aes_neon_32.S revision 1.10
1 /* $NetBSD: aes_neon_32.S,v 1.10 2020/09/10 11:30:28 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 RCSID("$NetBSD: aes_neon_32.S,v 1.10 2020/09/10 11:30:28 riastradh Exp $") 32 33 .fpu neon 34 35 .text 36 .p2align 2 37 .Lconstants_addr: 38 .long .Lconstants - . 39 40 .section .rodata 41 .p2align 5 42 .Lconstants: 43 44 .Linv_inva: /* inv and inva must be consecutive */ 45 .type inv,_ASM_TYPE_OBJECT 46 inv: 47 .byte 0x80,0x01,0x08,0x0D,0x0F,0x06,0x05,0x0E 48 .byte 0x02,0x0C,0x0B,0x0A,0x09,0x03,0x07,0x04 49 END(inv) 50 51 .type inva,_ASM_TYPE_OBJECT 52 inva: 53 .byte 0x80,0x07,0x0B,0x0F,0x06,0x0A,0x04,0x01 54 .byte 0x09,0x08,0x05,0x02,0x0C,0x0E,0x0D,0x03 55 END(inva) 56 57 .type mc_forward,_ASM_TYPE_OBJECT 58 mc_forward: 59 .byte 0x01,0x02,0x03,0x00,0x05,0x06,0x07,0x04 /* 0 */ 60 .byte 0x09,0x0A,0x0B,0x08,0x0D,0x0E,0x0F,0x0C 61 62 .byte 0x05,0x06,0x07,0x04,0x09,0x0A,0x0B,0x08 /* 1 */ 63 .byte 0x0D,0x0E,0x0F,0x0C,0x01,0x02,0x03,0x00 64 65 .byte 0x09,0x0A,0x0B,0x08,0x0D,0x0E,0x0F,0x0C /* 2 */ 66 .byte 0x01,0x02,0x03,0x00,0x05,0x06,0x07,0x04 67 68 .Lmc_forward_3: 69 .byte 0x0D,0x0E,0x0F,0x0C,0x01,0x02,0x03,0x00 /* 3 */ 70 .byte 0x05,0x06,0x07,0x04,0x09,0x0A,0x0B,0x08 71 END(mc_forward) 72 73 .type mc_backward,_ASM_TYPE_OBJECT 74 mc_backward: 75 .byte 0x03,0x00,0x01,0x02,0x07,0x04,0x05,0x06 /* 0 */ 76 .byte 0x0B,0x08,0x09,0x0A,0x0F,0x0C,0x0D,0x0E 77 78 .byte 0x0F,0x0C,0x0D,0x0E,0x03,0x00,0x01,0x02 /* 1 */ 79 .byte 0x07,0x04,0x05,0x06,0x0B,0x08,0x09,0x0A 80 81 .byte 0x0B,0x08,0x09,0x0A,0x0F,0x0C,0x0D,0x0E /* 2 */ 82 .byte 0x03,0x00,0x01,0x02,0x07,0x04,0x05,0x06 83 84 .byte 0x07,0x04,0x05,0x06,0x0B,0x08,0x09,0x0A /* 3 */ 85 .byte 0x0F,0x0C,0x0D,0x0E,0x03,0x00,0x01,0x02 86 END(mc_backward) 87 88 .type sr,_ASM_TYPE_OBJECT 89 sr: 90 .byte 0x00,0x01,0x02,0x03,0x04,0x05,0x06,0x07 /* 0 */ 91 .byte 0x08,0x09,0x0A,0x0B,0x0C,0x0D,0x0E,0x0F 92 93 .byte 0x00,0x05,0x0A,0x0F,0x04,0x09,0x0E,0x03 /* 1 */ 94 .byte 0x08,0x0D,0x02,0x07,0x0C,0x01,0x06,0x0B 95 96 .byte 0x00,0x09,0x02,0x0B,0x04,0x0D,0x06,0x0F /* 2 */ 97 .byte 0x08,0x01,0x0A,0x03,0x0C,0x05,0x0E,0x07 98 99 .byte 0x00,0x0D,0x0A,0x07,0x04,0x01,0x0E,0x0B /* 3 */ 100 .byte 0x08,0x05,0x02,0x0F,0x0C,0x09,0x06,0x03 101 END(sr) 102 103 .type ipt,_ASM_TYPE_OBJECT 104 ipt: 105 .byte 0x00,0x70,0x2A,0x5A,0x98,0xE8,0xB2,0xC2 /* lo */ 106 .byte 0x08,0x78,0x22,0x52,0x90,0xE0,0xBA,0xCA 107 .byte 0x00,0x4D,0x7C,0x31,0x7D,0x30,0x01,0x4C /* hi */ 108 .byte 0x81,0xCC,0xFD,0xB0,0xFC,0xB1,0x80,0xCD 109 END(ipt) 110 111 .type sb1,_ASM_TYPE_OBJECT 112 sb1: 113 .byte 0x00,0x3E,0x50,0xCB,0x8F,0xE1,0x9B,0xB1 /* 0 */ 114 .byte 0x44,0xF5,0x2A,0x14,0x6E,0x7A,0xDF,0xA5 115 .byte 0x00,0x23,0xE2,0xFA,0x15,0xD4,0x18,0x36 /* 1 */ 116 .byte 0xEF,0xD9,0x2E,0x0D,0xC1,0xCC,0xF7,0x3B 117 END(sb1) 118 119 .type sb2,_ASM_TYPE_OBJECT 120 sb2: 121 .byte 0x00,0x24,0x71,0x0B,0xC6,0x93,0x7A,0xE2 /* 0 */ 122 .byte 0xCD,0x2F,0x98,0xBC,0x55,0xE9,0xB7,0x5E 123 .byte 0x00,0x29,0xE1,0x0A,0x40,0x88,0xEB,0x69 /* 1 */ 124 .byte 0x4A,0x23,0x82,0xAB,0xC8,0x63,0xA1,0xC2 125 END(sb2) 126 127 .type sbo,_ASM_TYPE_OBJECT 128 sbo: 129 .byte 0x00,0xC7,0xBD,0x6F,0x17,0x6D,0xD2,0xD0 /* 0 */ 130 .byte 0x78,0xA8,0x02,0xC5,0x7A,0xBF,0xAA,0x15 131 .byte 0x00,0x6A,0xBB,0x5F,0xA5,0x74,0xE4,0xCF /* 1 */ 132 .byte 0xFA,0x35,0x2B,0x41,0xD1,0x90,0x1E,0x8E 133 END(sbo) 134 135 .type dipt,_ASM_TYPE_OBJECT 136 dipt: 137 .byte 0x00,0x5F,0x54,0x0B,0x04,0x5B,0x50,0x0F /* lo */ 138 .byte 0x1A,0x45,0x4E,0x11,0x1E,0x41,0x4A,0x15 139 .byte 0x00,0x65,0x05,0x60,0xE6,0x83,0xE3,0x86 /* hi */ 140 .byte 0x94,0xF1,0x91,0xF4,0x72,0x17,0x77,0x12 141 END(dipt) 142 143 .type dsb9,_ASM_TYPE_OBJECT 144 dsb9: 145 .byte 0x00,0xD6,0x86,0x9A,0x53,0x03,0x1C,0x85 /* 0 */ 146 .byte 0xC9,0x4C,0x99,0x4F,0x50,0x1F,0xD5,0xCA 147 .byte 0x00,0x49,0xD7,0xEC,0x89,0x17,0x3B,0xC0 /* 1 */ 148 .byte 0x65,0xA5,0xFB,0xB2,0x9E,0x2C,0x5E,0x72 149 END(dsb9) 150 151 .type dsbd,_ASM_TYPE_OBJECT 152 dsbd: 153 .byte 0x00,0xA2,0xB1,0xE6,0xDF,0xCC,0x57,0x7D /* 0 */ 154 .byte 0x39,0x44,0x2A,0x88,0x13,0x9B,0x6E,0xF5 155 .byte 0x00,0xCB,0xC6,0x24,0xF7,0xFA,0xE2,0x3C /* 1 */ 156 .byte 0xD3,0xEF,0xDE,0x15,0x0D,0x18,0x31,0x29 157 END(dsbd) 158 159 .type dsbb,_ASM_TYPE_OBJECT 160 dsbb: 161 .byte 0x00,0x42,0xB4,0x96,0x92,0x64,0x22,0xD0 /* 0 */ 162 .byte 0x04,0xD4,0xF2,0xB0,0xF6,0x46,0x26,0x60 163 .byte 0x00,0x67,0x59,0xCD,0xA6,0x98,0x94,0xC1 /* 1 */ 164 .byte 0x6B,0xAA,0x55,0x32,0x3E,0x0C,0xFF,0xF3 165 END(dsbb) 166 167 .type dsbe,_ASM_TYPE_OBJECT 168 dsbe: 169 .byte 0x00,0xD0,0xD4,0x26,0x96,0x92,0xF2,0x46 /* 0 */ 170 .byte 0xB0,0xF6,0xB4,0x64,0x04,0x60,0x42,0x22 171 .byte 0x00,0xC1,0xAA,0xFF,0xCD,0xA6,0x55,0x0C /* 1 */ 172 .byte 0x32,0x3E,0x59,0x98,0x6B,0xF3,0x67,0x94 173 END(dsbe) 174 175 .type dsbo,_ASM_TYPE_OBJECT 176 dsbo: 177 .byte 0x00,0x40,0xF9,0x7E,0x53,0xEA,0x87,0x13 /* 0 */ 178 .byte 0x2D,0x3E,0x94,0xD4,0xB9,0x6D,0xAA,0xC7 179 .byte 0x00,0x1D,0x44,0x93,0x0F,0x56,0xD7,0x12 /* 1 */ 180 .byte 0x9C,0x8E,0xC5,0xD8,0x59,0x81,0x4B,0xCA 181 END(dsbo) 182 183 /* 184 * aes_neon_enc1(enc, x, nrounds) 185 * 186 * With -mfloat-abi=hard: 187 * 188 * uint8x16_t@q0 189 * aes_neon_enc1(const struct aesenc *enc@r0, uint8x16_t x@q0, 190 * unsigned nrounds@r1) 191 * 192 * With -mfloat-abi=soft(fp) (i.e., __SOFTFP__): 193 * 194 * uint8x16_t@(r0,r1,r2,r3) 195 * aes_neon_enc1(const struct aesenc *enc@r0, 196 * uint8x16_t x@(r2,r3,sp[0],sp[4]), nrounds@sp[8]) 197 */ 198 ENTRY(aes_neon_enc1) 199 #ifdef __SOFTFP__ 200 #ifdef __ARM_BIG_ENDIAN 201 vmov d0, r3, r2 /* d0 := x lo */ 202 #else 203 vmov d0, r2, r3 /* d0 := x lo */ 204 #endif 205 vldr d1, [sp] /* d1 := x hi */ 206 ldr r1, [sp, #8] /* r1 := nrounds */ 207 #endif 208 push {r4, r5, r6, r8, r10, lr} 209 vpush {d8-d15} 210 211 /* 212 * r3: rmod4 213 * r4: mc_forward 214 * r5: mc_backward 215 * r6,r8,r10,ip: temporaries 216 * q0={d0-d1}: x/ak/A 217 * q1={d2-d3}: 0x0f0f... 218 * q2={d4-d5}: lo/k/j/io 219 * q3={d6-d7}: hi/i/jo 220 * q4={d8-d9}: iptlo 221 * q5={d10-d11}: ipthi 222 * q6={d12-d13}: sb1[0]/sbo[0] 223 * q7={d14-d15}: sb1[1]/sbo[1] 224 * q8={d16-d17}: sb2[0] 225 * q9={d18-d19}: sb2[1] 226 * q10={d20-d21}: inv 227 * q11={d22-d23}: inva 228 * q12={d24-d25}: ir/iak/iakr/sb1_0(io)/mc_backward[rmod4] 229 * q13={d26-d27}: jr/jak/jakr/sb1_1(jo)/mc_forward[rmod4] 230 * q14={d28-d29}: rk/A2/A2_B_D 231 * q15={d30-d31}: A2_B/sr[rmod4] 232 */ 233 234 /* ip := .Lconstants - .Lconstants_addr, r10 := .Lconstants_addr */ 235 ldr ip, .Lconstants_addr 236 adr r10, .Lconstants_addr 237 238 vld1.8 {q14}, [r0 :128]! /* q14 = *rk++ */ 239 movw r3, #0 240 vmov.i8 q1, #0x0f 241 242 /* ip := .Lconstants */ 243 add ip, ip, r10 244 245 /* (q4, q5) := (iptlo, ipthi) */ 246 add r6, ip, #(ipt - .Lconstants) 247 vld1.8 {q4-q5}, [r6 :256] 248 249 /* load the rest of the constants */ 250 add r4, ip, #(sb1 - .Lconstants) 251 add r6, ip, #(sb2 - .Lconstants) 252 add r8, ip, #(.Linv_inva - .Lconstants) 253 vld1.8 {q6-q7}, [r4 :256] /* q6 = sb1[0], q7 = sb1[1] */ 254 vld1.8 {q8-q9}, [r6 :256] /* q8 = sb2[0], q9 = sb2[1] */ 255 vld1.8 {q10-q11}, [r8 :256] /* q10 = inv, q11 = inva */ 256 257 /* (r4, r5) := (&mc_forward[0], &mc_backward[0]) */ 258 add r4, ip, #(mc_forward - .Lconstants) 259 add r5, ip, #(mc_backward - .Lconstants) 260 261 /* (q2, q3) := (lo, hi) */ 262 vshr.u8 q3, q0, #4 263 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 264 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 265 266 /* (q2, q3) := (iptlo(lo), ipthi(hi)) */ 267 vtbl.8 d4, {q4}, d4 268 vtbl.8 d5, {q4}, d5 269 vtbl.8 d6, {q5}, d6 270 vtbl.8 d7, {q5}, d7 271 272 /* q0 := rk[0] + iptlo(lo) + ipthi(hi) */ 273 veor q0, q14, q2 274 veor q0, q0, q3 275 276 b 2f 277 278 _ALIGN_TEXT 279 1: vld1.8 {q14}, [r0 :128]! /* q14 = *rk++ */ 280 281 /* q0 := A = rk[i] + sb1_0(io) + sb1_1(jo) */ 282 vtbl.8 d24, {q6}, d4 283 vtbl.8 d25, {q6}, d5 284 vtbl.8 d26, {q7}, d6 285 vtbl.8 d27, {q7}, d7 286 veor q0, q14, q12 287 veor q0, q0, q13 288 289 /* q14 := A2 = sb2_0[io] + sb2_1[jo] */ 290 vtbl.8 d24, {q8}, d4 291 vtbl.8 d25, {q8}, d5 292 vtbl.8 d26, {q9}, d6 293 vtbl.8 d27, {q9}, d7 294 veor q14, q12, q13 295 296 /* (q12, q13) := (mc_forward[rmod4], mc_backward[rmod4]) */ 297 add r6, r4, r3, lsl #4 298 add r8, r5, r3, lsl #4 299 vld1.8 {q12}, [r6 :128] 300 vld1.8 {q13}, [r8 :128] 301 302 /* q15 := A2_B = A2 + A(mcf) */ 303 vtbl.8 d30, {q0}, d24 304 vtbl.8 d31, {q0}, d25 305 veor q15, q15, q14 306 307 /* q14 := A2_B_D = A2_B + A(mcb) */ 308 vtbl.8 d28, {q0}, d26 309 vtbl.8 d29, {q0}, d27 310 veor q14, q14, q15 311 312 /* q0 := x = A2_B_D + A2_B(mcf) */ 313 vtbl.8 d0, {q15}, d24 314 vtbl.8 d1, {q15}, d25 315 veor q0, q0, q14 316 317 2: /* 318 * SubBytes 319 */ 320 321 /* (q2, q3) := (k, i) */ 322 vshr.u8 q3, q0, #4 323 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 324 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 325 326 /* q0 := a/k */ 327 vtbl.8 d0, {q11}, d4 328 vtbl.8 d1, {q11}, d5 329 330 /* q2 := j = i + k */ 331 veor q2, q3, q2 332 333 /* q12 := ir = 1/i */ 334 vtbl.8 d24, {q10}, d6 335 vtbl.8 d25, {q10}, d7 336 337 /* q13 := jr = 1/j */ 338 vtbl.8 d26, {q10}, d4 339 vtbl.8 d27, {q10}, d5 340 341 /* q12 := iak = 1/i + a/k */ 342 veor q12, q12, q0 343 344 /* q13 := jak = 1/j + a/k */ 345 veor q13, q13, q0 346 347 /* q12 := iakr = 1/(1/i + a/k) */ 348 vtbl.8 d24, {q10}, d24 349 vtbl.8 d25, {q10}, d25 350 351 /* q13 := jakr = 1/(1/j + a/k) */ 352 vtbl.8 d26, {q10}, d26 353 vtbl.8 d27, {q10}, d27 354 355 /* q2 := io = j + 1/(1/i + a/k) */ 356 veor q2, q2, q12 357 358 /* q3 := jo = i + 1/(1/j + a/k) */ 359 veor q3, q3, q13 360 361 /* advance round */ 362 add r3, r3, #1 363 subs r1, r1, #1 364 and r3, r3, #3 365 bne 1b 366 367 /* (q6, q7, q15) := (sbo[0], sbo[1], sr[rmod4]) */ 368 add r8, ip, #(sr - .Lconstants) 369 add r6, ip, #(sbo - .Lconstants) 370 add r8, r8, r3, lsl #4 371 vld1.8 {q6-q7}, [r6 :256] 372 vld1.8 {q15}, [r8 :128] 373 374 vld1.8 {q14}, [r0 :128]! /* q14 = *rk++ */ 375 376 /* (q2, q3) := (sbo_0(io), sbo_1(jo)) */ 377 vtbl.8 d4, {q6}, d4 378 vtbl.8 d5, {q6}, d5 379 vtbl.8 d6, {q7}, d6 380 vtbl.8 d7, {q7}, d7 381 382 /* q2 := x = rk[nr] + sbo_0(io) + sbo_1(jo) */ 383 veor q2, q2, q14 384 veor q2, q2, q3 385 386 /* q0 := x(sr[rmod4]) */ 387 vtbl.8 d0, {q2}, d30 388 vtbl.8 d1, {q2}, d31 389 390 vpop {d8-d15} 391 pop {r4, r5, r6, r8, r10, lr} 392 #ifdef __SOFTFP__ 393 #ifdef __ARM_BIG_ENDIAN 394 vmov r1, r0, d0 395 vmov r3, r2, d1 396 #else 397 vmov r0, r1, d0 398 vmov r2, r3, d1 399 #endif 400 #endif 401 bx lr 402 END(aes_neon_enc1) 403 404 /* 405 * aes_neon_dec1(dec, x, nrounds) 406 * 407 * With -mfloat-abi=hard: 408 * 409 * uint8x16_t@q0 410 * aes_neon_dec1(const struct aesdec *dec@r0, uint8x16_t x@q0, 411 * unsigned nrounds@r1) 412 * 413 * With -mfloat-abi=soft(fp) (here spelled `#ifdef _KERNEL'): 414 * 415 * uint8x16_t@(r0,r1,r2,r3) 416 * aes_neon_dec1(const struct aesdec *dec@r0, 417 * uint8x16_t x@(r2,r3,sp[0],sp[4]), nrounds@sp[8]) 418 */ 419 ENTRY(aes_neon_dec1) 420 #ifdef __SOFTFP__ 421 #ifdef __ARM_BIG_ENDIAN 422 vmov d0, r3, r2 /* d0 := x lo */ 423 #else 424 vmov d0, r2, r3 /* d0 := x lo */ 425 #endif 426 vldr d1, [sp] /* d1 := x hi */ 427 ldr r1, [sp, #8] /* r1 := nrounds */ 428 #endif 429 push {r4, r5, r6, r8, r10, lr} 430 vpush {d8-d15} 431 432 /* 433 * r3: 3 & ~(nrounds - 1) 434 * r4: dsbd 435 * r5: dsbe 436 * r6,r8,r10,ip: temporaries 437 * q0={d0-d1}: x/ak 438 * q1={d2-d3}: 0x0f0f... 439 * q2={d4-d5}: lo/k/j/io 440 * q3={d6-d7}: hi/i/jo 441 * q4={d8-d9}: diptlo/dsb9[0] 442 * q5={d10-d11}: dipthi/dsb9[1] 443 * q6={d12-d13}: dsbb[0]/dsbo[0] 444 * q7={d14-d15}: dsbb[1]/dsbo[1] 445 * q8={d16-d17}: dsbd[0]/dsbe[0] 446 * q9={d18-d19}: dsbd[1]/dsbe[0] 447 * q10={d20-d21}: inv 448 * q11={d22-d23}: inva 449 * q12={d24-d25}: ir/iak/iakr/dsbX_0(io) 450 * q13={d26-d27}: jr/jak/jakr/dsbX_1(jo) 451 * q14={d28-d29}: rk/xmc 452 * q15={d30-d31}: mc/sr[3 & ~(nrounds - 1)] 453 */ 454 455 /* ip := .Lconstants - .Lconstants_addr, r10 := .Lconstants_addr */ 456 ldr ip, .Lconstants_addr 457 adr r10, .Lconstants_addr 458 459 vld1.8 {q14}, [r0 :128]! /* q14 = *rk++ */ 460 rsb r3, r1, #0 /* r3 := ~(x - 1) = -x */ 461 vmov.i8 q1, #0x0f 462 and r3, r3, #3 /* r3 := 3 & ~(x - 1) */ 463 464 /* ip := .Lconstants */ 465 add ip, ip, r10 466 467 /* (q4, q5) := (diptlo, dipthi) */ 468 add r6, ip, #(dipt - .Lconstants) 469 vld1.8 {q4-q5}, [r6 :256] 470 471 /* load the rest of the constants */ 472 add r4, ip, #(dsbb - .Lconstants) 473 add r6, ip, #(.Linv_inva - .Lconstants) 474 add r8, ip, #(.Lmc_forward_3 - .Lconstants) 475 vld1.8 {q6-q7}, [r4 :256] /* q6 := dsbb[0], q7 := dsbb[1] */ 476 vld1.8 {q10-q11}, [r6 :256] /* q10 := inv, q11 := inva */ 477 vld1.8 {q15}, [r8 :128] /* q15 := mc_forward[3] */ 478 479 /* (q2, q3) := (lo, hi) */ 480 vshr.u8 q3, q0, #4 481 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 482 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 483 484 /* (q2, q3) := (diptlo(lo), dipthi(hi)) */ 485 vtbl.8 d4, {q4}, d4 486 vtbl.8 d5, {q4}, d5 487 vtbl.8 d6, {q5}, d6 488 vtbl.8 d7, {q5}, d7 489 490 /* load dsb9 */ 491 add r4, ip, #(dsb9 - .Lconstants) 492 vld1.8 {q4-q5}, [r4 :256] /* q4 := dsb9[0], q5 := dsb9[1] */ 493 494 /* r4 := dsbd, r5 := dsbe */ 495 add r4, ip, #(dsbd - .Lconstants) 496 add r5, ip, #(dsbe - .Lconstants) 497 498 /* q0 := rk[0] + diptlo(lo) + dipthi(hi) */ 499 veor q0, q14, q2 500 veor q0, q0, q3 501 502 b 2f 503 504 _ALIGN_TEXT 505 1: /* load dsbd */ 506 vld1.8 {q8-q9}, [r4 :256] /* q8 := dsbd[0], q9 := dsbd[1] */ 507 508 vld1.8 {q14}, [r0 :128]! /* q14 = *rk++ */ 509 510 /* q0 := rk[i] + dsb9_0(io) + dsb9_1(jo) */ 511 vtbl.8 d24, {q4}, d4 512 vtbl.8 d25, {q4}, d5 513 vtbl.8 d26, {q5}, d6 514 vtbl.8 d27, {q5}, d7 515 veor q0, q14, q12 516 veor q0, q0, q13 517 518 /* q14 := x(mc) */ 519 vtbl.8 d28, {q0}, d30 520 vtbl.8 d29, {q0}, d31 521 522 /* q0 := x(mc) + dsbd_0(io) + dsbd_1(jo) */ 523 vtbl.8 d24, {q8}, d4 524 vtbl.8 d25, {q8}, d5 525 vtbl.8 d26, {q9}, d6 526 vtbl.8 d27, {q9}, d7 527 veor q0, q14, q12 528 veor q0, q0, q13 529 530 /* load dsbe */ 531 vld1.8 {q8-q9}, [r5 :256] /* q8 := dsbe[0], q9 := dsbe[1] */ 532 533 /* q0 := x(mc) + dsbb_0(io) + dsbb_1(jo) */ 534 vtbl.8 d28, {q0}, d30 535 vtbl.8 d29, {q0}, d31 536 vtbl.8 d24, {q6}, d4 537 vtbl.8 d25, {q6}, d5 538 vtbl.8 d26, {q7}, d6 539 vtbl.8 d27, {q7}, d7 540 veor q0, q14, q12 541 veor q0, q0, q13 542 543 /* q0 := x(mc) + dsbe_0(io) + dsbe_1(jo) */ 544 vtbl.8 d28, {q0}, d30 545 vtbl.8 d29, {q0}, d31 546 vtbl.8 d24, {q8}, d4 547 vtbl.8 d25, {q8}, d5 548 vtbl.8 d26, {q9}, d6 549 vtbl.8 d27, {q9}, d7 550 veor q0, q14, q12 551 veor q0, q0, q13 552 553 /* q15 := mc := mc <<< 12*8 */ 554 vext.8 q15, q15, q15, #12 555 556 2: /* 557 * SubBytes 558 */ 559 560 /* (q2, q3) := (k, i) */ 561 vshr.u8 q3, q0, #4 562 vand q2, q0, q1 /* q2 := x & 0x0f0f... */ 563 vand q3, q3, q1 /* q3 := (x >> 4) & 0x0f0f... */ 564 565 /* q0 := a/k */ 566 vtbl.8 d0, {q11}, d4 567 vtbl.8 d1, {q11}, d5 568 569 /* q2 := j = i + k */ 570 veor q2, q3, q2 571 572 /* q12 := ir = 1/i */ 573 vtbl.8 d24, {q10}, d6 574 vtbl.8 d25, {q10}, d7 575 576 /* q13 := jr = 1/j */ 577 vtbl.8 d26, {q10}, d4 578 vtbl.8 d27, {q10}, d5 579 580 /* q12 := iak = 1/i + a/k */ 581 veor q12, q12, q0 582 583 /* q13 := jak = 1/j + a/k */ 584 veor q13, q13, q0 585 586 /* q12 := iakr = 1/(1/i + a/k) */ 587 vtbl.8 d24, {q10}, d24 588 vtbl.8 d25, {q10}, d25 589 590 /* q13 := jakr = 1/(1/j + a/k) */ 591 vtbl.8 d26, {q10}, d26 592 vtbl.8 d27, {q10}, d27 593 594 /* q2 := io = j + 1/(1/i + a/k) */ 595 veor q2, q2, q12 596 597 /* q3 := jo = i + 1/(1/j + a/k) */ 598 veor q3, q3, q13 599 600 /* advance round */ 601 subs r1, r1, #1 602 bne 1b 603 604 /* (q6, q7, q15) := (dsbo[0], dsbo[1], sr[i]) */ 605 add r8, ip, #(sr - .Lconstants) 606 add r6, ip, #(dsbo - .Lconstants) 607 add r8, r8, r3, lsl #4 608 vld1.8 {q6-q7}, [r6 :256] 609 vld1.8 {q15}, [r8 :128] 610 611 vld1.8 {q14}, [r0 :128]! /* q14 = *rk++ */ 612 613 /* (q2, q3) := (dsbo_0(io), dsbo_1(jo)) */ 614 vtbl.8 d4, {q6}, d4 615 vtbl.8 d5, {q6}, d5 616 vtbl.8 d6, {q7}, d6 617 vtbl.8 d7, {q7}, d7 618 619 /* q2 := x = rk[nr] + dsbo_0(io) + dsbo_1(jo) */ 620 veor q2, q2, q14 621 veor q2, q2, q3 622 623 /* q0 := x(sr[i]) */ 624 vtbl.8 d0, {q2}, d30 625 vtbl.8 d1, {q2}, d31 626 627 vpop {d8-d15} 628 pop {r4, r5, r6, r8, r10, lr} 629 #ifdef __SOFTFP__ 630 #ifdef __ARM_BIG_ENDIAN 631 vmov r1, r0, d0 632 vmov r3, r2, d1 633 #else 634 vmov r0, r1, d0 635 vmov r2, r3, d1 636 #endif 637 #endif 638 bx lr 639 END(aes_neon_dec1) 640
Indexes created Sun Oct 19 02:09:48 GMT 2025