* AC */ | |
| 1252 | 0x8000800080800080ull, /* AD */ | |
| 1253 | 0x8000800080808000ull, /* AE */ | |
| 1254 | 0x8000800080808080ull, /* AF */ | |
| 1255 | 0x8000808000000000ull, /* B0 */ | |
| 1256 | 0x8000808000000080ull, /* B1 */ | |
| 1257 | 0x8000808000008000ull, /* B2 */ | |
| 1258 | 0x8000808000008080ull, /* B3 */ | |
| 1259 | 0x8000808000800000ull, /* B4 */ | |
| 1260 | 0x8000808000800080ull, /* B5 */ | |
| 1261 | 0x8000808000808000ull, /* B6 */ | |
| 1262 | 0x8000808000808080ull, /* B7 */ | |
| 1263 | 0x8000808080000000ull, /* B8 */ | |
| 1264 | 0x8000808080000080ull, /* B9 */ | |
| 1265 | 0x8000808080008000ull, /* BA */ | |
| 1266 | 0x8000808080008080ull, /* BB */ | |
| 1267 | 0x8000808080800000ull, /* BC */ | |
| 1268 | 0x8000808080800080ull, /* BD */ | |
| 1269 | 0x8000808080808000ull, /* BE */ | |
| 1270 | 0x8000808080808080ull, /* BF */ | |
| 1271 | 0x8080000000000000ull, /* C0 */ | |
| 1272 | 0x8080000000000080ull, /* C1 */ | |
| 1273 | 0x8080000000008000ull, /* C2 */ | |
| 1274 | 0x8080000000008080ull, /* C3 */ | |
| 1275 | 0x8080000000800000ull, /* C4 */ | |
| 1276 | 0x8080000000800080ull, /* C5 */ | |
| 1277 | 0x8080000000808000ull, /* C6 */ | |
| 1278 | 0x8080000000808080ull, /* C7 */ | |
| 1279 | 0x8080000080000000ull, /* C8 */ | |
| 1280 | 0x8080000080000080ull, /* C9 */ | |
| 1281 | 0x8080000080008000ull, /* CA */ | |
| 1282 | 0x8080000080008080ull, /* CB */ | |
| 1283 | 0x8080000080800000ull, /* CC */ | |
| 1284 | 0x8080000080800080ull, /* CD */ | |
| 1285 | 0x8080000080808000ull, /* CE */ | |
| 1286 | 0x8080000080808080ull, /* CF */ | |
| 1287 | 0x8080008000000000ull, /* D0 */ | |
| 1288 | 0x8080008000000080ull, /* D1 */ | |
| 1289 | 0x8080008000008000ull, /* D2 */ | |
| 1290 | 0x8080008000008080ull, /* D3 */ | |
| 1291 | 0x8080008000800000ull, /* D4 */ | |
| 1292 | 0x8080008000800080ull, /* D5 */ | |
| 1293 | 0x8080008000808000ull, /* D6 */ | |
| 1294 | 0x8080008000808080ull, /* D7 */ | |
| 1295 | 0x8080008080000000ull, /* D8 */ | |
| 1296 | 0x8080008080000080ull, /* D9 */ | |
| 1297 | 0x8080008080008000ull, /* DA */ | |
| 1298 | 0x8080008080008080ull, /* DB */ | |
| 1299 | 0x8080008080800000ull, /* DC */ | |
| 1300 | 0x8080008080800080ull, /* DD */ | |
| 1301 | 0x8080008080808000ull, /* DE */ | |
| 1302 | 0x8080008080808080ull, /* DF */ | |
| 1303 | 0x8080800000000000ull, /* E0 */ | |
| 1304 | 0x8080800000000080ull, /* E1 */ | |
| 1305 | 0x8080800000008000ull, /* E2 */ | |
| 1306 | 0x8080800000008080ull, /* E3 */ | |
| 1307 | 0x8080800000800000ull, /* E4 */ | |
| 1308 | 0x8080800000800080ull, /* E5 */ | |
| 1309 | 0x8080800000808000ull, /* E6 */ | |
| 1310 | 0x8080800000808080ull, /* E7 */ | |
| 1311 | 0x8080800080000000ull, /* E8 */ | |
| 1312 | 0x8080800080000080ull, /* E9 */ | |
| 1313 | 0x8080800080008000ull, /* EA */ | |
| 1314 | 0x8080800080008080ull, /* EB */ | |
| 1315 | 0x8080800080800000ull, /* EC */ | |
| 1316 | 0x8080800080800080ull, /* ED */ | |
| 1317 | 0x8080800080808000ull, /* EE */ | |
| 1318 | 0x8080800080808080ull, /* EF */ | |
| 1319 | 0x8080808000000000ull, /* F0 */ | |
| 1320 | 0x8080808000000080ull, /* F1 */ | |
| 1321 | 0x8080808000008000ull, /* F2 */ | |
| 1322 | 0x8080808000008080ull, /* F3 */ | |
| 1323 | 0x8080808000800000ull, /* F4 */ | |
| 1324 | 0x8080808000800080ull, /* F5 */ | |
| 1325 | 0x8080808000808000ull, /* F6 */ | |
| 1326 | 0x8080808000808080ull, /* F7 */ | |
| 1327 | 0x8080808080000000ull, /* F8 */ | |
| 1328 | 0x8080808080000080ull, /* F9 */ | |
| 1329 | 0x8080808080008000ull, /* FA */ | |
| 1330 | 0x8080808080008080ull, /* FB */ | |
| 1331 | 0x8080808080800000ull, /* FC */ | |
| 1332 | 0x8080808080800080ull, /* FD */ | |
| 1333 | 0x8080808080808000ull, /* FE */ | |
| 1334 | 0x8080808080808080ull, /* FF */ | |
| 1335 | }; | |
| 1336 | ||
| 1337 | void helper_vgbbd(ppc_avr_t *r, ppc_avr_t *b) | |
| 1338 | { | |
| 1339 | int i; | |
| 1340 | uint64_t t[2] = { 0, 0 }; | |
| 1341 | ||
| 1342 | VECTOR_FOR_INORDER_I(i, u8) { | |
| 1343 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1344 | t[i>>3] |= VGBBD_MASKS[b->u8[i]] >> (i & 7); | |
| 1345 | #else | |
| 1346 | t[i>>3] |= VGBBD_MASKS[b->u8[i]] >> (7-(i & 7)); | |
| 1347 | #endif | |
| 1348 | } | |
| 1349 | ||
| 1350 | r->u64[0] = t[0]; | |
| 1351 | r->u64[1] = t[1]; | |
| 1352 | } | |
| 1353 | ||
| 64654ded BS |
1354 | #if defined(HOST_WORDS_BIGENDIAN) |
| 1355 | #define PKBIG 1 | |
| 1356 | #else | |
| 1357 | #define PKBIG 0 | |
| 1358 | #endif | |
| 1359 | void helper_vpkpx(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1360 | { | |
| 1361 | int i, j; | |
| 1362 | ppc_avr_t result; | |
| 1363 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1364 | const ppc_avr_t *x[2] = { a, b }; | |
| 1365 | #else | |
| 1366 | const ppc_avr_t *x[2] = { b, a }; | |
| 1367 | #endif | |
| 1368 | ||
| 1369 | VECTOR_FOR_INORDER_I(i, u64) { | |
| 1370 | VECTOR_FOR_INORDER_I(j, u32) { | |
| 1371 | uint32_t e = x[i]->u32[j]; | |
| 1372 | ||
| 1373 | result.u16[4*i+j] = (((e >> 9) & 0xfc00) | | |
| 1374 | ((e >> 6) & 0x3e0) | | |
| 1375 | ((e >> 3) & 0x1f)); | |
| 1376 | } | |
| 1377 | } | |
| 1378 | *r = result; | |
| 1379 | } | |
| 1380 | ||
| 1381 | #define VPK(suffix, from, to, cvt, dosat) \ | |
| d15f74fb BS |
1382 | void helper_vpk##suffix(CPUPPCState *env, ppc_avr_t *r, \ |
| 1383 | ppc_avr_t *a, ppc_avr_t *b) \ | |
| 64654ded BS |
1384 | { \ |
| 1385 | int i; \ | |
| 1386 | int sat = 0; \ | |
| 1387 | ppc_avr_t result; \ | |
| 1388 | ppc_avr_t *a0 = PKBIG ? a : b; \ | |
| 1389 | ppc_avr_t *a1 = PKBIG ? b : a; \ | |
| 1390 | \ | |
| 1391 | VECTOR_FOR_INORDER_I(i, from) { \ | |
| 1392 | result.to[i] = cvt(a0->from[i], &sat); \ | |
| 1393 | result.to[i+ARRAY_SIZE(r->from)] = cvt(a1->from[i], &sat); \ | |
| 1394 | } \ | |
| 1395 | *r = result; \ | |
| 1396 | if (dosat && sat) { \ | |
| 1397 | env->vscr |= (1 << VSCR_SAT); \ | |
| 1398 | } \ | |
| 1399 | } | |
| 1400 | #define I(x, y) (x) | |
| 1401 | VPK(shss, s16, s8, cvtshsb, 1) | |
| 1402 | VPK(shus, s16, u8, cvtshub, 1) | |
| 1403 | VPK(swss, s32, s16, cvtswsh, 1) | |
| 1404 | VPK(swus, s32, u16, cvtswuh, 1) | |
| 024215b2 TM |
1405 | VPK(sdss, s64, s32, cvtsdsw, 1) |
| 1406 | VPK(sdus, s64, u32, cvtsduw, 1) | |
| 64654ded BS |
1407 | VPK(uhus, u16, u8, cvtuhub, 1) |
| 1408 | VPK(uwus, u32, u16, cvtuwuh, 1) | |
| 024215b2 | 1409 | VPK(udus, u64, u32, cvtuduw, 1) |
| 64654ded BS |
1410 | VPK(uhum, u16, u8, I, 0) |
| 1411 | VPK(uwum, u32, u16, I, 0) | |
| 024215b2 | 1412 | VPK(udum, u64, u32, I, 0) |
| 64654ded BS |
1413 | #undef I |
| 1414 | #undef VPK | |
| 1415 | #undef PKBIG | |
| 1416 | ||
| d15f74fb | 1417 | void helper_vrefp(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *b) |
| 64654ded BS |
1418 | { |
| 1419 | int i; | |
| 1420 | ||
| 1421 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { | |
| ef9bd150 | 1422 | r->f[i] = float32_div(float32_one, b->f[i], &env->vec_status); |
| 64654ded BS |
1423 | } |
| 1424 | } | |
| 1425 | ||
| 1426 | #define VRFI(suffix, rounding) \ | |
| d15f74fb BS |
1427 | void helper_vrfi##suffix(CPUPPCState *env, ppc_avr_t *r, \ |
| 1428 | ppc_avr_t *b) \ | |
| 64654ded BS |
1429 | { \ |
| 1430 | int i; \ | |
| 1431 | float_status s = env->vec_status; \ | |
| 1432 | \ | |
| 1433 | set_float_rounding_mode(rounding, &s); \ | |
| 1434 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { \ | |
| ef9bd150 | 1435 | r->f[i] = float32_round_to_int (b->f[i], &s); \ |
| 64654ded BS |
1436 | } \ |
| 1437 | } | |
| 1438 | VRFI(n, float_round_nearest_even) | |
| 1439 | VRFI(m, float_round_down) | |
| 1440 | VRFI(p, float_round_up) | |
| 1441 | VRFI(z, float_round_to_zero) | |
| 1442 | #undef VRFI | |
| 1443 | ||
| 818692ff | 1444 | #define VROTATE(suffix, element, mask) \ |
| 64654ded BS |
1445 | void helper_vrl##suffix(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ |
| 1446 | { \ | |
| 1447 | int i; \ | |
| 1448 | \ | |
| 1449 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 64654ded BS |
1450 | unsigned int shift = b->element[i] & mask; \ |
| 1451 | r->element[i] = (a->element[i] << shift) | \ | |
| 1452 | (a->element[i] >> (sizeof(a->element[0]) * 8 - shift)); \ | |
| 1453 | } \ | |
| 1454 | } | |
| 818692ff TM |
1455 | VROTATE(b, u8, 0x7) |
| 1456 | VROTATE(h, u16, 0xF) | |
| 1457 | VROTATE(w, u32, 0x1F) | |
| 2fdf78e6 | 1458 | VROTATE(d, u64, 0x3F) |
| 64654ded BS |
1459 | #undef VROTATE |
| 1460 | ||
| d15f74fb | 1461 | void helper_vrsqrtefp(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *b) |
| 64654ded BS |
1462 | { |
| 1463 | int i; | |
| 1464 | ||
| 1465 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { | |
| ef9bd150 | 1466 | float32 t = float32_sqrt(b->f[i], &env->vec_status); |
| 64654ded | 1467 | |
| ef9bd150 | 1468 | r->f[i] = float32_div(float32_one, t, &env->vec_status); |
| 64654ded BS |
1469 | } |
| 1470 | } | |
| 1471 | ||
| d15f74fb BS |
1472 | void helper_vsel(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, |
| 1473 | ppc_avr_t *c) | |
| 64654ded BS |
1474 | { |
| 1475 | r->u64[0] = (a->u64[0] & ~c->u64[0]) | (b->u64[0] & c->u64[0]); | |
| 1476 | r->u64[1] = (a->u64[1] & ~c->u64[1]) | (b->u64[1] & c->u64[1]); | |
| 1477 | } | |
| 1478 | ||
| d15f74fb | 1479 | void helper_vexptefp(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *b) |
| 64654ded BS |
1480 | { |
| 1481 | int i; | |
| 1482 | ||
| 1483 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { | |
| ef9bd150 | 1484 | r->f[i] = float32_exp2(b->f[i], &env->vec_status); |
| 64654ded BS |
1485 | } |
| 1486 | } | |
| 1487 | ||
| d15f74fb | 1488 | void helper_vlogefp(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *b) |
| 64654ded BS |
1489 | { |
| 1490 | int i; | |
| 1491 | ||
| 1492 | for (i = 0; i < ARRAY_SIZE(r->f); i++) { | |
| ef9bd150 | 1493 | r->f[i] = float32_log2(b->f[i], &env->vec_status); |
| 64654ded BS |
1494 | } |
| 1495 | } | |
| 1496 | ||
| 1497 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1498 | #define LEFT 0 | |
| 1499 | #define RIGHT 1 | |
| 1500 | #else | |
| 1501 | #define LEFT 1 | |
| 1502 | #define RIGHT 0 | |
| 1503 | #endif | |
| 1504 | /* The specification says that the results are undefined if all of the | |
| 1505 | * shift counts are not identical. We check to make sure that they are | |
| 1506 | * to conform to what real hardware appears to do. */ | |
| 1507 | #define VSHIFT(suffix, leftp) \ | |
| 1508 | void helper_vs##suffix(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ | |
| 1509 | { \ | |
| 1510 | int shift = b->u8[LO_IDX*15] & 0x7; \ | |
| 1511 | int doit = 1; \ | |
| 1512 | int i; \ | |
| 1513 | \ | |
| 1514 | for (i = 0; i < ARRAY_SIZE(r->u8); i++) { \ | |
| 1515 | doit = doit && ((b->u8[i] & 0x7) == shift); \ | |
| 1516 | } \ | |
| 1517 | if (doit) { \ | |
| 1518 | if (shift == 0) { \ | |
| 1519 | *r = *a; \ | |
| 1520 | } else if (leftp) { \ | |
| 1521 | uint64_t carry = a->u64[LO_IDX] >> (64 - shift); \ | |
| 1522 | \ | |
| 1523 | r->u64[HI_IDX] = (a->u64[HI_IDX] << shift) | carry; \ | |
| 1524 | r->u64[LO_IDX] = a->u64[LO_IDX] << shift; \ | |
| 1525 | } else { \ | |
| 1526 | uint64_t carry = a->u64[HI_IDX] << (64 - shift); \ | |
| 1527 | \ | |
| 1528 | r->u64[LO_IDX] = (a->u64[LO_IDX] >> shift) | carry; \ | |
| 1529 | r->u64[HI_IDX] = a->u64[HI_IDX] >> shift; \ | |
| 1530 | } \ | |
| 1531 | } \ | |
| 1532 | } | |
| 1533 | VSHIFT(l, LEFT) | |
| 1534 | VSHIFT(r, RIGHT) | |
| 1535 | #undef VSHIFT | |
| 1536 | #undef LEFT | |
| 1537 | #undef RIGHT | |
| 1538 | ||
| 818692ff | 1539 | #define VSL(suffix, element, mask) \ |
| 64654ded BS |
1540 | void helper_vsl##suffix(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ |
| 1541 | { \ | |
| 1542 | int i; \ | |
| 1543 | \ | |
| 1544 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 64654ded BS |
1545 | unsigned int shift = b->element[i] & mask; \ |
| 1546 | \ | |
| 1547 | r->element[i] = a->element[i] << shift; \ | |
| 1548 | } \ | |
| 1549 | } | |
| 818692ff TM |
1550 | VSL(b, u8, 0x7) |
| 1551 | VSL(h, u16, 0x0F) | |
| 1552 | VSL(w, u32, 0x1F) | |
| 2fdf78e6 | 1553 | VSL(d, u64, 0x3F) |
| 64654ded BS |
1554 | #undef VSL |
| 1555 | ||
| 1556 | void helper_vsldoi(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, uint32_t shift) | |
| 1557 | { | |
| 1558 | int sh = shift & 0xf; | |
| 1559 | int i; | |
| 1560 | ppc_avr_t result; | |
| 1561 | ||
| 1562 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1563 | for (i = 0; i < ARRAY_SIZE(r->u8); i++) { | |
| 1564 | int index = sh + i; | |
| 1565 | if (index > 0xf) { | |
| 1566 | result.u8[i] = b->u8[index - 0x10]; | |
| 1567 | } else { | |
| 1568 | result.u8[i] = a->u8[index]; | |
| 1569 | } | |
| 1570 | } | |
| 1571 | #else | |
| 1572 | for (i = 0; i < ARRAY_SIZE(r->u8); i++) { | |
| 1573 | int index = (16 - sh) + i; | |
| 1574 | if (index > 0xf) { | |
| 1575 | result.u8[i] = a->u8[index - 0x10]; | |
| 1576 | } else { | |
| 1577 | result.u8[i] = b->u8[index]; | |
| 1578 | } | |
| 1579 | } | |
| 1580 | #endif | |
| 1581 | *r = result; | |
| 1582 | } | |
| 1583 | ||
| 1584 | void helper_vslo(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1585 | { | |
| 1586 | int sh = (b->u8[LO_IDX*0xf] >> 3) & 0xf; | |
| 1587 | ||
| 1588 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1589 | memmove(&r->u8[0], &a->u8[sh], 16 - sh); | |
| 1590 | memset(&r->u8[16-sh], 0, sh); | |
| 1591 | #else | |
| 1592 | memmove(&r->u8[sh], &a->u8[0], 16 - sh); | |
| 1593 | memset(&r->u8[0], 0, sh); | |
| 1594 | #endif | |
| 1595 | } | |
| 1596 | ||
| 1597 | /* Experimental testing shows that hardware masks the immediate. */ | |
| 1598 | #define _SPLAT_MASKED(element) (splat & (ARRAY_SIZE(r->element) - 1)) | |
| 1599 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1600 | #define SPLAT_ELEMENT(element) _SPLAT_MASKED(element) | |
| 1601 | #else | |
| 1602 | #define SPLAT_ELEMENT(element) \ | |
| 1603 | (ARRAY_SIZE(r->element) - 1 - _SPLAT_MASKED(element)) | |
| 1604 | #endif | |
| 1605 | #define VSPLT(suffix, element) \ | |
| 1606 | void helper_vsplt##suffix(ppc_avr_t *r, ppc_avr_t *b, uint32_t splat) \ | |
| 1607 | { \ | |
| 1608 | uint32_t s = b->element[SPLAT_ELEMENT(element)]; \ | |
| 1609 | int i; \ | |
| 1610 | \ | |
| 1611 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 1612 | r->element[i] = s; \ | |
| 1613 | } \ | |
| 1614 | } | |
| 1615 | VSPLT(b, u8) | |
| 1616 | VSPLT(h, u16) | |
| 1617 | VSPLT(w, u32) | |
| 1618 | #undef VSPLT | |
| 1619 | #undef SPLAT_ELEMENT | |
| 1620 | #undef _SPLAT_MASKED | |
| 1621 | ||
| 1622 | #define VSPLTI(suffix, element, splat_type) \ | |
| 1623 | void helper_vspltis##suffix(ppc_avr_t *r, uint32_t splat) \ | |
| 1624 | { \ | |
| 1625 | splat_type x = (int8_t)(splat << 3) >> 3; \ | |
| 1626 | int i; \ | |
| 1627 | \ | |
| 1628 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 1629 | r->element[i] = x; \ | |
| 1630 | } \ | |
| 1631 | } | |
| 1632 | VSPLTI(b, s8, int8_t) | |
| 1633 | VSPLTI(h, s16, int16_t) | |
| 1634 | VSPLTI(w, s32, int32_t) | |
| 1635 | #undef VSPLTI | |
| 1636 | ||
| 818692ff | 1637 | #define VSR(suffix, element, mask) \ |
| 64654ded BS |
1638 | void helper_vsr##suffix(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \ |
| 1639 | { \ | |
| 1640 | int i; \ | |
| 1641 | \ | |
| 1642 | for (i = 0; i < ARRAY_SIZE(r->element); i++) { \ | |
| 64654ded | 1643 | unsigned int shift = b->element[i] & mask; \ |
| 64654ded BS |
1644 | r->element[i] = a->element[i] >> shift; \ |
| 1645 | } \ | |
| 1646 | } | |
| 818692ff TM |
1647 | VSR(ab, s8, 0x7) |
| 1648 | VSR(ah, s16, 0xF) | |
| 1649 | VSR(aw, s32, 0x1F) | |
| 2fdf78e6 | 1650 | VSR(ad, s64, 0x3F) |
| 818692ff TM |
1651 | VSR(b, u8, 0x7) |
| 1652 | VSR(h, u16, 0xF) | |
| 1653 | VSR(w, u32, 0x1F) | |
| 2fdf78e6 | 1654 | VSR(d, u64, 0x3F) |
| 64654ded BS |
1655 | #undef VSR |
| 1656 | ||
| 1657 | void helper_vsro(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1658 | { | |
| 1659 | int sh = (b->u8[LO_IDX * 0xf] >> 3) & 0xf; | |
| 1660 | ||
| 1661 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1662 | memmove(&r->u8[sh], &a->u8[0], 16 - sh); | |
| 1663 | memset(&r->u8[0], 0, sh); | |
| 1664 | #else | |
| 1665 | memmove(&r->u8[0], &a->u8[sh], 16 - sh); | |
| 1666 | memset(&r->u8[16 - sh], 0, sh); | |
| 1667 | #endif | |
| 1668 | } | |
| 1669 | ||
| 1670 | void helper_vsubcuw(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1671 | { | |
| 1672 | int i; | |
| 1673 | ||
| 1674 | for (i = 0; i < ARRAY_SIZE(r->u32); i++) { | |
| 1675 | r->u32[i] = a->u32[i] >= b->u32[i]; | |
| 1676 | } | |
| 1677 | } | |
| 1678 | ||
| d15f74fb | 1679 | void helper_vsumsws(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) |
| 64654ded BS |
1680 | { |
| 1681 | int64_t t; | |
| 1682 | int i, upper; | |
| 1683 | ppc_avr_t result; | |
| 1684 | int sat = 0; | |
| 1685 | ||
| 1686 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1687 | upper = ARRAY_SIZE(r->s32)-1; | |
| 1688 | #else | |
| 1689 | upper = 0; | |
| 1690 | #endif | |
| 1691 | t = (int64_t)b->s32[upper]; | |
| 1692 | for (i = 0; i < ARRAY_SIZE(r->s32); i++) { | |
| 1693 | t += a->s32[i]; | |
| 1694 | result.s32[i] = 0; | |
| 1695 | } | |
| 1696 | result.s32[upper] = cvtsdsw(t, &sat); | |
| 1697 | *r = result; | |
| 1698 | ||
| 1699 | if (sat) { | |
| 1700 | env->vscr |= (1 << VSCR_SAT); | |
| 1701 | } | |
| 1702 | } | |
| 1703 | ||
| d15f74fb | 1704 | void helper_vsum2sws(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) |
| 64654ded BS |
1705 | { |
| 1706 | int i, j, upper; | |
| 1707 | ppc_avr_t result; | |
| 1708 | int sat = 0; | |
| 1709 | ||
| 1710 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1711 | upper = 1; | |
| 1712 | #else | |
| 1713 | upper = 0; | |
| 1714 | #endif | |
| 1715 | for (i = 0; i < ARRAY_SIZE(r->u64); i++) { | |
| 1716 | int64_t t = (int64_t)b->s32[upper + i * 2]; | |
| 1717 | ||
| 1718 | result.u64[i] = 0; | |
| 1719 | for (j = 0; j < ARRAY_SIZE(r->u64); j++) { | |
| 1720 | t += a->s32[2 * i + j]; | |
| 1721 | } | |
| 1722 | result.s32[upper + i * 2] = cvtsdsw(t, &sat); | |
| 1723 | } | |
| 1724 | ||
| 1725 | *r = result; | |
| 1726 | if (sat) { | |
| 1727 | env->vscr |= (1 << VSCR_SAT); | |
| 1728 | } | |
| 1729 | } | |
| 1730 | ||
| d15f74fb | 1731 | void helper_vsum4sbs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) |
| 64654ded BS |
1732 | { |
| 1733 | int i, j; | |
| 1734 | int sat = 0; | |
| 1735 | ||
| 1736 | for (i = 0; i < ARRAY_SIZE(r->s32); i++) { | |
| 1737 | int64_t t = (int64_t)b->s32[i]; | |
| 1738 | ||
| 1739 | for (j = 0; j < ARRAY_SIZE(r->s32); j++) { | |
| 1740 | t += a->s8[4 * i + j]; | |
| 1741 | } | |
| 1742 | r->s32[i] = cvtsdsw(t, &sat); | |
| 1743 | } | |
| 1744 | ||
| 1745 | if (sat) { | |
| 1746 | env->vscr |= (1 << VSCR_SAT); | |
| 1747 | } | |
| 1748 | } | |
| 1749 | ||
| d15f74fb | 1750 | void helper_vsum4shs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) |
| 64654ded BS |
1751 | { |
| 1752 | int sat = 0; | |
| 1753 | int i; | |
| 1754 | ||
| 1755 | for (i = 0; i < ARRAY_SIZE(r->s32); i++) { | |
| 1756 | int64_t t = (int64_t)b->s32[i]; | |
| 1757 | ||
| 1758 | t += a->s16[2 * i] + a->s16[2 * i + 1]; | |
| 1759 | r->s32[i] = cvtsdsw(t, &sat); | |
| 1760 | } | |
| 1761 | ||
| 1762 | if (sat) { | |
| 1763 | env->vscr |= (1 << VSCR_SAT); | |
| 1764 | } | |
| 1765 | } | |
| 1766 | ||
| d15f74fb | 1767 | void helper_vsum4ubs(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) |
| 64654ded BS |
1768 | { |
| 1769 | int i, j; | |
| 1770 | int sat = 0; | |
| 1771 | ||
| 1772 | for (i = 0; i < ARRAY_SIZE(r->u32); i++) { | |
| 1773 | uint64_t t = (uint64_t)b->u32[i]; | |
| 1774 | ||
| 1775 | for (j = 0; j < ARRAY_SIZE(r->u32); j++) { | |
| 1776 | t += a->u8[4 * i + j]; | |
| 1777 | } | |
| 1778 | r->u32[i] = cvtuduw(t, &sat); | |
| 1779 | } | |
| 1780 | ||
| 1781 | if (sat) { | |
| 1782 | env->vscr |= (1 << VSCR_SAT); | |
| 1783 | } | |
| 1784 | } | |
| 1785 | ||
| 1786 | #if defined(HOST_WORDS_BIGENDIAN) | |
| 1787 | #define UPKHI 1 | |
| 1788 | #define UPKLO 0 | |
| 1789 | #else | |
| 1790 | #define UPKHI 0 | |
| 1791 | #define UPKLO 1 | |
| 1792 | #endif | |
| 1793 | #define VUPKPX(suffix, hi) \ | |
| 1794 | void helper_vupk##suffix(ppc_avr_t *r, ppc_avr_t *b) \ | |
| 1795 | { \ | |
| 1796 | int i; \ | |
| 1797 | ppc_avr_t result; \ | |
| 1798 | \ | |
| 1799 | for (i = 0; i < ARRAY_SIZE(r->u32); i++) { \ | |
| 1800 | uint16_t e = b->u16[hi ? i : i+4]; \ | |
| 1801 | uint8_t a = (e >> 15) ? 0xff : 0; \ | |
| 1802 | uint8_t r = (e >> 10) & 0x1f; \ | |
| 1803 | uint8_t g = (e >> 5) & 0x1f; \ | |
| 1804 | uint8_t b = e & 0x1f; \ | |
| 1805 | \ | |
| 1806 | result.u32[i] = (a << 24) | (r << 16) | (g << 8) | b; \ | |
| 1807 | } \ | |
| 1808 | *r = result; \ | |
| 1809 | } | |
| 1810 | VUPKPX(lpx, UPKLO) | |
| 1811 | VUPKPX(hpx, UPKHI) | |
| 1812 | #undef VUPKPX | |
| 1813 | ||
| 1814 | #define VUPK(suffix, unpacked, packee, hi) \ | |
| 1815 | void helper_vupk##suffix(ppc_avr_t *r, ppc_avr_t *b) \ | |
| 1816 | { \ | |
| 1817 | int i; \ | |
| 1818 | ppc_avr_t result; \ | |
| 1819 | \ | |
| 1820 | if (hi) { \ | |
| 1821 | for (i = 0; i < ARRAY_SIZE(r->unpacked); i++) { \ | |
| 1822 | result.unpacked[i] = b->packee[i]; \ | |
| 1823 | } \ | |
| 1824 | } else { \ | |
| 1825 | for (i = ARRAY_SIZE(r->unpacked); i < ARRAY_SIZE(r->packee); \ | |
| 1826 | i++) { \ | |
| 1827 | result.unpacked[i - ARRAY_SIZE(r->unpacked)] = b->packee[i]; \ | |
| 1828 | } \ | |
| 1829 | } \ | |
| 1830 | *r = result; \ | |
| 1831 | } | |
| 1832 | VUPK(hsb, s16, s8, UPKHI) | |
| 1833 | VUPK(hsh, s32, s16, UPKHI) | |
| 4430e076 | 1834 | VUPK(hsw, s64, s32, UPKHI) |
| 64654ded BS |
1835 | VUPK(lsb, s16, s8, UPKLO) |
| 1836 | VUPK(lsh, s32, s16, UPKLO) | |
| 4430e076 | 1837 | VUPK(lsw, s64, s32, UPKLO) |
| 64654ded BS |
1838 | #undef VUPK |
| 1839 | #undef UPKHI | |
| 1840 | #undef UPKLO | |
| 1841 | ||
| f293f04a TM |
1842 | #define VGENERIC_DO(name, element) \ |
| 1843 | void helper_v##name(ppc_avr_t *r, ppc_avr_t *b) \ | |
| 1844 | { \ | |
| 1845 | int i; \ | |
| 1846 | \ | |
| 1847 | VECTOR_FOR_INORDER_I(i, element) { \ | |
| 1848 | r->element[i] = name(b->element[i]); \ | |
| 1849 | } \ | |
| 1850 | } | |
| 1851 | ||
| 1852 | #define clzb(v) ((v) ? clz32((uint32_t)(v) << 24) : 8) | |
| 1853 | #define clzh(v) ((v) ? clz32((uint32_t)(v) << 16) : 16) | |
| 1854 | #define clzw(v) clz32((v)) | |
| 1855 | #define clzd(v) clz64((v)) | |
| 1856 | ||
| 1857 | VGENERIC_DO(clzb, u8) | |
| 1858 | VGENERIC_DO(clzh, u16) | |
| 1859 | VGENERIC_DO(clzw, u32) | |
| 1860 | VGENERIC_DO(clzd, u64) | |
| 1861 | ||
| 1862 | #undef clzb | |
| 1863 | #undef clzh | |
| 1864 | #undef clzw | |
| 1865 | #undef clzd | |
| 1866 | ||
| e13500b3 TM |
1867 | #define popcntb(v) ctpop8(v) |
| 1868 | #define popcnth(v) ctpop16(v) | |
| 1869 | #define popcntw(v) ctpop32(v) | |
| 1870 | #define popcntd(v) ctpop64(v) | |
| 1871 | ||
| 1872 | VGENERIC_DO(popcntb, u8) | |
| 1873 | VGENERIC_DO(popcnth, u16) | |
| 1874 | VGENERIC_DO(popcntw, u32) | |
| 1875 | VGENERIC_DO(popcntd, u64) | |
| 1876 | ||
| 1877 | #undef popcntb | |
| 1878 | #undef popcnth | |
| 1879 | #undef popcntw | |
| 1880 | #undef popcntd | |
| f293f04a TM |
1881 | |
| 1882 | #undef VGENERIC_DO | |
| 1883 | ||
| b41da4eb TM |
1884 | #if defined(HOST_WORDS_BIGENDIAN) |
| 1885 | #define QW_ONE { .u64 = { 0, 1 } } | |
| 1886 | #else | |
| 1887 | #define QW_ONE { .u64 = { 1, 0 } } | |
| 1888 | #endif | |
| 1889 | ||
| 1890 | #ifndef CONFIG_INT128 | |
| 1891 | ||
| 1892 | static inline void avr_qw_not(ppc_avr_t *t, ppc_avr_t a) | |
| 1893 | { | |
| 1894 | t->u64[0] = ~a.u64[0]; | |
| 1895 | t->u64[1] = ~a.u64[1]; | |
| 1896 | } | |
| 1897 | ||
| 1898 | static int avr_qw_cmpu(ppc_avr_t a, ppc_avr_t b) | |
| 1899 | { | |
| 1900 | if (a.u64[HI_IDX] < b.u64[HI_IDX]) { | |
| 1901 | return -1; | |
| 1902 | } else if (a.u64[HI_IDX] > b.u64[HI_IDX]) { | |
| 1903 | return 1; | |
| 1904 | } else if (a.u64[LO_IDX] < b.u64[LO_IDX]) { | |
| 1905 | return -1; | |
| 1906 | } else if (a.u64[LO_IDX] > b.u64[LO_IDX]) { | |
| 1907 | return 1; | |
| 1908 | } else { | |
| 1909 | return 0; | |
| 1910 | } | |
| 1911 | } | |
| 1912 | ||
| 1913 | static void avr_qw_add(ppc_avr_t *t, ppc_avr_t a, ppc_avr_t b) | |
| 1914 | { | |
| 1915 | t->u64[LO_IDX] = a.u64[LO_IDX] + b.u64[LO_IDX]; | |
| 1916 | t->u64[HI_IDX] = a.u64[HI_IDX] + b.u64[HI_IDX] + | |
| 1917 | (~a.u64[LO_IDX] < b.u64[LO_IDX]); | |
| 1918 | } | |
| 1919 | ||
| 1920 | static int avr_qw_addc(ppc_avr_t *t, ppc_avr_t a, ppc_avr_t b) | |
| 1921 | { | |
| 1922 | ppc_avr_t not_a; | |
| 1923 | t->u64[LO_IDX] = a.u64[LO_IDX] + b.u64[LO_IDX]; | |
| 1924 | t->u64[HI_IDX] = a.u64[HI_IDX] + b.u64[HI_IDX] + | |
| 1925 | (~a.u64[LO_IDX] < b.u64[LO_IDX]); | |
| 1926 | avr_qw_not(¬_a, a); | |
| 1927 | return avr_qw_cmpu(not_a, b) < 0; | |
| 1928 | } | |
| 1929 | ||
| 1930 | #endif | |
| 1931 | ||
| 1932 | void helper_vadduqm(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1933 | { | |
| 1934 | #ifdef CONFIG_INT128 | |
| 1935 | r->u128 = a->u128 + b->u128; | |
| 1936 | #else | |
| 1937 | avr_qw_add(r, *a, *b); | |
| 1938 | #endif | |
| 1939 | } | |
| 1940 | ||
| 1941 | void helper_vaddeuqm(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c) | |
| 1942 | { | |
| 1943 | #ifdef CONFIG_INT128 | |
| 1944 | r->u128 = a->u128 + b->u128 + (c->u128 & 1); | |
| 1945 | #else | |
| 1946 | ||
| 1947 | if (c->u64[LO_IDX] & 1) { | |
| 1948 | ppc_avr_t tmp; | |
| 1949 | ||
| 1950 | tmp.u64[HI_IDX] = 0; | |
| 1951 | tmp.u64[LO_IDX] = c->u64[LO_IDX] & 1; | |
| 1952 | avr_qw_add(&tmp, *a, tmp); | |
| 1953 | avr_qw_add(r, tmp, *b); | |
| 1954 | } else { | |
| 1955 | avr_qw_add(r, *a, *b); | |
| 1956 | } | |
| 1957 | #endif | |
| 1958 | } | |
| 1959 | ||
| 1960 | void helper_vaddcuq(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 1961 | { | |
| 1962 | #ifdef CONFIG_INT128 | |
| 1963 | r->u128 = (~a->u128 < b->u128); | |
| 1964 | #else | |
| 1965 | ppc_avr_t not_a; | |
| 1966 | ||
| 1967 | avr_qw_not(¬_a, *a); | |
| 1968 | ||
| 1969 | r->u64[HI_IDX] = 0; | |
| 1970 | r->u64[LO_IDX] = (avr_qw_cmpu(not_a, *b) < 0); | |
| 1971 | #endif | |
| 1972 | } | |
| 1973 | ||
| 1974 | void helper_vaddecuq(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c) | |
| 1975 | { | |
| 1976 | #ifdef CONFIG_INT128 | |
| 1977 | int carry_out = (~a->u128 < b->u128); | |
| 1978 | if (!carry_out && (c->u128 & 1)) { | |
| 1979 | carry_out = ((a->u128 + b->u128 + 1) == 0) && | |
| 1980 | ((a->u128 != 0) || (b->u128 != 0)); | |
| 1981 | } | |
| 1982 | r->u128 = carry_out; | |
| 1983 | #else | |
| 1984 | ||
| 1985 | int carry_in = c->u64[LO_IDX] & 1; | |
| 1986 | int carry_out = 0; | |
| 1987 | ppc_avr_t tmp; | |
| 1988 | ||
| 1989 | carry_out = avr_qw_addc(&tmp, *a, *b); | |
| 1990 | ||
| 1991 | if (!carry_out && carry_in) { | |
| 1992 | ppc_avr_t one = QW_ONE; | |
| 1993 | carry_out = avr_qw_addc(&tmp, tmp, one); | |
| 1994 | } | |
| 1995 | r->u64[HI_IDX] = 0; | |
| 1996 | r->u64[LO_IDX] = carry_out; | |
| 1997 | #endif | |
| 1998 | } | |
| 1999 | ||
| 2000 | void helper_vsubuqm(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 2001 | { | |
| 2002 | #ifdef CONFIG_INT128 | |
| 2003 | r->u128 = a->u128 - b->u128; | |
| 2004 | #else | |
| 2005 | ppc_avr_t tmp; | |
| 2006 | ppc_avr_t one = QW_ONE; | |
| 2007 | ||
| 2008 | avr_qw_not(&tmp, *b); | |
| 2009 | avr_qw_add(&tmp, *a, tmp); | |
| 2010 | avr_qw_add(r, tmp, one); | |
| 2011 | #endif | |
| 2012 | } | |
| 2013 | ||
| 2014 | void helper_vsubeuqm(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c) | |
| 2015 | { | |
| 2016 | #ifdef CONFIG_INT128 | |
| 2017 | r->u128 = a->u128 + ~b->u128 + (c->u128 & 1); | |
| 2018 | #else | |
| 2019 | ppc_avr_t tmp, sum; | |
| 2020 | ||
| 2021 | avr_qw_not(&tmp, *b); | |
| 2022 | avr_qw_add(&sum, *a, tmp); | |
| 2023 | ||
| 2024 | tmp.u64[HI_IDX] = 0; | |
| 2025 | tmp.u64[LO_IDX] = c->u64[LO_IDX] & 1; | |
| 2026 | avr_qw_add(r, sum, tmp); | |
| 2027 | #endif | |
| 2028 | } | |
| 2029 | ||
| 2030 | void helper_vsubcuq(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) | |
| 2031 | { | |
| 2032 | #ifdef CONFIG_INT128 | |
| 2033 | r->u128 = (~a->u128 < ~b->u128) || | |
| 2034 | (a->u128 + ~b->u128 == (__uint128_t)-1); | |
| 2035 | #else | |
| 2036 | int carry = (avr_qw_cmpu(*a, *b) > 0); | |
| 2037 | if (!carry) { | |
| 2038 | ppc_avr_t tmp; | |
| 2039 | avr_qw_not(&tmp, *b); | |
| 2040 | avr_qw_add(&tmp, *a, tmp); | |
| 2041 | carry = ((tmp.s64[HI_IDX] == -1ull) && (tmp.s64[LO_IDX] == -1ull)); | |
| 2042 | } | |
| 2043 | r->u64[HI_IDX] = 0; | |
| 2044 | r->u64[LO_IDX] = carry; | |
| 2045 | #endif | |
| 2046 | } | |
| 2047 | ||
| 2048 | void helper_vsubecuq(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c) | |
| 2049 | { | |
| 2050 | #ifdef CONFIG_INT128 | |
| 2051 | r->u128 = | |
| 2052 | (~a->u128 < ~b->u128) || | |
| 2053 | ((c->u128 & 1) && (a->u128 + ~b->u128 == (__uint128_t)-1)); | |
| 2054 | #else | |
| 2055 | int carry_in = c->u64[LO_IDX] & 1; | |
| 2056 | int carry_out = (avr_qw_cmpu(*a, *b) > 0); | |
| 2057 | if (!carry_out && carry_in) { | |
| 2058 | ppc_avr_t tmp; | |
| 2059 | avr_qw_not(&tmp, *b); | |
| 2060 | avr_qw_add(&tmp, *a, tmp); | |
| 2061 | carry_out = ((tmp.u64[HI_IDX] == -1ull) && (tmp.u64[LO_IDX] == -1ull)); | |
| 2062 | } | |
| 2063 | ||
| 2064 | r->u64[HI_IDX] = 0; | |
| 2065 | r->u64[LO_IDX] = carry_out; | |
| 2066 | #endif | |
| 2067 | } | |
| 2068 | ||
| f293f04a | 2069 | |
| 64654ded BS |
2070 | #undef VECTOR_FOR_INORDER_I |
| 2071 | #undef HI_IDX | |
| 2072 | #undef LO_IDX | |
| 2073 | ||
| 2074 | /*****************************************************************************/ | |
| 2075 | /* SPE extension helpers */ | |
| 2076 | /* Use a table to make this quicker */ | |
| ea6c0dac | 2077 | static const uint8_t hbrev[16] = { |
| 64654ded BS |
2078 | 0x0, 0x8, 0x4, 0xC, 0x2, 0xA, 0x6, 0xE, |
| 2079 | 0x1, 0x9, 0x5, 0xD, 0x3, 0xB, 0x7, 0xF, | |
| 2080 | }; | |
| 2081 | ||
| 2082 | static inline uint8_t byte_reverse(uint8_t val) | |
| 2083 | { | |
| 2084 | return hbrev[val >> 4] | (hbrev[val & 0xF] << 4); | |
| 2085 | } | |
| 2086 | ||
| 2087 | static inline uint32_t word_reverse(uint32_t val) | |
| 2088 | { | |
| 2089 | return byte_reverse(val >> 24) | (byte_reverse(val >> 16) << 8) | | |
| 2090 | (byte_reverse(val >> 8) << 16) | (byte_reverse(val) << 24); | |
| 2091 | } | |
| 2092 | ||
| 2093 | #define MASKBITS 16 /* Random value - to be fixed (implementation dependent) */ | |
| 2094 | target_ulong helper_brinc(target_ulong arg1, target_ulong arg2) | |
| 2095 | { | |
| 2096 | uint32_t a, b, d, mask; | |
| 2097 | ||
| 2098 | mask = UINT32_MAX >> (32 - MASKBITS); | |
| 2099 | a = arg1 & mask; | |
| 2100 | b = arg2 & mask; | |
| 2101 | d = word_reverse(1 + word_reverse(a | ~b)); | |
| 2102 | return (arg1 & ~mask) | (d & b); | |
| 2103 | } | |
| 2104 | ||
| 2105 | uint32_t helper_cntlsw32(uint32_t val) | |
| 2106 | { | |
| 2107 | if (val & 0x80000000) { | |
| 2108 | return clz32(~val); | |
| 2109 | } else { | |
| 2110 | return clz32(val); | |
| 2111 | } | |
| 2112 | } | |
| 2113 | ||
| 2114 | uint32_t helper_cntlzw32(uint32_t val) | |
| 2115 | { | |
| 2116 | return clz32(val); | |
| 2117 | } | |
| 2118 | ||
| 2119 | /* 440 specific */ | |
| d15f74fb BS |
2120 | target_ulong helper_dlmzb(CPUPPCState *env, target_ulong high, |
| 2121 | target_ulong low, uint32_t update_Rc) | |
| 64654ded BS |
2122 | { |
| 2123 | target_ulong mask; | |
| 2124 | int i; | |
| 2125 | ||
| 2126 | i = 1; | |
| 2127 | for (mask = 0xFF000000; mask != 0; mask = mask >> 8) { | |
| 2128 | if ((high & mask) == 0) { | |
| 2129 | if (update_Rc) { | |
| 2130 | env->crf[0] = 0x4; | |
| 2131 | } | |
| 2132 | goto done; | |
| 2133 | } | |
| 2134 | i++; | |
| 2135 | } | |
| 2136 | for (mask = 0xFF000000; mask != 0; mask = mask >> 8) { | |
| 2137 | if ((low & mask) == 0) { | |
| 2138 | if (update_Rc) { | |
| 2139 | env->crf[0] = 0x8; | |
| 2140 | } | |
| 2141 | goto done; | |
| 2142 | } | |
| 2143 | i++; | |
| 2144 | } | |
| 2145 | if (update_Rc) { | |
| 2146 | env->crf[0] = 0x2; | |
| 2147 | } | |
| 2148 | done: | |
| 2149 | env->xer = (env->xer & ~0x7F) | i; | |
| 2150 | if (update_Rc) { | |
| 2151 | env->crf[0] |= xer_so; | |
| 2152 | } | |
| 2153 | return i; | |
| 2154 | } |