projects / mirror_zfs.git / blame
| Commit | Line | Data |
|---|---|---|
| 31b160f0 AF |
1 | # Copyright 2013-2016 The OpenSSL Project Authors. All Rights Reserved. |
| 2 | # | |
| 3 | # Licensed under the Apache License 2.0 (the "License"). You may not use | |
| 4 | # this file except in compliance with the License. You can obtain a copy | |
| 5 | # in the file LICENSE in the source distribution or at | |
| 6 | # https://www.openssl.org/source/license.html | |
| 7 | ||
| 8 | # | |
| 9 | # ==================================================================== | |
| 10 | # Written by Andy Polyakov <appro@openssl.org> for the OpenSSL | |
| 11 | # project. The module is, however, dual licensed under OpenSSL and | |
| 12 | # CRYPTOGAMS licenses depending on where you obtain it. For further | |
| 13 | # details see http://www.openssl.org/~appro/cryptogams/. | |
| 14 | # ==================================================================== | |
| 15 | # | |
| 16 | # | |
| 17 | # AES-NI-CTR+GHASH stitch. | |
| 18 | # | |
| 19 | # February 2013 | |
| 20 | # | |
| 21 | # OpenSSL GCM implementation is organized in such way that its | |
| 22 | # performance is rather close to the sum of its streamed components, | |
| 23 | # in the context parallelized AES-NI CTR and modulo-scheduled | |
| 24 | # PCLMULQDQ-enabled GHASH. Unfortunately, as no stitch implementation | |
| 25 | # was observed to perform significantly better than the sum of the | |
| 26 | # components on contemporary CPUs, the effort was deemed impossible to | |
| 27 | # justify. This module is based on combination of Intel submissions, | |
| 28 | # [1] and [2], with MOVBE twist suggested by Ilya Albrekht and Max | |
| 29 | # Locktyukhin of Intel Corp. who verified that it reduces shuffles | |
| 30 | # pressure with notable relative improvement, achieving 1.0 cycle per | |
| 31 | # byte processed with 128-bit key on Haswell processor, 0.74 - on | |
| 32 | # Broadwell, 0.63 - on Skylake... [Mentioned results are raw profiled | |
| 33 | # measurements for favourable packet size, one divisible by 96. | |
| 34 | # Applications using the EVP interface will observe a few percent | |
| 35 | # worse performance.] | |
| 36 | # | |
| 37 | # Knights Landing processes 1 byte in 1.25 cycles (measured with EVP). | |
| 38 | # | |
| 39 | # [1] http://rt.openssl.org/Ticket/Display.html?id=2900&user=guest&pass=guest | |
| 40 | # [2] http://www.intel.com/content/dam/www/public/us/en/documents/software-support/enabling-high-performance-gcm.pdf | |
| 41 | ||
| 42 | # Generated once from | |
| 43 | # https://github.com/openssl/openssl/blob/5ffc3324/crypto/modes/asm/aesni-gcm-x86_64.pl | |
| 44 | # and modified for ICP. Modification are kept at a bare minimum to ease later | |
| 45 | # upstream merges. | |
| 46 | ||
| 47 | #if defined(__x86_64__) && defined(HAVE_AVX) && \ | |
| 5b3b7955 AF |
48 | defined(HAVE_AES) && defined(HAVE_PCLMULQDQ) |
| 49 | ||
| b844489e AL |
50 | #define _ASM |
| 51 | #include <sys/asm_linkage.h> | |
| 52 | ||
| 5b3b7955 | 53 | .extern gcm_avx_can_use_movbe |
| 31b160f0 AF |
54 | |
| 55 | .text | |
| 56 | ||
| 5b3b7955 | 57 | #ifdef HAVE_MOVBE |
| 31b160f0 AF |
58 | .type _aesni_ctr32_ghash_6x,@function |
| 59 | .align 32 | |
| 60 | _aesni_ctr32_ghash_6x: | |
| d9655c5b | 61 | .cfi_startproc |
| 31b160f0 AF |
62 | vmovdqu 32(%r11),%xmm2 |
| 63 | subq $6,%rdx | |
| 64 | vpxor %xmm4,%xmm4,%xmm4 | |
| 65 | vmovdqu 0-128(%rcx),%xmm15 | |
| 66 | vpaddb %xmm2,%xmm1,%xmm10 | |
| 67 | vpaddb %xmm2,%xmm10,%xmm11 | |
| 68 | vpaddb %xmm2,%xmm11,%xmm12 | |
| 69 | vpaddb %xmm2,%xmm12,%xmm13 | |
| 70 | vpaddb %xmm2,%xmm13,%xmm14 | |
| 71 | vpxor %xmm15,%xmm1,%xmm9 | |
| 72 | vmovdqu %xmm4,16+8(%rsp) | |
| 73 | jmp .Loop6x | |
| 74 | ||
| 75 | .align 32 | |
| 76 | .Loop6x: | |
| 77 | addl $100663296,%ebx | |
| 78 | jc .Lhandle_ctr32 | |
| 79 | vmovdqu 0-32(%r9),%xmm3 | |
| 80 | vpaddb %xmm2,%xmm14,%xmm1 | |
| 81 | vpxor %xmm15,%xmm10,%xmm10 | |
| 82 | vpxor %xmm15,%xmm11,%xmm11 | |
| 83 | ||
| 84 | .Lresume_ctr32: | |
| 85 | vmovdqu %xmm1,(%r8) | |
| 86 | vpclmulqdq $0x10,%xmm3,%xmm7,%xmm5 | |
| 87 | vpxor %xmm15,%xmm12,%xmm12 | |
| 88 | vmovups 16-128(%rcx),%xmm2 | |
| 89 | vpclmulqdq $0x01,%xmm3,%xmm7,%xmm6 | |
| 90 | xorq %r12,%r12 | |
| 91 | cmpq %r14,%r15 | |
| 92 | ||
| 93 | vaesenc %xmm2,%xmm9,%xmm9 | |
| 94 | vmovdqu 48+8(%rsp),%xmm0 | |
| 95 | vpxor %xmm15,%xmm13,%xmm13 | |
| 96 | vpclmulqdq $0x00,%xmm3,%xmm7,%xmm1 | |
| 97 | vaesenc %xmm2,%xmm10,%xmm10 | |
| 98 | vpxor %xmm15,%xmm14,%xmm14 | |
| 99 | setnc %r12b | |
| 100 | vpclmulqdq $0x11,%xmm3,%xmm7,%xmm7 | |
| 101 | vaesenc %xmm2,%xmm11,%xmm11 | |
| 102 | vmovdqu 16-32(%r9),%xmm3 | |
| 103 | negq %r12 | |
| 104 | vaesenc %xmm2,%xmm12,%xmm12 | |
| 105 | vpxor %xmm5,%xmm6,%xmm6 | |
| 106 | vpclmulqdq $0x00,%xmm3,%xmm0,%xmm5 | |
| 107 | vpxor %xmm4,%xmm8,%xmm8 | |
| 108 | vaesenc %xmm2,%xmm13,%xmm13 | |
| 109 | vpxor %xmm5,%xmm1,%xmm4 | |
| 110 | andq $0x60,%r12 | |
| 111 | vmovups 32-128(%rcx),%xmm15 | |
| 112 | vpclmulqdq $0x10,%xmm3,%xmm0,%xmm1 | |
| 113 | vaesenc %xmm2,%xmm14,%xmm14 | |
| 114 | ||
| 115 | vpclmulqdq $0x01,%xmm3,%xmm0,%xmm2 | |
| 116 | leaq (%r14,%r12,1),%r14 | |
| 117 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 118 | vpxor 16+8(%rsp),%xmm8,%xmm8 | |
| 119 | vpclmulqdq $0x11,%xmm3,%xmm0,%xmm3 | |
| 120 | vmovdqu 64+8(%rsp),%xmm0 | |
| 121 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 122 | movbeq 88(%r14),%r13 | |
| 123 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 124 | movbeq 80(%r14),%r12 | |
| 125 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 126 | movq %r13,32+8(%rsp) | |
| 127 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 128 | movq %r12,40+8(%rsp) | |
| 129 | vmovdqu 48-32(%r9),%xmm5 | |
| 130 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 131 | ||
| 132 | vmovups 48-128(%rcx),%xmm15 | |
| 133 | vpxor %xmm1,%xmm6,%xmm6 | |
| 134 | vpclmulqdq $0x00,%xmm5,%xmm0,%xmm1 | |
| 135 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 136 | vpxor %xmm2,%xmm6,%xmm6 | |
| 137 | vpclmulqdq $0x10,%xmm5,%xmm0,%xmm2 | |
| 138 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 139 | vpxor %xmm3,%xmm7,%xmm7 | |
| 140 | vpclmulqdq $0x01,%xmm5,%xmm0,%xmm3 | |
| 141 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 142 | vpclmulqdq $0x11,%xmm5,%xmm0,%xmm5 | |
| 143 | vmovdqu 80+8(%rsp),%xmm0 | |
| 144 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 145 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 146 | vpxor %xmm1,%xmm4,%xmm4 | |
| 147 | vmovdqu 64-32(%r9),%xmm1 | |
| 148 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 149 | ||
| 150 | vmovups 64-128(%rcx),%xmm15 | |
| 151 | vpxor %xmm2,%xmm6,%xmm6 | |
| 152 | vpclmulqdq $0x00,%xmm1,%xmm0,%xmm2 | |
| 153 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 154 | vpxor %xmm3,%xmm6,%xmm6 | |
| 155 | vpclmulqdq $0x10,%xmm1,%xmm0,%xmm3 | |
| 156 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 157 | movbeq 72(%r14),%r13 | |
| 158 | vpxor %xmm5,%xmm7,%xmm7 | |
| 159 | vpclmulqdq $0x01,%xmm1,%xmm0,%xmm5 | |
| 160 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 161 | movbeq 64(%r14),%r12 | |
| 162 | vpclmulqdq $0x11,%xmm1,%xmm0,%xmm1 | |
| 163 | vmovdqu 96+8(%rsp),%xmm0 | |
| 164 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 165 | movq %r13,48+8(%rsp) | |
| 166 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 167 | movq %r12,56+8(%rsp) | |
| 168 | vpxor %xmm2,%xmm4,%xmm4 | |
| 169 | vmovdqu 96-32(%r9),%xmm2 | |
| 170 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 171 | ||
| 172 | vmovups 80-128(%rcx),%xmm15 | |
| 173 | vpxor %xmm3,%xmm6,%xmm6 | |
| 174 | vpclmulqdq $0x00,%xmm2,%xmm0,%xmm3 | |
| 175 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 176 | vpxor %xmm5,%xmm6,%xmm6 | |
| 177 | vpclmulqdq $0x10,%xmm2,%xmm0,%xmm5 | |
| 178 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 179 | movbeq 56(%r14),%r13 | |
| 180 | vpxor %xmm1,%xmm7,%xmm7 | |
| 181 | vpclmulqdq $0x01,%xmm2,%xmm0,%xmm1 | |
| 182 | vpxor 112+8(%rsp),%xmm8,%xmm8 | |
| 183 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 184 | movbeq 48(%r14),%r12 | |
| 185 | vpclmulqdq $0x11,%xmm2,%xmm0,%xmm2 | |
| 186 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 187 | movq %r13,64+8(%rsp) | |
| 188 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 189 | movq %r12,72+8(%rsp) | |
| 190 | vpxor %xmm3,%xmm4,%xmm4 | |
| 191 | vmovdqu 112-32(%r9),%xmm3 | |
| 192 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 193 | ||
| 194 | vmovups 96-128(%rcx),%xmm15 | |
| 195 | vpxor %xmm5,%xmm6,%xmm6 | |
| 196 | vpclmulqdq $0x10,%xmm3,%xmm8,%xmm5 | |
| 197 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 198 | vpxor %xmm1,%xmm6,%xmm6 | |
| 199 | vpclmulqdq $0x01,%xmm3,%xmm8,%xmm1 | |
| 200 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 201 | movbeq 40(%r14),%r13 | |
| 202 | vpxor %xmm2,%xmm7,%xmm7 | |
| 203 | vpclmulqdq $0x00,%xmm3,%xmm8,%xmm2 | |
| 204 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 205 | movbeq 32(%r14),%r12 | |
| 206 | vpclmulqdq $0x11,%xmm3,%xmm8,%xmm8 | |
| 207 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 208 | movq %r13,80+8(%rsp) | |
| 209 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 210 | movq %r12,88+8(%rsp) | |
| 211 | vpxor %xmm5,%xmm6,%xmm6 | |
| 212 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 213 | vpxor %xmm1,%xmm6,%xmm6 | |
| 214 | ||
| 215 | vmovups 112-128(%rcx),%xmm15 | |
| 216 | vpslldq $8,%xmm6,%xmm5 | |
| 217 | vpxor %xmm2,%xmm4,%xmm4 | |
| 218 | vmovdqu 16(%r11),%xmm3 | |
| 219 | ||
| 220 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 221 | vpxor %xmm8,%xmm7,%xmm7 | |
| 222 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 223 | vpxor %xmm5,%xmm4,%xmm4 | |
| 224 | movbeq 24(%r14),%r13 | |
| 225 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 226 | movbeq 16(%r14),%r12 | |
| 227 | vpalignr $8,%xmm4,%xmm4,%xmm0 | |
| 228 | vpclmulqdq $0x10,%xmm3,%xmm4,%xmm4 | |
| 229 | movq %r13,96+8(%rsp) | |
| 230 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 231 | movq %r12,104+8(%rsp) | |
| 232 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 233 | vmovups 128-128(%rcx),%xmm1 | |
| 234 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 235 | ||
| 236 | vaesenc %xmm1,%xmm9,%xmm9 | |
| 237 | vmovups 144-128(%rcx),%xmm15 | |
| 238 | vaesenc %xmm1,%xmm10,%xmm10 | |
| 239 | vpsrldq $8,%xmm6,%xmm6 | |
| 240 | vaesenc %xmm1,%xmm11,%xmm11 | |
| 241 | vpxor %xmm6,%xmm7,%xmm7 | |
| 242 | vaesenc %xmm1,%xmm12,%xmm12 | |
| 243 | vpxor %xmm0,%xmm4,%xmm4 | |
| 244 | movbeq 8(%r14),%r13 | |
| 245 | vaesenc %xmm1,%xmm13,%xmm13 | |
| 246 | movbeq 0(%r14),%r12 | |
| 247 | vaesenc %xmm1,%xmm14,%xmm14 | |
| 248 | vmovups 160-128(%rcx),%xmm1 | |
| 249 | cmpl $12,%ebp // ICP uses 10,12,14 not 9,11,13 for rounds. | |
| 250 | jb .Lenc_tail | |
| 251 | ||
| 252 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 253 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 254 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 255 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 256 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 257 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 258 | ||
| 259 | vaesenc %xmm1,%xmm9,%xmm9 | |
| 260 | vaesenc %xmm1,%xmm10,%xmm10 | |
| 261 | vaesenc %xmm1,%xmm11,%xmm11 | |
| 262 | vaesenc %xmm1,%xmm12,%xmm12 | |
| 263 | vaesenc %xmm1,%xmm13,%xmm13 | |
| 264 | vmovups 176-128(%rcx),%xmm15 | |
| 265 | vaesenc %xmm1,%xmm14,%xmm14 | |
| 266 | vmovups 192-128(%rcx),%xmm1 | |
| 267 | cmpl $14,%ebp // ICP does not zero key schedule. | |
| 268 | jb .Lenc_tail | |
| 269 | ||
| 270 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 271 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 272 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 273 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 274 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 275 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 276 | ||
| 277 | vaesenc %xmm1,%xmm9,%xmm9 | |
| 278 | vaesenc %xmm1,%xmm10,%xmm10 | |
| 279 | vaesenc %xmm1,%xmm11,%xmm11 | |
| 280 | vaesenc %xmm1,%xmm12,%xmm12 | |
| 281 | vaesenc %xmm1,%xmm13,%xmm13 | |
| 282 | vmovups 208-128(%rcx),%xmm15 | |
| 283 | vaesenc %xmm1,%xmm14,%xmm14 | |
| 284 | vmovups 224-128(%rcx),%xmm1 | |
| 285 | jmp .Lenc_tail | |
| 286 | ||
| 287 | .align 32 | |
| 288 | .Lhandle_ctr32: | |
| 289 | vmovdqu (%r11),%xmm0 | |
| 290 | vpshufb %xmm0,%xmm1,%xmm6 | |
| 291 | vmovdqu 48(%r11),%xmm5 | |
| 292 | vpaddd 64(%r11),%xmm6,%xmm10 | |
| 293 | vpaddd %xmm5,%xmm6,%xmm11 | |
| 294 | vmovdqu 0-32(%r9),%xmm3 | |
| 295 | vpaddd %xmm5,%xmm10,%xmm12 | |
| 296 | vpshufb %xmm0,%xmm10,%xmm10 | |
| 297 | vpaddd %xmm5,%xmm11,%xmm13 | |
| 298 | vpshufb %xmm0,%xmm11,%xmm11 | |
| 299 | vpxor %xmm15,%xmm10,%xmm10 | |
| 300 | vpaddd %xmm5,%xmm12,%xmm14 | |
| 301 | vpshufb %xmm0,%xmm12,%xmm12 | |
| 302 | vpxor %xmm15,%xmm11,%xmm11 | |
| 303 | vpaddd %xmm5,%xmm13,%xmm1 | |
| 304 | vpshufb %xmm0,%xmm13,%xmm13 | |
| 305 | vpshufb %xmm0,%xmm14,%xmm14 | |
| 306 | vpshufb %xmm0,%xmm1,%xmm1 | |
| 307 | jmp .Lresume_ctr32 | |
| 308 | ||
| 309 | .align 32 | |
| 310 | .Lenc_tail: | |
| 311 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 312 | vmovdqu %xmm7,16+8(%rsp) | |
| 313 | vpalignr $8,%xmm4,%xmm4,%xmm8 | |
| 314 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 315 | vpclmulqdq $0x10,%xmm3,%xmm4,%xmm4 | |
| 316 | vpxor 0(%rdi),%xmm1,%xmm2 | |
| 317 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 318 | vpxor 16(%rdi),%xmm1,%xmm0 | |
| 319 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 320 | vpxor 32(%rdi),%xmm1,%xmm5 | |
| 321 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 322 | vpxor 48(%rdi),%xmm1,%xmm6 | |
| 323 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 324 | vpxor 64(%rdi),%xmm1,%xmm7 | |
| 325 | vpxor 80(%rdi),%xmm1,%xmm3 | |
| 326 | vmovdqu (%r8),%xmm1 | |
| 327 | ||
| 328 | vaesenclast %xmm2,%xmm9,%xmm9 | |
| 329 | vmovdqu 32(%r11),%xmm2 | |
| 330 | vaesenclast %xmm0,%xmm10,%xmm10 | |
| 331 | vpaddb %xmm2,%xmm1,%xmm0 | |
| 332 | movq %r13,112+8(%rsp) | |
| 333 | leaq 96(%rdi),%rdi | |
| 334 | vaesenclast %xmm5,%xmm11,%xmm11 | |
| 335 | vpaddb %xmm2,%xmm0,%xmm5 | |
| 336 | movq %r12,120+8(%rsp) | |
| 337 | leaq 96(%rsi),%rsi | |
| 338 | vmovdqu 0-128(%rcx),%xmm15 | |
| 339 | vaesenclast %xmm6,%xmm12,%xmm12 | |
| 340 | vpaddb %xmm2,%xmm5,%xmm6 | |
| 341 | vaesenclast %xmm7,%xmm13,%xmm13 | |
| 342 | vpaddb %xmm2,%xmm6,%xmm7 | |
| 343 | vaesenclast %xmm3,%xmm14,%xmm14 | |
| 344 | vpaddb %xmm2,%xmm7,%xmm3 | |
| 345 | ||
| 346 | addq $0x60,%r10 | |
| 347 | subq $0x6,%rdx | |
| 348 | jc .L6x_done | |
| 349 | ||
| 350 | vmovups %xmm9,-96(%rsi) | |
| 351 | vpxor %xmm15,%xmm1,%xmm9 | |
| 352 | vmovups %xmm10,-80(%rsi) | |
| 353 | vmovdqa %xmm0,%xmm10 | |
| 354 | vmovups %xmm11,-64(%rsi) | |
| 355 | vmovdqa %xmm5,%xmm11 | |
| 356 | vmovups %xmm12,-48(%rsi) | |
| 357 | vmovdqa %xmm6,%xmm12 | |
| 358 | vmovups %xmm13,-32(%rsi) | |
| 359 | vmovdqa %xmm7,%xmm13 | |
| 360 | vmovups %xmm14,-16(%rsi) | |
| 361 | vmovdqa %xmm3,%xmm14 | |
| 362 | vmovdqu 32+8(%rsp),%xmm7 | |
| 363 | jmp .Loop6x | |
| 364 | ||
| 365 | .L6x_done: | |
| 366 | vpxor 16+8(%rsp),%xmm8,%xmm8 | |
| 367 | vpxor %xmm4,%xmm8,%xmm8 | |
| 368 | ||
| b844489e | 369 | RET |
| d9655c5b | 370 | .cfi_endproc |
| 31b160f0 | 371 | .size _aesni_ctr32_ghash_6x,.-_aesni_ctr32_ghash_6x |
| 5b3b7955 AF |
372 | #endif /* ifdef HAVE_MOVBE */ |
| 373 | ||
| 374 | .type _aesni_ctr32_ghash_no_movbe_6x,@function | |
| 375 | .align 32 | |
| 376 | _aesni_ctr32_ghash_no_movbe_6x: | |
| d9655c5b | 377 | .cfi_startproc |
| 5b3b7955 AF |
378 | vmovdqu 32(%r11),%xmm2 |
| 379 | subq $6,%rdx | |
| 380 | vpxor %xmm4,%xmm4,%xmm4 | |
| 381 | vmovdqu 0-128(%rcx),%xmm15 | |
| 382 | vpaddb %xmm2,%xmm1,%xmm10 | |
| 383 | vpaddb %xmm2,%xmm10,%xmm11 | |
| 384 | vpaddb %xmm2,%xmm11,%xmm12 | |
| 385 | vpaddb %xmm2,%xmm12,%xmm13 | |
| 386 | vpaddb %xmm2,%xmm13,%xmm14 | |
| 387 | vpxor %xmm15,%xmm1,%xmm9 | |
| 388 | vmovdqu %xmm4,16+8(%rsp) | |
| 389 | jmp .Loop6x_nmb | |
| 390 | ||
| 391 | .align 32 | |
| 392 | .Loop6x_nmb: | |
| 393 | addl $100663296,%ebx | |
| 394 | jc .Lhandle_ctr32_nmb | |
| 395 | vmovdqu 0-32(%r9),%xmm3 | |
| 396 | vpaddb %xmm2,%xmm14,%xmm1 | |
| 397 | vpxor %xmm15,%xmm10,%xmm10 | |
| 398 | vpxor %xmm15,%xmm11,%xmm11 | |
| 399 | ||
| 400 | .Lresume_ctr32_nmb: | |
| 401 | vmovdqu %xmm1,(%r8) | |
| 402 | vpclmulqdq $0x10,%xmm3,%xmm7,%xmm5 | |
| 403 | vpxor %xmm15,%xmm12,%xmm12 | |
| 404 | vmovups 16-128(%rcx),%xmm2 | |
| 405 | vpclmulqdq $0x01,%xmm3,%xmm7,%xmm6 | |
| 406 | xorq %r12,%r12 | |
| 407 | cmpq %r14,%r15 | |
| 408 | ||
| 409 | vaesenc %xmm2,%xmm9,%xmm9 | |
| 410 | vmovdqu 48+8(%rsp),%xmm0 | |
| 411 | vpxor %xmm15,%xmm13,%xmm13 | |
| 412 | vpclmulqdq $0x00,%xmm3,%xmm7,%xmm1 | |
| 413 | vaesenc %xmm2,%xmm10,%xmm10 | |
| 414 | vpxor %xmm15,%xmm14,%xmm14 | |
| 415 | setnc %r12b | |
| 416 | vpclmulqdq $0x11,%xmm3,%xmm7,%xmm7 | |
| 417 | vaesenc %xmm2,%xmm11,%xmm11 | |
| 418 | vmovdqu 16-32(%r9),%xmm3 | |
| 419 | negq %r12 | |
| 420 | vaesenc %xmm2,%xmm12,%xmm12 | |
| 421 | vpxor %xmm5,%xmm6,%xmm6 | |
| 422 | vpclmulqdq $0x00,%xmm3,%xmm0,%xmm5 | |
| 423 | vpxor %xmm4,%xmm8,%xmm8 | |
| 424 | vaesenc %xmm2,%xmm13,%xmm13 | |
| 425 | vpxor %xmm5,%xmm1,%xmm4 | |
| 426 | andq $0x60,%r12 | |
| 427 | vmovups 32-128(%rcx),%xmm15 | |
| 428 | vpclmulqdq $0x10,%xmm3,%xmm0,%xmm1 | |
| 429 | vaesenc %xmm2,%xmm14,%xmm14 | |
| 430 | ||
| 431 | vpclmulqdq $0x01,%xmm3,%xmm0,%xmm2 | |
| 432 | leaq (%r14,%r12,1),%r14 | |
| 433 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 434 | vpxor 16+8(%rsp),%xmm8,%xmm8 | |
| 435 | vpclmulqdq $0x11,%xmm3,%xmm0,%xmm3 | |
| 436 | vmovdqu 64+8(%rsp),%xmm0 | |
| 437 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 438 | movq 88(%r14),%r13 | |
| 439 | bswapq %r13 | |
| 440 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 441 | movq 80(%r14),%r12 | |
| 442 | bswapq %r12 | |
| 443 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 444 | movq %r13,32+8(%rsp) | |
| 445 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 446 | movq %r12,40+8(%rsp) | |
| 447 | vmovdqu 48-32(%r9),%xmm5 | |
| 448 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 449 | ||
| 450 | vmovups 48-128(%rcx),%xmm15 | |
| 451 | vpxor %xmm1,%xmm6,%xmm6 | |
| 452 | vpclmulqdq $0x00,%xmm5,%xmm0,%xmm1 | |
| 453 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 454 | vpxor %xmm2,%xmm6,%xmm6 | |
| 455 | vpclmulqdq $0x10,%xmm5,%xmm0,%xmm2 | |
| 456 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 457 | vpxor %xmm3,%xmm7,%xmm7 | |
| 458 | vpclmulqdq $0x01,%xmm5,%xmm0,%xmm3 | |
| 459 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 460 | vpclmulqdq $0x11,%xmm5,%xmm0,%xmm5 | |
| 461 | vmovdqu 80+8(%rsp),%xmm0 | |
| 462 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 463 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 464 | vpxor %xmm1,%xmm4,%xmm4 | |
| 465 | vmovdqu 64-32(%r9),%xmm1 | |
| 466 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 467 | ||
| 468 | vmovups 64-128(%rcx),%xmm15 | |
| 469 | vpxor %xmm2,%xmm6,%xmm6 | |
| 470 | vpclmulqdq $0x00,%xmm1,%xmm0,%xmm2 | |
| 471 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 472 | vpxor %xmm3,%xmm6,%xmm6 | |
| 473 | vpclmulqdq $0x10,%xmm1,%xmm0,%xmm3 | |
| 474 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 475 | movq 72(%r14),%r13 | |
| 476 | bswapq %r13 | |
| 477 | vpxor %xmm5,%xmm7,%xmm7 | |
| 478 | vpclmulqdq $0x01,%xmm1,%xmm0,%xmm5 | |
| 479 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 480 | movq 64(%r14),%r12 | |
| 481 | bswapq %r12 | |
| 482 | vpclmulqdq $0x11,%xmm1,%xmm0,%xmm1 | |
| 483 | vmovdqu 96+8(%rsp),%xmm0 | |
| 484 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 485 | movq %r13,48+8(%rsp) | |
| 486 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 487 | movq %r12,56+8(%rsp) | |
| 488 | vpxor %xmm2,%xmm4,%xmm4 | |
| 489 | vmovdqu 96-32(%r9),%xmm2 | |
| 490 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 491 | ||
| 492 | vmovups 80-128(%rcx),%xmm15 | |
| 493 | vpxor %xmm3,%xmm6,%xmm6 | |
| 494 | vpclmulqdq $0x00,%xmm2,%xmm0,%xmm3 | |
| 495 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 496 | vpxor %xmm5,%xmm6,%xmm6 | |
| 497 | vpclmulqdq $0x10,%xmm2,%xmm0,%xmm5 | |
| 498 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 499 | movq 56(%r14),%r13 | |
| 500 | bswapq %r13 | |
| 501 | vpxor %xmm1,%xmm7,%xmm7 | |
| 502 | vpclmulqdq $0x01,%xmm2,%xmm0,%xmm1 | |
| 503 | vpxor 112+8(%rsp),%xmm8,%xmm8 | |
| 504 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 505 | movq 48(%r14),%r12 | |
| 506 | bswapq %r12 | |
| 507 | vpclmulqdq $0x11,%xmm2,%xmm0,%xmm2 | |
| 508 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 509 | movq %r13,64+8(%rsp) | |
| 510 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 511 | movq %r12,72+8(%rsp) | |
| 512 | vpxor %xmm3,%xmm4,%xmm4 | |
| 513 | vmovdqu 112-32(%r9),%xmm3 | |
| 514 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 515 | ||
| 516 | vmovups 96-128(%rcx),%xmm15 | |
| 517 | vpxor %xmm5,%xmm6,%xmm6 | |
| 518 | vpclmulqdq $0x10,%xmm3,%xmm8,%xmm5 | |
| 519 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 520 | vpxor %xmm1,%xmm6,%xmm6 | |
| 521 | vpclmulqdq $0x01,%xmm3,%xmm8,%xmm1 | |
| 522 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 523 | movq 40(%r14),%r13 | |
| 524 | bswapq %r13 | |
| 525 | vpxor %xmm2,%xmm7,%xmm7 | |
| 526 | vpclmulqdq $0x00,%xmm3,%xmm8,%xmm2 | |
| 527 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 528 | movq 32(%r14),%r12 | |
| 529 | bswapq %r12 | |
| 530 | vpclmulqdq $0x11,%xmm3,%xmm8,%xmm8 | |
| 531 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 532 | movq %r13,80+8(%rsp) | |
| 533 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 534 | movq %r12,88+8(%rsp) | |
| 535 | vpxor %xmm5,%xmm6,%xmm6 | |
| 536 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 537 | vpxor %xmm1,%xmm6,%xmm6 | |
| 538 | ||
| 539 | vmovups 112-128(%rcx),%xmm15 | |
| 540 | vpslldq $8,%xmm6,%xmm5 | |
| 541 | vpxor %xmm2,%xmm4,%xmm4 | |
| 542 | vmovdqu 16(%r11),%xmm3 | |
| 543 | ||
| 544 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 545 | vpxor %xmm8,%xmm7,%xmm7 | |
| 546 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 547 | vpxor %xmm5,%xmm4,%xmm4 | |
| 548 | movq 24(%r14),%r13 | |
| 549 | bswapq %r13 | |
| 550 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 551 | movq 16(%r14),%r12 | |
| 552 | bswapq %r12 | |
| 553 | vpalignr $8,%xmm4,%xmm4,%xmm0 | |
| 554 | vpclmulqdq $0x10,%xmm3,%xmm4,%xmm4 | |
| 555 | movq %r13,96+8(%rsp) | |
| 556 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 557 | movq %r12,104+8(%rsp) | |
| 558 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 559 | vmovups 128-128(%rcx),%xmm1 | |
| 560 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 561 | ||
| 562 | vaesenc %xmm1,%xmm9,%xmm9 | |
| 563 | vmovups 144-128(%rcx),%xmm15 | |
| 564 | vaesenc %xmm1,%xmm10,%xmm10 | |
| 565 | vpsrldq $8,%xmm6,%xmm6 | |
| 566 | vaesenc %xmm1,%xmm11,%xmm11 | |
| 567 | vpxor %xmm6,%xmm7,%xmm7 | |
| 568 | vaesenc %xmm1,%xmm12,%xmm12 | |
| 569 | vpxor %xmm0,%xmm4,%xmm4 | |
| 570 | movq 8(%r14),%r13 | |
| 571 | bswapq %r13 | |
| 572 | vaesenc %xmm1,%xmm13,%xmm13 | |
| 573 | movq 0(%r14),%r12 | |
| 574 | bswapq %r12 | |
| 575 | vaesenc %xmm1,%xmm14,%xmm14 | |
| 576 | vmovups 160-128(%rcx),%xmm1 | |
| 577 | cmpl $12,%ebp // ICP uses 10,12,14 not 9,11,13 for rounds. | |
| 578 | jb .Lenc_tail_nmb | |
| 579 | ||
| 580 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 581 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 582 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 583 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 584 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 585 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 586 | ||
| 587 | vaesenc %xmm1,%xmm9,%xmm9 | |
| 588 | vaesenc %xmm1,%xmm10,%xmm10 | |
| 589 | vaesenc %xmm1,%xmm11,%xmm11 | |
| 590 | vaesenc %xmm1,%xmm12,%xmm12 | |
| 591 | vaesenc %xmm1,%xmm13,%xmm13 | |
| 592 | vmovups 176-128(%rcx),%xmm15 | |
| 593 | vaesenc %xmm1,%xmm14,%xmm14 | |
| 594 | vmovups 192-128(%rcx),%xmm1 | |
| 595 | cmpl $14,%ebp // ICP does not zero key schedule. | |
| 596 | jb .Lenc_tail_nmb | |
| 597 | ||
| 598 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 599 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 600 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 601 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 602 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 603 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 604 | ||
| 605 | vaesenc %xmm1,%xmm9,%xmm9 | |
| 606 | vaesenc %xmm1,%xmm10,%xmm10 | |
| 607 | vaesenc %xmm1,%xmm11,%xmm11 | |
| 608 | vaesenc %xmm1,%xmm12,%xmm12 | |
| 609 | vaesenc %xmm1,%xmm13,%xmm13 | |
| 610 | vmovups 208-128(%rcx),%xmm15 | |
| 611 | vaesenc %xmm1,%xmm14,%xmm14 | |
| 612 | vmovups 224-128(%rcx),%xmm1 | |
| 613 | jmp .Lenc_tail_nmb | |
| 614 | ||
| 615 | .align 32 | |
| 616 | .Lhandle_ctr32_nmb: | |
| 617 | vmovdqu (%r11),%xmm0 | |
| 618 | vpshufb %xmm0,%xmm1,%xmm6 | |
| 619 | vmovdqu 48(%r11),%xmm5 | |
| 620 | vpaddd 64(%r11),%xmm6,%xmm10 | |
| 621 | vpaddd %xmm5,%xmm6,%xmm11 | |
| 622 | vmovdqu 0-32(%r9),%xmm3 | |
| 623 | vpaddd %xmm5,%xmm10,%xmm12 | |
| 624 | vpshufb %xmm0,%xmm10,%xmm10 | |
| 625 | vpaddd %xmm5,%xmm11,%xmm13 | |
| 626 | vpshufb %xmm0,%xmm11,%xmm11 | |
| 627 | vpxor %xmm15,%xmm10,%xmm10 | |
| 628 | vpaddd %xmm5,%xmm12,%xmm14 | |
| 629 | vpshufb %xmm0,%xmm12,%xmm12 | |
| 630 | vpxor %xmm15,%xmm11,%xmm11 | |
| 631 | vpaddd %xmm5,%xmm13,%xmm1 | |
| 632 | vpshufb %xmm0,%xmm13,%xmm13 | |
| 633 | vpshufb %xmm0,%xmm14,%xmm14 | |
| 634 | vpshufb %xmm0,%xmm1,%xmm1 | |
| 635 | jmp .Lresume_ctr32_nmb | |
| 636 | ||
| 637 | .align 32 | |
| 638 | .Lenc_tail_nmb: | |
| 639 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 640 | vmovdqu %xmm7,16+8(%rsp) | |
| 641 | vpalignr $8,%xmm4,%xmm4,%xmm8 | |
| 642 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 643 | vpclmulqdq $0x10,%xmm3,%xmm4,%xmm4 | |
| 644 | vpxor 0(%rdi),%xmm1,%xmm2 | |
| 645 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 646 | vpxor 16(%rdi),%xmm1,%xmm0 | |
| 647 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 648 | vpxor 32(%rdi),%xmm1,%xmm5 | |
| 649 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 650 | vpxor 48(%rdi),%xmm1,%xmm6 | |
| 651 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 652 | vpxor 64(%rdi),%xmm1,%xmm7 | |
| 653 | vpxor 80(%rdi),%xmm1,%xmm3 | |
| 654 | vmovdqu (%r8),%xmm1 | |
| 655 | ||
| 656 | vaesenclast %xmm2,%xmm9,%xmm9 | |
| 657 | vmovdqu 32(%r11),%xmm2 | |
| 658 | vaesenclast %xmm0,%xmm10,%xmm10 | |
| 659 | vpaddb %xmm2,%xmm1,%xmm0 | |
| 660 | movq %r13,112+8(%rsp) | |
| 661 | leaq 96(%rdi),%rdi | |
| 662 | vaesenclast %xmm5,%xmm11,%xmm11 | |
| 663 | vpaddb %xmm2,%xmm0,%xmm5 | |
| 664 | movq %r12,120+8(%rsp) | |
| 665 | leaq 96(%rsi),%rsi | |
| 666 | vmovdqu 0-128(%rcx),%xmm15 | |
| 667 | vaesenclast %xmm6,%xmm12,%xmm12 | |
| 668 | vpaddb %xmm2,%xmm5,%xmm6 | |
| 669 | vaesenclast %xmm7,%xmm13,%xmm13 | |
| 670 | vpaddb %xmm2,%xmm6,%xmm7 | |
| 671 | vaesenclast %xmm3,%xmm14,%xmm14 | |
| 672 | vpaddb %xmm2,%xmm7,%xmm3 | |
| 673 | ||
| 674 | addq $0x60,%r10 | |
| 675 | subq $0x6,%rdx | |
| 676 | jc .L6x_done_nmb | |
| 677 | ||
| 678 | vmovups %xmm9,-96(%rsi) | |
| 679 | vpxor %xmm15,%xmm1,%xmm9 | |
| 680 | vmovups %xmm10,-80(%rsi) | |
| 681 | vmovdqa %xmm0,%xmm10 | |
| 682 | vmovups %xmm11,-64(%rsi) | |
| 683 | vmovdqa %xmm5,%xmm11 | |
| 684 | vmovups %xmm12,-48(%rsi) | |
| 685 | vmovdqa %xmm6,%xmm12 | |
| 686 | vmovups %xmm13,-32(%rsi) | |
| 687 | vmovdqa %xmm7,%xmm13 | |
| 688 | vmovups %xmm14,-16(%rsi) | |
| 689 | vmovdqa %xmm3,%xmm14 | |
| 690 | vmovdqu 32+8(%rsp),%xmm7 | |
| 691 | jmp .Loop6x_nmb | |
| 692 | ||
| 693 | .L6x_done_nmb: | |
| 694 | vpxor 16+8(%rsp),%xmm8,%xmm8 | |
| 695 | vpxor %xmm4,%xmm8,%xmm8 | |
| 696 | ||
| b844489e | 697 | RET |
| d9655c5b | 698 | .cfi_endproc |
| 5b3b7955 AF |
699 | .size _aesni_ctr32_ghash_no_movbe_6x,.-_aesni_ctr32_ghash_no_movbe_6x |
| 700 | ||
| 31b160f0 AF |
701 | .globl aesni_gcm_decrypt |
| 702 | .type aesni_gcm_decrypt,@function | |
| 703 | .align 32 | |
| 704 | aesni_gcm_decrypt: | |
| 705 | .cfi_startproc | |
| 706 | xorq %r10,%r10 | |
| 707 | cmpq $0x60,%rdx | |
| 708 | jb .Lgcm_dec_abort | |
| 709 | ||
| 710 | leaq (%rsp),%rax | |
| 711 | .cfi_def_cfa_register %rax | |
| 712 | pushq %rbx | |
| 713 | .cfi_offset %rbx,-16 | |
| 714 | pushq %rbp | |
| 715 | .cfi_offset %rbp,-24 | |
| 716 | pushq %r12 | |
| 717 | .cfi_offset %r12,-32 | |
| 718 | pushq %r13 | |
| 719 | .cfi_offset %r13,-40 | |
| 720 | pushq %r14 | |
| 721 | .cfi_offset %r14,-48 | |
| 722 | pushq %r15 | |
| 723 | .cfi_offset %r15,-56 | |
| e8beeaa1 AF |
724 | pushq %r9 |
| 725 | .cfi_offset %r9,-64 | |
| 31b160f0 AF |
726 | vzeroupper |
| 727 | ||
| 728 | vmovdqu (%r8),%xmm1 | |
| 729 | addq $-128,%rsp | |
| 730 | movl 12(%r8),%ebx | |
| 731 | leaq .Lbswap_mask(%rip),%r11 | |
| 732 | leaq -128(%rcx),%r14 | |
| 733 | movq $0xf80,%r15 | |
| 734 | vmovdqu (%r9),%xmm8 | |
| 735 | andq $-128,%rsp | |
| 736 | vmovdqu (%r11),%xmm0 | |
| 737 | leaq 128(%rcx),%rcx | |
| e8beeaa1 AF |
738 | movq 32(%r9),%r9 |
| 739 | leaq 32(%r9),%r9 | |
| 31b160f0 AF |
740 | movl 504-128(%rcx),%ebp // ICP has a larger offset for rounds. |
| 741 | vpshufb %xmm0,%xmm8,%xmm8 | |
| 742 | ||
| 743 | andq %r15,%r14 | |
| 744 | andq %rsp,%r15 | |
| 745 | subq %r14,%r15 | |
| 746 | jc .Ldec_no_key_aliasing | |
| 747 | cmpq $768,%r15 | |
| 748 | jnc .Ldec_no_key_aliasing | |
| 749 | subq %r15,%rsp | |
| 750 | .Ldec_no_key_aliasing: | |
| 751 | ||
| 752 | vmovdqu 80(%rdi),%xmm7 | |
| 753 | leaq (%rdi),%r14 | |
| 754 | vmovdqu 64(%rdi),%xmm4 | |
| 755 | leaq -192(%rdi,%rdx,1),%r15 | |
| 756 | vmovdqu 48(%rdi),%xmm5 | |
| 757 | shrq $4,%rdx | |
| 758 | xorq %r10,%r10 | |
| 759 | vmovdqu 32(%rdi),%xmm6 | |
| 760 | vpshufb %xmm0,%xmm7,%xmm7 | |
| 761 | vmovdqu 16(%rdi),%xmm2 | |
| 762 | vpshufb %xmm0,%xmm4,%xmm4 | |
| 763 | vmovdqu (%rdi),%xmm3 | |
| 764 | vpshufb %xmm0,%xmm5,%xmm5 | |
| 765 | vmovdqu %xmm4,48(%rsp) | |
| 766 | vpshufb %xmm0,%xmm6,%xmm6 | |
| 767 | vmovdqu %xmm5,64(%rsp) | |
| 768 | vpshufb %xmm0,%xmm2,%xmm2 | |
| 769 | vmovdqu %xmm6,80(%rsp) | |
| 770 | vpshufb %xmm0,%xmm3,%xmm3 | |
| 771 | vmovdqu %xmm2,96(%rsp) | |
| 772 | vmovdqu %xmm3,112(%rsp) | |
| 773 | ||
| 5b3b7955 AF |
774 | #ifdef HAVE_MOVBE |
| 775 | #ifdef _KERNEL | |
| 776 | testl $1,gcm_avx_can_use_movbe(%rip) | |
| 777 | #else | |
| 778 | testl $1,gcm_avx_can_use_movbe@GOTPCREL(%rip) | |
| 779 | #endif | |
| 780 | jz 1f | |
| 31b160f0 | 781 | call _aesni_ctr32_ghash_6x |
| 5b3b7955 AF |
782 | jmp 2f |
| 783 | 1: | |
| 784 | #endif | |
| 785 | call _aesni_ctr32_ghash_no_movbe_6x | |
| 786 | 2: | |
| 31b160f0 AF |
787 | vmovups %xmm9,-96(%rsi) |
| 788 | vmovups %xmm10,-80(%rsi) | |
| 789 | vmovups %xmm11,-64(%rsi) | |
| 790 | vmovups %xmm12,-48(%rsi) | |
| 791 | vmovups %xmm13,-32(%rsi) | |
| 792 | vmovups %xmm14,-16(%rsi) | |
| 793 | ||
| 794 | vpshufb (%r11),%xmm8,%xmm8 | |
| e8beeaa1 AF |
795 | movq -56(%rax),%r9 |
| 796 | .cfi_restore %r9 | |
| 797 | vmovdqu %xmm8,(%r9) | |
| 31b160f0 AF |
798 | |
| 799 | vzeroupper | |
| 800 | movq -48(%rax),%r15 | |
| 801 | .cfi_restore %r15 | |
| 802 | movq -40(%rax),%r14 | |
| 803 | .cfi_restore %r14 | |
| 804 | movq -32(%rax),%r13 | |
| 805 | .cfi_restore %r13 | |
| 806 | movq -24(%rax),%r12 | |
| 807 | .cfi_restore %r12 | |
| 808 | movq -16(%rax),%rbp | |
| 809 | .cfi_restore %rbp | |
| 810 | movq -8(%rax),%rbx | |
| 811 | .cfi_restore %rbx | |
| 812 | leaq (%rax),%rsp | |
| 813 | .cfi_def_cfa_register %rsp | |
| 814 | .Lgcm_dec_abort: | |
| 815 | movq %r10,%rax | |
| b844489e | 816 | RET |
| 31b160f0 AF |
817 | .cfi_endproc |
| 818 | .size aesni_gcm_decrypt,.-aesni_gcm_decrypt | |
| 819 | .type _aesni_ctr32_6x,@function | |
| 820 | .align 32 | |
| 821 | _aesni_ctr32_6x: | |
| d9655c5b | 822 | .cfi_startproc |
| 31b160f0 AF |
823 | vmovdqu 0-128(%rcx),%xmm4 |
| 824 | vmovdqu 32(%r11),%xmm2 | |
| 825 | leaq -2(%rbp),%r13 // ICP uses 10,12,14 not 9,11,13 for rounds. | |
| 826 | vmovups 16-128(%rcx),%xmm15 | |
| 827 | leaq 32-128(%rcx),%r12 | |
| 828 | vpxor %xmm4,%xmm1,%xmm9 | |
| 829 | addl $100663296,%ebx | |
| 830 | jc .Lhandle_ctr32_2 | |
| 831 | vpaddb %xmm2,%xmm1,%xmm10 | |
| 832 | vpaddb %xmm2,%xmm10,%xmm11 | |
| 833 | vpxor %xmm4,%xmm10,%xmm10 | |
| 834 | vpaddb %xmm2,%xmm11,%xmm12 | |
| 835 | vpxor %xmm4,%xmm11,%xmm11 | |
| 836 | vpaddb %xmm2,%xmm12,%xmm13 | |
| 837 | vpxor %xmm4,%xmm12,%xmm12 | |
| 838 | vpaddb %xmm2,%xmm13,%xmm14 | |
| 839 | vpxor %xmm4,%xmm13,%xmm13 | |
| 840 | vpaddb %xmm2,%xmm14,%xmm1 | |
| 841 | vpxor %xmm4,%xmm14,%xmm14 | |
| 842 | jmp .Loop_ctr32 | |
| 843 | ||
| 844 | .align 16 | |
| 845 | .Loop_ctr32: | |
| 846 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 847 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 848 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 849 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 850 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 851 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 852 | vmovups (%r12),%xmm15 | |
| 853 | leaq 16(%r12),%r12 | |
| 854 | decl %r13d | |
| 855 | jnz .Loop_ctr32 | |
| 856 | ||
| 857 | vmovdqu (%r12),%xmm3 | |
| 858 | vaesenc %xmm15,%xmm9,%xmm9 | |
| 859 | vpxor 0(%rdi),%xmm3,%xmm4 | |
| 860 | vaesenc %xmm15,%xmm10,%xmm10 | |
| 861 | vpxor 16(%rdi),%xmm3,%xmm5 | |
| 862 | vaesenc %xmm15,%xmm11,%xmm11 | |
| 863 | vpxor 32(%rdi),%xmm3,%xmm6 | |
| 864 | vaesenc %xmm15,%xmm12,%xmm12 | |
| 865 | vpxor 48(%rdi),%xmm3,%xmm8 | |
| 866 | vaesenc %xmm15,%xmm13,%xmm13 | |
| 867 | vpxor 64(%rdi),%xmm3,%xmm2 | |
| 868 | vaesenc %xmm15,%xmm14,%xmm14 | |
| 869 | vpxor 80(%rdi),%xmm3,%xmm3 | |
| 870 | leaq 96(%rdi),%rdi | |
| 871 | ||
| 872 | vaesenclast %xmm4,%xmm9,%xmm9 | |
| 873 | vaesenclast %xmm5,%xmm10,%xmm10 | |
| 874 | vaesenclast %xmm6,%xmm11,%xmm11 | |
| 875 | vaesenclast %xmm8,%xmm12,%xmm12 | |
| 876 | vaesenclast %xmm2,%xmm13,%xmm13 | |
| 877 | vaesenclast %xmm3,%xmm14,%xmm14 | |
| 878 | vmovups %xmm9,0(%rsi) | |
| 879 | vmovups %xmm10,16(%rsi) | |
| 880 | vmovups %xmm11,32(%rsi) | |
| 881 | vmovups %xmm12,48(%rsi) | |
| 882 | vmovups %xmm13,64(%rsi) | |
| 883 | vmovups %xmm14,80(%rsi) | |
| 884 | leaq 96(%rsi),%rsi | |
| 885 | ||
| b844489e | 886 | RET |
| 31b160f0 AF |
887 | .align 32 |
| 888 | .Lhandle_ctr32_2: | |
| 889 | vpshufb %xmm0,%xmm1,%xmm6 | |
| 890 | vmovdqu 48(%r11),%xmm5 | |
| 891 | vpaddd 64(%r11),%xmm6,%xmm10 | |
| 892 | vpaddd %xmm5,%xmm6,%xmm11 | |
| 893 | vpaddd %xmm5,%xmm10,%xmm12 | |
| 894 | vpshufb %xmm0,%xmm10,%xmm10 | |
| 895 | vpaddd %xmm5,%xmm11,%xmm13 | |
| 896 | vpshufb %xmm0,%xmm11,%xmm11 | |
| 897 | vpxor %xmm4,%xmm10,%xmm10 | |
| 898 | vpaddd %xmm5,%xmm12,%xmm14 | |
| 899 | vpshufb %xmm0,%xmm12,%xmm12 | |
| 900 | vpxor %xmm4,%xmm11,%xmm11 | |
| 901 | vpaddd %xmm5,%xmm13,%xmm1 | |
| 902 | vpshufb %xmm0,%xmm13,%xmm13 | |
| 903 | vpxor %xmm4,%xmm12,%xmm12 | |
| 904 | vpshufb %xmm0,%xmm14,%xmm14 | |
| 905 | vpxor %xmm4,%xmm13,%xmm13 | |
| 906 | vpshufb %xmm0,%xmm1,%xmm1 | |
| 907 | vpxor %xmm4,%xmm14,%xmm14 | |
| 908 | jmp .Loop_ctr32 | |
| d9655c5b | 909 | .cfi_endproc |
| 31b160f0 AF |
910 | .size _aesni_ctr32_6x,.-_aesni_ctr32_6x |
| 911 | ||
| 912 | .globl aesni_gcm_encrypt | |
| 913 | .type aesni_gcm_encrypt,@function | |
| 914 | .align 32 | |
| 915 | aesni_gcm_encrypt: | |
| 916 | .cfi_startproc | |
| 917 | xorq %r10,%r10 | |
| 918 | cmpq $288,%rdx | |
| 919 | jb .Lgcm_enc_abort | |
| 920 | ||
| 921 | leaq (%rsp),%rax | |
| 922 | .cfi_def_cfa_register %rax | |
| 923 | pushq %rbx | |
| 924 | .cfi_offset %rbx,-16 | |
| 925 | pushq %rbp | |
| 926 | .cfi_offset %rbp,-24 | |
| 927 | pushq %r12 | |
| 928 | .cfi_offset %r12,-32 | |
| 929 | pushq %r13 | |
| 930 | .cfi_offset %r13,-40 | |
| 931 | pushq %r14 | |
| 932 | .cfi_offset %r14,-48 | |
| 933 | pushq %r15 | |
| 934 | .cfi_offset %r15,-56 | |
| e8beeaa1 AF |
935 | pushq %r9 |
| 936 | .cfi_offset %r9,-64 | |
| 31b160f0 AF |
937 | vzeroupper |
| 938 | ||
| 939 | vmovdqu (%r8),%xmm1 | |
| 940 | addq $-128,%rsp | |
| 941 | movl 12(%r8),%ebx | |
| 942 | leaq .Lbswap_mask(%rip),%r11 | |
| 943 | leaq -128(%rcx),%r14 | |
| 944 | movq $0xf80,%r15 | |
| 945 | leaq 128(%rcx),%rcx | |
| 946 | vmovdqu (%r11),%xmm0 | |
| 947 | andq $-128,%rsp | |
| 948 | movl 504-128(%rcx),%ebp // ICP has an larger offset for rounds. | |
| 949 | ||
| 950 | andq %r15,%r14 | |
| 951 | andq %rsp,%r15 | |
| 952 | subq %r14,%r15 | |
| 953 | jc .Lenc_no_key_aliasing | |
| 954 | cmpq $768,%r15 | |
| 955 | jnc .Lenc_no_key_aliasing | |
| 956 | subq %r15,%rsp | |
| 957 | .Lenc_no_key_aliasing: | |
| 958 | ||
| 959 | leaq (%rsi),%r14 | |
| 960 | leaq -192(%rsi,%rdx,1),%r15 | |
| 961 | shrq $4,%rdx | |
| 962 | ||
| 963 | call _aesni_ctr32_6x | |
| 964 | vpshufb %xmm0,%xmm9,%xmm8 | |
| 965 | vpshufb %xmm0,%xmm10,%xmm2 | |
| 966 | vmovdqu %xmm8,112(%rsp) | |
| 967 | vpshufb %xmm0,%xmm11,%xmm4 | |
| 968 | vmovdqu %xmm2,96(%rsp) | |
| 969 | vpshufb %xmm0,%xmm12,%xmm5 | |
| 970 | vmovdqu %xmm4,80(%rsp) | |
| 971 | vpshufb %xmm0,%xmm13,%xmm6 | |
| 972 | vmovdqu %xmm5,64(%rsp) | |
| 973 | vpshufb %xmm0,%xmm14,%xmm7 | |
| 974 | vmovdqu %xmm6,48(%rsp) | |
| 975 | ||
| 976 | call _aesni_ctr32_6x | |
| 977 | ||
| 978 | vmovdqu (%r9),%xmm8 | |
| e8beeaa1 AF |
979 | movq 32(%r9),%r9 |
| 980 | leaq 32(%r9),%r9 | |
| 31b160f0 AF |
981 | subq $12,%rdx |
| 982 | movq $192,%r10 | |
| 983 | vpshufb %xmm0,%xmm8,%xmm8 | |
| 984 | ||
| 5b3b7955 AF |
985 | #ifdef HAVE_MOVBE |
| 986 | #ifdef _KERNEL | |
| 987 | testl $1,gcm_avx_can_use_movbe(%rip) | |
| 988 | #else | |
| 989 | testl $1,gcm_avx_can_use_movbe@GOTPCREL(%rip) | |
| 990 | #endif | |
| 991 | jz 1f | |
| 31b160f0 | 992 | call _aesni_ctr32_ghash_6x |
| 5b3b7955 AF |
993 | jmp 2f |
| 994 | 1: | |
| 995 | #endif | |
| 996 | call _aesni_ctr32_ghash_no_movbe_6x | |
| 997 | 2: | |
| 31b160f0 AF |
998 | vmovdqu 32(%rsp),%xmm7 |
| 999 | vmovdqu (%r11),%xmm0 | |
| 1000 | vmovdqu 0-32(%r9),%xmm3 | |
| 1001 | vpunpckhqdq %xmm7,%xmm7,%xmm1 | |
| 1002 | vmovdqu 32-32(%r9),%xmm15 | |
| 1003 | vmovups %xmm9,-96(%rsi) | |
| 1004 | vpshufb %xmm0,%xmm9,%xmm9 | |
| 1005 | vpxor %xmm7,%xmm1,%xmm1 | |
| 1006 | vmovups %xmm10,-80(%rsi) | |
| 1007 | vpshufb %xmm0,%xmm10,%xmm10 | |
| 1008 | vmovups %xmm11,-64(%rsi) | |
| 1009 | vpshufb %xmm0,%xmm11,%xmm11 | |
| 1010 | vmovups %xmm12,-48(%rsi) | |
| 1011 | vpshufb %xmm0,%xmm12,%xmm12 | |
| 1012 | vmovups %xmm13,-32(%rsi) | |
| 1013 | vpshufb %xmm0,%xmm13,%xmm13 | |
| 1014 | vmovups %xmm14,-16(%rsi) | |
| 1015 | vpshufb %xmm0,%xmm14,%xmm14 | |
| 1016 | vmovdqu %xmm9,16(%rsp) | |
| 1017 | vmovdqu 48(%rsp),%xmm6 | |
| 1018 | vmovdqu 16-32(%r9),%xmm0 | |
| 1019 | vpunpckhqdq %xmm6,%xmm6,%xmm2 | |
| 1020 | vpclmulqdq $0x00,%xmm3,%xmm7,%xmm5 | |
| 1021 | vpxor %xmm6,%xmm2,%xmm2 | |
| 1022 | vpclmulqdq $0x11,%xmm3,%xmm7,%xmm7 | |
| 1023 | vpclmulqdq $0x00,%xmm15,%xmm1,%xmm1 | |
| 1024 | ||
| 1025 | vmovdqu 64(%rsp),%xmm9 | |
| 1026 | vpclmulqdq $0x00,%xmm0,%xmm6,%xmm4 | |
| 1027 | vmovdqu 48-32(%r9),%xmm3 | |
| 1028 | vpxor %xmm5,%xmm4,%xmm4 | |
| 1029 | vpunpckhqdq %xmm9,%xmm9,%xmm5 | |
| 1030 | vpclmulqdq $0x11,%xmm0,%xmm6,%xmm6 | |
| 1031 | vpxor %xmm9,%xmm5,%xmm5 | |
| 1032 | vpxor %xmm7,%xmm6,%xmm6 | |
| 1033 | vpclmulqdq $0x10,%xmm15,%xmm2,%xmm2 | |
| 1034 | vmovdqu 80-32(%r9),%xmm15 | |
| 1035 | vpxor %xmm1,%xmm2,%xmm2 | |
| 1036 | ||
| 1037 | vmovdqu 80(%rsp),%xmm1 | |
| 1038 | vpclmulqdq $0x00,%xmm3,%xmm9,%xmm7 | |
| 1039 | vmovdqu 64-32(%r9),%xmm0 | |
| 1040 | vpxor %xmm4,%xmm7,%xmm7 | |
| 1041 | vpunpckhqdq %xmm1,%xmm1,%xmm4 | |
| 1042 | vpclmulqdq $0x11,%xmm3,%xmm9,%xmm9 | |
| 1043 | vpxor %xmm1,%xmm4,%xmm4 | |
| 1044 | vpxor %xmm6,%xmm9,%xmm9 | |
| 1045 | vpclmulqdq $0x00,%xmm15,%xmm5,%xmm5 | |
| 1046 | vpxor %xmm2,%xmm5,%xmm5 | |
| 1047 | ||
| 1048 | vmovdqu 96(%rsp),%xmm2 | |
| 1049 | vpclmulqdq $0x00,%xmm0,%xmm1,%xmm6 | |
| 1050 | vmovdqu 96-32(%r9),%xmm3 | |
| 1051 | vpxor %xmm7,%xmm6,%xmm6 | |
| 1052 | vpunpckhqdq %xmm2,%xmm2,%xmm7 | |
| 1053 | vpclmulqdq $0x11,%xmm0,%xmm1,%xmm1 | |
| 1054 | vpxor %xmm2,%xmm7,%xmm7 | |
| 1055 | vpxor %xmm9,%xmm1,%xmm1 | |
| 1056 | vpclmulqdq $0x10,%xmm15,%xmm4,%xmm4 | |
| 1057 | vmovdqu 128-32(%r9),%xmm15 | |
| 1058 | vpxor %xmm5,%xmm4,%xmm4 | |
| 1059 | ||
| 1060 | vpxor 112(%rsp),%xmm8,%xmm8 | |
| 1061 | vpclmulqdq $0x00,%xmm3,%xmm2,%xmm5 | |
| 1062 | vmovdqu 112-32(%r9),%xmm0 | |
| 1063 | vpunpckhqdq %xmm8,%xmm8,%xmm9 | |
| 1064 | vpxor %xmm6,%xmm5,%xmm5 | |
| 1065 | vpclmulqdq $0x11,%xmm3,%xmm2,%xmm2 | |
| 1066 | vpxor %xmm8,%xmm9,%xmm9 | |
| 1067 | vpxor %xmm1,%xmm2,%xmm2 | |
| 1068 | vpclmulqdq $0x00,%xmm15,%xmm7,%xmm7 | |
| 1069 | vpxor %xmm4,%xmm7,%xmm4 | |
| 1070 | ||
| 1071 | vpclmulqdq $0x00,%xmm0,%xmm8,%xmm6 | |
| 1072 | vmovdqu 0-32(%r9),%xmm3 | |
| 1073 | vpunpckhqdq %xmm14,%xmm14,%xmm1 | |
| 1074 | vpclmulqdq $0x11,%xmm0,%xmm8,%xmm8 | |
| 1075 | vpxor %xmm14,%xmm1,%xmm1 | |
| 1076 | vpxor %xmm5,%xmm6,%xmm5 | |
| 1077 | vpclmulqdq $0x10,%xmm15,%xmm9,%xmm9 | |
| 1078 | vmovdqu 32-32(%r9),%xmm15 | |
| 1079 | vpxor %xmm2,%xmm8,%xmm7 | |
| 1080 | vpxor %xmm4,%xmm9,%xmm6 | |
| 1081 | ||
| 1082 | vmovdqu 16-32(%r9),%xmm0 | |
| 1083 | vpxor %xmm5,%xmm7,%xmm9 | |
| 1084 | vpclmulqdq $0x00,%xmm3,%xmm14,%xmm4 | |
| 1085 | vpxor %xmm9,%xmm6,%xmm6 | |
| 1086 | vpunpckhqdq %xmm13,%xmm13,%xmm2 | |
| 1087 | vpclmulqdq $0x11,%xmm3,%xmm14,%xmm14 | |
| 1088 | vpxor %xmm13,%xmm2,%xmm2 | |
| 1089 | vpslldq $8,%xmm6,%xmm9 | |
| 1090 | vpclmulqdq $0x00,%xmm15,%xmm1,%xmm1 | |
| 1091 | vpxor %xmm9,%xmm5,%xmm8 | |
| 1092 | vpsrldq $8,%xmm6,%xmm6 | |
| 1093 | vpxor %xmm6,%xmm7,%xmm7 | |
| 1094 | ||
| 1095 | vpclmulqdq $0x00,%xmm0,%xmm13,%xmm5 | |
| 1096 | vmovdqu 48-32(%r9),%xmm3 | |
| 1097 | vpxor %xmm4,%xmm5,%xmm5 | |
| 1098 | vpunpckhqdq %xmm12,%xmm12,%xmm9 | |
| 1099 | vpclmulqdq $0x11,%xmm0,%xmm13,%xmm13 | |
| 1100 | vpxor %xmm12,%xmm9,%xmm9 | |
| 1101 | vpxor %xmm14,%xmm13,%xmm13 | |
| 1102 | vpalignr $8,%xmm8,%xmm8,%xmm14 | |
| 1103 | vpclmulqdq $0x10,%xmm15,%xmm2,%xmm2 | |
| 1104 | vmovdqu 80-32(%r9),%xmm15 | |
| 1105 | vpxor %xmm1,%xmm2,%xmm2 | |
| 1106 | ||
| 1107 | vpclmulqdq $0x00,%xmm3,%xmm12,%xmm4 | |
| 1108 | vmovdqu 64-32(%r9),%xmm0 | |
| 1109 | vpxor %xmm5,%xmm4,%xmm4 | |
| 1110 | vpunpckhqdq %xmm11,%xmm11,%xmm1 | |
| 1111 | vpclmulqdq $0x11,%xmm3,%xmm12,%xmm12 | |
| 1112 | vpxor %xmm11,%xmm1,%xmm1 | |
| 1113 | vpxor %xmm13,%xmm12,%xmm12 | |
| 1114 | vxorps 16(%rsp),%xmm7,%xmm7 | |
| 1115 | vpclmulqdq $0x00,%xmm15,%xmm9,%xmm9 | |
| 1116 | vpxor %xmm2,%xmm9,%xmm9 | |
| 1117 | ||
| 1118 | vpclmulqdq $0x10,16(%r11),%xmm8,%xmm8 | |
| 1119 | vxorps %xmm14,%xmm8,%xmm8 | |
| 1120 | ||
| 1121 | vpclmulqdq $0x00,%xmm0,%xmm11,%xmm5 | |
| 1122 | vmovdqu 96-32(%r9),%xmm3 | |
| 1123 | vpxor %xmm4,%xmm5,%xmm5 | |
| 1124 | vpunpckhqdq %xmm10,%xmm10,%xmm2 | |
| 1125 | vpclmulqdq $0x11,%xmm0,%xmm11,%xmm11 | |
| 1126 | vpxor %xmm10,%xmm2,%xmm2 | |
| 1127 | vpalignr $8,%xmm8,%xmm8,%xmm14 | |
| 1128 | vpxor %xmm12,%xmm11,%xmm11 | |
| 1129 | vpclmulqdq $0x10,%xmm15,%xmm1,%xmm1 | |
| 1130 | vmovdqu 128-32(%r9),%xmm15 | |
| 1131 | vpxor %xmm9,%xmm1,%xmm1 | |
| 1132 | ||
| 1133 | vxorps %xmm7,%xmm14,%xmm14 | |
| 1134 | vpclmulqdq $0x10,16(%r11),%xmm8,%xmm8 | |
| 1135 | vxorps %xmm14,%xmm8,%xmm8 | |
| 1136 | ||
| 1137 | vpclmulqdq $0x00,%xmm3,%xmm10,%xmm4 | |
| 1138 | vmovdqu 112-32(%r9),%xmm0 | |
| 1139 | vpxor %xmm5,%xmm4,%xmm4 | |
| 1140 | vpunpckhqdq %xmm8,%xmm8,%xmm9 | |
| 1141 | vpclmulqdq $0x11,%xmm3,%xmm10,%xmm10 | |
| 1142 | vpxor %xmm8,%xmm9,%xmm9 | |
| 1143 | vpxor %xmm11,%xmm10,%xmm10 | |
| 1144 | vpclmulqdq $0x00,%xmm15,%xmm2,%xmm2 | |
| 1145 | vpxor %xmm1,%xmm2,%xmm2 | |
| 1146 | ||
| 1147 | vpclmulqdq $0x00,%xmm0,%xmm8,%xmm5 | |
| 1148 | vpclmulqdq $0x11,%xmm0,%xmm8,%xmm7 | |
| 1149 | vpxor %xmm4,%xmm5,%xmm5 | |
| 1150 | vpclmulqdq $0x10,%xmm15,%xmm9,%xmm6 | |
| 1151 | vpxor %xmm10,%xmm7,%xmm7 | |
| 1152 | vpxor %xmm2,%xmm6,%xmm6 | |
| 1153 | ||
| 1154 | vpxor %xmm5,%xmm7,%xmm4 | |
| 1155 | vpxor %xmm4,%xmm6,%xmm6 | |
| 1156 | vpslldq $8,%xmm6,%xmm1 | |
| 1157 | vmovdqu 16(%r11),%xmm3 | |
| 1158 | vpsrldq $8,%xmm6,%xmm6 | |
| 1159 | vpxor %xmm1,%xmm5,%xmm8 | |
| 1160 | vpxor %xmm6,%xmm7,%xmm7 | |
| 1161 | ||
| 1162 | vpalignr $8,%xmm8,%xmm8,%xmm2 | |
| 1163 | vpclmulqdq $0x10,%xmm3,%xmm8,%xmm8 | |
| 1164 | vpxor %xmm2,%xmm8,%xmm8 | |
| 1165 | ||
| 1166 | vpalignr $8,%xmm8,%xmm8,%xmm2 | |
| 1167 | vpclmulqdq $0x10,%xmm3,%xmm8,%xmm8 | |
| 1168 | vpxor %xmm7,%xmm2,%xmm2 | |
| 1169 | vpxor %xmm2,%xmm8,%xmm8 | |
| 1170 | vpshufb (%r11),%xmm8,%xmm8 | |
| e8beeaa1 AF |
1171 | movq -56(%rax),%r9 |
| 1172 | .cfi_restore %r9 | |
| 1173 | vmovdqu %xmm8,(%r9) | |
| 31b160f0 AF |
1174 | |
| 1175 | vzeroupper | |
| 1176 | movq -48(%rax),%r15 | |
| 1177 | .cfi_restore %r15 | |
| 1178 | movq -40(%rax),%r14 | |
| 1179 | .cfi_restore %r14 | |
| 1180 | movq -32(%rax),%r13 | |
| 1181 | .cfi_restore %r13 | |
| 1182 | movq -24(%rax),%r12 | |
| 1183 | .cfi_restore %r12 | |
| 1184 | movq -16(%rax),%rbp | |
| 1185 | .cfi_restore %rbp | |
| 1186 | movq -8(%rax),%rbx | |
| 1187 | .cfi_restore %rbx | |
| 1188 | leaq (%rax),%rsp | |
| 1189 | .cfi_def_cfa_register %rsp | |
| 1190 | .Lgcm_enc_abort: | |
| 1191 | movq %r10,%rax | |
| b844489e | 1192 | RET |
| 31b160f0 AF |
1193 | .cfi_endproc |
| 1194 | .size aesni_gcm_encrypt,.-aesni_gcm_encrypt | |
| 1195 | ||
| 1196 | /* Some utility routines */ | |
| 1197 | ||
| 1198 | /* | |
| 1199 | * clear all fpu registers | |
| 1200 | * void clear_fpu_regs_avx(void); | |
| 1201 | */ | |
| 1202 | .globl clear_fpu_regs_avx | |
| 1203 | .type clear_fpu_regs_avx,@function | |
| 1204 | .align 32 | |
| 1205 | clear_fpu_regs_avx: | |
| 1206 | vzeroall | |
| 43569ee3 | 1207 | RET |
| 31b160f0 AF |
1208 | .size clear_fpu_regs_avx,.-clear_fpu_regs_avx |
| 1209 | ||
| 1210 | /* | |
| 1211 | * void gcm_xor_avx(const uint8_t *src, uint8_t *dst); | |
| 1212 | * | |
| 1213 | * XORs one pair of unaligned 128-bit blocks from `src' and `dst' and | |
| 1214 | * stores the result at `dst'. The XOR is performed using FPU registers, | |
| 1215 | * so make sure FPU state is saved when running this in the kernel. | |
| 1216 | */ | |
| 1217 | .globl gcm_xor_avx | |
| 1218 | .type gcm_xor_avx,@function | |
| 1219 | .align 32 | |
| 1220 | gcm_xor_avx: | |
| 1221 | movdqu (%rdi), %xmm0 | |
| 1222 | movdqu (%rsi), %xmm1 | |
| 1223 | pxor %xmm1, %xmm0 | |
| 1224 | movdqu %xmm0, (%rsi) | |
| 43569ee3 | 1225 | RET |
| 31b160f0 AF |
1226 | .size gcm_xor_avx,.-gcm_xor_avx |
| 1227 | ||
| 1228 | /* | |
| 1229 | * Toggle a boolean_t value atomically and return the new value. | |
| 1230 | * boolean_t atomic_toggle_boolean_nv(volatile boolean_t *); | |
| 1231 | */ | |
| 1232 | .globl atomic_toggle_boolean_nv | |
| 1233 | .type atomic_toggle_boolean_nv,@function | |
| 1234 | .align 32 | |
| 1235 | atomic_toggle_boolean_nv: | |
| 1236 | xorl %eax, %eax | |
| 1237 | lock | |
| 1238 | xorl $1, (%rdi) | |
| 1239 | jz 1f | |
| 1240 | movl $1, %eax | |
| 1241 | 1: | |
| 43569ee3 | 1242 | RET |
| 31b160f0 AF |
1243 | .size atomic_toggle_boolean_nv,.-atomic_toggle_boolean_nv |
| 1244 | ||
| 61cca6fa | 1245 | .pushsection .rodata |
| 31b160f0 AF |
1246 | .align 64 |
| 1247 | .Lbswap_mask: | |
| 1248 | .byte 15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0 | |
| 1249 | .Lpoly: | |
| 1250 | .byte 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xc2 | |
| 1251 | .Lone_msb: | |
| 1252 | .byte 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1 | |
| 1253 | .Ltwo_lsb: | |
| 1254 | .byte 2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 | |
| 1255 | .Lone_lsb: | |
| 1256 | .byte 1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 | |
| 1257 | .byte 65,69,83,45,78,73,32,71,67,77,32,109,111,100,117,108,101,32,102,111,114,32,120,56,54,95,54,52,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0 | |
| 1258 | .align 64 | |
| 61cca6fa | 1259 | .popsection |
| 31b160f0 AF |
1260 | |
| 1261 | /* Mark the stack non-executable. */ | |
| 1262 | #if defined(__linux__) && defined(__ELF__) | |
| 1263 | .section .note.GNU-stack,"",%progbits | |
| 1264 | #endif | |
| 1265 | ||
| 1266 | #endif /* defined(__x86_64__) && defined(HAVE_AVX) && defined(HAVE_AES) ... */ |