| 1 |
// This file is generated from a similarly-named Perl script in the BoringSSL |
| 2 |
// source tree. Do not edit by hand. |
| 3 |
|
| 4 |
#if !defined(__has_feature) |
| 5 |
#define __has_feature(x) 0 |
| 6 |
#endif |
| 7 |
#if __has_feature(memory_sanitizer) && !defined(OPENSSL_NO_ASM) |
| 8 |
#define OPENSSL_NO_ASM |
| 9 |
#endif |
| 10 |
|
| 11 |
#if !defined(OPENSSL_NO_ASM) && defined(__AARCH64EL__) && defined(__APPLE__) |
| 12 |
#if defined(BORINGSSL_PREFIX) |
| 13 |
#include <boringssl_prefix_symbols_asm.h> |
| 14 |
#endif |
| 15 |
#include <openssl/arm_arch.h> |
| 16 |
#if __ARM_MAX_ARCH__ >= 8 |
| 17 |
|
| 18 |
|
| 19 |
.text |
| 20 |
.globl _aes_gcm_enc_kernel |
| 21 |
.private_extern _aes_gcm_enc_kernel |
| 22 |
|
| 23 |
.align 4 |
| 24 |
_aes_gcm_enc_kernel: |
| 25 |
AARCH64_SIGN_LINK_REGISTER |
| 26 |
stp x29, x30, [sp, #-128]! |
| 27 |
mov x29, sp |
| 28 |
stp x19, x20, [sp, #16] |
| 29 |
mov x16, x4 |
| 30 |
mov x8, x5 |
| 31 |
stp x21, x22, [sp, #32] |
| 32 |
stp x23, x24, [sp, #48] |
| 33 |
stp d8, d9, [sp, #64] |
| 34 |
stp d10, d11, [sp, #80] |
| 35 |
stp d12, d13, [sp, #96] |
| 36 |
stp d14, d15, [sp, #112] |
| 37 |
ldr w17, [x8, #240] |
| 38 |
add x19, x8, x17, lsl #4 // borrow input_l1 for last key |
| 39 |
ldp x13, x14, [x19] // load round N keys |
| 40 |
ldr q31, [x19, #-16] // load round N-1 keys |
| 41 |
add x4, x0, x1, lsr #3 // end_input_ptr |
| 42 |
lsr x5, x1, #3 // byte_len |
| 43 |
mov x15, x5 |
| 44 |
ldp x10, x11, [x16] // ctr96_b64, ctr96_t32 |
| 45 |
ld1 { v0.16b}, [x16] // special case vector load initial counter so we can start first AES block as quickly as possible |
| 46 |
sub x5, x5, #1 // byte_len - 1 |
| 47 |
ldr q18, [x8, #0] // load rk0 |
| 48 |
and x5, x5, #0xffffffffffffffc0 // number of bytes to be processed in main loop (at least 1 byte must be handled by tail) |
| 49 |
ldr q25, [x8, #112] // load rk7 |
| 50 |
add x5, x5, x0 |
| 51 |
lsr x12, x11, #32 |
| 52 |
fmov d2, x10 // CTR block 2 |
| 53 |
orr w11, w11, w11 |
| 54 |
rev w12, w12 // rev_ctr32 |
| 55 |
fmov d1, x10 // CTR block 1 |
| 56 |
aese v0.16b, v18.16b |
| 57 |
aesmc v0.16b, v0.16b // AES block 0 - round 0 |
| 58 |
add w12, w12, #1 // increment rev_ctr32 |
| 59 |
rev w9, w12 // CTR block 1 |
| 60 |
fmov d3, x10 // CTR block 3 |
| 61 |
orr x9, x11, x9, lsl #32 // CTR block 1 |
| 62 |
add w12, w12, #1 // CTR block 1 |
| 63 |
ldr q19, [x8, #16] // load rk1 |
| 64 |
fmov v1.d[1], x9 // CTR block 1 |
| 65 |
rev w9, w12 // CTR block 2 |
| 66 |
add w12, w12, #1 // CTR block 2 |
| 67 |
orr x9, x11, x9, lsl #32 // CTR block 2 |
| 68 |
ldr q20, [x8, #32] // load rk2 |
| 69 |
fmov v2.d[1], x9 // CTR block 2 |
| 70 |
rev w9, w12 // CTR block 3 |
| 71 |
aese v0.16b, v19.16b |
| 72 |
aesmc v0.16b, v0.16b // AES block 0 - round 1 |
| 73 |
orr x9, x11, x9, lsl #32 // CTR block 3 |
| 74 |
fmov v3.d[1], x9 // CTR block 3 |
| 75 |
aese v1.16b, v18.16b |
| 76 |
aesmc v1.16b, v1.16b // AES block 1 - round 0 |
| 77 |
ldr q21, [x8, #48] // load rk3 |
| 78 |
aese v0.16b, v20.16b |
| 79 |
aesmc v0.16b, v0.16b // AES block 0 - round 2 |
| 80 |
ldr q24, [x8, #96] // load rk6 |
| 81 |
aese v2.16b, v18.16b |
| 82 |
aesmc v2.16b, v2.16b // AES block 2 - round 0 |
| 83 |
ldr q23, [x8, #80] // load rk5 |
| 84 |
aese v1.16b, v19.16b |
| 85 |
aesmc v1.16b, v1.16b // AES block 1 - round 1 |
| 86 |
ldr q14, [x3, #80] // load h3l | h3h |
| 87 |
ext v14.16b, v14.16b, v14.16b, #8 |
| 88 |
aese v3.16b, v18.16b |
| 89 |
aesmc v3.16b, v3.16b // AES block 3 - round 0 |
| 90 |
aese v2.16b, v19.16b |
| 91 |
aesmc v2.16b, v2.16b // AES block 2 - round 1 |
| 92 |
ldr q22, [x8, #64] // load rk4 |
| 93 |
aese v1.16b, v20.16b |
| 94 |
aesmc v1.16b, v1.16b // AES block 1 - round 2 |
| 95 |
ldr q13, [x3, #64] // load h2l | h2h |
| 96 |
ext v13.16b, v13.16b, v13.16b, #8 |
| 97 |
aese v3.16b, v19.16b |
| 98 |
aesmc v3.16b, v3.16b // AES block 3 - round 1 |
| 99 |
ldr q30, [x8, #192] // load rk12 |
| 100 |
aese v2.16b, v20.16b |
| 101 |
aesmc v2.16b, v2.16b // AES block 2 - round 2 |
| 102 |
ldr q15, [x3, #112] // load h4l | h4h |
| 103 |
ext v15.16b, v15.16b, v15.16b, #8 |
| 104 |
aese v1.16b, v21.16b |
| 105 |
aesmc v1.16b, v1.16b // AES block 1 - round 3 |
| 106 |
ldr q29, [x8, #176] // load rk11 |
| 107 |
aese v3.16b, v20.16b |
| 108 |
aesmc v3.16b, v3.16b // AES block 3 - round 2 |
| 109 |
ldr q26, [x8, #128] // load rk8 |
| 110 |
aese v2.16b, v21.16b |
| 111 |
aesmc v2.16b, v2.16b // AES block 2 - round 3 |
| 112 |
add w12, w12, #1 // CTR block 3 |
| 113 |
aese v0.16b, v21.16b |
| 114 |
aesmc v0.16b, v0.16b // AES block 0 - round 3 |
| 115 |
aese v3.16b, v21.16b |
| 116 |
aesmc v3.16b, v3.16b // AES block 3 - round 3 |
| 117 |
ld1 { v11.16b}, [x3] |
| 118 |
ext v11.16b, v11.16b, v11.16b, #8 |
| 119 |
rev64 v11.16b, v11.16b |
| 120 |
aese v2.16b, v22.16b |
| 121 |
aesmc v2.16b, v2.16b // AES block 2 - round 4 |
| 122 |
aese v0.16b, v22.16b |
| 123 |
aesmc v0.16b, v0.16b // AES block 0 - round 4 |
| 124 |
aese v1.16b, v22.16b |
| 125 |
aesmc v1.16b, v1.16b // AES block 1 - round 4 |
| 126 |
aese v3.16b, v22.16b |
| 127 |
aesmc v3.16b, v3.16b // AES block 3 - round 4 |
| 128 |
cmp x17, #12 // setup flags for AES-128/192/256 check |
| 129 |
aese v0.16b, v23.16b |
| 130 |
aesmc v0.16b, v0.16b // AES block 0 - round 5 |
| 131 |
aese v1.16b, v23.16b |
| 132 |
aesmc v1.16b, v1.16b // AES block 1 - round 5 |
| 133 |
aese v3.16b, v23.16b |
| 134 |
aesmc v3.16b, v3.16b // AES block 3 - round 5 |
| 135 |
aese v2.16b, v23.16b |
| 136 |
aesmc v2.16b, v2.16b // AES block 2 - round 5 |
| 137 |
aese v1.16b, v24.16b |
| 138 |
aesmc v1.16b, v1.16b // AES block 1 - round 6 |
| 139 |
trn2 v17.2d, v14.2d, v15.2d // h4l | h3l |
| 140 |
aese v3.16b, v24.16b |
| 141 |
aesmc v3.16b, v3.16b // AES block 3 - round 6 |
| 142 |
ldr q27, [x8, #144] // load rk9 |
| 143 |
aese v0.16b, v24.16b |
| 144 |
aesmc v0.16b, v0.16b // AES block 0 - round 6 |
| 145 |
ldr q12, [x3, #32] // load h1l | h1h |
| 146 |
ext v12.16b, v12.16b, v12.16b, #8 |
| 147 |
aese v2.16b, v24.16b |
| 148 |
aesmc v2.16b, v2.16b // AES block 2 - round 6 |
| 149 |
ldr q28, [x8, #160] // load rk10 |
| 150 |
aese v1.16b, v25.16b |
| 151 |
aesmc v1.16b, v1.16b // AES block 1 - round 7 |
| 152 |
trn1 v9.2d, v14.2d, v15.2d // h4h | h3h |
| 153 |
aese v0.16b, v25.16b |
| 154 |
aesmc v0.16b, v0.16b // AES block 0 - round 7 |
| 155 |
aese v2.16b, v25.16b |
| 156 |
aesmc v2.16b, v2.16b // AES block 2 - round 7 |
| 157 |
aese v3.16b, v25.16b |
| 158 |
aesmc v3.16b, v3.16b // AES block 3 - round 7 |
| 159 |
trn2 v16.2d, v12.2d, v13.2d // h2l | h1l |
| 160 |
aese v1.16b, v26.16b |
| 161 |
aesmc v1.16b, v1.16b // AES block 1 - round 8 |
| 162 |
aese v2.16b, v26.16b |
| 163 |
aesmc v2.16b, v2.16b // AES block 2 - round 8 |
| 164 |
aese v3.16b, v26.16b |
| 165 |
aesmc v3.16b, v3.16b // AES block 3 - round 8 |
| 166 |
aese v0.16b, v26.16b |
| 167 |
aesmc v0.16b, v0.16b // AES block 0 - round 8 |
| 168 |
b.lt Lenc_finish_first_blocks // branch if AES-128 |
| 169 |
|
| 170 |
aese v1.16b, v27.16b |
| 171 |
aesmc v1.16b, v1.16b // AES block 1 - round 9 |
| 172 |
aese v2.16b, v27.16b |
| 173 |
aesmc v2.16b, v2.16b // AES block 2 - round 9 |
| 174 |
aese v3.16b, v27.16b |
| 175 |
aesmc v3.16b, v3.16b // AES block 3 - round 9 |
| 176 |
aese v0.16b, v27.16b |
| 177 |
aesmc v0.16b, v0.16b // AES block 0 - round 9 |
| 178 |
aese v1.16b, v28.16b |
| 179 |
aesmc v1.16b, v1.16b // AES block 1 - round 10 |
| 180 |
aese v2.16b, v28.16b |
| 181 |
aesmc v2.16b, v2.16b // AES block 2 - round 10 |
| 182 |
aese v3.16b, v28.16b |
| 183 |
aesmc v3.16b, v3.16b // AES block 3 - round 10 |
| 184 |
aese v0.16b, v28.16b |
| 185 |
aesmc v0.16b, v0.16b // AES block 0 - round 10 |
| 186 |
b.eq Lenc_finish_first_blocks // branch if AES-192 |
| 187 |
|
| 188 |
aese v1.16b, v29.16b |
| 189 |
aesmc v1.16b, v1.16b // AES block 1 - round 11 |
| 190 |
aese v2.16b, v29.16b |
| 191 |
aesmc v2.16b, v2.16b // AES block 2 - round 11 |
| 192 |
aese v0.16b, v29.16b |
| 193 |
aesmc v0.16b, v0.16b // AES block 0 - round 11 |
| 194 |
aese v3.16b, v29.16b |
| 195 |
aesmc v3.16b, v3.16b // AES block 3 - round 11 |
| 196 |
aese v1.16b, v30.16b |
| 197 |
aesmc v1.16b, v1.16b // AES block 1 - round 12 |
| 198 |
aese v2.16b, v30.16b |
| 199 |
aesmc v2.16b, v2.16b // AES block 2 - round 12 |
| 200 |
aese v0.16b, v30.16b |
| 201 |
aesmc v0.16b, v0.16b // AES block 0 - round 12 |
| 202 |
aese v3.16b, v30.16b |
| 203 |
aesmc v3.16b, v3.16b // AES block 3 - round 12 |
| 204 |
|
| 205 |
Lenc_finish_first_blocks: |
| 206 |
cmp x0, x5 // check if we have <= 4 blocks |
| 207 |
eor v17.16b, v17.16b, v9.16b // h4k | h3k |
| 208 |
aese v2.16b, v31.16b // AES block 2 - round N-1 |
| 209 |
trn1 v8.2d, v12.2d, v13.2d // h2h | h1h |
| 210 |
aese v1.16b, v31.16b // AES block 1 - round N-1 |
| 211 |
aese v0.16b, v31.16b // AES block 0 - round N-1 |
| 212 |
aese v3.16b, v31.16b // AES block 3 - round N-1 |
| 213 |
eor v16.16b, v16.16b, v8.16b // h2k | h1k |
| 214 |
b.ge Lenc_tail // handle tail |
| 215 |
|
| 216 |
ldp x19, x20, [x0, #16] // AES block 1 - load plaintext |
| 217 |
rev w9, w12 // CTR block 4 |
| 218 |
ldp x6, x7, [x0, #0] // AES block 0 - load plaintext |
| 219 |
ldp x23, x24, [x0, #48] // AES block 3 - load plaintext |
| 220 |
ldp x21, x22, [x0, #32] // AES block 2 - load plaintext |
| 221 |
add x0, x0, #64 // AES input_ptr update |
| 222 |
eor x19, x19, x13 // AES block 1 - round N low |
| 223 |
eor x20, x20, x14 // AES block 1 - round N high |
| 224 |
fmov d5, x19 // AES block 1 - mov low |
| 225 |
eor x6, x6, x13 // AES block 0 - round N low |
| 226 |
eor x7, x7, x14 // AES block 0 - round N high |
| 227 |
eor x24, x24, x14 // AES block 3 - round N high |
| 228 |
fmov d4, x6 // AES block 0 - mov low |
| 229 |
cmp x0, x5 // check if we have <= 8 blocks |
| 230 |
fmov v4.d[1], x7 // AES block 0 - mov high |
| 231 |
eor x23, x23, x13 // AES block 3 - round N low |
| 232 |
eor x21, x21, x13 // AES block 2 - round N low |
| 233 |
fmov v5.d[1], x20 // AES block 1 - mov high |
| 234 |
fmov d6, x21 // AES block 2 - mov low |
| 235 |
add w12, w12, #1 // CTR block 4 |
| 236 |
orr x9, x11, x9, lsl #32 // CTR block 4 |
| 237 |
fmov d7, x23 // AES block 3 - mov low |
| 238 |
eor x22, x22, x14 // AES block 2 - round N high |
| 239 |
fmov v6.d[1], x22 // AES block 2 - mov high |
| 240 |
eor v4.16b, v4.16b, v0.16b // AES block 0 - result |
| 241 |
fmov d0, x10 // CTR block 4 |
| 242 |
fmov v0.d[1], x9 // CTR block 4 |
| 243 |
rev w9, w12 // CTR block 5 |
| 244 |
add w12, w12, #1 // CTR block 5 |
| 245 |
eor v5.16b, v5.16b, v1.16b // AES block 1 - result |
| 246 |
fmov d1, x10 // CTR block 5 |
| 247 |
orr x9, x11, x9, lsl #32 // CTR block 5 |
| 248 |
fmov v1.d[1], x9 // CTR block 5 |
| 249 |
rev w9, w12 // CTR block 6 |
| 250 |
st1 { v4.16b}, [x2], #16 // AES block 0 - store result |
| 251 |
fmov v7.d[1], x24 // AES block 3 - mov high |
| 252 |
orr x9, x11, x9, lsl #32 // CTR block 6 |
| 253 |
eor v6.16b, v6.16b, v2.16b // AES block 2 - result |
| 254 |
st1 { v5.16b}, [x2], #16 // AES block 1 - store result |
| 255 |
add w12, w12, #1 // CTR block 6 |
| 256 |
fmov d2, x10 // CTR block 6 |
| 257 |
fmov v2.d[1], x9 // CTR block 6 |
| 258 |
st1 { v6.16b}, [x2], #16 // AES block 2 - store result |
| 259 |
rev w9, w12 // CTR block 7 |
| 260 |
orr x9, x11, x9, lsl #32 // CTR block 7 |
| 261 |
eor v7.16b, v7.16b, v3.16b // AES block 3 - result |
| 262 |
st1 { v7.16b}, [x2], #16 // AES block 3 - store result |
| 263 |
b.ge Lenc_prepretail // do prepretail |
| 264 |
|
| 265 |
Lenc_main_loop: // main loop start |
| 266 |
aese v0.16b, v18.16b |
| 267 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 0 |
| 268 |
rev64 v4.16b, v4.16b // GHASH block 4k (only t0 is free) |
| 269 |
aese v1.16b, v18.16b |
| 270 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 0 |
| 271 |
fmov d3, x10 // CTR block 4k+3 |
| 272 |
aese v2.16b, v18.16b |
| 273 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 0 |
| 274 |
ext v11.16b, v11.16b, v11.16b, #8 // PRE 0 |
| 275 |
aese v0.16b, v19.16b |
| 276 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 1 |
| 277 |
fmov v3.d[1], x9 // CTR block 4k+3 |
| 278 |
aese v1.16b, v19.16b |
| 279 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 1 |
| 280 |
ldp x23, x24, [x0, #48] // AES block 4k+7 - load plaintext |
| 281 |
aese v2.16b, v19.16b |
| 282 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 1 |
| 283 |
ldp x21, x22, [x0, #32] // AES block 4k+6 - load plaintext |
| 284 |
aese v0.16b, v20.16b |
| 285 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 2 |
| 286 |
eor v4.16b, v4.16b, v11.16b // PRE 1 |
| 287 |
aese v1.16b, v20.16b |
| 288 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 2 |
| 289 |
aese v3.16b, v18.16b |
| 290 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 0 |
| 291 |
eor x23, x23, x13 // AES block 4k+7 - round N low |
| 292 |
aese v0.16b, v21.16b |
| 293 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 3 |
| 294 |
mov d10, v17.d[1] // GHASH block 4k - mid |
| 295 |
pmull2 v9.1q, v4.2d, v15.2d // GHASH block 4k - high |
| 296 |
eor x22, x22, x14 // AES block 4k+6 - round N high |
| 297 |
mov d8, v4.d[1] // GHASH block 4k - mid |
| 298 |
aese v3.16b, v19.16b |
| 299 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 1 |
| 300 |
rev64 v5.16b, v5.16b // GHASH block 4k+1 (t0 and t1 free) |
| 301 |
aese v0.16b, v22.16b |
| 302 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 4 |
| 303 |
pmull v11.1q, v4.1d, v15.1d // GHASH block 4k - low |
| 304 |
eor v8.8b, v8.8b, v4.8b // GHASH block 4k - mid |
| 305 |
aese v2.16b, v20.16b |
| 306 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 2 |
| 307 |
aese v0.16b, v23.16b |
| 308 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 5 |
| 309 |
rev64 v7.16b, v7.16b // GHASH block 4k+3 (t0, t1, t2 and t3 free) |
| 310 |
pmull2 v4.1q, v5.2d, v14.2d // GHASH block 4k+1 - high |
| 311 |
pmull v10.1q, v8.1d, v10.1d // GHASH block 4k - mid |
| 312 |
rev64 v6.16b, v6.16b // GHASH block 4k+2 (t0, t1, and t2 free) |
| 313 |
pmull v8.1q, v5.1d, v14.1d // GHASH block 4k+1 - low |
| 314 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+1 - high |
| 315 |
mov d4, v5.d[1] // GHASH block 4k+1 - mid |
| 316 |
aese v1.16b, v21.16b |
| 317 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 3 |
| 318 |
aese v3.16b, v20.16b |
| 319 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 2 |
| 320 |
eor v11.16b, v11.16b, v8.16b // GHASH block 4k+1 - low |
| 321 |
aese v2.16b, v21.16b |
| 322 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 3 |
| 323 |
aese v1.16b, v22.16b |
| 324 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 4 |
| 325 |
mov d8, v6.d[1] // GHASH block 4k+2 - mid |
| 326 |
aese v3.16b, v21.16b |
| 327 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 3 |
| 328 |
eor v4.8b, v4.8b, v5.8b // GHASH block 4k+1 - mid |
| 329 |
aese v2.16b, v22.16b |
| 330 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 4 |
| 331 |
aese v0.16b, v24.16b |
| 332 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 6 |
| 333 |
eor v8.8b, v8.8b, v6.8b // GHASH block 4k+2 - mid |
| 334 |
aese v3.16b, v22.16b |
| 335 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 4 |
| 336 |
pmull v4.1q, v4.1d, v17.1d // GHASH block 4k+1 - mid |
| 337 |
aese v0.16b, v25.16b |
| 338 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 7 |
| 339 |
aese v3.16b, v23.16b |
| 340 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 5 |
| 341 |
ins v8.d[1], v8.d[0] // GHASH block 4k+2 - mid |
| 342 |
aese v1.16b, v23.16b |
| 343 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 5 |
| 344 |
aese v0.16b, v26.16b |
| 345 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 8 |
| 346 |
aese v2.16b, v23.16b |
| 347 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 5 |
| 348 |
aese v1.16b, v24.16b |
| 349 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 6 |
| 350 |
eor v10.16b, v10.16b, v4.16b // GHASH block 4k+1 - mid |
| 351 |
pmull2 v4.1q, v6.2d, v13.2d // GHASH block 4k+2 - high |
| 352 |
pmull v5.1q, v6.1d, v13.1d // GHASH block 4k+2 - low |
| 353 |
aese v1.16b, v25.16b |
| 354 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 7 |
| 355 |
pmull v6.1q, v7.1d, v12.1d // GHASH block 4k+3 - low |
| 356 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+2 - high |
| 357 |
aese v3.16b, v24.16b |
| 358 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 6 |
| 359 |
ldp x19, x20, [x0, #16] // AES block 4k+5 - load plaintext |
| 360 |
aese v1.16b, v26.16b |
| 361 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 8 |
| 362 |
mov d4, v7.d[1] // GHASH block 4k+3 - mid |
| 363 |
aese v2.16b, v24.16b |
| 364 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 6 |
| 365 |
eor v11.16b, v11.16b, v5.16b // GHASH block 4k+2 - low |
| 366 |
pmull2 v8.1q, v8.2d, v16.2d // GHASH block 4k+2 - mid |
| 367 |
pmull2 v5.1q, v7.2d, v12.2d // GHASH block 4k+3 - high |
| 368 |
eor v4.8b, v4.8b, v7.8b // GHASH block 4k+3 - mid |
| 369 |
aese v2.16b, v25.16b |
| 370 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 7 |
| 371 |
eor x19, x19, x13 // AES block 4k+5 - round N low |
| 372 |
aese v2.16b, v26.16b |
| 373 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 8 |
| 374 |
eor v10.16b, v10.16b, v8.16b // GHASH block 4k+2 - mid |
| 375 |
aese v3.16b, v25.16b |
| 376 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 7 |
| 377 |
eor x21, x21, x13 // AES block 4k+6 - round N low |
| 378 |
aese v3.16b, v26.16b |
| 379 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 8 |
| 380 |
movi v8.8b, #0xc2 |
| 381 |
pmull v4.1q, v4.1d, v16.1d // GHASH block 4k+3 - mid |
| 382 |
eor v9.16b, v9.16b, v5.16b // GHASH block 4k+3 - high |
| 383 |
cmp x17, #12 // setup flags for AES-128/192/256 check |
| 384 |
fmov d5, x19 // AES block 4k+5 - mov low |
| 385 |
ldp x6, x7, [x0, #0] // AES block 4k+4 - load plaintext |
| 386 |
b.lt Lenc_main_loop_continue // branch if AES-128 |
| 387 |
|
| 388 |
aese v1.16b, v27.16b |
| 389 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 9 |
| 390 |
aese v0.16b, v27.16b |
| 391 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 9 |
| 392 |
aese v2.16b, v27.16b |
| 393 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 9 |
| 394 |
aese v3.16b, v27.16b |
| 395 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 9 |
| 396 |
aese v0.16b, v28.16b |
| 397 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 10 |
| 398 |
aese v1.16b, v28.16b |
| 399 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 10 |
| 400 |
aese v2.16b, v28.16b |
| 401 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 10 |
| 402 |
aese v3.16b, v28.16b |
| 403 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 10 |
| 404 |
b.eq Lenc_main_loop_continue // branch if AES-192 |
| 405 |
|
| 406 |
aese v0.16b, v29.16b |
| 407 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 11 |
| 408 |
aese v1.16b, v29.16b |
| 409 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 11 |
| 410 |
aese v2.16b, v29.16b |
| 411 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 11 |
| 412 |
aese v3.16b, v29.16b |
| 413 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 11 |
| 414 |
aese v1.16b, v30.16b |
| 415 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 12 |
| 416 |
aese v0.16b, v30.16b |
| 417 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 12 |
| 418 |
aese v2.16b, v30.16b |
| 419 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 12 |
| 420 |
aese v3.16b, v30.16b |
| 421 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 12 |
| 422 |
|
| 423 |
Lenc_main_loop_continue: |
| 424 |
shl d8, d8, #56 // mod_constant |
| 425 |
eor v11.16b, v11.16b, v6.16b // GHASH block 4k+3 - low |
| 426 |
eor v10.16b, v10.16b, v4.16b // GHASH block 4k+3 - mid |
| 427 |
add w12, w12, #1 // CTR block 4k+3 |
| 428 |
eor v4.16b, v11.16b, v9.16b // MODULO - karatsuba tidy up |
| 429 |
add x0, x0, #64 // AES input_ptr update |
| 430 |
pmull v7.1q, v9.1d, v8.1d // MODULO - top 64b align with mid |
| 431 |
rev w9, w12 // CTR block 4k+8 |
| 432 |
ext v9.16b, v9.16b, v9.16b, #8 // MODULO - other top alignment |
| 433 |
eor x6, x6, x13 // AES block 4k+4 - round N low |
| 434 |
eor v10.16b, v10.16b, v4.16b // MODULO - karatsuba tidy up |
| 435 |
eor x7, x7, x14 // AES block 4k+4 - round N high |
| 436 |
fmov d4, x6 // AES block 4k+4 - mov low |
| 437 |
orr x9, x11, x9, lsl #32 // CTR block 4k+8 |
| 438 |
eor v7.16b, v9.16b, v7.16b // MODULO - fold into mid |
| 439 |
eor x20, x20, x14 // AES block 4k+5 - round N high |
| 440 |
eor x24, x24, x14 // AES block 4k+7 - round N high |
| 441 |
add w12, w12, #1 // CTR block 4k+8 |
| 442 |
aese v0.16b, v31.16b // AES block 4k+4 - round N-1 |
| 443 |
fmov v4.d[1], x7 // AES block 4k+4 - mov high |
| 444 |
eor v10.16b, v10.16b, v7.16b // MODULO - fold into mid |
| 445 |
fmov d7, x23 // AES block 4k+7 - mov low |
| 446 |
aese v1.16b, v31.16b // AES block 4k+5 - round N-1 |
| 447 |
fmov v5.d[1], x20 // AES block 4k+5 - mov high |
| 448 |
fmov d6, x21 // AES block 4k+6 - mov low |
| 449 |
cmp x0, x5 // LOOP CONTROL |
| 450 |
fmov v6.d[1], x22 // AES block 4k+6 - mov high |
| 451 |
pmull v9.1q, v10.1d, v8.1d // MODULO - mid 64b align with low |
| 452 |
eor v4.16b, v4.16b, v0.16b // AES block 4k+4 - result |
| 453 |
fmov d0, x10 // CTR block 4k+8 |
| 454 |
fmov v0.d[1], x9 // CTR block 4k+8 |
| 455 |
rev w9, w12 // CTR block 4k+9 |
| 456 |
add w12, w12, #1 // CTR block 4k+9 |
| 457 |
eor v5.16b, v5.16b, v1.16b // AES block 4k+5 - result |
| 458 |
fmov d1, x10 // CTR block 4k+9 |
| 459 |
orr x9, x11, x9, lsl #32 // CTR block 4k+9 |
| 460 |
fmov v1.d[1], x9 // CTR block 4k+9 |
| 461 |
aese v2.16b, v31.16b // AES block 4k+6 - round N-1 |
| 462 |
rev w9, w12 // CTR block 4k+10 |
| 463 |
st1 { v4.16b}, [x2], #16 // AES block 4k+4 - store result |
| 464 |
orr x9, x11, x9, lsl #32 // CTR block 4k+10 |
| 465 |
eor v11.16b, v11.16b, v9.16b // MODULO - fold into low |
| 466 |
fmov v7.d[1], x24 // AES block 4k+7 - mov high |
| 467 |
ext v10.16b, v10.16b, v10.16b, #8 // MODULO - other mid alignment |
| 468 |
st1 { v5.16b}, [x2], #16 // AES block 4k+5 - store result |
| 469 |
add w12, w12, #1 // CTR block 4k+10 |
| 470 |
aese v3.16b, v31.16b // AES block 4k+7 - round N-1 |
| 471 |
eor v6.16b, v6.16b, v2.16b // AES block 4k+6 - result |
| 472 |
fmov d2, x10 // CTR block 4k+10 |
| 473 |
st1 { v6.16b}, [x2], #16 // AES block 4k+6 - store result |
| 474 |
fmov v2.d[1], x9 // CTR block 4k+10 |
| 475 |
rev w9, w12 // CTR block 4k+11 |
| 476 |
eor v11.16b, v11.16b, v10.16b // MODULO - fold into low |
| 477 |
orr x9, x11, x9, lsl #32 // CTR block 4k+11 |
| 478 |
eor v7.16b, v7.16b, v3.16b // AES block 4k+7 - result |
| 479 |
st1 { v7.16b}, [x2], #16 // AES block 4k+7 - store result |
| 480 |
b.lt Lenc_main_loop |
| 481 |
|
| 482 |
Lenc_prepretail: // PREPRETAIL |
| 483 |
aese v1.16b, v18.16b |
| 484 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 0 |
| 485 |
rev64 v6.16b, v6.16b // GHASH block 4k+2 (t0, t1, and t2 free) |
| 486 |
aese v2.16b, v18.16b |
| 487 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 0 |
| 488 |
fmov d3, x10 // CTR block 4k+3 |
| 489 |
aese v0.16b, v18.16b |
| 490 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 0 |
| 491 |
rev64 v4.16b, v4.16b // GHASH block 4k (only t0 is free) |
| 492 |
fmov v3.d[1], x9 // CTR block 4k+3 |
| 493 |
ext v11.16b, v11.16b, v11.16b, #8 // PRE 0 |
| 494 |
aese v2.16b, v19.16b |
| 495 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 1 |
| 496 |
aese v0.16b, v19.16b |
| 497 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 1 |
| 498 |
eor v4.16b, v4.16b, v11.16b // PRE 1 |
| 499 |
rev64 v5.16b, v5.16b // GHASH block 4k+1 (t0 and t1 free) |
| 500 |
aese v2.16b, v20.16b |
| 501 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 2 |
| 502 |
aese v3.16b, v18.16b |
| 503 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 0 |
| 504 |
mov d10, v17.d[1] // GHASH block 4k - mid |
| 505 |
aese v1.16b, v19.16b |
| 506 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 1 |
| 507 |
pmull v11.1q, v4.1d, v15.1d // GHASH block 4k - low |
| 508 |
mov d8, v4.d[1] // GHASH block 4k - mid |
| 509 |
pmull2 v9.1q, v4.2d, v15.2d // GHASH block 4k - high |
| 510 |
aese v2.16b, v21.16b |
| 511 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 3 |
| 512 |
aese v1.16b, v20.16b |
| 513 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 2 |
| 514 |
eor v8.8b, v8.8b, v4.8b // GHASH block 4k - mid |
| 515 |
aese v0.16b, v20.16b |
| 516 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 2 |
| 517 |
aese v3.16b, v19.16b |
| 518 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 1 |
| 519 |
aese v1.16b, v21.16b |
| 520 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 3 |
| 521 |
pmull v10.1q, v8.1d, v10.1d // GHASH block 4k - mid |
| 522 |
pmull2 v4.1q, v5.2d, v14.2d // GHASH block 4k+1 - high |
| 523 |
pmull v8.1q, v5.1d, v14.1d // GHASH block 4k+1 - low |
| 524 |
aese v3.16b, v20.16b |
| 525 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 2 |
| 526 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+1 - high |
| 527 |
mov d4, v5.d[1] // GHASH block 4k+1 - mid |
| 528 |
aese v0.16b, v21.16b |
| 529 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 3 |
| 530 |
eor v11.16b, v11.16b, v8.16b // GHASH block 4k+1 - low |
| 531 |
aese v3.16b, v21.16b |
| 532 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 3 |
| 533 |
eor v4.8b, v4.8b, v5.8b // GHASH block 4k+1 - mid |
| 534 |
mov d8, v6.d[1] // GHASH block 4k+2 - mid |
| 535 |
aese v0.16b, v22.16b |
| 536 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 4 |
| 537 |
rev64 v7.16b, v7.16b // GHASH block 4k+3 (t0, t1, t2 and t3 free) |
| 538 |
aese v3.16b, v22.16b |
| 539 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 4 |
| 540 |
pmull v4.1q, v4.1d, v17.1d // GHASH block 4k+1 - mid |
| 541 |
eor v8.8b, v8.8b, v6.8b // GHASH block 4k+2 - mid |
| 542 |
add w12, w12, #1 // CTR block 4k+3 |
| 543 |
pmull v5.1q, v6.1d, v13.1d // GHASH block 4k+2 - low |
| 544 |
aese v3.16b, v23.16b |
| 545 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 5 |
| 546 |
aese v2.16b, v22.16b |
| 547 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 4 |
| 548 |
eor v10.16b, v10.16b, v4.16b // GHASH block 4k+1 - mid |
| 549 |
pmull2 v4.1q, v6.2d, v13.2d // GHASH block 4k+2 - high |
| 550 |
eor v11.16b, v11.16b, v5.16b // GHASH block 4k+2 - low |
| 551 |
ins v8.d[1], v8.d[0] // GHASH block 4k+2 - mid |
| 552 |
aese v2.16b, v23.16b |
| 553 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 5 |
| 554 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+2 - high |
| 555 |
mov d4, v7.d[1] // GHASH block 4k+3 - mid |
| 556 |
aese v1.16b, v22.16b |
| 557 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 4 |
| 558 |
pmull2 v8.1q, v8.2d, v16.2d // GHASH block 4k+2 - mid |
| 559 |
eor v4.8b, v4.8b, v7.8b // GHASH block 4k+3 - mid |
| 560 |
pmull2 v5.1q, v7.2d, v12.2d // GHASH block 4k+3 - high |
| 561 |
aese v1.16b, v23.16b |
| 562 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 5 |
| 563 |
pmull v4.1q, v4.1d, v16.1d // GHASH block 4k+3 - mid |
| 564 |
eor v10.16b, v10.16b, v8.16b // GHASH block 4k+2 - mid |
| 565 |
aese v0.16b, v23.16b |
| 566 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 5 |
| 567 |
aese v1.16b, v24.16b |
| 568 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 6 |
| 569 |
aese v2.16b, v24.16b |
| 570 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 6 |
| 571 |
aese v0.16b, v24.16b |
| 572 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 6 |
| 573 |
movi v8.8b, #0xc2 |
| 574 |
aese v3.16b, v24.16b |
| 575 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 6 |
| 576 |
aese v1.16b, v25.16b |
| 577 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 7 |
| 578 |
eor v9.16b, v9.16b, v5.16b // GHASH block 4k+3 - high |
| 579 |
aese v0.16b, v25.16b |
| 580 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 7 |
| 581 |
aese v3.16b, v25.16b |
| 582 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 7 |
| 583 |
shl d8, d8, #56 // mod_constant |
| 584 |
aese v1.16b, v26.16b |
| 585 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 8 |
| 586 |
eor v10.16b, v10.16b, v4.16b // GHASH block 4k+3 - mid |
| 587 |
pmull v6.1q, v7.1d, v12.1d // GHASH block 4k+3 - low |
| 588 |
aese v3.16b, v26.16b |
| 589 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 8 |
| 590 |
cmp x17, #12 // setup flags for AES-128/192/256 check |
| 591 |
aese v0.16b, v26.16b |
| 592 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 8 |
| 593 |
eor v11.16b, v11.16b, v6.16b // GHASH block 4k+3 - low |
| 594 |
aese v2.16b, v25.16b |
| 595 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 7 |
| 596 |
eor v10.16b, v10.16b, v9.16b // karatsuba tidy up |
| 597 |
aese v2.16b, v26.16b |
| 598 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 8 |
| 599 |
pmull v4.1q, v9.1d, v8.1d |
| 600 |
ext v9.16b, v9.16b, v9.16b, #8 |
| 601 |
eor v10.16b, v10.16b, v11.16b |
| 602 |
b.lt Lenc_finish_prepretail // branch if AES-128 |
| 603 |
|
| 604 |
aese v1.16b, v27.16b |
| 605 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 9 |
| 606 |
aese v3.16b, v27.16b |
| 607 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 9 |
| 608 |
aese v0.16b, v27.16b |
| 609 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 9 |
| 610 |
aese v2.16b, v27.16b |
| 611 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 9 |
| 612 |
aese v3.16b, v28.16b |
| 613 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 10 |
| 614 |
aese v1.16b, v28.16b |
| 615 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 10 |
| 616 |
aese v0.16b, v28.16b |
| 617 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 10 |
| 618 |
aese v2.16b, v28.16b |
| 619 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 10 |
| 620 |
b.eq Lenc_finish_prepretail // branch if AES-192 |
| 621 |
|
| 622 |
aese v1.16b, v29.16b |
| 623 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 11 |
| 624 |
aese v0.16b, v29.16b |
| 625 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 11 |
| 626 |
aese v3.16b, v29.16b |
| 627 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 11 |
| 628 |
aese v2.16b, v29.16b |
| 629 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 11 |
| 630 |
aese v1.16b, v30.16b |
| 631 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 12 |
| 632 |
aese v0.16b, v30.16b |
| 633 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 12 |
| 634 |
aese v3.16b, v30.16b |
| 635 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 12 |
| 636 |
aese v2.16b, v30.16b |
| 637 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 12 |
| 638 |
|
| 639 |
Lenc_finish_prepretail: |
| 640 |
eor v10.16b, v10.16b, v4.16b |
| 641 |
eor v10.16b, v10.16b, v9.16b |
| 642 |
pmull v4.1q, v10.1d, v8.1d |
| 643 |
ext v10.16b, v10.16b, v10.16b, #8 |
| 644 |
aese v1.16b, v31.16b // AES block 4k+5 - round N-1 |
| 645 |
eor v11.16b, v11.16b, v4.16b |
| 646 |
aese v3.16b, v31.16b // AES block 4k+7 - round N-1 |
| 647 |
aese v0.16b, v31.16b // AES block 4k+4 - round N-1 |
| 648 |
aese v2.16b, v31.16b // AES block 4k+6 - round N-1 |
| 649 |
eor v11.16b, v11.16b, v10.16b |
| 650 |
|
| 651 |
Lenc_tail: // TAIL |
| 652 |
ext v8.16b, v11.16b, v11.16b, #8 // prepare final partial tag |
| 653 |
sub x5, x4, x0 // main_end_input_ptr is number of bytes left to process |
| 654 |
ldp x6, x7, [x0], #16 // AES block 4k+4 - load plaintext |
| 655 |
eor x6, x6, x13 // AES block 4k+4 - round N low |
| 656 |
eor x7, x7, x14 // AES block 4k+4 - round N high |
| 657 |
cmp x5, #48 |
| 658 |
fmov d4, x6 // AES block 4k+4 - mov low |
| 659 |
fmov v4.d[1], x7 // AES block 4k+4 - mov high |
| 660 |
eor v5.16b, v4.16b, v0.16b // AES block 4k+4 - result |
| 661 |
b.gt Lenc_blocks_more_than_3 |
| 662 |
cmp x5, #32 |
| 663 |
mov v3.16b, v2.16b |
| 664 |
movi v11.8b, #0 |
| 665 |
movi v9.8b, #0 |
| 666 |
sub w12, w12, #1 |
| 667 |
mov v2.16b, v1.16b |
| 668 |
movi v10.8b, #0 |
| 669 |
b.gt Lenc_blocks_more_than_2 |
| 670 |
mov v3.16b, v1.16b |
| 671 |
sub w12, w12, #1 |
| 672 |
cmp x5, #16 |
| 673 |
b.gt Lenc_blocks_more_than_1 |
| 674 |
sub w12, w12, #1 |
| 675 |
b Lenc_blocks_less_than_1 |
| 676 |
Lenc_blocks_more_than_3: // blocks left > 3 |
| 677 |
st1 { v5.16b}, [x2], #16 // AES final-3 block - store result |
| 678 |
ldp x6, x7, [x0], #16 // AES final-2 block - load input low & high |
| 679 |
rev64 v4.16b, v5.16b // GHASH final-3 block |
| 680 |
eor x6, x6, x13 // AES final-2 block - round N low |
| 681 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 682 |
eor x7, x7, x14 // AES final-2 block - round N high |
| 683 |
mov d22, v4.d[1] // GHASH final-3 block - mid |
| 684 |
fmov d5, x6 // AES final-2 block - mov low |
| 685 |
fmov v5.d[1], x7 // AES final-2 block - mov high |
| 686 |
eor v22.8b, v22.8b, v4.8b // GHASH final-3 block - mid |
| 687 |
movi v8.8b, #0 // suppress further partial tag feed in |
| 688 |
mov d10, v17.d[1] // GHASH final-3 block - mid |
| 689 |
pmull v11.1q, v4.1d, v15.1d // GHASH final-3 block - low |
| 690 |
pmull2 v9.1q, v4.2d, v15.2d // GHASH final-3 block - high |
| 691 |
pmull v10.1q, v22.1d, v10.1d // GHASH final-3 block - mid |
| 692 |
eor v5.16b, v5.16b, v1.16b // AES final-2 block - result |
| 693 |
Lenc_blocks_more_than_2: // blocks left > 2 |
| 694 |
st1 { v5.16b}, [x2], #16 // AES final-2 block - store result |
| 695 |
ldp x6, x7, [x0], #16 // AES final-1 block - load input low & high |
| 696 |
rev64 v4.16b, v5.16b // GHASH final-2 block |
| 697 |
eor x6, x6, x13 // AES final-1 block - round N low |
| 698 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 699 |
fmov d5, x6 // AES final-1 block - mov low |
| 700 |
eor x7, x7, x14 // AES final-1 block - round N high |
| 701 |
fmov v5.d[1], x7 // AES final-1 block - mov high |
| 702 |
movi v8.8b, #0 // suppress further partial tag feed in |
| 703 |
pmull2 v20.1q, v4.2d, v14.2d // GHASH final-2 block - high |
| 704 |
mov d22, v4.d[1] // GHASH final-2 block - mid |
| 705 |
pmull v21.1q, v4.1d, v14.1d // GHASH final-2 block - low |
| 706 |
eor v22.8b, v22.8b, v4.8b // GHASH final-2 block - mid |
| 707 |
eor v5.16b, v5.16b, v2.16b // AES final-1 block - result |
| 708 |
eor v9.16b, v9.16b, v20.16b // GHASH final-2 block - high |
| 709 |
pmull v22.1q, v22.1d, v17.1d // GHASH final-2 block - mid |
| 710 |
eor v11.16b, v11.16b, v21.16b // GHASH final-2 block - low |
| 711 |
eor v10.16b, v10.16b, v22.16b // GHASH final-2 block - mid |
| 712 |
Lenc_blocks_more_than_1: // blocks left > 1 |
| 713 |
st1 { v5.16b}, [x2], #16 // AES final-1 block - store result |
| 714 |
rev64 v4.16b, v5.16b // GHASH final-1 block |
| 715 |
ldp x6, x7, [x0], #16 // AES final block - load input low & high |
| 716 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 717 |
movi v8.8b, #0 // suppress further partial tag feed in |
| 718 |
eor x6, x6, x13 // AES final block - round N low |
| 719 |
mov d22, v4.d[1] // GHASH final-1 block - mid |
| 720 |
pmull2 v20.1q, v4.2d, v13.2d // GHASH final-1 block - high |
| 721 |
eor x7, x7, x14 // AES final block - round N high |
| 722 |
eor v22.8b, v22.8b, v4.8b // GHASH final-1 block - mid |
| 723 |
eor v9.16b, v9.16b, v20.16b // GHASH final-1 block - high |
| 724 |
ins v22.d[1], v22.d[0] // GHASH final-1 block - mid |
| 725 |
fmov d5, x6 // AES final block - mov low |
| 726 |
fmov v5.d[1], x7 // AES final block - mov high |
| 727 |
pmull2 v22.1q, v22.2d, v16.2d // GHASH final-1 block - mid |
| 728 |
pmull v21.1q, v4.1d, v13.1d // GHASH final-1 block - low |
| 729 |
eor v5.16b, v5.16b, v3.16b // AES final block - result |
| 730 |
eor v10.16b, v10.16b, v22.16b // GHASH final-1 block - mid |
| 731 |
eor v11.16b, v11.16b, v21.16b // GHASH final-1 block - low |
| 732 |
Lenc_blocks_less_than_1: // blocks left <= 1 |
| 733 |
and x1, x1, #127 // bit_length %= 128 |
| 734 |
mvn x13, xzr // rkN_l = 0xffffffffffffffff |
| 735 |
sub x1, x1, #128 // bit_length -= 128 |
| 736 |
neg x1, x1 // bit_length = 128 - #bits in input (in range [1,128]) |
| 737 |
ld1 { v18.16b}, [x2] // load existing bytes where the possibly partial last block is to be stored |
| 738 |
mvn x14, xzr // rkN_h = 0xffffffffffffffff |
| 739 |
and x1, x1, #127 // bit_length %= 128 |
| 740 |
lsr x14, x14, x1 // rkN_h is mask for top 64b of last block |
| 741 |
cmp x1, #64 |
| 742 |
csel x6, x13, x14, lt |
| 743 |
csel x7, x14, xzr, lt |
| 744 |
fmov d0, x6 // ctr0b is mask for last block |
| 745 |
fmov v0.d[1], x7 |
| 746 |
and v5.16b, v5.16b, v0.16b // possibly partial last block has zeroes in highest bits |
| 747 |
rev64 v4.16b, v5.16b // GHASH final block |
| 748 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 749 |
bif v5.16b, v18.16b, v0.16b // insert existing bytes in top end of result before storing |
| 750 |
pmull2 v20.1q, v4.2d, v12.2d // GHASH final block - high |
| 751 |
mov d8, v4.d[1] // GHASH final block - mid |
| 752 |
rev w9, w12 |
| 753 |
pmull v21.1q, v4.1d, v12.1d // GHASH final block - low |
| 754 |
eor v9.16b, v9.16b, v20.16b // GHASH final block - high |
| 755 |
eor v8.8b, v8.8b, v4.8b // GHASH final block - mid |
| 756 |
pmull v8.1q, v8.1d, v16.1d // GHASH final block - mid |
| 757 |
eor v11.16b, v11.16b, v21.16b // GHASH final block - low |
| 758 |
eor v10.16b, v10.16b, v8.16b // GHASH final block - mid |
| 759 |
movi v8.8b, #0xc2 |
| 760 |
eor v4.16b, v11.16b, v9.16b // MODULO - karatsuba tidy up |
| 761 |
shl d8, d8, #56 // mod_constant |
| 762 |
eor v10.16b, v10.16b, v4.16b // MODULO - karatsuba tidy up |
| 763 |
pmull v7.1q, v9.1d, v8.1d // MODULO - top 64b align with mid |
| 764 |
ext v9.16b, v9.16b, v9.16b, #8 // MODULO - other top alignment |
| 765 |
eor v10.16b, v10.16b, v7.16b // MODULO - fold into mid |
| 766 |
eor v10.16b, v10.16b, v9.16b // MODULO - fold into mid |
| 767 |
pmull v9.1q, v10.1d, v8.1d // MODULO - mid 64b align with low |
| 768 |
ext v10.16b, v10.16b, v10.16b, #8 // MODULO - other mid alignment |
| 769 |
str w9, [x16, #12] // store the updated counter |
| 770 |
st1 { v5.16b}, [x2] // store all 16B |
| 771 |
eor v11.16b, v11.16b, v9.16b // MODULO - fold into low |
| 772 |
eor v11.16b, v11.16b, v10.16b // MODULO - fold into low |
| 773 |
ext v11.16b, v11.16b, v11.16b, #8 |
| 774 |
rev64 v11.16b, v11.16b |
| 775 |
mov x0, x15 |
| 776 |
st1 { v11.16b }, [x3] |
| 777 |
ldp x19, x20, [sp, #16] |
| 778 |
ldp x21, x22, [sp, #32] |
| 779 |
ldp x23, x24, [sp, #48] |
| 780 |
ldp d8, d9, [sp, #64] |
| 781 |
ldp d10, d11, [sp, #80] |
| 782 |
ldp d12, d13, [sp, #96] |
| 783 |
ldp d14, d15, [sp, #112] |
| 784 |
ldp x29, x30, [sp], #128 |
| 785 |
AARCH64_VALIDATE_LINK_REGISTER |
| 786 |
ret |
| 787 |
|
| 788 |
.globl _aes_gcm_dec_kernel |
| 789 |
.private_extern _aes_gcm_dec_kernel |
| 790 |
|
| 791 |
.align 4 |
| 792 |
_aes_gcm_dec_kernel: |
| 793 |
AARCH64_SIGN_LINK_REGISTER |
| 794 |
stp x29, x30, [sp, #-128]! |
| 795 |
mov x29, sp |
| 796 |
stp x19, x20, [sp, #16] |
| 797 |
mov x16, x4 |
| 798 |
mov x8, x5 |
| 799 |
stp x21, x22, [sp, #32] |
| 800 |
stp x23, x24, [sp, #48] |
| 801 |
stp d8, d9, [sp, #64] |
| 802 |
stp d10, d11, [sp, #80] |
| 803 |
stp d12, d13, [sp, #96] |
| 804 |
stp d14, d15, [sp, #112] |
| 805 |
ldr w17, [x8, #240] |
| 806 |
add x19, x8, x17, lsl #4 // borrow input_l1 for last key |
| 807 |
ldp x13, x14, [x19] // load round N keys |
| 808 |
ldr q31, [x19, #-16] // load round N-1 keys |
| 809 |
lsr x5, x1, #3 // byte_len |
| 810 |
mov x15, x5 |
| 811 |
ldp x10, x11, [x16] // ctr96_b64, ctr96_t32 |
| 812 |
ldr q26, [x8, #128] // load rk8 |
| 813 |
sub x5, x5, #1 // byte_len - 1 |
| 814 |
ldr q25, [x8, #112] // load rk7 |
| 815 |
and x5, x5, #0xffffffffffffffc0 // number of bytes to be processed in main loop (at least 1 byte must be handled by tail) |
| 816 |
add x4, x0, x1, lsr #3 // end_input_ptr |
| 817 |
ldr q24, [x8, #96] // load rk6 |
| 818 |
lsr x12, x11, #32 |
| 819 |
ldr q23, [x8, #80] // load rk5 |
| 820 |
orr w11, w11, w11 |
| 821 |
ldr q21, [x8, #48] // load rk3 |
| 822 |
add x5, x5, x0 |
| 823 |
rev w12, w12 // rev_ctr32 |
| 824 |
add w12, w12, #1 // increment rev_ctr32 |
| 825 |
fmov d3, x10 // CTR block 3 |
| 826 |
rev w9, w12 // CTR block 1 |
| 827 |
add w12, w12, #1 // CTR block 1 |
| 828 |
fmov d1, x10 // CTR block 1 |
| 829 |
orr x9, x11, x9, lsl #32 // CTR block 1 |
| 830 |
ld1 { v0.16b}, [x16] // special case vector load initial counter so we can start first AES block as quickly as possible |
| 831 |
fmov v1.d[1], x9 // CTR block 1 |
| 832 |
rev w9, w12 // CTR block 2 |
| 833 |
add w12, w12, #1 // CTR block 2 |
| 834 |
fmov d2, x10 // CTR block 2 |
| 835 |
orr x9, x11, x9, lsl #32 // CTR block 2 |
| 836 |
fmov v2.d[1], x9 // CTR block 2 |
| 837 |
rev w9, w12 // CTR block 3 |
| 838 |
orr x9, x11, x9, lsl #32 // CTR block 3 |
| 839 |
ldr q18, [x8, #0] // load rk0 |
| 840 |
fmov v3.d[1], x9 // CTR block 3 |
| 841 |
add w12, w12, #1 // CTR block 3 |
| 842 |
ldr q22, [x8, #64] // load rk4 |
| 843 |
ldr q19, [x8, #16] // load rk1 |
| 844 |
aese v0.16b, v18.16b |
| 845 |
aesmc v0.16b, v0.16b // AES block 0 - round 0 |
| 846 |
ldr q14, [x3, #80] // load h3l | h3h |
| 847 |
ext v14.16b, v14.16b, v14.16b, #8 |
| 848 |
aese v3.16b, v18.16b |
| 849 |
aesmc v3.16b, v3.16b // AES block 3 - round 0 |
| 850 |
ldr q15, [x3, #112] // load h4l | h4h |
| 851 |
ext v15.16b, v15.16b, v15.16b, #8 |
| 852 |
aese v1.16b, v18.16b |
| 853 |
aesmc v1.16b, v1.16b // AES block 1 - round 0 |
| 854 |
ldr q13, [x3, #64] // load h2l | h2h |
| 855 |
ext v13.16b, v13.16b, v13.16b, #8 |
| 856 |
aese v2.16b, v18.16b |
| 857 |
aesmc v2.16b, v2.16b // AES block 2 - round 0 |
| 858 |
ldr q20, [x8, #32] // load rk2 |
| 859 |
aese v0.16b, v19.16b |
| 860 |
aesmc v0.16b, v0.16b // AES block 0 - round 1 |
| 861 |
aese v1.16b, v19.16b |
| 862 |
aesmc v1.16b, v1.16b // AES block 1 - round 1 |
| 863 |
ld1 { v11.16b}, [x3] |
| 864 |
ext v11.16b, v11.16b, v11.16b, #8 |
| 865 |
rev64 v11.16b, v11.16b |
| 866 |
aese v2.16b, v19.16b |
| 867 |
aesmc v2.16b, v2.16b // AES block 2 - round 1 |
| 868 |
ldr q27, [x8, #144] // load rk9 |
| 869 |
aese v3.16b, v19.16b |
| 870 |
aesmc v3.16b, v3.16b // AES block 3 - round 1 |
| 871 |
ldr q30, [x8, #192] // load rk12 |
| 872 |
aese v0.16b, v20.16b |
| 873 |
aesmc v0.16b, v0.16b // AES block 0 - round 2 |
| 874 |
ldr q12, [x3, #32] // load h1l | h1h |
| 875 |
ext v12.16b, v12.16b, v12.16b, #8 |
| 876 |
aese v2.16b, v20.16b |
| 877 |
aesmc v2.16b, v2.16b // AES block 2 - round 2 |
| 878 |
ldr q28, [x8, #160] // load rk10 |
| 879 |
aese v3.16b, v20.16b |
| 880 |
aesmc v3.16b, v3.16b // AES block 3 - round 2 |
| 881 |
aese v0.16b, v21.16b |
| 882 |
aesmc v0.16b, v0.16b // AES block 0 - round 3 |
| 883 |
aese v1.16b, v20.16b |
| 884 |
aesmc v1.16b, v1.16b // AES block 1 - round 2 |
| 885 |
aese v3.16b, v21.16b |
| 886 |
aesmc v3.16b, v3.16b // AES block 3 - round 3 |
| 887 |
aese v0.16b, v22.16b |
| 888 |
aesmc v0.16b, v0.16b // AES block 0 - round 4 |
| 889 |
aese v2.16b, v21.16b |
| 890 |
aesmc v2.16b, v2.16b // AES block 2 - round 3 |
| 891 |
aese v1.16b, v21.16b |
| 892 |
aesmc v1.16b, v1.16b // AES block 1 - round 3 |
| 893 |
aese v3.16b, v22.16b |
| 894 |
aesmc v3.16b, v3.16b // AES block 3 - round 4 |
| 895 |
aese v2.16b, v22.16b |
| 896 |
aesmc v2.16b, v2.16b // AES block 2 - round 4 |
| 897 |
aese v1.16b, v22.16b |
| 898 |
aesmc v1.16b, v1.16b // AES block 1 - round 4 |
| 899 |
aese v3.16b, v23.16b |
| 900 |
aesmc v3.16b, v3.16b // AES block 3 - round 5 |
| 901 |
aese v0.16b, v23.16b |
| 902 |
aesmc v0.16b, v0.16b // AES block 0 - round 5 |
| 903 |
aese v1.16b, v23.16b |
| 904 |
aesmc v1.16b, v1.16b // AES block 1 - round 5 |
| 905 |
aese v2.16b, v23.16b |
| 906 |
aesmc v2.16b, v2.16b // AES block 2 - round 5 |
| 907 |
aese v0.16b, v24.16b |
| 908 |
aesmc v0.16b, v0.16b // AES block 0 - round 6 |
| 909 |
aese v3.16b, v24.16b |
| 910 |
aesmc v3.16b, v3.16b // AES block 3 - round 6 |
| 911 |
cmp x17, #12 // setup flags for AES-128/192/256 check |
| 912 |
aese v1.16b, v24.16b |
| 913 |
aesmc v1.16b, v1.16b // AES block 1 - round 6 |
| 914 |
aese v2.16b, v24.16b |
| 915 |
aesmc v2.16b, v2.16b // AES block 2 - round 6 |
| 916 |
aese v0.16b, v25.16b |
| 917 |
aesmc v0.16b, v0.16b // AES block 0 - round 7 |
| 918 |
aese v1.16b, v25.16b |
| 919 |
aesmc v1.16b, v1.16b // AES block 1 - round 7 |
| 920 |
aese v3.16b, v25.16b |
| 921 |
aesmc v3.16b, v3.16b // AES block 3 - round 7 |
| 922 |
aese v0.16b, v26.16b |
| 923 |
aesmc v0.16b, v0.16b // AES block 0 - round 8 |
| 924 |
aese v2.16b, v25.16b |
| 925 |
aesmc v2.16b, v2.16b // AES block 2 - round 7 |
| 926 |
aese v3.16b, v26.16b |
| 927 |
aesmc v3.16b, v3.16b // AES block 3 - round 8 |
| 928 |
aese v1.16b, v26.16b |
| 929 |
aesmc v1.16b, v1.16b // AES block 1 - round 8 |
| 930 |
ldr q29, [x8, #176] // load rk11 |
| 931 |
aese v2.16b, v26.16b |
| 932 |
aesmc v2.16b, v2.16b // AES block 2 - round 8 |
| 933 |
b.lt Ldec_finish_first_blocks // branch if AES-128 |
| 934 |
|
| 935 |
aese v0.16b, v27.16b |
| 936 |
aesmc v0.16b, v0.16b // AES block 0 - round 9 |
| 937 |
aese v1.16b, v27.16b |
| 938 |
aesmc v1.16b, v1.16b // AES block 1 - round 9 |
| 939 |
aese v3.16b, v27.16b |
| 940 |
aesmc v3.16b, v3.16b // AES block 3 - round 9 |
| 941 |
aese v2.16b, v27.16b |
| 942 |
aesmc v2.16b, v2.16b // AES block 2 - round 9 |
| 943 |
aese v0.16b, v28.16b |
| 944 |
aesmc v0.16b, v0.16b // AES block 0 - round 10 |
| 945 |
aese v1.16b, v28.16b |
| 946 |
aesmc v1.16b, v1.16b // AES block 1 - round 10 |
| 947 |
aese v3.16b, v28.16b |
| 948 |
aesmc v3.16b, v3.16b // AES block 3 - round 10 |
| 949 |
aese v2.16b, v28.16b |
| 950 |
aesmc v2.16b, v2.16b // AES block 2 - round 10 |
| 951 |
b.eq Ldec_finish_first_blocks // branch if AES-192 |
| 952 |
|
| 953 |
aese v0.16b, v29.16b |
| 954 |
aesmc v0.16b, v0.16b // AES block 0 - round 11 |
| 955 |
aese v3.16b, v29.16b |
| 956 |
aesmc v3.16b, v3.16b // AES block 3 - round 11 |
| 957 |
aese v1.16b, v29.16b |
| 958 |
aesmc v1.16b, v1.16b // AES block 1 - round 11 |
| 959 |
aese v2.16b, v29.16b |
| 960 |
aesmc v2.16b, v2.16b // AES block 2 - round 11 |
| 961 |
aese v1.16b, v30.16b |
| 962 |
aesmc v1.16b, v1.16b // AES block 1 - round 12 |
| 963 |
aese v0.16b, v30.16b |
| 964 |
aesmc v0.16b, v0.16b // AES block 0 - round 12 |
| 965 |
aese v2.16b, v30.16b |
| 966 |
aesmc v2.16b, v2.16b // AES block 2 - round 12 |
| 967 |
aese v3.16b, v30.16b |
| 968 |
aesmc v3.16b, v3.16b // AES block 3 - round 12 |
| 969 |
|
| 970 |
Ldec_finish_first_blocks: |
| 971 |
cmp x0, x5 // check if we have <= 4 blocks |
| 972 |
trn1 v9.2d, v14.2d, v15.2d // h4h | h3h |
| 973 |
trn2 v17.2d, v14.2d, v15.2d // h4l | h3l |
| 974 |
trn1 v8.2d, v12.2d, v13.2d // h2h | h1h |
| 975 |
trn2 v16.2d, v12.2d, v13.2d // h2l | h1l |
| 976 |
eor v17.16b, v17.16b, v9.16b // h4k | h3k |
| 977 |
aese v1.16b, v31.16b // AES block 1 - round N-1 |
| 978 |
aese v2.16b, v31.16b // AES block 2 - round N-1 |
| 979 |
eor v16.16b, v16.16b, v8.16b // h2k | h1k |
| 980 |
aese v3.16b, v31.16b // AES block 3 - round N-1 |
| 981 |
aese v0.16b, v31.16b // AES block 0 - round N-1 |
| 982 |
b.ge Ldec_tail // handle tail |
| 983 |
|
| 984 |
ldr q4, [x0, #0] // AES block 0 - load ciphertext |
| 985 |
ldr q5, [x0, #16] // AES block 1 - load ciphertext |
| 986 |
rev w9, w12 // CTR block 4 |
| 987 |
eor v0.16b, v4.16b, v0.16b // AES block 0 - result |
| 988 |
eor v1.16b, v5.16b, v1.16b // AES block 1 - result |
| 989 |
rev64 v5.16b, v5.16b // GHASH block 1 |
| 990 |
ldr q7, [x0, #48] // AES block 3 - load ciphertext |
| 991 |
mov x7, v0.d[1] // AES block 0 - mov high |
| 992 |
mov x6, v0.d[0] // AES block 0 - mov low |
| 993 |
rev64 v4.16b, v4.16b // GHASH block 0 |
| 994 |
add w12, w12, #1 // CTR block 4 |
| 995 |
fmov d0, x10 // CTR block 4 |
| 996 |
orr x9, x11, x9, lsl #32 // CTR block 4 |
| 997 |
fmov v0.d[1], x9 // CTR block 4 |
| 998 |
rev w9, w12 // CTR block 5 |
| 999 |
add w12, w12, #1 // CTR block 5 |
| 1000 |
mov x19, v1.d[0] // AES block 1 - mov low |
| 1001 |
orr x9, x11, x9, lsl #32 // CTR block 5 |
| 1002 |
mov x20, v1.d[1] // AES block 1 - mov high |
| 1003 |
eor x7, x7, x14 // AES block 0 - round N high |
| 1004 |
eor x6, x6, x13 // AES block 0 - round N low |
| 1005 |
stp x6, x7, [x2], #16 // AES block 0 - store result |
| 1006 |
fmov d1, x10 // CTR block 5 |
| 1007 |
ldr q6, [x0, #32] // AES block 2 - load ciphertext |
| 1008 |
add x0, x0, #64 // AES input_ptr update |
| 1009 |
fmov v1.d[1], x9 // CTR block 5 |
| 1010 |
rev w9, w12 // CTR block 6 |
| 1011 |
add w12, w12, #1 // CTR block 6 |
| 1012 |
eor x19, x19, x13 // AES block 1 - round N low |
| 1013 |
orr x9, x11, x9, lsl #32 // CTR block 6 |
| 1014 |
eor x20, x20, x14 // AES block 1 - round N high |
| 1015 |
stp x19, x20, [x2], #16 // AES block 1 - store result |
| 1016 |
eor v2.16b, v6.16b, v2.16b // AES block 2 - result |
| 1017 |
cmp x0, x5 // check if we have <= 8 blocks |
| 1018 |
b.ge Ldec_prepretail // do prepretail |
| 1019 |
|
| 1020 |
Ldec_main_loop: // main loop start |
| 1021 |
mov x21, v2.d[0] // AES block 4k+2 - mov low |
| 1022 |
ext v11.16b, v11.16b, v11.16b, #8 // PRE 0 |
| 1023 |
eor v3.16b, v7.16b, v3.16b // AES block 4k+3 - result |
| 1024 |
aese v0.16b, v18.16b |
| 1025 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 0 |
| 1026 |
mov x22, v2.d[1] // AES block 4k+2 - mov high |
| 1027 |
aese v1.16b, v18.16b |
| 1028 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 0 |
| 1029 |
fmov d2, x10 // CTR block 4k+6 |
| 1030 |
fmov v2.d[1], x9 // CTR block 4k+6 |
| 1031 |
eor v4.16b, v4.16b, v11.16b // PRE 1 |
| 1032 |
rev w9, w12 // CTR block 4k+7 |
| 1033 |
aese v0.16b, v19.16b |
| 1034 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 1 |
| 1035 |
mov x24, v3.d[1] // AES block 4k+3 - mov high |
| 1036 |
aese v1.16b, v19.16b |
| 1037 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 1 |
| 1038 |
mov x23, v3.d[0] // AES block 4k+3 - mov low |
| 1039 |
pmull2 v9.1q, v4.2d, v15.2d // GHASH block 4k - high |
| 1040 |
mov d8, v4.d[1] // GHASH block 4k - mid |
| 1041 |
fmov d3, x10 // CTR block 4k+7 |
| 1042 |
aese v0.16b, v20.16b |
| 1043 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 2 |
| 1044 |
orr x9, x11, x9, lsl #32 // CTR block 4k+7 |
| 1045 |
aese v2.16b, v18.16b |
| 1046 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 0 |
| 1047 |
fmov v3.d[1], x9 // CTR block 4k+7 |
| 1048 |
aese v1.16b, v20.16b |
| 1049 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 2 |
| 1050 |
eor v8.8b, v8.8b, v4.8b // GHASH block 4k - mid |
| 1051 |
aese v0.16b, v21.16b |
| 1052 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 3 |
| 1053 |
eor x22, x22, x14 // AES block 4k+2 - round N high |
| 1054 |
aese v2.16b, v19.16b |
| 1055 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 1 |
| 1056 |
mov d10, v17.d[1] // GHASH block 4k - mid |
| 1057 |
aese v1.16b, v21.16b |
| 1058 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 3 |
| 1059 |
rev64 v6.16b, v6.16b // GHASH block 4k+2 |
| 1060 |
aese v3.16b, v18.16b |
| 1061 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 0 |
| 1062 |
eor x21, x21, x13 // AES block 4k+2 - round N low |
| 1063 |
aese v2.16b, v20.16b |
| 1064 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 2 |
| 1065 |
stp x21, x22, [x2], #16 // AES block 4k+2 - store result |
| 1066 |
pmull v11.1q, v4.1d, v15.1d // GHASH block 4k - low |
| 1067 |
pmull2 v4.1q, v5.2d, v14.2d // GHASH block 4k+1 - high |
| 1068 |
aese v2.16b, v21.16b |
| 1069 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 3 |
| 1070 |
rev64 v7.16b, v7.16b // GHASH block 4k+3 |
| 1071 |
pmull v10.1q, v8.1d, v10.1d // GHASH block 4k - mid |
| 1072 |
eor x23, x23, x13 // AES block 4k+3 - round N low |
| 1073 |
pmull v8.1q, v5.1d, v14.1d // GHASH block 4k+1 - low |
| 1074 |
eor x24, x24, x14 // AES block 4k+3 - round N high |
| 1075 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+1 - high |
| 1076 |
aese v2.16b, v22.16b |
| 1077 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 4 |
| 1078 |
aese v3.16b, v19.16b |
| 1079 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 1 |
| 1080 |
mov d4, v5.d[1] // GHASH block 4k+1 - mid |
| 1081 |
aese v0.16b, v22.16b |
| 1082 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 4 |
| 1083 |
eor v11.16b, v11.16b, v8.16b // GHASH block 4k+1 - low |
| 1084 |
aese v2.16b, v23.16b |
| 1085 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 5 |
| 1086 |
add w12, w12, #1 // CTR block 4k+7 |
| 1087 |
aese v3.16b, v20.16b |
| 1088 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 2 |
| 1089 |
mov d8, v6.d[1] // GHASH block 4k+2 - mid |
| 1090 |
aese v1.16b, v22.16b |
| 1091 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 4 |
| 1092 |
eor v4.8b, v4.8b, v5.8b // GHASH block 4k+1 - mid |
| 1093 |
pmull v5.1q, v6.1d, v13.1d // GHASH block 4k+2 - low |
| 1094 |
aese v3.16b, v21.16b |
| 1095 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 3 |
| 1096 |
eor v8.8b, v8.8b, v6.8b // GHASH block 4k+2 - mid |
| 1097 |
aese v1.16b, v23.16b |
| 1098 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 5 |
| 1099 |
aese v0.16b, v23.16b |
| 1100 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 5 |
| 1101 |
eor v11.16b, v11.16b, v5.16b // GHASH block 4k+2 - low |
| 1102 |
pmull v4.1q, v4.1d, v17.1d // GHASH block 4k+1 - mid |
| 1103 |
rev w9, w12 // CTR block 4k+8 |
| 1104 |
aese v1.16b, v24.16b |
| 1105 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 6 |
| 1106 |
ins v8.d[1], v8.d[0] // GHASH block 4k+2 - mid |
| 1107 |
aese v0.16b, v24.16b |
| 1108 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 6 |
| 1109 |
add w12, w12, #1 // CTR block 4k+8 |
| 1110 |
aese v3.16b, v22.16b |
| 1111 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 4 |
| 1112 |
aese v1.16b, v25.16b |
| 1113 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 7 |
| 1114 |
eor v10.16b, v10.16b, v4.16b // GHASH block 4k+1 - mid |
| 1115 |
aese v0.16b, v25.16b |
| 1116 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 7 |
| 1117 |
pmull2 v4.1q, v6.2d, v13.2d // GHASH block 4k+2 - high |
| 1118 |
mov d6, v7.d[1] // GHASH block 4k+3 - mid |
| 1119 |
aese v3.16b, v23.16b |
| 1120 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 5 |
| 1121 |
pmull2 v8.1q, v8.2d, v16.2d // GHASH block 4k+2 - mid |
| 1122 |
aese v0.16b, v26.16b |
| 1123 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 8 |
| 1124 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+2 - high |
| 1125 |
aese v3.16b, v24.16b |
| 1126 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 6 |
| 1127 |
pmull v4.1q, v7.1d, v12.1d // GHASH block 4k+3 - low |
| 1128 |
orr x9, x11, x9, lsl #32 // CTR block 4k+8 |
| 1129 |
eor v10.16b, v10.16b, v8.16b // GHASH block 4k+2 - mid |
| 1130 |
pmull2 v5.1q, v7.2d, v12.2d // GHASH block 4k+3 - high |
| 1131 |
cmp x17, #12 // setup flags for AES-128/192/256 check |
| 1132 |
eor v6.8b, v6.8b, v7.8b // GHASH block 4k+3 - mid |
| 1133 |
aese v1.16b, v26.16b |
| 1134 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 8 |
| 1135 |
aese v2.16b, v24.16b |
| 1136 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 6 |
| 1137 |
eor v9.16b, v9.16b, v5.16b // GHASH block 4k+3 - high |
| 1138 |
pmull v6.1q, v6.1d, v16.1d // GHASH block 4k+3 - mid |
| 1139 |
movi v8.8b, #0xc2 |
| 1140 |
aese v2.16b, v25.16b |
| 1141 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 7 |
| 1142 |
eor v11.16b, v11.16b, v4.16b // GHASH block 4k+3 - low |
| 1143 |
aese v3.16b, v25.16b |
| 1144 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 7 |
| 1145 |
shl d8, d8, #56 // mod_constant |
| 1146 |
aese v2.16b, v26.16b |
| 1147 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 8 |
| 1148 |
eor v10.16b, v10.16b, v6.16b // GHASH block 4k+3 - mid |
| 1149 |
aese v3.16b, v26.16b |
| 1150 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 8 |
| 1151 |
b.lt Ldec_main_loop_continue // branch if AES-128 |
| 1152 |
|
| 1153 |
aese v0.16b, v27.16b |
| 1154 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 9 |
| 1155 |
aese v2.16b, v27.16b |
| 1156 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 9 |
| 1157 |
aese v1.16b, v27.16b |
| 1158 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 9 |
| 1159 |
aese v3.16b, v27.16b |
| 1160 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 9 |
| 1161 |
aese v0.16b, v28.16b |
| 1162 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 10 |
| 1163 |
aese v1.16b, v28.16b |
| 1164 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 10 |
| 1165 |
aese v2.16b, v28.16b |
| 1166 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 10 |
| 1167 |
aese v3.16b, v28.16b |
| 1168 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 10 |
| 1169 |
b.eq Ldec_main_loop_continue // branch if AES-192 |
| 1170 |
|
| 1171 |
aese v0.16b, v29.16b |
| 1172 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 11 |
| 1173 |
aese v1.16b, v29.16b |
| 1174 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 11 |
| 1175 |
aese v2.16b, v29.16b |
| 1176 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 11 |
| 1177 |
aese v3.16b, v29.16b |
| 1178 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 11 |
| 1179 |
aese v0.16b, v30.16b |
| 1180 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 12 |
| 1181 |
aese v1.16b, v30.16b |
| 1182 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 12 |
| 1183 |
aese v2.16b, v30.16b |
| 1184 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 12 |
| 1185 |
aese v3.16b, v30.16b |
| 1186 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 12 |
| 1187 |
|
| 1188 |
Ldec_main_loop_continue: |
| 1189 |
pmull v7.1q, v9.1d, v8.1d // MODULO - top 64b align with mid |
| 1190 |
eor v6.16b, v11.16b, v9.16b // MODULO - karatsuba tidy up |
| 1191 |
ldr q4, [x0, #0] // AES block 4k+4 - load ciphertext |
| 1192 |
aese v0.16b, v31.16b // AES block 4k+4 - round N-1 |
| 1193 |
ext v9.16b, v9.16b, v9.16b, #8 // MODULO - other top alignment |
| 1194 |
eor v10.16b, v10.16b, v6.16b // MODULO - karatsuba tidy up |
| 1195 |
ldr q5, [x0, #16] // AES block 4k+5 - load ciphertext |
| 1196 |
eor v0.16b, v4.16b, v0.16b // AES block 4k+4 - result |
| 1197 |
stp x23, x24, [x2], #16 // AES block 4k+3 - store result |
| 1198 |
eor v10.16b, v10.16b, v7.16b // MODULO - fold into mid |
| 1199 |
ldr q7, [x0, #48] // AES block 4k+7 - load ciphertext |
| 1200 |
ldr q6, [x0, #32] // AES block 4k+6 - load ciphertext |
| 1201 |
mov x7, v0.d[1] // AES block 4k+4 - mov high |
| 1202 |
eor v10.16b, v10.16b, v9.16b // MODULO - fold into mid |
| 1203 |
aese v1.16b, v31.16b // AES block 4k+5 - round N-1 |
| 1204 |
add x0, x0, #64 // AES input_ptr update |
| 1205 |
mov x6, v0.d[0] // AES block 4k+4 - mov low |
| 1206 |
fmov d0, x10 // CTR block 4k+8 |
| 1207 |
fmov v0.d[1], x9 // CTR block 4k+8 |
| 1208 |
pmull v8.1q, v10.1d, v8.1d // MODULO - mid 64b align with low |
| 1209 |
eor v1.16b, v5.16b, v1.16b // AES block 4k+5 - result |
| 1210 |
rev w9, w12 // CTR block 4k+9 |
| 1211 |
aese v2.16b, v31.16b // AES block 4k+6 - round N-1 |
| 1212 |
orr x9, x11, x9, lsl #32 // CTR block 4k+9 |
| 1213 |
cmp x0, x5 // LOOP CONTROL |
| 1214 |
add w12, w12, #1 // CTR block 4k+9 |
| 1215 |
eor x6, x6, x13 // AES block 4k+4 - round N low |
| 1216 |
eor x7, x7, x14 // AES block 4k+4 - round N high |
| 1217 |
mov x20, v1.d[1] // AES block 4k+5 - mov high |
| 1218 |
eor v2.16b, v6.16b, v2.16b // AES block 4k+6 - result |
| 1219 |
eor v11.16b, v11.16b, v8.16b // MODULO - fold into low |
| 1220 |
mov x19, v1.d[0] // AES block 4k+5 - mov low |
| 1221 |
fmov d1, x10 // CTR block 4k+9 |
| 1222 |
ext v10.16b, v10.16b, v10.16b, #8 // MODULO - other mid alignment |
| 1223 |
fmov v1.d[1], x9 // CTR block 4k+9 |
| 1224 |
rev w9, w12 // CTR block 4k+10 |
| 1225 |
add w12, w12, #1 // CTR block 4k+10 |
| 1226 |
aese v3.16b, v31.16b // AES block 4k+7 - round N-1 |
| 1227 |
orr x9, x11, x9, lsl #32 // CTR block 4k+10 |
| 1228 |
rev64 v5.16b, v5.16b // GHASH block 4k+5 |
| 1229 |
eor x20, x20, x14 // AES block 4k+5 - round N high |
| 1230 |
stp x6, x7, [x2], #16 // AES block 4k+4 - store result |
| 1231 |
eor x19, x19, x13 // AES block 4k+5 - round N low |
| 1232 |
stp x19, x20, [x2], #16 // AES block 4k+5 - store result |
| 1233 |
rev64 v4.16b, v4.16b // GHASH block 4k+4 |
| 1234 |
eor v11.16b, v11.16b, v10.16b // MODULO - fold into low |
| 1235 |
b.lt Ldec_main_loop |
| 1236 |
|
| 1237 |
Ldec_prepretail: // PREPRETAIL |
| 1238 |
ext v11.16b, v11.16b, v11.16b, #8 // PRE 0 |
| 1239 |
mov x21, v2.d[0] // AES block 4k+2 - mov low |
| 1240 |
eor v3.16b, v7.16b, v3.16b // AES block 4k+3 - result |
| 1241 |
aese v0.16b, v18.16b |
| 1242 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 0 |
| 1243 |
mov x22, v2.d[1] // AES block 4k+2 - mov high |
| 1244 |
aese v1.16b, v18.16b |
| 1245 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 0 |
| 1246 |
fmov d2, x10 // CTR block 4k+6 |
| 1247 |
fmov v2.d[1], x9 // CTR block 4k+6 |
| 1248 |
rev w9, w12 // CTR block 4k+7 |
| 1249 |
eor v4.16b, v4.16b, v11.16b // PRE 1 |
| 1250 |
rev64 v6.16b, v6.16b // GHASH block 4k+2 |
| 1251 |
orr x9, x11, x9, lsl #32 // CTR block 4k+7 |
| 1252 |
mov x23, v3.d[0] // AES block 4k+3 - mov low |
| 1253 |
aese v1.16b, v19.16b |
| 1254 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 1 |
| 1255 |
mov x24, v3.d[1] // AES block 4k+3 - mov high |
| 1256 |
pmull v11.1q, v4.1d, v15.1d // GHASH block 4k - low |
| 1257 |
mov d8, v4.d[1] // GHASH block 4k - mid |
| 1258 |
fmov d3, x10 // CTR block 4k+7 |
| 1259 |
pmull2 v9.1q, v4.2d, v15.2d // GHASH block 4k - high |
| 1260 |
fmov v3.d[1], x9 // CTR block 4k+7 |
| 1261 |
aese v2.16b, v18.16b |
| 1262 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 0 |
| 1263 |
mov d10, v17.d[1] // GHASH block 4k - mid |
| 1264 |
aese v0.16b, v19.16b |
| 1265 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 1 |
| 1266 |
eor v8.8b, v8.8b, v4.8b // GHASH block 4k - mid |
| 1267 |
pmull2 v4.1q, v5.2d, v14.2d // GHASH block 4k+1 - high |
| 1268 |
aese v2.16b, v19.16b |
| 1269 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 1 |
| 1270 |
rev64 v7.16b, v7.16b // GHASH block 4k+3 |
| 1271 |
aese v3.16b, v18.16b |
| 1272 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 0 |
| 1273 |
pmull v10.1q, v8.1d, v10.1d // GHASH block 4k - mid |
| 1274 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+1 - high |
| 1275 |
pmull v8.1q, v5.1d, v14.1d // GHASH block 4k+1 - low |
| 1276 |
aese v3.16b, v19.16b |
| 1277 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 1 |
| 1278 |
mov d4, v5.d[1] // GHASH block 4k+1 - mid |
| 1279 |
aese v0.16b, v20.16b |
| 1280 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 2 |
| 1281 |
aese v1.16b, v20.16b |
| 1282 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 2 |
| 1283 |
eor v11.16b, v11.16b, v8.16b // GHASH block 4k+1 - low |
| 1284 |
aese v2.16b, v20.16b |
| 1285 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 2 |
| 1286 |
aese v0.16b, v21.16b |
| 1287 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 3 |
| 1288 |
mov d8, v6.d[1] // GHASH block 4k+2 - mid |
| 1289 |
aese v3.16b, v20.16b |
| 1290 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 2 |
| 1291 |
eor v4.8b, v4.8b, v5.8b // GHASH block 4k+1 - mid |
| 1292 |
pmull v5.1q, v6.1d, v13.1d // GHASH block 4k+2 - low |
| 1293 |
aese v0.16b, v22.16b |
| 1294 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 4 |
| 1295 |
aese v3.16b, v21.16b |
| 1296 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 3 |
| 1297 |
eor v8.8b, v8.8b, v6.8b // GHASH block 4k+2 - mid |
| 1298 |
pmull v4.1q, v4.1d, v17.1d // GHASH block 4k+1 - mid |
| 1299 |
aese v0.16b, v23.16b |
| 1300 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 5 |
| 1301 |
eor v11.16b, v11.16b, v5.16b // GHASH block 4k+2 - low |
| 1302 |
aese v3.16b, v22.16b |
| 1303 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 4 |
| 1304 |
pmull2 v5.1q, v7.2d, v12.2d // GHASH block 4k+3 - high |
| 1305 |
eor v10.16b, v10.16b, v4.16b // GHASH block 4k+1 - mid |
| 1306 |
pmull2 v4.1q, v6.2d, v13.2d // GHASH block 4k+2 - high |
| 1307 |
aese v3.16b, v23.16b |
| 1308 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 5 |
| 1309 |
ins v8.d[1], v8.d[0] // GHASH block 4k+2 - mid |
| 1310 |
aese v2.16b, v21.16b |
| 1311 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 3 |
| 1312 |
aese v1.16b, v21.16b |
| 1313 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 3 |
| 1314 |
eor v9.16b, v9.16b, v4.16b // GHASH block 4k+2 - high |
| 1315 |
pmull v4.1q, v7.1d, v12.1d // GHASH block 4k+3 - low |
| 1316 |
aese v2.16b, v22.16b |
| 1317 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 4 |
| 1318 |
mov d6, v7.d[1] // GHASH block 4k+3 - mid |
| 1319 |
aese v1.16b, v22.16b |
| 1320 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 4 |
| 1321 |
pmull2 v8.1q, v8.2d, v16.2d // GHASH block 4k+2 - mid |
| 1322 |
aese v2.16b, v23.16b |
| 1323 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 5 |
| 1324 |
eor v6.8b, v6.8b, v7.8b // GHASH block 4k+3 - mid |
| 1325 |
aese v1.16b, v23.16b |
| 1326 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 5 |
| 1327 |
aese v3.16b, v24.16b |
| 1328 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 6 |
| 1329 |
eor v10.16b, v10.16b, v8.16b // GHASH block 4k+2 - mid |
| 1330 |
aese v2.16b, v24.16b |
| 1331 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 6 |
| 1332 |
aese v0.16b, v24.16b |
| 1333 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 6 |
| 1334 |
movi v8.8b, #0xc2 |
| 1335 |
aese v1.16b, v24.16b |
| 1336 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 6 |
| 1337 |
eor v11.16b, v11.16b, v4.16b // GHASH block 4k+3 - low |
| 1338 |
pmull v6.1q, v6.1d, v16.1d // GHASH block 4k+3 - mid |
| 1339 |
aese v3.16b, v25.16b |
| 1340 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 7 |
| 1341 |
cmp x17, #12 // setup flags for AES-128/192/256 check |
| 1342 |
eor v9.16b, v9.16b, v5.16b // GHASH block 4k+3 - high |
| 1343 |
aese v1.16b, v25.16b |
| 1344 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 7 |
| 1345 |
aese v0.16b, v25.16b |
| 1346 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 7 |
| 1347 |
eor v10.16b, v10.16b, v6.16b // GHASH block 4k+3 - mid |
| 1348 |
aese v3.16b, v26.16b |
| 1349 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 8 |
| 1350 |
aese v2.16b, v25.16b |
| 1351 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 7 |
| 1352 |
eor v6.16b, v11.16b, v9.16b // MODULO - karatsuba tidy up |
| 1353 |
aese v1.16b, v26.16b |
| 1354 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 8 |
| 1355 |
aese v0.16b, v26.16b |
| 1356 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 8 |
| 1357 |
shl d8, d8, #56 // mod_constant |
| 1358 |
aese v2.16b, v26.16b |
| 1359 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 8 |
| 1360 |
b.lt Ldec_finish_prepretail // branch if AES-128 |
| 1361 |
|
| 1362 |
aese v1.16b, v27.16b |
| 1363 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 9 |
| 1364 |
aese v2.16b, v27.16b |
| 1365 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 9 |
| 1366 |
aese v3.16b, v27.16b |
| 1367 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 9 |
| 1368 |
aese v0.16b, v27.16b |
| 1369 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 9 |
| 1370 |
aese v2.16b, v28.16b |
| 1371 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 10 |
| 1372 |
aese v3.16b, v28.16b |
| 1373 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 10 |
| 1374 |
aese v0.16b, v28.16b |
| 1375 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 10 |
| 1376 |
aese v1.16b, v28.16b |
| 1377 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 10 |
| 1378 |
b.eq Ldec_finish_prepretail // branch if AES-192 |
| 1379 |
|
| 1380 |
aese v2.16b, v29.16b |
| 1381 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 11 |
| 1382 |
aese v0.16b, v29.16b |
| 1383 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 11 |
| 1384 |
aese v1.16b, v29.16b |
| 1385 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 11 |
| 1386 |
aese v2.16b, v30.16b |
| 1387 |
aesmc v2.16b, v2.16b // AES block 4k+6 - round 12 |
| 1388 |
aese v3.16b, v29.16b |
| 1389 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 11 |
| 1390 |
aese v1.16b, v30.16b |
| 1391 |
aesmc v1.16b, v1.16b // AES block 4k+5 - round 12 |
| 1392 |
aese v0.16b, v30.16b |
| 1393 |
aesmc v0.16b, v0.16b // AES block 4k+4 - round 12 |
| 1394 |
aese v3.16b, v30.16b |
| 1395 |
aesmc v3.16b, v3.16b // AES block 4k+7 - round 12 |
| 1396 |
|
| 1397 |
Ldec_finish_prepretail: |
| 1398 |
eor v10.16b, v10.16b, v6.16b // MODULO - karatsuba tidy up |
| 1399 |
pmull v7.1q, v9.1d, v8.1d // MODULO - top 64b align with mid |
| 1400 |
ext v9.16b, v9.16b, v9.16b, #8 // MODULO - other top alignment |
| 1401 |
eor v10.16b, v10.16b, v7.16b // MODULO - fold into mid |
| 1402 |
eor x22, x22, x14 // AES block 4k+2 - round N high |
| 1403 |
eor x23, x23, x13 // AES block 4k+3 - round N low |
| 1404 |
eor v10.16b, v10.16b, v9.16b // MODULO - fold into mid |
| 1405 |
add w12, w12, #1 // CTR block 4k+7 |
| 1406 |
eor x21, x21, x13 // AES block 4k+2 - round N low |
| 1407 |
pmull v8.1q, v10.1d, v8.1d // MODULO - mid 64b align with low |
| 1408 |
eor x24, x24, x14 // AES block 4k+3 - round N high |
| 1409 |
stp x21, x22, [x2], #16 // AES block 4k+2 - store result |
| 1410 |
ext v10.16b, v10.16b, v10.16b, #8 // MODULO - other mid alignment |
| 1411 |
stp x23, x24, [x2], #16 // AES block 4k+3 - store result |
| 1412 |
|
| 1413 |
eor v11.16b, v11.16b, v8.16b // MODULO - fold into low |
| 1414 |
aese v1.16b, v31.16b // AES block 4k+5 - round N-1 |
| 1415 |
aese v0.16b, v31.16b // AES block 4k+4 - round N-1 |
| 1416 |
aese v3.16b, v31.16b // AES block 4k+7 - round N-1 |
| 1417 |
aese v2.16b, v31.16b // AES block 4k+6 - round N-1 |
| 1418 |
eor v11.16b, v11.16b, v10.16b // MODULO - fold into low |
| 1419 |
|
| 1420 |
Ldec_tail: // TAIL |
| 1421 |
sub x5, x4, x0 // main_end_input_ptr is number of bytes left to process |
| 1422 |
ld1 { v5.16b}, [x0], #16 // AES block 4k+4 - load ciphertext |
| 1423 |
eor v0.16b, v5.16b, v0.16b // AES block 4k+4 - result |
| 1424 |
mov x6, v0.d[0] // AES block 4k+4 - mov low |
| 1425 |
mov x7, v0.d[1] // AES block 4k+4 - mov high |
| 1426 |
ext v8.16b, v11.16b, v11.16b, #8 // prepare final partial tag |
| 1427 |
cmp x5, #48 |
| 1428 |
eor x6, x6, x13 // AES block 4k+4 - round N low |
| 1429 |
eor x7, x7, x14 // AES block 4k+4 - round N high |
| 1430 |
b.gt Ldec_blocks_more_than_3 |
| 1431 |
sub w12, w12, #1 |
| 1432 |
mov v3.16b, v2.16b |
| 1433 |
movi v10.8b, #0 |
| 1434 |
movi v11.8b, #0 |
| 1435 |
cmp x5, #32 |
| 1436 |
movi v9.8b, #0 |
| 1437 |
mov v2.16b, v1.16b |
| 1438 |
b.gt Ldec_blocks_more_than_2 |
| 1439 |
sub w12, w12, #1 |
| 1440 |
mov v3.16b, v1.16b |
| 1441 |
cmp x5, #16 |
| 1442 |
b.gt Ldec_blocks_more_than_1 |
| 1443 |
sub w12, w12, #1 |
| 1444 |
b Ldec_blocks_less_than_1 |
| 1445 |
Ldec_blocks_more_than_3: // blocks left > 3 |
| 1446 |
rev64 v4.16b, v5.16b // GHASH final-3 block |
| 1447 |
ld1 { v5.16b}, [x0], #16 // AES final-2 block - load ciphertext |
| 1448 |
stp x6, x7, [x2], #16 // AES final-3 block - store result |
| 1449 |
mov d10, v17.d[1] // GHASH final-3 block - mid |
| 1450 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 1451 |
eor v0.16b, v5.16b, v1.16b // AES final-2 block - result |
| 1452 |
mov d22, v4.d[1] // GHASH final-3 block - mid |
| 1453 |
mov x6, v0.d[0] // AES final-2 block - mov low |
| 1454 |
mov x7, v0.d[1] // AES final-2 block - mov high |
| 1455 |
eor v22.8b, v22.8b, v4.8b // GHASH final-3 block - mid |
| 1456 |
movi v8.8b, #0 // suppress further partial tag feed in |
| 1457 |
pmull2 v9.1q, v4.2d, v15.2d // GHASH final-3 block - high |
| 1458 |
pmull v10.1q, v22.1d, v10.1d // GHASH final-3 block - mid |
| 1459 |
eor x6, x6, x13 // AES final-2 block - round N low |
| 1460 |
pmull v11.1q, v4.1d, v15.1d // GHASH final-3 block - low |
| 1461 |
eor x7, x7, x14 // AES final-2 block - round N high |
| 1462 |
Ldec_blocks_more_than_2: // blocks left > 2 |
| 1463 |
rev64 v4.16b, v5.16b // GHASH final-2 block |
| 1464 |
ld1 { v5.16b}, [x0], #16 // AES final-1 block - load ciphertext |
| 1465 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 1466 |
stp x6, x7, [x2], #16 // AES final-2 block - store result |
| 1467 |
eor v0.16b, v5.16b, v2.16b // AES final-1 block - result |
| 1468 |
mov d22, v4.d[1] // GHASH final-2 block - mid |
| 1469 |
pmull v21.1q, v4.1d, v14.1d // GHASH final-2 block - low |
| 1470 |
pmull2 v20.1q, v4.2d, v14.2d // GHASH final-2 block - high |
| 1471 |
eor v22.8b, v22.8b, v4.8b // GHASH final-2 block - mid |
| 1472 |
mov x6, v0.d[0] // AES final-1 block - mov low |
| 1473 |
mov x7, v0.d[1] // AES final-1 block - mov high |
| 1474 |
eor v11.16b, v11.16b, v21.16b // GHASH final-2 block - low |
| 1475 |
movi v8.8b, #0 // suppress further partial tag feed in |
| 1476 |
pmull v22.1q, v22.1d, v17.1d // GHASH final-2 block - mid |
| 1477 |
eor v9.16b, v9.16b, v20.16b // GHASH final-2 block - high |
| 1478 |
eor x6, x6, x13 // AES final-1 block - round N low |
| 1479 |
eor v10.16b, v10.16b, v22.16b // GHASH final-2 block - mid |
| 1480 |
eor x7, x7, x14 // AES final-1 block - round N high |
| 1481 |
Ldec_blocks_more_than_1: // blocks left > 1 |
| 1482 |
stp x6, x7, [x2], #16 // AES final-1 block - store result |
| 1483 |
rev64 v4.16b, v5.16b // GHASH final-1 block |
| 1484 |
ld1 { v5.16b}, [x0], #16 // AES final block - load ciphertext |
| 1485 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 1486 |
movi v8.8b, #0 // suppress further partial tag feed in |
| 1487 |
mov d22, v4.d[1] // GHASH final-1 block - mid |
| 1488 |
eor v0.16b, v5.16b, v3.16b // AES final block - result |
| 1489 |
pmull2 v20.1q, v4.2d, v13.2d // GHASH final-1 block - high |
| 1490 |
eor v22.8b, v22.8b, v4.8b // GHASH final-1 block - mid |
| 1491 |
pmull v21.1q, v4.1d, v13.1d // GHASH final-1 block - low |
| 1492 |
mov x6, v0.d[0] // AES final block - mov low |
| 1493 |
ins v22.d[1], v22.d[0] // GHASH final-1 block - mid |
| 1494 |
mov x7, v0.d[1] // AES final block - mov high |
| 1495 |
pmull2 v22.1q, v22.2d, v16.2d // GHASH final-1 block - mid |
| 1496 |
eor x6, x6, x13 // AES final block - round N low |
| 1497 |
eor v11.16b, v11.16b, v21.16b // GHASH final-1 block - low |
| 1498 |
eor v9.16b, v9.16b, v20.16b // GHASH final-1 block - high |
| 1499 |
eor v10.16b, v10.16b, v22.16b // GHASH final-1 block - mid |
| 1500 |
eor x7, x7, x14 // AES final block - round N high |
| 1501 |
Ldec_blocks_less_than_1: // blocks left <= 1 |
| 1502 |
and x1, x1, #127 // bit_length %= 128 |
| 1503 |
mvn x14, xzr // rkN_h = 0xffffffffffffffff |
| 1504 |
sub x1, x1, #128 // bit_length -= 128 |
| 1505 |
mvn x13, xzr // rkN_l = 0xffffffffffffffff |
| 1506 |
ldp x4, x5, [x2] // load existing bytes we need to not overwrite |
| 1507 |
neg x1, x1 // bit_length = 128 - #bits in input (in range [1,128]) |
| 1508 |
and x1, x1, #127 // bit_length %= 128 |
| 1509 |
lsr x14, x14, x1 // rkN_h is mask for top 64b of last block |
| 1510 |
cmp x1, #64 |
| 1511 |
csel x9, x13, x14, lt |
| 1512 |
csel x10, x14, xzr, lt |
| 1513 |
fmov d0, x9 // ctr0b is mask for last block |
| 1514 |
and x6, x6, x9 |
| 1515 |
mov v0.d[1], x10 |
| 1516 |
bic x4, x4, x9 // mask out low existing bytes |
| 1517 |
rev w9, w12 |
| 1518 |
bic x5, x5, x10 // mask out high existing bytes |
| 1519 |
orr x6, x6, x4 |
| 1520 |
and x7, x7, x10 |
| 1521 |
orr x7, x7, x5 |
| 1522 |
and v5.16b, v5.16b, v0.16b // possibly partial last block has zeroes in highest bits |
| 1523 |
rev64 v4.16b, v5.16b // GHASH final block |
| 1524 |
eor v4.16b, v4.16b, v8.16b // feed in partial tag |
| 1525 |
pmull v21.1q, v4.1d, v12.1d // GHASH final block - low |
| 1526 |
mov d8, v4.d[1] // GHASH final block - mid |
| 1527 |
eor v8.8b, v8.8b, v4.8b // GHASH final block - mid |
| 1528 |
pmull2 v20.1q, v4.2d, v12.2d // GHASH final block - high |
| 1529 |
pmull v8.1q, v8.1d, v16.1d // GHASH final block - mid |
| 1530 |
eor v9.16b, v9.16b, v20.16b // GHASH final block - high |
| 1531 |
eor v11.16b, v11.16b, v21.16b // GHASH final block - low |
| 1532 |
eor v10.16b, v10.16b, v8.16b // GHASH final block - mid |
| 1533 |
movi v8.8b, #0xc2 |
| 1534 |
eor v6.16b, v11.16b, v9.16b // MODULO - karatsuba tidy up |
| 1535 |
shl d8, d8, #56 // mod_constant |
| 1536 |
eor v10.16b, v10.16b, v6.16b // MODULO - karatsuba tidy up |
| 1537 |
pmull v7.1q, v9.1d, v8.1d // MODULO - top 64b align with mid |
| 1538 |
ext v9.16b, v9.16b, v9.16b, #8 // MODULO - other top alignment |
| 1539 |
eor v10.16b, v10.16b, v7.16b // MODULO - fold into mid |
| 1540 |
eor v10.16b, v10.16b, v9.16b // MODULO - fold into mid |
| 1541 |
pmull v8.1q, v10.1d, v8.1d // MODULO - mid 64b align with low |
| 1542 |
ext v10.16b, v10.16b, v10.16b, #8 // MODULO - other mid alignment |
| 1543 |
eor v11.16b, v11.16b, v8.16b // MODULO - fold into low |
| 1544 |
stp x6, x7, [x2] |
| 1545 |
str w9, [x16, #12] // store the updated counter |
| 1546 |
eor v11.16b, v11.16b, v10.16b // MODULO - fold into low |
| 1547 |
ext v11.16b, v11.16b, v11.16b, #8 |
| 1548 |
rev64 v11.16b, v11.16b |
| 1549 |
mov x0, x15 |
| 1550 |
st1 { v11.16b }, [x3] |
| 1551 |
ldp x19, x20, [sp, #16] |
| 1552 |
ldp x21, x22, [sp, #32] |
| 1553 |
ldp x23, x24, [sp, #48] |
| 1554 |
ldp d8, d9, [sp, #64] |
| 1555 |
ldp d10, d11, [sp, #80] |
| 1556 |
ldp d12, d13, [sp, #96] |
| 1557 |
ldp d14, d15, [sp, #112] |
| 1558 |
ldp x29, x30, [sp], #128 |
| 1559 |
AARCH64_VALIDATE_LINK_REGISTER |
| 1560 |
ret |
| 1561 |
|
| 1562 |
#endif |
| 1563 |
#endif // !OPENSSL_NO_ASM && defined(__AARCH64EL__) && defined(__APPLE__) |
| 1564 |
#if defined(__ELF__) |
| 1565 |
// See https://www.airs.com/blog/archives/518. |
| 1566 |
.section .note.GNU-stack,"",%progbits |
| 1567 |
#endif |