aes_neon_32.S revision 1.1
1 /* $NetBSD: aes_neon_32.S,v 1.1 2020/06/29 23:57:56 riastradh Exp $ */ 2 3 /*- 4 * Copyright (c) 2020 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 17 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 18 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 19 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 20 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 21 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 22 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 23 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 24 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 25 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 26 * POSSIBILITY OF SUCH DAMAGE. 27 */ 28 29 #include <arm/asm.h> 30 31 .fpu neon 32 33 .section .rodata 34 .p2align 4 35 36 .type inv,_ASM_TYPE_OBJECT 37 inv: 38 .byte 0x80,0x01,0x08,0x0D,0x0F,0x06,0x05,0x0E 39 .byte 0x02,0x0C,0x0B,0x0A,0x09,0x03,0x07,0x04 40 END(inv) 41 42 .type inva,_ASM_TYPE_OBJECT 43 inva: 44 .byte 0x80,0x07,0x0B,0x0F,0x06,0x0A,0x04,0x01 45 .byte 0x09,0x08,0x05,0x02,0x0C,0x0E,0x0D,0x03 46 END(inva) 47 48 .type mc_forward,_ASM_TYPE_OBJECT 49 mc_forward: 50 .byte 0x01,0x02,0x03,0x00,0x05,0x06,0x07,0x04 /* 0 */ 51 .byte 0x09,0x0A,0x0B,0x08,0x0D,0x0E,0x0F,0x0C 52 53 .byte 0x05,0x06,0x07,0x04,0x09,0x0A,0x0B,0x08 /* 1 */ 54 .byte 0x0D,0x0E,0x0F,0x0C,0x01,0x02,0x03,0x00 55 56 .byte 0x09,0x0A,0x0B,0x08,0x0D,0x0E,0x0F,0x0C /* 2 */ 57 .byte 0x01,0x02,0x03,0x00,0x05,0x06,0x07,0x04 58 59 .Lmc_forward_3: 60 .byte 0x0D,0x0E,0x0F,0x0C,0x01,0x02,0x03,0x00 /* 3 */ 61 .byte 0x05,0x06,0x07,0x04,0x09,0x0A,0x0B,0x08 62 END(mc_forward) 63 64 .type mc_backward,_ASM_TYPE_OBJECT 65 mc_backward: 66 .byte 0x03,0x00,0x01,0x02,0x07,0x04,0x05,0x06 /* 0 */ 67 .byte 0x0B,0x08,0x09,0x0A,0x0F,0x0C,0x0D,0x0E 68 69 .byte 0x0F,0x0C,0x0D,0x0E,0x03,0x00,0x01,0x02 /* 1 */ 70 .byte 0x07,0x04,0x05,0x06,0x0B,0x08,0x09,0x0A 71 72 .byte 0x0B,0x08,0x09,0x0A,0x0F,0x0C,0x0D,0x0E /* 2 */ 73 .byte 0x03,0x00,0x01,0x02,0x07,0x04,0x05,0x06 74 75 .byte 0x07,0x04,0x05,0x06,0x0B,0x08,0x09,0x0A /* 3 */ 76 .byte 0x0F,0x0C,0x0D,0x0E,0x03,0x00,0x01,0x02 77 END(mc_backward) 78 79 .type sr,_ASM_TYPE_OBJECT 80 sr: 81 .byte 0x00,0x01,0x02,0x03,0x04,0x05,0x06,0x07 /* 0 */ 82 .byte 0x08,0x09,0x0A,0x0B,0x0C,0x0D,0x0E,0x0F 83 84 .byte 0x00,0x05,0x0A,0x0F,0x04,0x09,0x0E,0x03 /* 1 */ 85 .byte 0x08,0x0D,0x02,0x07,0x0C,0x01,0x06,0x0B 86 87 .byte 0x00,0x09,0x02,0x0B,0x04,0x0D,0x06,0x0F /* 2 */ 88 .byte 0x08,0x01,0x0A,0x03,0x0C,0x05,0x0E,0x07 89 90 .byte 0x00,0x0D,0x0A,0x07,0x04,0x01,0x0E,0x0B /* 3 */ 91 .byte 0x08,0x05,0x02,0x0F,0x0C,0x09,0x06,0x03 92 END(sr) 93 94 .type iptlo,_ASM_TYPE_OBJECT 95 iptlo: 96 .byte 0x00,0x70,0x2A,0x5A,0x98,0xE8,0xB2,0xC2 97 .byte 0x08,0x78,0x22,0x52,0x90,0xE0,0xBA,0xCA 98 END(iptlo) 99 100 .type ipthi,_ASM_TYPE_OBJECT 101 ipthi: 102 .byte 0x00,0x4D,0x7C,0x31,0x7D,0x30,0x01,0x4C 103 .byte 0x81,0xCC,0xFD,0xB0,0xFC,0xB1,0x80,0xCD 104 END(ipthi) 105 106 .type sb1_0,_ASM_TYPE_OBJECT 107 sb1_0: 108 .byte 0x00,0x3E,0x50,0xCB,0x8F,0xE1,0x9B,0xB1 109 .byte 0x44,0xF5,0x2A,0x14,0x6E,0x7A,0xDF,0xA5 110 END(sb1_0) 111 112 .type sb1_1,_ASM_TYPE_OBJECT 113 sb1_1: 114 .byte 0x00,0x23,0xE2,0xFA,0x15,0xD4,0x18,0x36 115 .byte 0xEF,0xD9,0x2E,0x0D,0xC1,0xCC,0xF7,0x3B 116 END(sb1_1) 117 118 .type sb2_0,_ASM_TYPE_OBJECT 119 sb2_0: 120 .byte 0x00,0x24,0x71,0x0B,0xC6,0x93,0x7A,0xE2 121 .byte 0xCD,0x2F,0x98,0xBC,0x55,0xE9,0xB7,0x5E 122 END(sb2_0) 123 124 .type sb2_1,_ASM_TYPE_OBJECT 125 sb2_1: 126 .byte 0x00,0x29,0xE1,0x0A,0x40,0x88,0xEB,0x69 127 .byte 0x4A,0x23,0x82,0xAB,0xC8,0x63,0xA1,0xC2 128 END(sb2_1) 129 130 .type sbo_0,_ASM_TYPE_OBJECT 131 sbo_0: 132 .byte 0x00,0xC7,0xBD,0x6F,0x17,0x6D,0xD2,0xD0 133 .byte 0x78,0xA8,0x02,0xC5,0x7A,0xBF,0xAA,0x15 134 END(sbo_0) 135 136 .type sbo_1,_ASM_TYPE_OBJECT 137 sbo_1: 138 .byte 0x00,0x6A,0xBB,0x5F,0xA5,0x74,0xE4,0xCF 139 .byte 0xFA,0x35,0x2B,0x41,0xD1,0x90,0x1E,0x8E 140 END(sbo_1) 141 142 .type diptlo,_ASM_TYPE_OBJECT 143 diptlo: 144 .byte 0x00,0x5F,0x54,0x0B,0x04,0x5B,0x50,0x0F 145 .byte 0x1A,0x45,0x4E,0x11,0x1E,0x41,0x4A,0x15 146 END(diptlo) 147 148 .type dipthi,_ASM_TYPE_OBJECT 149 dipthi: 150 .byte 0x00,0x65,0x05,0x60,0xE6,0x83,0xE3,0x86 151 .byte 0x94,0xF1,0x91,0xF4,0x72,0x17,0x77,0x12 152 END(dipthi) 153 154 .type dsb9_0,_ASM_TYPE_OBJECT 155 dsb9_0: 156 .byte 0x00,0xD6,0x86,0x9A,0x53,0x03,0x1C,0x85 157 .byte 0xC9,0x4C,0x99,0x4F,0x50,0x1F,0xD5,0xCA 158 END(dsb9_0) 159 160 .type dsb9_1,_ASM_TYPE_OBJECT 161 dsb9_1: 162 .byte 0x00,0x49,0xD7,0xEC,0x89,0x17,0x3B,0xC0 163 .byte 0x65,0xA5,0xFB,0xB2,0x9E,0x2C,0x5E,0x72 164 END(dsb9_1) 165 166 .type dsbd_0,_ASM_TYPE_OBJECT 167 dsbd_0: 168 .byte 0x00,0xA2,0xB1,0xE6,0xDF,0xCC,0x57,0x7D 169 .byte 0x39,0x44,0x2A,0x88,0x13,0x9B,0x6E,0xF5 170 END(dsbd_0) 171 172 .type dsbd_1,_ASM_TYPE_OBJECT 173 dsbd_1: 174 .byte 0x00,0xCB,0xC6,0x24,0xF7,0xFA,0xE2,0x3C 175 .byte 0xD3,0xEF,0xDE,0x15,0x0D,0x18,0x31,0x29 176 END(dsbd_1) 177 178 .type dsbb_0,_ASM_TYPE_OBJECT 179 dsbb_0: 180 .byte 0x00,0x42,0xB4,0x96,0x92,0x64,0x22,0xD0 181 .byte 0x04,0xD4,0xF2,0xB0,0xF6,0x46,0x26,0x60 182 END(dsbb_0) 183 184 .type dsbb_1,_ASM_TYPE_OBJECT 185 dsbb_1: 186 .byte 0x00,0x67,0x59,0xCD,0xA6,0x98,0x94,0xC1 187 .byte 0x6B,0xAA,0x55,0x32,0x3E,0x0C,0xFF,0xF3 188 END(dsbb_1) 189 190 .type dsbe_0,_ASM_TYPE_OBJECT 191 dsbe_0: 192 .byte 0x00,0xD0,0xD4,0x26,0x96,0x92,0xF2,0x46 193 .byte 0xB0,0xF6,0xB4,0x64,0x04,0x60,0x42,0x22 194 END(dsbe_0) 195 196 .type dsbe_1,_ASM_TYPE_OBJECT 197 dsbe_1: 198 .byte 0x00,0xC1,0xAA,0xFF,0xCD,0xA6,0x55,0x0C 199 .byte 0x32,0x3E,0x59,0x98,0x6B,0xF3,0x67,0x94 200 END(dsbe_1) 201 202 .type dsbo_0,_ASM_TYPE_OBJECT 203 dsbo_0: 204 .byte 0x00,0x40,0xF9,0x7E,0x53,0xEA,0x87,0x13 205 .byte 0x2D,0x3E,0x94,0xD4,0xB9,0x6D,0xAA,0xC7 206 END(dsbo_0) 207 208 .type dsbo_1,_ASM_TYPE_OBJECT 209 dsbo_1: 210 .byte 0x00,0x1D,0x44,0x93,0x0F,0x56,0xD7,0x12 211 .byte 0x9C,0x8E,0xC5,0xD8,0x59,0x81,0x4B,0xCA 212 END(dsbo_1) 213 214 /* 215 * aes_neon_enc1(enc, x, nrounds) 216 * 217 * With -mfloat-abi=hard: 218 * 219 * uint8x16_t@q0 220 * aes_neon_enc1(const struct aesenc *enc@r0, uint8x16_t x@q0, 221 * unsigned nrounds@r1) 222 * 223 * With -mfloat-abi=soft(fp) (here spelled `#ifdef _KERNEL'): 224 * 225 * uint8x16_t@(r0,r1,r2,r3) 226 * aes_neon_enc1(const struct aesenc *enc@r0, 227 * uint8x16_t x@(r2,r3,sp[0],sp[4]), nrounds@sp[8]) 228 */ 229 ENTRY(aes_neon_enc1) 230 #ifdef _KERNEL 231 vmov d0, r2, r3 /* d0 := x lo */ 232 vldr d1, [sp] /* d1 := x hi */ 233 ldr r1, [sp, #8] /* r1 := nrounds */ 234 #endif 235 push {r4, r5, r6, r7, r8, r10, r11, lr} 236 vpush {d8-d15} 237 238 /* 239 * r3: rmod4 240 * r4: mc_forward 241 * r5: mc_backward 242 * r6,r7,r8,r10,r11: temporaries 243 * q0={d0-d1}: x/ak/A 244 * q1={d2-d3}: 0x0f0f... 245 * q2={d4-d5}: lo/k/j/io 246 * q3={d6-d7}: hi/i/jo 247 * q4={d8-d9}: iptlo 248 * q5={d10-d11}: ipthi 249 * q6={d12-d13}: sb1[0]/sbo[0] 250 * q7={d14-d15}: sb1[1]/sbo[1] 251 * q8={d16-d17}: sb2[0] 252 * q9={d18-d19}: sb2[1] 253 * q10={d20-d21}: inv 254 * q11={d22-d23}: inva 255 * q12={d24-d25}: ir/iak/iakr/sb1_0(io)/mc_backward[rmod4] 256 * q13={d26-d27}: jr/jak/jakr/sb1_1(jo)/mc_forward[rmod4] 257 * q14={d28-d29}: rk/A2/A2_B_D 258 * q15={d30-d31}: A2_B/sr[rmod4] 259 */ 260 261 vld1.64 {d28-d29}, [r0 :128]! /* q14 = *rk++ */ 262 movw r3, #0 263 vmov.i8 q1, #0x0f 264 265 /* (q4, q5) := (iptlo, ipthi) */ 266 ldr r6, =iptlo 267 ldr r7, =ipthi 268 vld1.64 {d8-d9}, [r6 :128] 269 vld1.64 {d10-d11}, [r7 :128] 270 271 /* load the rest of the constants */ 272 ldr r4, =sb1_0 273 ldr r5, =sb1_1 274 ldr r6, =sb2_0 275 ldr r7, =sb2_1 276 ldr r8, =inv 277 ldr r10, =inva 278 vld1.64 {d12-d13}, [r4 :128] /* q6 = sb1[0] */ 279 vld1.64 {d14-d15}, [r5 :128] /* q7 = sb1[1] */ 280 vld1.64 {d16-d17}, [r6 :128] /* q8 = sb2[0] */ 281 vld1.64 {d18-d19}, [r7 :128] /* q9 = sb2[1] */ 282 vld1.64 {d20-d21}, [r8 :128] /* q10 = inv */ 283 vld1.64 {d22-d23}, [r10 :128] /* q11 = inva */ 284 285 /* (r4, r5) := (&mc_forward[0], &mc_backward[0]) */ 286 ldr r4, =mc_forward 287 ldr r5, =mc_backward 288 289 /* (q2, q3) := (lo, hi) */ 290 vshr.u8 q3, q0, #4 291 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 292 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 293 294 /* (q2, q3) := (iptlo(lo), ipthi(hi)) */ 295 vtbl.8 d4, {d8-d9}, d4 296 vtbl.8 d5, {d8-d9}, d5 297 vtbl.8 d6, {d10-d11}, d6 298 vtbl.8 d7, {d10-d11}, d7 299 300 /* q0 := rk[0] + iptlo(lo) + ipthi(hi) */ 301 veor q0, q14, q2 302 veor q0, q0, q3 303 304 b 2f 305 306 1: vld1.64 {d28-d29}, [r0 :128]! /* q14 = *rk++ */ 307 308 /* q0 := A = rk[i] + sb1_0(io) + sb1_1(jo) */ 309 vtbl.8 d24, {d12-d13}, d4 310 vtbl.8 d25, {d12-d13}, d5 311 vtbl.8 d26, {d14-d15}, d6 312 vtbl.8 d27, {d14-d15}, d7 313 veor q0, q14, q12 314 veor q0, q0, q13 315 316 /* q14 := A2 = sb2_0[io] + sb2_1[jo] */ 317 vtbl.8 d24, {d16-d17}, d4 318 vtbl.8 d25, {d16-d17}, d5 319 vtbl.8 d26, {d18-d19}, d6 320 vtbl.8 d27, {d18-d19}, d7 321 veor q14, q12, q13 322 323 /* (q12, q13) := (mc_forward[rmod4], mc_backward[rmod4]) */ 324 add r6, r4, r3, lsl #4 325 add r7, r5, r3, lsl #4 326 vld1.64 {d24-d25}, [r6] 327 vld1.64 {d26-d27}, [r7] 328 329 /* q15 := A2_B = A2 + A(mcf) */ 330 vtbl.8 d30, {d0-d1}, d24 331 vtbl.8 d31, {d0-d1}, d25 332 veor q15, q15, q14 333 334 /* q14 := A2_B_D = A2_B + A(mcb) */ 335 vtbl.8 d28, {d0-d1}, d26 336 vtbl.8 d29, {d0-d1}, d27 337 veor q14, q14, q15 338 339 /* q0 := x = A2_B_D + A2_B(mcf) */ 340 vtbl.8 d0, {d30-d31}, d24 341 vtbl.8 d1, {d30-d31}, d25 342 veor q0, q0, q14 343 344 2: /* 345 * SubBytes 346 */ 347 348 /* (q2, q3) := (k, i) */ 349 vshr.u8 q3, q0, #4 350 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 351 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 352 353 /* q0 := a/k */ 354 vtbl.8 d0, {d22-d23}, d4 355 vtbl.8 d1, {d22-d23}, d5 356 357 /* q2 := j = i + k */ 358 veor q2, q3, q2 359 360 /* q12 := ir = 1/i */ 361 vtbl.8 d24, {d20-d21}, d6 362 vtbl.8 d25, {d20-d21}, d7 363 364 /* q13 := jr = 1/j */ 365 vtbl.8 d26, {d20-d21}, d4 366 vtbl.8 d27, {d20-d21}, d5 367 368 /* q12 := iak = 1/i + a/k */ 369 veor q12, q12, q0 370 371 /* q13 := jak = 1/j + a/k */ 372 veor q13, q13, q0 373 374 /* q12 := iakr = 1/(1/i + a/k) */ 375 vtbl.8 d24, {d20-d21}, d24 376 vtbl.8 d25, {d20-d21}, d25 377 378 /* q13 := jakr = 1/(1/j + a/k) */ 379 vtbl.8 d26, {d20-d21}, d26 380 vtbl.8 d27, {d20-d21}, d27 381 382 /* q2 := io = j + 1/(1/i + a/k) */ 383 veor q2, q2, q12 384 385 /* q3 := jo = i + 1/(1/j + a/k) */ 386 veor q3, q3, q13 387 388 /* advance round */ 389 add r3, r3, #1 390 subs r1, r1, #1 391 and r3, r3, #3 392 bne 1b 393 394 /* (q6, q7, q15) := (sbo[0], sbo[1], sr[rmod4]) */ 395 ldr r8, =sr 396 ldr r6, =sbo_0 397 ldr r7, =sbo_1 398 add r8, r8, r3, lsl #4 399 vld1.64 {d12-d13}, [r6 :128] 400 vld1.64 {d14-d15}, [r7 :128] 401 vld1.64 {d30-d31}, [r8 :128] 402 403 vld1.64 {d28-d29}, [r0 :128]! /* q14 = *rk++ */ 404 405 /* (q2, q3) := (sbo_0(io), sbo_1(jo)) */ 406 vtbl.8 d4, {d12-d13}, d4 407 vtbl.8 d5, {d12-d13}, d5 408 vtbl.8 d6, {d14-d15}, d6 409 vtbl.8 d7, {d14-d15}, d7 410 411 /* q2 := x = rk[nr] + sbo_0(io) + sbo_1(jo) */ 412 veor q2, q2, q14 413 veor q2, q2, q3 414 415 /* q0 := x(sr[rmod4]) */ 416 vtbl.8 d0, {d4-d5}, d30 417 vtbl.8 d1, {d4-d5}, d31 418 419 vpop {d8-d15} 420 pop {r4, r5, r6, r7, r8, r10, r11, lr} 421 #ifdef _KERNEL 422 vmov r0, r1, d0 423 vmov r2, r3, d1 424 #endif 425 bx lr 426 END(aes_neon_enc1) 427 428 /* 429 * aes_neon_dec1(dec, x, nrounds) 430 * 431 * With -mfloat-abi=hard: 432 * 433 * uint8x16_t@q0 434 * aes_neon_dec1(const struct aesdec *dec@r0, uint8x16_t x@q0, 435 * unsigned nrounds@r1) 436 * 437 * With -mfloat-abi=soft(fp) (here spelled `#ifdef _KERNEL'): 438 * 439 * uint8x16_t@(r0,r1,r2,r3) 440 * aes_neon_dec1(const struct aesdec *dec@r0, 441 * uint8x16_t x@(r2,r3,sp[0],sp[4]), nrounds@sp[8]) 442 */ 443 ENTRY(aes_neon_dec1) 444 #ifdef _KERNEL 445 vmov d0, r2, r3 /* d0 := x lo */ 446 vldr d1, [sp] /* d1 := x hi */ 447 ldr r1, [sp, #8] /* r1 := nrounds */ 448 #endif 449 push {r4, r5, r6, r7, r8, r10, r11, lr} 450 vpush {d8-d15} 451 452 /* 453 * r3: 3 & ~(nrounds - 1) 454 * q0={d0-d1}: x/ak 455 * q1={d2-d3}: 0x0f0f... 456 * q2={d4-d5}: lo/k/j/io 457 * q3={d6-d7}: hi/i/jo 458 * q4={d8-d9}: diptlo/dsb9[0] 459 * q5={d10-d11}: dipthi/dsb9[1] 460 * q6={d12-d13}: dsbb[0]/dsbo[0] 461 * q7={d14-d15}: dsbb[1]/dsbo[1] 462 * q8={d16-d17}: dsbd[0]/dsbe[0] 463 * q9={d18-d19}: dsbd[1]/dsbe[0] 464 * q10={d20-d21}: inv 465 * q11={d22-d23}: inva 466 * q12={d24-d25}: ir/iak/iakr/dsbX_0(io) 467 * q13={d26-d27}: jr/jak/jakr/dsbX_1(jo) 468 * q14={d28-d29}: rk/xmc 469 * q15={d30-d31}: mc/sr[3 & ~(nrounds - 1)] 470 */ 471 472 vld1.64 {d28-d29}, [r0 :128]! /* q14 = *rk++ */ 473 rsb r3, r1, #0 /* r3 := ~(x - 1) = -x */ 474 vmov.i8 q1, #0x0f 475 and r3, r3, #3 /* r3 := 3 & ~(x - 1) */ 476 477 /* (q4, q5) := (diptlo, dipthi) */ 478 ldr r6, =diptlo 479 ldr r7, =dipthi 480 vld1.64 {d8-d9}, [r6 :128] 481 vld1.64 {d10-d11}, [r7 :128] 482 483 /* load the rest of the constants */ 484 ldr r4, =dsbb_0 485 ldr r5, =dsbb_1 486 ldr r6, =inv 487 ldr r7, =inva 488 ldr r8, =.Lmc_forward_3 489 vld1.64 {d12-d13}, [r4 :128] /* q6 := dsbb[0] */ 490 vld1.64 {d14-d15}, [r5 :128] /* q7 := dsbb[1] */ 491 vld1.64 {d20-d21}, [r6 :128] /* q10 := inv */ 492 vld1.64 {d22-d23}, [r7 :128] /* q11 := inva */ 493 vld1.64 {d30-d31}, [r8 :128] /* q15 := mc_forward[3] */ 494 495 /* (q2, q3) := (lo, hi) */ 496 vshr.u8 q3, q0, #4 497 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 498 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 499 500 /* (q2, q3) := (diptlo(lo), dipthi(hi)) */ 501 vtbl.8 d4, {d8-d9}, d4 502 vtbl.8 d5, {d8-d9}, d5 503 vtbl.8 d6, {d10-d11}, d6 504 vtbl.8 d7, {d10-d11}, d7 505 506 /* load dsb9 */ 507 ldr r4, =dsb9_0 508 ldr r5, =dsb9_1 509 vld1.64 {d8-d9}, [r4 :128] /* q4 := dsb9[0] */ 510 vld1.64 {d10-d11}, [r5 :128] /* q5 := dsb9[1] */ 511 512 /* q0 := rk[0] + diptlo(lo) + dipthi(hi) */ 513 veor q0, q14, q2 514 veor q0, q0, q3 515 516 b 2f 517 518 1: /* load dsbd */ 519 ldr r4, =dsbd_0 520 vld1.64 {d16-d17}, [r4 :128]! /* q8 := dsbd[0] */ 521 vld1.64 {d18-d19}, [r4 :128] /* q9 := dsbd[1] */ 522 523 vld1.64 {d28-d29}, [r0 :128]! /* q14 = *rk++ */ 524 525 /* q0 := rk[i] + dsb9_0(io) + dsb9_1(jo) */ 526 vtbl.8 d24, {d8-d9}, d4 527 vtbl.8 d25, {d8-d9}, d5 528 vtbl.8 d26, {d10-d11}, d6 529 vtbl.8 d27, {d10-d11}, d7 530 veor q0, q14, q12 531 veor q0, q0, q13 532 533 /* q14 := x(mc) */ 534 vtbl.8 d28, {d0-d1}, d30 535 vtbl.8 d29, {d0-d1}, d31 536 537 /* q0 := x(mc) + dsbd_0(io) + dsbd_1(jo) */ 538 vtbl.8 d24, {d16-d17}, d4 539 vtbl.8 d25, {d16-d17}, d5 540 vtbl.8 d26, {d18-d19}, d6 541 vtbl.8 d27, {d18-d19}, d7 542 veor q0, q14, q12 543 veor q0, q0, q13 544 545 /* load dsbe */ 546 ldr r4, =dsbe_0 547 vld1.64 {d16-d17}, [r4 :128]! /* q8 := dsbe[0] */ 548 vld1.64 {d18-d19}, [r4 :128] /* q9 := dsbe[1] */ 549 550 /* q0 := x(mc) + dsbb_0(io) + dsbb_1(jo) */ 551 vtbl.8 d28, {d0-d1}, d30 552 vtbl.8 d29, {d0-d1}, d31 553 vtbl.8 d24, {d12-d13}, d4 554 vtbl.8 d25, {d12-d13}, d5 555 vtbl.8 d26, {d14-d15}, d6 556 vtbl.8 d27, {d14-d15}, d7 557 veor q0, q14, q12 558 veor q0, q0, q13 559 560 /* q0 := x(mc) + dsbe_0(io) + dsbe_1(jo) */ 561 vtbl.8 d28, {d0-d1}, d30 562 vtbl.8 d29, {d0-d1}, d31 563 vtbl.8 d24, {d16-d17}, d4 564 vtbl.8 d25, {d16-d17}, d5 565 vtbl.8 d26, {d18-d19}, d6 566 vtbl.8 d27, {d18-d19}, d7 567 veor q0, q14, q12 568 veor q0, q0, q13 569 570 /* q15 := mc := mc <<< 12*8 */ 571 vext.8 q15, q15, q15, #12 572 573 2: /* 574 * SubBytes 575 */ 576 577 /* (q2, q3) := (k, i) */ 578 vshr.u8 q3, q0, #4 579 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 580 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 581 582 /* q0 := a/k */ 583 vtbl.8 d0, {d22-d23}, d4 584 vtbl.8 d1, {d22-d23}, d5 585 586 /* q2 := j = i + k */ 587 veor q2, q3, q2 588 589 /* q12 := ir = 1/i */ 590 vtbl.8 d24, {d20-d21}, d6 591 vtbl.8 d25, {d20-d21}, d7 592 593 /* q13 := jr = 1/j */ 594 vtbl.8 d26, {d20-d21}, d4 595 vtbl.8 d27, {d20-d21}, d5 596 597 /* q12 := iak = 1/i + a/k */ 598 veor q12, q12, q0 599 600 /* q13 := jak = 1/j + a/k */ 601 veor q13, q13, q0 602 603 /* q12 := iakr = 1/(1/i + a/k) */ 604 vtbl.8 d24, {d20-d21}, d24 605 vtbl.8 d25, {d20-d21}, d25 606 607 /* q13 := jakr = 1/(1/j + a/k) */ 608 vtbl.8 d26, {d20-d21}, d26 609 vtbl.8 d27, {d20-d21}, d27 610 611 /* q2 := io = j + 1/(1/i + a/k) */ 612 veor q2, q2, q12 613 614 /* q3 := jo = i + 1/(1/j + a/k) */ 615 veor q3, q3, q13 616 617 /* advance round */ 618 subs r1, r1, #1 619 bne 1b 620 621 /* (q6, q7, q15) := (dsbo[0], dsbo[1], sr[i]) */ 622 ldr r8, =sr 623 ldr r6, =dsbo_0 624 ldr r7, =dsbo_1 625 add r8, r8, r3, lsl #4 626 vld1.64 {d12-d13}, [r6 :128] 627 vld1.64 {d14-d15}, [r7 :128] 628 vld1.64 {d30-d31}, [r8 :128] 629 630 vld1.64 {d28-d29}, [r0 :128]! /* q14 = *rk++ */ 631 632 /* (q2, q3) := (dsbo_0(io), dsbo_1(jo)) */ 633 vtbl.8 d4, {d12-d13}, d4 634 vtbl.8 d5, {d12-d13}, d5 635 vtbl.8 d6, {d14-d15}, d6 636 vtbl.8 d7, {d14-d15}, d7 637 638 /* q2 := x = rk[nr] + dsbo_0(io) + dsbo_1(jo) */ 639 veor q2, q2, q14 640 veor q2, q2, q3 641 642 /* q0 := x(sr[i]) */ 643 vtbl.8 d0, {d4-d5}, d30 644 vtbl.8 d1, {d4-d5}, d31 645 646 vpop {d8-d15} 647 pop {r4, r5, r6, r7, r8, r10, r11, lr} 648 #ifdef _KERNEL 649 vmov r0, r1, d0 650 vmov r2, r3, d1 651 #endif 652 bx lr 653 END(aes_neon_dec1) 654
Indexes created Wed Oct 15 07:09:58 GMT 2025