sm3-neon-core.S 21.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * sm3-neon-core.S - SM3 secure hash using NEON instructions
 *
 * Linux/arm64 port of the libgcrypt SM3 implementation for AArch64
 *
 * Copyright (C) 2021 Jussi Kivilinna <jussi.kivilinna@iki.fi>
 * Copyright (c) 2022 Tianjia Zhang <tianjia.zhang@linux.alibaba.com>
 */

#include <linux/linkage.h>
12
#include <linux/cfi_types.h>
13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
#include <asm/assembler.h>

/* Context structure */

#define state_h0 0
#define state_h1 4
#define state_h2 8
#define state_h3 12
#define state_h4 16
#define state_h5 20
#define state_h6 24
#define state_h7 28

/* Stack structure */

#define STACK_W_SIZE        (32 * 2 * 3)

#define STACK_W             (0)
#define STACK_SIZE          (STACK_W + STACK_W_SIZE)

/* Register macros */

#define RSTATE x0
#define RDATA  x1
#define RNBLKS x2
#define RKPTR  x28
#define RFRAME x29

#define ra w3
#define rb w4
#define rc w5
#define rd w6
#define re w7
#define rf w8
#define rg w9
#define rh w10

#define t0 w11
#define t1 w12
#define t2 w13
#define t3 w14
#define t4 w15
#define t5 w16
#define t6 w17

#define k_even w19
#define k_odd w20

#define addr0 x21
#define addr1 x22

#define s0 w23
#define s1 w24
#define s2 w25
#define s3 w26

#define W0 v0
#define W1 v1
#define W2 v2
#define W3 v3
#define W4 v4
#define W5 v5

#define XTMP0 v6
#define XTMP1 v7
#define XTMP2 v16
#define XTMP3 v17
#define XTMP4 v18
#define XTMP5 v19
#define XTMP6 v20

/* Helper macros. */

#define _(...) /*_*/

#define clear_vec(x) \
	movi	x.8h, #0;

#define rolw(o, a, n) \
	ror	o, a, #(32 - n);

/* Round function macros. */

#define GG1_1(x, y, z, o, t) \
	eor	o, x, y;
#define GG1_2(x, y, z, o, t) \
	eor	o, o, z;
#define GG1_3(x, y, z, o, t)

#define FF1_1(x, y, z, o, t) GG1_1(x, y, z, o, t)
#define FF1_2(x, y, z, o, t)
#define FF1_3(x, y, z, o, t) GG1_2(x, y, z, o, t)

#define GG2_1(x, y, z, o, t) \
	bic	o, z, x;
#define GG2_2(x, y, z, o, t) \
	and	t, y, x;
#define GG2_3(x, y, z, o, t) \
	eor	o, o, t;

#define FF2_1(x, y, z, o, t) \
	eor	o, x, y;
#define FF2_2(x, y, z, o, t) \
	and	t, x, y; \
	and	o, o, z;
#define FF2_3(x, y, z, o, t) \
	eor	o, o, t;

#define R(i, a, b, c, d, e, f, g, h, k, K_LOAD, round, widx, wtype, IOP, iop_param) \
	K_LOAD(round);                                                        \
	ldr	t5, [sp, #(wtype##_W1_ADDR(round, widx))];                    \
	rolw(t0, a, 12);                              /* rol(a, 12) => t0 */  \
      IOP(1, iop_param);                                                      \
	FF##i##_1(a, b, c, t1, t2);                                           \
	ldr	t6, [sp, #(wtype##_W1W2_ADDR(round, widx))];                  \
	add	k, k, e;                                                      \
      IOP(2, iop_param);                                                      \
	GG##i##_1(e, f, g, t3, t4);                                           \
	FF##i##_2(a, b, c, t1, t2);                                           \
      IOP(3, iop_param);                                                      \
	add	k, k, t0;                                                     \
	add	h, h, t5;                                                     \
	add	d, d, t6;                     /* w1w2 + d => d */             \
      IOP(4, iop_param);                                                      \
	rolw(k, k, 7);                        /* rol (t0 + e + t), 7) => k */ \
	GG##i##_2(e, f, g, t3, t4);                                           \
	add	h, h, k;                      /* h + w1 + k => h */           \
      IOP(5, iop_param);                                                      \
	FF##i##_3(a, b, c, t1, t2);                                           \
	eor	t0, t0, k;                    /* k ^ t0 => t0 */              \
	GG##i##_3(e, f, g, t3, t4);                                           \
	add	d, d, t1;                     /* FF(a,b,c) + d => d */        \
      IOP(6, iop_param);                                                      \
	add	t3, t3, h;                    /* GG(e,f,g) + h => t3 */       \
	rolw(b, b, 9);                        /* rol(b, 9) => b */            \
	eor	h, t3, t3, ror #(32-9);                                       \
      IOP(7, iop_param);                                                      \
	add	d, d, t0;                     /* t0 + d => d */               \
	rolw(f, f, 19);                       /* rol(f, 19) => f */           \
      IOP(8, iop_param);                                                      \
	eor	h, h, t3, ror #(32-17);       /* P0(t3) => h */

#define R1(a, b, c, d, e, f, g, h, k, K_LOAD, round, widx, wtype, IOP, iop_param) \
	R(1, ##a, ##b, ##c, ##d, ##e, ##f, ##g, ##h, ##k, K_LOAD, round, widx, wtype, IOP, iop_param)

#define R2(a, b, c, d, e, f, g, h, k, K_LOAD, round, widx, wtype, IOP, iop_param) \
	R(2, ##a, ##b, ##c, ##d, ##e, ##f, ##g, ##h, ##k, K_LOAD, round, widx, wtype, IOP, iop_param)

#define KL(round) \
	ldp	k_even, k_odd, [RKPTR, #(4*(round))];

/* Input expansion macros. */

/* Byte-swapped input address. */
#define IW_W_ADDR(round, widx, offs) \
	(STACK_W + ((round) / 4) * 64 + (offs) + ((widx) * 4))

/* Expanded input address. */
#define XW_W_ADDR(round, widx, offs) \
	(STACK_W + ((((round) / 3) - 4) % 2) * 64 + (offs) + ((widx) * 4))

/* Rounds 1-12, byte-swapped input block addresses. */
#define IW_W1_ADDR(round, widx)   IW_W_ADDR(round, widx, 32)
#define IW_W1W2_ADDR(round, widx) IW_W_ADDR(round, widx, 48)

/* Rounds 1-12, expanded input block addresses. */
#define XW_W1_ADDR(round, widx)   XW_W_ADDR(round, widx, 0)
#define XW_W1W2_ADDR(round, widx) XW_W_ADDR(round, widx, 16)

/* Input block loading.
 * Interleaving within round function needed for in-order CPUs. */
#define LOAD_W_VEC_1_1() \
	add	addr0, sp, #IW_W1_ADDR(0, 0);
#define LOAD_W_VEC_1_2() \
	add	addr1, sp, #IW_W1_ADDR(4, 0);
#define LOAD_W_VEC_1_3() \
	ld1	{W0.16b}, [RDATA], #16;
#define LOAD_W_VEC_1_4() \
	ld1	{W1.16b}, [RDATA], #16;
#define LOAD_W_VEC_1_5() \
	ld1	{W2.16b}, [RDATA], #16;
#define LOAD_W_VEC_1_6() \
	ld1	{W3.16b}, [RDATA], #16;
#define LOAD_W_VEC_1_7() \
	rev32	XTMP0.16b, W0.16b;
#define LOAD_W_VEC_1_8() \
	rev32	XTMP1.16b, W1.16b;
#define LOAD_W_VEC_2_1() \
	rev32	XTMP2.16b, W2.16b;
#define LOAD_W_VEC_2_2() \
	rev32	XTMP3.16b, W3.16b;
#define LOAD_W_VEC_2_3() \
	eor	XTMP4.16b, XTMP1.16b, XTMP0.16b;
#define LOAD_W_VEC_2_4() \
	eor	XTMP5.16b, XTMP2.16b, XTMP1.16b;
#define LOAD_W_VEC_2_5() \
	st1	{XTMP0.16b}, [addr0], #16;
#define LOAD_W_VEC_2_6() \
	st1	{XTMP4.16b}, [addr0]; \
	add	addr0, sp, #IW_W1_ADDR(8, 0);
#define LOAD_W_VEC_2_7() \
	eor	XTMP6.16b, XTMP3.16b, XTMP2.16b;
#define LOAD_W_VEC_2_8() \
	ext	W0.16b, XTMP0.16b, XTMP0.16b, #8;  /* W0: xx, w0, xx, xx */
#define LOAD_W_VEC_3_1() \
	mov	W2.16b, XTMP1.16b;                 /* W2: xx, w6, w5, w4 */
#define LOAD_W_VEC_3_2() \
	st1	{XTMP1.16b}, [addr1], #16;
#define LOAD_W_VEC_3_3() \
	st1	{XTMP5.16b}, [addr1]; \
	ext	W1.16b, XTMP0.16b, XTMP0.16b, #4;  /* W1: xx, w3, w2, w1 */
#define LOAD_W_VEC_3_4() \
	ext	W3.16b, XTMP1.16b, XTMP2.16b, #12; /* W3: xx, w9, w8, w7 */
#define LOAD_W_VEC_3_5() \
	ext	W4.16b, XTMP2.16b, XTMP3.16b, #8;  /* W4: xx, w12, w11, w10 */
#define LOAD_W_VEC_3_6() \
	st1	{XTMP2.16b}, [addr0], #16;
#define LOAD_W_VEC_3_7() \
	st1	{XTMP6.16b}, [addr0];
#define LOAD_W_VEC_3_8() \
	ext	W5.16b, XTMP3.16b, XTMP3.16b, #4;  /* W5: xx, w15, w14, w13 */

#define LOAD_W_VEC_1(iop_num, ...) \
	LOAD_W_VEC_1_##iop_num()
#define LOAD_W_VEC_2(iop_num, ...) \
	LOAD_W_VEC_2_##iop_num()
#define LOAD_W_VEC_3(iop_num, ...) \
	LOAD_W_VEC_3_##iop_num()

/* Message scheduling. Note: 3 words per vector register.
 * Interleaving within round function needed for in-order CPUs. */
#define SCHED_W_1_1(round, w0, w1, w2, w3, w4, w5) \
	/* Load (w[i - 16]) => XTMP0 */            \
	/* Load (w[i - 13]) => XTMP5 */            \
	ext	XTMP0.16b, w0.16b, w0.16b, #12;    /* XTMP0: w0, xx, xx, xx */
#define SCHED_W_1_2(round, w0, w1, w2, w3, w4, w5) \
	ext	XTMP5.16b, w1.16b, w1.16b, #12;
#define SCHED_W_1_3(round, w0, w1, w2, w3, w4, w5) \
	ext	XTMP0.16b, XTMP0.16b, w1.16b, #12; /* XTMP0: xx, w2, w1, w0 */
#define SCHED_W_1_4(round, w0, w1, w2, w3, w4, w5) \
	ext	XTMP5.16b, XTMP5.16b, w2.16b, #12;
#define SCHED_W_1_5(round, w0, w1, w2, w3, w4, w5) \
	/* w[i - 9] == w3 */                       \
	/* W3 ^ XTMP0 => XTMP0 */                  \
	eor	XTMP0.16b, XTMP0.16b, w3.16b;
#define SCHED_W_1_6(round, w0, w1, w2, w3, w4, w5) \
	/* w[i - 3] == w5 */                       \
	/* rol(XMM5, 15) ^ XTMP0 => XTMP0 */       \
	/* rol(XTMP5, 7) => XTMP1 */               \
	add	addr0, sp, #XW_W1_ADDR((round), 0); \
	shl	XTMP2.4s, w5.4s, #15;
#define SCHED_W_1_7(round, w0, w1, w2, w3, w4, w5) \
	shl	XTMP1.4s, XTMP5.4s, #7;
#define SCHED_W_1_8(round, w0, w1, w2, w3, w4, w5) \
	sri	XTMP2.4s, w5.4s, #(32-15);
#define SCHED_W_2_1(round, w0, w1, w2, w3, w4, w5) \
	sri	XTMP1.4s, XTMP5.4s, #(32-7);
#define SCHED_W_2_2(round, w0, w1, w2, w3, w4, w5) \
	eor	XTMP0.16b, XTMP0.16b, XTMP2.16b;
#define SCHED_W_2_3(round, w0, w1, w2, w3, w4, w5) \
	/* w[i - 6] == W4 */                       \
	/* W4 ^ XTMP1 => XTMP1 */                  \
	eor	XTMP1.16b, XTMP1.16b, w4.16b;
#define SCHED_W_2_4(round, w0, w1, w2, w3, w4, w5) \
	/* P1(XTMP0) ^ XTMP1 => W0 */              \
	shl	XTMP3.4s, XTMP0.4s, #15;
#define SCHED_W_2_5(round, w0, w1, w2, w3, w4, w5) \
	shl	XTMP4.4s, XTMP0.4s, #23;
#define SCHED_W_2_6(round, w0, w1, w2, w3, w4, w5) \
	eor	w0.16b, XTMP1.16b, XTMP0.16b;
#define SCHED_W_2_7(round, w0, w1, w2, w3, w4, w5) \
	sri	XTMP3.4s, XTMP0.4s, #(32-15);
#define SCHED_W_2_8(round, w0, w1, w2, w3, w4, w5) \
	sri	XTMP4.4s, XTMP0.4s, #(32-23);
#define SCHED_W_3_1(round, w0, w1, w2, w3, w4, w5) \
	eor	w0.16b, w0.16b, XTMP3.16b;
#define SCHED_W_3_2(round, w0, w1, w2, w3, w4, w5) \
	/* Load (w[i - 3]) => XTMP2 */             \
	ext	XTMP2.16b, w4.16b, w4.16b, #12;
#define SCHED_W_3_3(round, w0, w1, w2, w3, w4, w5) \
	eor	w0.16b, w0.16b, XTMP4.16b;
#define SCHED_W_3_4(round, w0, w1, w2, w3, w4, w5) \
	ext	XTMP2.16b, XTMP2.16b, w5.16b, #12;
#define SCHED_W_3_5(round, w0, w1, w2, w3, w4, w5) \
	/* W1 ^ W2 => XTMP3 */                     \
	eor	XTMP3.16b, XTMP2.16b, w0.16b;
#define SCHED_W_3_6(round, w0, w1, w2, w3, w4, w5)
#define SCHED_W_3_7(round, w0, w1, w2, w3, w4, w5) \
	st1	{XTMP2.16b-XTMP3.16b}, [addr0];
#define SCHED_W_3_8(round, w0, w1, w2, w3, w4, w5)

#define SCHED_W_W0W1W2W3W4W5_1(iop_num, round) \
	SCHED_W_1_##iop_num(round, W0, W1, W2, W3, W4, W5)
#define SCHED_W_W0W1W2W3W4W5_2(iop_num, round) \
	SCHED_W_2_##iop_num(round, W0, W1, W2, W3, W4, W5)
#define SCHED_W_W0W1W2W3W4W5_3(iop_num, round) \
	SCHED_W_3_##iop_num(round, W0, W1, W2, W3, W4, W5)

#define SCHED_W_W1W2W3W4W5W0_1(iop_num, round) \
	SCHED_W_1_##iop_num(round, W1, W2, W3, W4, W5, W0)
#define SCHED_W_W1W2W3W4W5W0_2(iop_num, round) \
	SCHED_W_2_##iop_num(round, W1, W2, W3, W4, W5, W0)
#define SCHED_W_W1W2W3W4W5W0_3(iop_num, round) \
	SCHED_W_3_##iop_num(round, W1, W2, W3, W4, W5, W0)

#define SCHED_W_W2W3W4W5W0W1_1(iop_num, round) \
	SCHED_W_1_##iop_num(round, W2, W3, W4, W5, W0, W1)
#define SCHED_W_W2W3W4W5W0W1_2(iop_num, round) \
	SCHED_W_2_##iop_num(round, W2, W3, W4, W5, W0, W1)
#define SCHED_W_W2W3W4W5W0W1_3(iop_num, round) \
	SCHED_W_3_##iop_num(round, W2, W3, W4, W5, W0, W1)

#define SCHED_W_W3W4W5W0W1W2_1(iop_num, round) \
	SCHED_W_1_##iop_num(round, W3, W4, W5, W0, W1, W2)
#define SCHED_W_W3W4W5W0W1W2_2(iop_num, round) \
	SCHED_W_2_##iop_num(round, W3, W4, W5, W0, W1, W2)
#define SCHED_W_W3W4W5W0W1W2_3(iop_num, round) \
	SCHED_W_3_##iop_num(round, W3, W4, W5, W0, W1, W2)

#define SCHED_W_W4W5W0W1W2W3_1(iop_num, round) \
	SCHED_W_1_##iop_num(round, W4, W5, W0, W1, W2, W3)
#define SCHED_W_W4W5W0W1W2W3_2(iop_num, round) \
	SCHED_W_2_##iop_num(round, W4, W5, W0, W1, W2, W3)
#define SCHED_W_W4W5W0W1W2W3_3(iop_num, round) \
	SCHED_W_3_##iop_num(round, W4, W5, W0, W1, W2, W3)

#define SCHED_W_W5W0W1W2W3W4_1(iop_num, round) \
	SCHED_W_1_##iop_num(round, W5, W0, W1, W2, W3, W4)
#define SCHED_W_W5W0W1W2W3W4_2(iop_num, round) \
	SCHED_W_2_##iop_num(round, W5, W0, W1, W2, W3, W4)
#define SCHED_W_W5W0W1W2W3W4_3(iop_num, round) \
	SCHED_W_3_##iop_num(round, W5, W0, W1, W2, W3, W4)


	/*
	 * Transform blocks*64 bytes (blocks*16 32-bit words) at 'src'.
	 *
	 * void sm3_neon_transform(struct sm3_state *sst, u8 const *src,
	 *                         int blocks)
	 */
	.text
.align 3
355
SYM_TYPED_FUNC_START(sm3_neon_transform)
356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601
	ldp		ra, rb, [RSTATE, #0]
	ldp		rc, rd, [RSTATE, #8]
	ldp		re, rf, [RSTATE, #16]
	ldp		rg, rh, [RSTATE, #24]

	stp		x28, x29, [sp, #-16]!
	stp		x19, x20, [sp, #-16]!
	stp		x21, x22, [sp, #-16]!
	stp		x23, x24, [sp, #-16]!
	stp		x25, x26, [sp, #-16]!
	mov		RFRAME, sp

	sub		addr0, sp, #STACK_SIZE
	adr_l		RKPTR, .LKtable
	and		sp, addr0, #(~63)

	/* Preload first block. */
	LOAD_W_VEC_1(1, 0)
	LOAD_W_VEC_1(2, 0)
	LOAD_W_VEC_1(3, 0)
	LOAD_W_VEC_1(4, 0)
	LOAD_W_VEC_1(5, 0)
	LOAD_W_VEC_1(6, 0)
	LOAD_W_VEC_1(7, 0)
	LOAD_W_VEC_1(8, 0)
	LOAD_W_VEC_2(1, 0)
	LOAD_W_VEC_2(2, 0)
	LOAD_W_VEC_2(3, 0)
	LOAD_W_VEC_2(4, 0)
	LOAD_W_VEC_2(5, 0)
	LOAD_W_VEC_2(6, 0)
	LOAD_W_VEC_2(7, 0)
	LOAD_W_VEC_2(8, 0)
	LOAD_W_VEC_3(1, 0)
	LOAD_W_VEC_3(2, 0)
	LOAD_W_VEC_3(3, 0)
	LOAD_W_VEC_3(4, 0)
	LOAD_W_VEC_3(5, 0)
	LOAD_W_VEC_3(6, 0)
	LOAD_W_VEC_3(7, 0)
	LOAD_W_VEC_3(8, 0)

.balign 16
.Loop:
	/* Transform 0-3 */
	R1(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 0, 0, IW, _, 0)
	R1(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  1, 1, IW, _, 0)
	R1(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 2, 2, IW, _, 0)
	R1(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  3, 3, IW, _, 0)

	/* Transform 4-7 + Precalc 12-14 */
	R1(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 4, 0, IW, _, 0)
	R1(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  5, 1, IW, _, 0)
	R1(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 6, 2, IW, SCHED_W_W0W1W2W3W4W5_1, 12)
	R1(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  7, 3, IW, SCHED_W_W0W1W2W3W4W5_2, 12)

	/* Transform 8-11 + Precalc 12-17 */
	R1(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 8, 0, IW, SCHED_W_W0W1W2W3W4W5_3, 12)
	R1(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  9, 1, IW, SCHED_W_W1W2W3W4W5W0_1, 15)
	R1(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 10, 2, IW, SCHED_W_W1W2W3W4W5W0_2, 15)
	R1(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  11, 3, IW, SCHED_W_W1W2W3W4W5W0_3, 15)

	/* Transform 12-14 + Precalc 18-20 */
	R1(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 12, 0, XW, SCHED_W_W2W3W4W5W0W1_1, 18)
	R1(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  13, 1, XW, SCHED_W_W2W3W4W5W0W1_2, 18)
	R1(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 14, 2, XW, SCHED_W_W2W3W4W5W0W1_3, 18)

	/* Transform 15-17 + Precalc 21-23 */
	R1(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  15, 0, XW, SCHED_W_W3W4W5W0W1W2_1, 21)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 16, 1, XW, SCHED_W_W3W4W5W0W1W2_2, 21)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  17, 2, XW, SCHED_W_W3W4W5W0W1W2_3, 21)

	/* Transform 18-20 + Precalc 24-26 */
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 18, 0, XW, SCHED_W_W4W5W0W1W2W3_1, 24)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  19, 1, XW, SCHED_W_W4W5W0W1W2W3_2, 24)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 20, 2, XW, SCHED_W_W4W5W0W1W2W3_3, 24)

	/* Transform 21-23 + Precalc 27-29 */
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  21, 0, XW, SCHED_W_W5W0W1W2W3W4_1, 27)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 22, 1, XW, SCHED_W_W5W0W1W2W3W4_2, 27)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  23, 2, XW, SCHED_W_W5W0W1W2W3W4_3, 27)

	/* Transform 24-26 + Precalc 30-32 */
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 24, 0, XW, SCHED_W_W0W1W2W3W4W5_1, 30)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  25, 1, XW, SCHED_W_W0W1W2W3W4W5_2, 30)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 26, 2, XW, SCHED_W_W0W1W2W3W4W5_3, 30)

	/* Transform 27-29 + Precalc 33-35 */
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  27, 0, XW, SCHED_W_W1W2W3W4W5W0_1, 33)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 28, 1, XW, SCHED_W_W1W2W3W4W5W0_2, 33)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  29, 2, XW, SCHED_W_W1W2W3W4W5W0_3, 33)

	/* Transform 30-32 + Precalc 36-38 */
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 30, 0, XW, SCHED_W_W2W3W4W5W0W1_1, 36)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  31, 1, XW, SCHED_W_W2W3W4W5W0W1_2, 36)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 32, 2, XW, SCHED_W_W2W3W4W5W0W1_3, 36)

	/* Transform 33-35 + Precalc 39-41 */
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  33, 0, XW, SCHED_W_W3W4W5W0W1W2_1, 39)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 34, 1, XW, SCHED_W_W3W4W5W0W1W2_2, 39)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  35, 2, XW, SCHED_W_W3W4W5W0W1W2_3, 39)

	/* Transform 36-38 + Precalc 42-44 */
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 36, 0, XW, SCHED_W_W4W5W0W1W2W3_1, 42)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  37, 1, XW, SCHED_W_W4W5W0W1W2W3_2, 42)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 38, 2, XW, SCHED_W_W4W5W0W1W2W3_3, 42)

	/* Transform 39-41 + Precalc 45-47 */
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  39, 0, XW, SCHED_W_W5W0W1W2W3W4_1, 45)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 40, 1, XW, SCHED_W_W5W0W1W2W3W4_2, 45)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  41, 2, XW, SCHED_W_W5W0W1W2W3W4_3, 45)

	/* Transform 42-44 + Precalc 48-50 */
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 42, 0, XW, SCHED_W_W0W1W2W3W4W5_1, 48)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  43, 1, XW, SCHED_W_W0W1W2W3W4W5_2, 48)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 44, 2, XW, SCHED_W_W0W1W2W3W4W5_3, 48)

	/* Transform 45-47 + Precalc 51-53 */
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  45, 0, XW, SCHED_W_W1W2W3W4W5W0_1, 51)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 46, 1, XW, SCHED_W_W1W2W3W4W5W0_2, 51)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  47, 2, XW, SCHED_W_W1W2W3W4W5W0_3, 51)

	/* Transform 48-50 + Precalc 54-56 */
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 48, 0, XW, SCHED_W_W2W3W4W5W0W1_1, 54)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  49, 1, XW, SCHED_W_W2W3W4W5W0W1_2, 54)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 50, 2, XW, SCHED_W_W2W3W4W5W0W1_3, 54)

	/* Transform 51-53 + Precalc 57-59 */
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  51, 0, XW, SCHED_W_W3W4W5W0W1W2_1, 57)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 52, 1, XW, SCHED_W_W3W4W5W0W1W2_2, 57)
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  53, 2, XW, SCHED_W_W3W4W5W0W1W2_3, 57)

	/* Transform 54-56 + Precalc 60-62 */
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 54, 0, XW, SCHED_W_W4W5W0W1W2W3_1, 60)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  55, 1, XW, SCHED_W_W4W5W0W1W2W3_2, 60)
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 56, 2, XW, SCHED_W_W4W5W0W1W2W3_3, 60)

	/* Transform 57-59 + Precalc 63 */
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  57, 0, XW, SCHED_W_W5W0W1W2W3W4_1, 63)
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 58, 1, XW, SCHED_W_W5W0W1W2W3W4_2, 63)
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  59, 2, XW, SCHED_W_W5W0W1W2W3W4_3, 63)

	/* Transform 60 */
	R2(ra, rb, rc, rd, re, rf, rg, rh, k_even, KL, 60, 0, XW, _, _)
	subs		RNBLKS, RNBLKS, #1
	b.eq		.Lend

	/* Transform 61-63 + Preload next block */
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  61, 1, XW, LOAD_W_VEC_1, _)
	ldp		s0, s1, [RSTATE, #0]
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 62, 2, XW, LOAD_W_VEC_2, _)
	ldp		s2, s3, [RSTATE, #8]
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  63, 0, XW, LOAD_W_VEC_3, _)

	/* Update the chaining variables. */
	eor		ra, ra, s0
	eor		rb, rb, s1
	ldp		s0, s1, [RSTATE, #16]
	eor		rc, rc, s2
	ldp		k_even, k_odd, [RSTATE, #24]
	eor		rd, rd, s3
	eor		re, re, s0
	stp		ra, rb, [RSTATE, #0]
	eor		rf, rf, s1
	stp		rc, rd, [RSTATE, #8]
	eor		rg, rg, k_even
	stp		re, rf, [RSTATE, #16]
	eor		rh, rh, k_odd
	stp		rg, rh, [RSTATE, #24]
	b		.Loop

.Lend:
	/* Transform 61-63 */
	R2(rd, ra, rb, rc, rh, re, rf, rg, k_odd,  _,  61, 1, XW, _, _)
	ldp		s0, s1, [RSTATE, #0]
	R2(rc, rd, ra, rb, rg, rh, re, rf, k_even, KL, 62, 2, XW, _, _)
	ldp		s2, s3, [RSTATE, #8]
	R2(rb, rc, rd, ra, rf, rg, rh, re, k_odd,  _,  63, 0, XW, _, _)

	/* Update the chaining variables. */
	eor		ra, ra, s0
	clear_vec(W0)
	eor		rb, rb, s1
	clear_vec(W1)
	ldp		s0, s1, [RSTATE, #16]
	clear_vec(W2)
	eor		rc, rc, s2
	clear_vec(W3)
	ldp		k_even, k_odd, [RSTATE, #24]
	clear_vec(W4)
	eor		rd, rd, s3
	clear_vec(W5)
	eor		re, re, s0
	clear_vec(XTMP0)
	stp		ra, rb, [RSTATE, #0]
	clear_vec(XTMP1)
	eor		rf, rf, s1
	clear_vec(XTMP2)
	stp		rc, rd, [RSTATE, #8]
	clear_vec(XTMP3)
	eor		rg, rg, k_even
	clear_vec(XTMP4)
	stp		re, rf, [RSTATE, #16]
	clear_vec(XTMP5)
	eor		rh, rh, k_odd
	clear_vec(XTMP6)
	stp		rg, rh, [RSTATE, #24]

	/* Clear message expansion area */
	add		addr0, sp, #STACK_W
	st1		{W0.16b-W3.16b}, [addr0], #64
	st1		{W0.16b-W3.16b}, [addr0], #64
	st1		{W0.16b-W3.16b}, [addr0]

	mov		sp, RFRAME

	ldp		x25, x26, [sp], #16
	ldp		x23, x24, [sp], #16
	ldp		x21, x22, [sp], #16
	ldp		x19, x20, [sp], #16
	ldp		x28, x29, [sp], #16

	ret
SYM_FUNC_END(sm3_neon_transform)


	.section	".rodata", "a"

	.align 4
.LKtable:
	.long 0x79cc4519, 0xf3988a32, 0xe7311465, 0xce6228cb
	.long 0x9cc45197, 0x3988a32f, 0x7311465e, 0xe6228cbc
	.long 0xcc451979, 0x988a32f3, 0x311465e7, 0x6228cbce
	.long 0xc451979c, 0x88a32f39, 0x11465e73, 0x228cbce6
	.long 0x9d8a7a87, 0x3b14f50f, 0x7629ea1e, 0xec53d43c
	.long 0xd8a7a879, 0xb14f50f3, 0x629ea1e7, 0xc53d43ce
	.long 0x8a7a879d, 0x14f50f3b, 0x29ea1e76, 0x53d43cec
	.long 0xa7a879d8, 0x4f50f3b1, 0x9ea1e762, 0x3d43cec5
	.long 0x7a879d8a, 0xf50f3b14, 0xea1e7629, 0xd43cec53
	.long 0xa879d8a7, 0x50f3b14f, 0xa1e7629e, 0x43cec53d
	.long 0x879d8a7a, 0x0f3b14f5, 0x1e7629ea, 0x3cec53d4
	.long 0x79d8a7a8, 0xf3b14f50, 0xe7629ea1, 0xcec53d43
	.long 0x9d8a7a87, 0x3b14f50f, 0x7629ea1e, 0xec53d43c
	.long 0xd8a7a879, 0xb14f50f3, 0x629ea1e7, 0xc53d43ce
	.long 0x8a7a879d, 0x14f50f3b, 0x29ea1e76, 0x53d43cec
	.long 0xa7a879d8, 0x4f50f3b1, 0x9ea1e762, 0x3d43cec5