arch/s390/include/asm/vx-insn.h

   1 /*
   2  * Support for Vector Instructions
   3  *
   4  * Assembler macros to generate .byte/.word code for particular
   5  * vector instructions that are supported by recent binutils (>= 2.26) only.
   6  *
   7  * Copyright IBM Corp. 2015
   8  * Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
   9  */
  10
  11 #ifndef __ASM_S390_VX_INSN_H
  12 #define __ASM_S390_VX_INSN_H
  13
  14 #ifdef __ASSEMBLY__
  15
  16
  17 /* Macros to generate vector instruction byte code */
  18
  19 /* GR_NUM - Retrieve general-purpose register number
  20  *
  21  * @opd:        Operand to store register number
  22  * @r64:        String designation register in the format "%rN"
  23  */
  24 .macro  GR_NUM  opd gr
  25         \opd = 255
  26         .ifc \gr,%r0
  27                 \opd = 0
  28         .endif
  29         .ifc \gr,%r1
  30                 \opd = 1
  31         .endif
  32         .ifc \gr,%r2
  33                 \opd = 2
  34         .endif
  35         .ifc \gr,%r3
  36                 \opd = 3
  37         .endif
  38         .ifc \gr,%r4
  39                 \opd = 4
  40         .endif
  41         .ifc \gr,%r5
  42                 \opd = 5
  43         .endif
  44         .ifc \gr,%r6
  45                 \opd = 6
  46         .endif
  47         .ifc \gr,%r7
  48                 \opd = 7
  49         .endif
  50         .ifc \gr,%r8
  51                 \opd = 8
  52         .endif
  53         .ifc \gr,%r9
  54                 \opd = 9
  55         .endif
  56         .ifc \gr,%r10
  57                 \opd = 10
  58         .endif
  59         .ifc \gr,%r11
  60                 \opd = 11
  61         .endif
  62         .ifc \gr,%r12
  63                 \opd = 12
  64         .endif
  65         .ifc \gr,%r13
  66                 \opd = 13
  67         .endif
  68         .ifc \gr,%r14
  69                 \opd = 14
  70         .endif
  71         .ifc \gr,%r15
  72                 \opd = 15
  73         .endif
  74         .if \opd == 255
  75                 \opd = \gr
  76         .endif
  77 .endm
  78
  79 /* VX_NUM - Retrieve vector register number
  80  *
  81  * @opd:        Operand to store register number
  82  * @vxr:        String designation register in the format "%vN"
  83  *
  84  * The vector register number is used for as input number to the
  85  * instruction and, as well as, to compute the RXB field of the
  86  * instruction.
  87  */
  88 .macro  VX_NUM  opd vxr
  89         \opd = 255
  90         .ifc \vxr,%v0
  91                 \opd = 0
  92         .endif
  93         .ifc \vxr,%v1
  94                 \opd = 1
  95         .endif
  96         .ifc \vxr,%v2
  97                 \opd = 2
  98         .endif
  99         .ifc \vxr,%v3
 100                 \opd = 3
 101         .endif
 102         .ifc \vxr,%v4
 103                 \opd = 4
 104         .endif
 105         .ifc \vxr,%v5
 106                 \opd = 5
 107         .endif
 108         .ifc \vxr,%v6
 109                 \opd = 6
 110         .endif
 111         .ifc \vxr,%v7
 112                 \opd = 7
 113         .endif
 114         .ifc \vxr,%v8
 115                 \opd = 8
 116         .endif
 117         .ifc \vxr,%v9
 118                 \opd = 9
 119         .endif
 120         .ifc \vxr,%v10
 121                 \opd = 10
 122         .endif
 123         .ifc \vxr,%v11
 124                 \opd = 11
 125         .endif
 126         .ifc \vxr,%v12
 127                 \opd = 12
 128         .endif
 129         .ifc \vxr,%v13
 130                 \opd = 13
 131         .endif
 132         .ifc \vxr,%v14
 133                 \opd = 14
 134         .endif
 135         .ifc \vxr,%v15
 136                 \opd = 15
 137         .endif
 138         .ifc \vxr,%v16
 139                 \opd = 16
 140         .endif
 141         .ifc \vxr,%v17
 142                 \opd = 17
 143         .endif
 144         .ifc \vxr,%v18
 145                 \opd = 18
 146         .endif
 147         .ifc \vxr,%v19
 148                 \opd = 19
 149         .endif
 150         .ifc \vxr,%v20
 151                 \opd = 20
 152         .endif
 153         .ifc \vxr,%v21
 154                 \opd = 21
 155         .endif
 156         .ifc \vxr,%v22
 157                 \opd = 22
 158         .endif
 159         .ifc \vxr,%v23
 160                 \opd = 23
 161         .endif
 162         .ifc \vxr,%v24
 163                 \opd = 24
 164         .endif
 165         .ifc \vxr,%v25
 166                 \opd = 25
 167         .endif
 168         .ifc \vxr,%v26
 169                 \opd = 26
 170         .endif
 171         .ifc \vxr,%v27
 172                 \opd = 27
 173         .endif
 174         .ifc \vxr,%v28
 175                 \opd = 28
 176         .endif
 177         .ifc \vxr,%v29
 178                 \opd = 29
 179         .endif
 180         .ifc \vxr,%v30
 181                 \opd = 30
 182         .endif
 183         .ifc \vxr,%v31
 184                 \opd = 31
 185         .endif
 186         .if \opd == 255
 187                 \opd = \vxr
 188         .endif
 189 .endm
 190
 191 /* RXB - Compute most significant bit used vector registers
 192  *
 193  * @rxb:        Operand to store computed RXB value
 194  * @v1:         First vector register designated operand
 195  * @v2:         Second vector register designated operand
 196  * @v3:         Third vector register designated operand
 197  * @v4:         Fourth vector register designated operand
 198  */
 199 .macro  RXB     rxb v1 v2=0 v3=0 v4=0
 200         \rxb = 0
 201         .if \v1 & 0x10
 202                 \rxb = \rxb | 0x08
 203         .endif
 204         .if \v2 & 0x10
 205                 \rxb = \rxb | 0x04
 206         .endif
 207         .if \v3 & 0x10
 208                 \rxb = \rxb | 0x02
 209         .endif
 210         .if \v4 & 0x10
 211                 \rxb = \rxb | 0x01
 212         .endif
 213 .endm
 214
 215 /* MRXB - Generate Element Size Control and RXB value
 216  *
 217  * @m:          Element size control
 218  * @v1:         First vector register designated operand (for RXB)
 219  * @v2:         Second vector register designated operand (for RXB)
 220  * @v3:         Third vector register designated operand (for RXB)
 221  * @v4:         Fourth vector register designated operand (for RXB)
 222  */
 223 .macro  MRXB    m v1 v2=0 v3=0 v4=0
 224         rxb = 0
 225         RXB     rxb, \v1, \v2, \v3, \v4
 226         .byte   (\m << 4) | rxb
 227 .endm
 228
 229 /* MRXBOPC - Generate Element Size Control, RXB, and final Opcode fields
 230  *
 231  * @m:          Element size control
 232  * @opc:        Opcode
 233  * @v1:         First vector register designated operand (for RXB)
 234  * @v2:         Second vector register designated operand (for RXB)
 235  * @v3:         Third vector register designated operand (for RXB)
 236  * @v4:         Fourth vector register designated operand (for RXB)
 237  */
 238 .macro  MRXBOPC m opc v1 v2=0 v3=0 v4=0
 239         MRXB    \m, \v1, \v2, \v3, \v4
 240         .byte   \opc
 241 .endm
 242
 243 /* Vector support instructions */
 244
 245 /* VECTOR GENERATE BYTE MASK */
 246 .macro  VGBM    vr imm2
 247         VX_NUM  v1, \vr
 248         .word   (0xE700 | ((v1&15) << 4))
 249         .word   \imm2
 250         MRXBOPC 0, 0x44, v1
 251 .endm
 252 .macro  VZERO   vxr
 253         VGBM    \vxr, 0
 254 .endm
 255 .macro  VONE    vxr
 256         VGBM    \vxr, 0xFFFF
 257 .endm
 258
 259 /* VECTOR LOAD VR ELEMENT FROM GR */
 260 .macro  VLVG    v, gr, disp, m
 261         VX_NUM  v1, \v
 262         GR_NUM  b2, "%r0"
 263         GR_NUM  r3, \gr
 264         .word   0xE700 | ((v1&15) << 4) | r3
 265         .word   (b2 << 12) | (\disp)
 266         MRXBOPC \m, 0x22, v1
 267 .endm
 268 .macro  VLVGB   v, gr, index, base
 269         VLVG    \v, \gr, \index, \base, 0
 270 .endm
 271 .macro  VLVGH   v, gr, index
 272         VLVG    \v, \gr, \index, 1
 273 .endm
 274 .macro  VLVGF   v, gr, index
 275         VLVG    \v, \gr, \index, 2
 276 .endm
 277 .macro  VLVGG   v, gr, index
 278         VLVG    \v, \gr, \index, 3
 279 .endm
 280
 281 /* VECTOR LOAD REGISTER */
 282 .macro  VLR     v1, v2
 283         VX_NUM  v1, \v1
 284         VX_NUM  v2, \v2
 285         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 286         .word   0
 287         MRXBOPC 0, 0x56, v1, v2
 288 .endm
 289
 290 /* VECTOR LOAD */
 291 .macro  VL      v, disp, index="%r0", base
 292         VX_NUM  v1, \v
 293         GR_NUM  x2, \index
 294         GR_NUM  b2, \base
 295         .word   0xE700 | ((v1&15) << 4) | x2
 296         .word   (b2 << 12) | (\disp)
 297         MRXBOPC 0, 0x06, v1
 298 .endm
 299
 300 /* VECTOR LOAD ELEMENT */
 301 .macro  VLEx    vr1, disp, index="%r0", base, m3, opc
 302         VX_NUM  v1, \vr1
 303         GR_NUM  x2, \index
 304         GR_NUM  b2, \base
 305         .word   0xE700 | ((v1&15) << 4) | x2
 306         .word   (b2 << 12) | (\disp)
 307         MRXBOPC \m3, \opc, v1
 308 .endm
 309 .macro  VLEB    vr1, disp, index="%r0", base, m3
 310         VLEx    \vr1, \disp, \index, \base, \m3, 0x00
 311 .endm
 312 .macro  VLEH    vr1, disp, index="%r0", base, m3
 313         VLEx    \vr1, \disp, \index, \base, \m3, 0x01
 314 .endm
 315 .macro  VLEF    vr1, disp, index="%r0", base, m3
 316         VLEx    \vr1, \disp, \index, \base, \m3, 0x03
 317 .endm
 318 .macro  VLEG    vr1, disp, index="%r0", base, m3
 319         VLEx    \vr1, \disp, \index, \base, \m3, 0x02
 320 .endm
 321
 322 /* VECTOR LOAD ELEMENT IMMEDIATE */
 323 .macro  VLEIx   vr1, imm2, m3, opc
 324         VX_NUM  v1, \vr1
 325         .word   0xE700 | ((v1&15) << 4)
 326         .word   \imm2
 327         MRXBOPC \m3, \opc, v1
 328 .endm
 329 .macro  VLEIB   vr1, imm2, index
 330         VLEIx   \vr1, \imm2, \index, 0x40
 331 .endm
 332 .macro  VLEIH   vr1, imm2, index
 333         VLEIx   \vr1, \imm2, \index, 0x41
 334 .endm
 335 .macro  VLEIF   vr1, imm2, index
 336         VLEIx   \vr1, \imm2, \index, 0x43
 337 .endm
 338 .macro  VLEIG   vr1, imm2, index
 339         VLEIx   \vr1, \imm2, \index, 0x42
 340 .endm
 341
 342 /* VECTOR LOAD GR FROM VR ELEMENT */
 343 .macro  VLGV    gr, vr, disp, base="%r0", m
 344         GR_NUM  r1, \gr
 345         GR_NUM  b2, \base
 346         VX_NUM  v3, \vr
 347         .word   0xE700 | (r1 << 4) | (v3&15)
 348         .word   (b2 << 12) | (\disp)
 349         MRXBOPC \m, 0x21, v3
 350 .endm
 351 .macro  VLGVB   gr, vr, disp, base="%r0"
 352         VLGV    \gr, \vr, \disp, \base, 0
 353 .endm
 354 .macro  VLGVH   gr, vr, disp, base="%r0"
 355         VLGV    \gr, \vr, \disp, \base, 1
 356 .endm
 357 .macro  VLGVF   gr, vr, disp, base="%r0"
 358         VLGV    \gr, \vr, \disp, \base, 2
 359 .endm
 360 .macro  VLGVG   gr, vr, disp, base="%r0"
 361         VLGV    \gr, \vr, \disp, \base, 3
 362 .endm
 363
 364 /* VECTOR LOAD MULTIPLE */
 365 .macro  VLM     vfrom, vto, disp, base
 366         VX_NUM  v1, \vfrom
 367         VX_NUM  v3, \vto
 368         GR_NUM  b2, \base           /* Base register */
 369         .word   0xE700 | ((v1&15) << 4) | (v3&15)
 370         .word   (b2 << 12) | (\disp)
 371         MRXBOPC 0, 0x36, v1, v3
 372 .endm
 373
 374 /* VECTOR STORE MULTIPLE */
 375 .macro  VSTM    vfrom, vto, disp, base
 376         VX_NUM  v1, \vfrom
 377         VX_NUM  v3, \vto
 378         GR_NUM  b2, \base           /* Base register */
 379         .word   0xE700 | ((v1&15) << 4) | (v3&15)
 380         .word   (b2 << 12) | (\disp)
 381         MRXBOPC 0, 0x3E, v1, v3
 382 .endm
 383
 384 /* VECTOR PERMUTE */
 385 .macro  VPERM   vr1, vr2, vr3, vr4
 386         VX_NUM  v1, \vr1
 387         VX_NUM  v2, \vr2
 388         VX_NUM  v3, \vr3
 389         VX_NUM  v4, \vr4
 390         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 391         .word   ((v3&15) << 12)
 392         MRXBOPC (v4&15), 0x8C, v1, v2, v3, v4
 393 .endm
 394
 395 /* VECTOR UNPACK LOGICAL LOW */
 396 .macro  VUPLL   vr1, vr2, m3
 397         VX_NUM  v1, \vr1
 398         VX_NUM  v2, \vr2
 399         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 400         .word   0x0000
 401         MRXBOPC \m3, 0xD4, v1, v2
 402 .endm
 403 .macro  VUPLLB  vr1, vr2
 404         VUPLL   \vr1, \vr2, 0
 405 .endm
 406 .macro  VUPLLH  vr1, vr2
 407         VUPLL   \vr1, \vr2, 1
 408 .endm
 409 .macro  VUPLLF  vr1, vr2
 410         VUPLL   \vr1, \vr2, 2
 411 .endm
 412
 413
 414 /* Vector integer instructions */
 415
 416 /* VECTOR AND */
 417 .macro  VN      vr1, vr2, vr3
 418         VX_NUM  v1, \vr1
 419         VX_NUM  v2, \vr2
 420         VX_NUM  v3, \vr3
 421         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 422         .word   ((v3&15) << 12)
 423         MRXBOPC 0, 0x68, v1, v2, v3
 424 .endm
 425
 426 /* VECTOR EXCLUSIVE OR */
 427 .macro  VX      vr1, vr2, vr3
 428         VX_NUM  v1, \vr1
 429         VX_NUM  v2, \vr2
 430         VX_NUM  v3, \vr3
 431         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 432         .word   ((v3&15) << 12)
 433         MRXBOPC 0, 0x6D, v1, v2, v3
 434 .endm
 435
 436 /* VECTOR GALOIS FIELD MULTIPLY SUM */
 437 .macro  VGFM    vr1, vr2, vr3, m4
 438         VX_NUM  v1, \vr1
 439         VX_NUM  v2, \vr2
 440         VX_NUM  v3, \vr3
 441         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 442         .word   ((v3&15) << 12)
 443         MRXBOPC \m4, 0xB4, v1, v2, v3
 444 .endm
 445 .macro  VGFMB   vr1, vr2, vr3
 446         VGFM    \vr1, \vr2, \vr3, 0
 447 .endm
 448 .macro  VGFMH   vr1, vr2, vr3
 449         VGFM    \vr1, \vr2, \vr3, 1
 450 .endm
 451 .macro  VGFMF   vr1, vr2, vr3
 452         VGFM    \vr1, \vr2, \vr3, 2
 453 .endm
 454 .macro  VGFMG   vr1, vr2, vr3
 455         VGFM    \vr1, \vr2, \vr3, 3
 456 .endm
 457
 458 /* VECTOR GALOIS FIELD MULTIPLY SUM AND ACCUMULATE */
 459 .macro  VGFMA   vr1, vr2, vr3, vr4, m5
 460         VX_NUM  v1, \vr1
 461         VX_NUM  v2, \vr2
 462         VX_NUM  v3, \vr3
 463         VX_NUM  v4, \vr4
 464         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 465         .word   ((v3&15) << 12) | (\m5 << 8)
 466         MRXBOPC (v4&15), 0xBC, v1, v2, v3, v4
 467 .endm
 468 .macro  VGFMAB  vr1, vr2, vr3, vr4
 469         VGFMA   \vr1, \vr2, \vr3, \vr4, 0
 470 .endm
 471 .macro  VGFMAH  vr1, vr2, vr3, vr4
 472         VGFMA   \vr1, \vr2, \vr3, \vr4, 1
 473 .endm
 474 .macro  VGFMAF  vr1, vr2, vr3, vr4
 475         VGFMA   \vr1, \vr2, \vr3, \vr4, 2
 476 .endm
 477 .macro  VGFMAG  vr1, vr2, vr3, vr4
 478         VGFMA   \vr1, \vr2, \vr3, \vr4, 3
 479 .endm
 480
 481 /* VECTOR SHIFT RIGHT LOGICAL BY BYTE */
 482 .macro  VSRLB   vr1, vr2, vr3
 483         VX_NUM  v1, \vr1
 484         VX_NUM  v2, \vr2
 485         VX_NUM  v3, \vr3
 486         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 487         .word   ((v3&15) << 12)
 488         MRXBOPC 0, 0x7D, v1, v2, v3
 489 .endm
 490
 491 /* VECTOR REPLICATE IMMEDIATE */
 492 .macro  VREPI   vr1, imm2, m3
 493         VX_NUM  v1, \vr1
 494         .word   0xE700 | ((v1&15) << 4)
 495         .word   \imm2
 496         MRXBOPC \m3, 0x45, v1
 497 .endm
 498 .macro  VREPIB  vr1, imm2
 499         VREPI   \vr1, \imm2, 0
 500 .endm
 501 .macro  VREPIH  vr1, imm2
 502         VREPI   \vr1, \imm2, 1
 503 .endm
 504 .macro  VREPIF  vr1, imm2
 505         VREPI   \vr1, \imm2, 2
 506 .endm
 507 .macro  VREPIG  vr1, imm2
 508         VREP    \vr1, \imm2, 3
 509 .endm
 510
 511 /* VECTOR ADD */
 512 .macro  VA      vr1, vr2, vr3, m4
 513         VX_NUM  v1, \vr1
 514         VX_NUM  v2, \vr2
 515         VX_NUM  v3, \vr3
 516         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 517         .word   ((v3&15) << 12)
 518         MRXBOPC \m4, 0xF3, v1, v2, v3
 519 .endm
 520 .macro  VAB     vr1, vr2, vr3
 521         VA      \vr1, \vr2, \vr3, 0
 522 .endm
 523 .macro  VAH     vr1, vr2, vr3
 524         VA      \vr1, \vr2, \vr3, 1
 525 .endm
 526 .macro  VAF     vr1, vr2, vr3
 527         VA      \vr1, \vr2, \vr3, 2
 528 .endm
 529 .macro  VAG     vr1, vr2, vr3
 530         VA      \vr1, \vr2, \vr3, 3
 531 .endm
 532 .macro  VAQ     vr1, vr2, vr3
 533         VA      \vr1, \vr2, \vr3, 4
 534 .endm
 535
 536 /* VECTOR ELEMENT SHIFT RIGHT ARITHMETIC */
 537 .macro  VESRAV  vr1, vr2, vr3, m4
 538         VX_NUM  v1, \vr1
 539         VX_NUM  v2, \vr2
 540         VX_NUM  v3, \vr3
 541         .word   0xE700 | ((v1&15) << 4) | (v2&15)
 542         .word   ((v3&15) << 12)
 543         MRXBOPC \m4, 0x7A, v1, v2, v3
 544 .endm
 545
 546 .macro  VESRAVB vr1, vr2, vr3
 547         VESRAV  \vr1, \vr2, \vr3, 0
 548 .endm
 549 .macro  VESRAVH vr1, vr2, vr3
 550         VESRAV  \vr1, \vr2, \vr3, 1
 551 .endm
 552 .macro  VESRAVF vr1, vr2, vr3
 553         VESRAV  \vr1, \vr2, \vr3, 2
 554 .endm
 555 .macro  VESRAVG vr1, vr2, vr3
 556         VESRAV  \vr1, \vr2, \vr3, 3
 557 .endm
 558
 559 #endif  /* __ASSEMBLY__ */
 560 #endif  /* __ASM_S390_VX_INSN_H */