Alexandre Savard | 1b09e31 | 2012-08-07 20:33:29 -0400 | [diff] [blame^] | 1 | #!/usr/bin/env perl |
| 2 | |
| 3 | # ==================================================================== |
| 4 | # Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL |
| 5 | # project. The module is, however, dual licensed under OpenSSL and |
| 6 | # CRYPTOGAMS licenses depending on where you obtain it. For further |
| 7 | # details see http://www.openssl.org/~appro/cryptogams/. |
| 8 | # ==================================================================== |
| 9 | |
| 10 | # AES for s390x. |
| 11 | |
| 12 | # April 2007. |
| 13 | # |
| 14 | # Software performance improvement over gcc-generated code is ~70% and |
| 15 | # in absolute terms is ~73 cycles per byte processed with 128-bit key. |
| 16 | # You're likely to exclaim "why so slow?" Keep in mind that z-CPUs are |
| 17 | # *strictly* in-order execution and issued instruction [in this case |
| 18 | # load value from memory is critical] has to complete before execution |
| 19 | # flow proceeds. S-boxes are compressed to 2KB[+256B]. |
| 20 | # |
| 21 | # As for hardware acceleration support. It's basically a "teaser," as |
| 22 | # it can and should be improved in several ways. Most notably support |
| 23 | # for CBC is not utilized, nor multiple blocks are ever processed. |
| 24 | # Then software key schedule can be postponed till hardware support |
| 25 | # detection... Performance improvement over assembler is reportedly |
| 26 | # ~2.5x, but can reach >8x [naturally on larger chunks] if proper |
| 27 | # support is implemented. |
| 28 | |
| 29 | # May 2007. |
| 30 | # |
| 31 | # Implement AES_set_[en|de]crypt_key. Key schedule setup is avoided |
| 32 | # for 128-bit keys, if hardware support is detected. |
| 33 | |
| 34 | # Januray 2009. |
| 35 | # |
| 36 | # Add support for hardware AES192/256 and reschedule instructions to |
| 37 | # minimize/avoid Address Generation Interlock hazard and to favour |
| 38 | # dual-issue z10 pipeline. This gave ~25% improvement on z10 and |
| 39 | # almost 50% on z9. The gain is smaller on z10, because being dual- |
| 40 | # issue z10 makes it improssible to eliminate the interlock condition: |
| 41 | # critial path is not long enough. Yet it spends ~24 cycles per byte |
| 42 | # processed with 128-bit key. |
| 43 | # |
| 44 | # Unlike previous version hardware support detection takes place only |
| 45 | # at the moment of key schedule setup, which is denoted in key->rounds. |
| 46 | # This is done, because deferred key setup can't be made MT-safe, not |
| 47 | # for keys longer than 128 bits. |
| 48 | # |
| 49 | # Add AES_cbc_encrypt, which gives incredible performance improvement, |
| 50 | # it was measured to be ~6.6x. It's less than previously mentioned 8x, |
| 51 | # because software implementation was optimized. |
| 52 | |
| 53 | # May 2010. |
| 54 | # |
| 55 | # Add AES_ctr32_encrypt. If hardware-assisted, it provides up to 4.3x |
| 56 | # performance improvement over "generic" counter mode routine relying |
| 57 | # on single-block, also hardware-assisted, AES_encrypt. "Up to" refers |
| 58 | # to the fact that exact throughput value depends on current stack |
| 59 | # frame alignment within 4KB page. In worst case you get ~75% of the |
| 60 | # maximum, but *on average* it would be as much as ~98%. Meaning that |
| 61 | # worst case is unlike, it's like hitting ravine on plateau. |
| 62 | |
| 63 | # November 2010. |
| 64 | # |
| 65 | # Adapt for -m31 build. If kernel supports what's called "highgprs" |
| 66 | # feature on Linux [see /proc/cpuinfo], it's possible to use 64-bit |
| 67 | # instructions and achieve "64-bit" performance even in 31-bit legacy |
| 68 | # application context. The feature is not specific to any particular |
| 69 | # processor, as long as it's "z-CPU". Latter implies that the code |
| 70 | # remains z/Architecture specific. On z990 it was measured to perform |
| 71 | # 2x better than code generated by gcc 4.3. |
| 72 | |
| 73 | # December 2010. |
| 74 | # |
| 75 | # Add support for z196 "cipher message with counter" instruction. |
| 76 | # Note however that it's disengaged, because it was measured to |
| 77 | # perform ~12% worse than vanilla km-based code... |
| 78 | |
| 79 | # February 2011. |
| 80 | # |
| 81 | # Add AES_xts_[en|de]crypt. This includes support for z196 km-xts-aes |
| 82 | # instructions, which deliver ~70% improvement at 8KB block size over |
| 83 | # vanilla km-based code, 37% - at most like 512-bytes block size. |
| 84 | |
| 85 | $flavour = shift; |
| 86 | |
| 87 | if ($flavour =~ /3[12]/) { |
| 88 | $SIZE_T=4; |
| 89 | $g=""; |
| 90 | } else { |
| 91 | $SIZE_T=8; |
| 92 | $g="g"; |
| 93 | } |
| 94 | |
| 95 | while (($output=shift) && ($output!~/^\w[\w\-]*\.\w+$/)) {} |
| 96 | open STDOUT,">$output"; |
| 97 | |
| 98 | $softonly=0; # allow hardware support |
| 99 | |
| 100 | $t0="%r0"; $mask="%r0"; |
| 101 | $t1="%r1"; |
| 102 | $t2="%r2"; $inp="%r2"; |
| 103 | $t3="%r3"; $out="%r3"; $bits="%r3"; |
| 104 | $key="%r4"; |
| 105 | $i1="%r5"; |
| 106 | $i2="%r6"; |
| 107 | $i3="%r7"; |
| 108 | $s0="%r8"; |
| 109 | $s1="%r9"; |
| 110 | $s2="%r10"; |
| 111 | $s3="%r11"; |
| 112 | $tbl="%r12"; |
| 113 | $rounds="%r13"; |
| 114 | $ra="%r14"; |
| 115 | $sp="%r15"; |
| 116 | |
| 117 | $stdframe=16*$SIZE_T+4*8; |
| 118 | |
| 119 | sub _data_word() |
| 120 | { my $i; |
| 121 | while(defined($i=shift)) { $code.=sprintf".long\t0x%08x,0x%08x\n",$i,$i; } |
| 122 | } |
| 123 | |
| 124 | $code=<<___; |
| 125 | .text |
| 126 | |
| 127 | .type AES_Te,\@object |
| 128 | .align 256 |
| 129 | AES_Te: |
| 130 | ___ |
| 131 | &_data_word( |
| 132 | 0xc66363a5, 0xf87c7c84, 0xee777799, 0xf67b7b8d, |
| 133 | 0xfff2f20d, 0xd66b6bbd, 0xde6f6fb1, 0x91c5c554, |
| 134 | 0x60303050, 0x02010103, 0xce6767a9, 0x562b2b7d, |
| 135 | 0xe7fefe19, 0xb5d7d762, 0x4dababe6, 0xec76769a, |
| 136 | 0x8fcaca45, 0x1f82829d, 0x89c9c940, 0xfa7d7d87, |
| 137 | 0xeffafa15, 0xb25959eb, 0x8e4747c9, 0xfbf0f00b, |
| 138 | 0x41adadec, 0xb3d4d467, 0x5fa2a2fd, 0x45afafea, |
| 139 | 0x239c9cbf, 0x53a4a4f7, 0xe4727296, 0x9bc0c05b, |
| 140 | 0x75b7b7c2, 0xe1fdfd1c, 0x3d9393ae, 0x4c26266a, |
| 141 | 0x6c36365a, 0x7e3f3f41, 0xf5f7f702, 0x83cccc4f, |
| 142 | 0x6834345c, 0x51a5a5f4, 0xd1e5e534, 0xf9f1f108, |
| 143 | 0xe2717193, 0xabd8d873, 0x62313153, 0x2a15153f, |
| 144 | 0x0804040c, 0x95c7c752, 0x46232365, 0x9dc3c35e, |
| 145 | 0x30181828, 0x379696a1, 0x0a05050f, 0x2f9a9ab5, |
| 146 | 0x0e070709, 0x24121236, 0x1b80809b, 0xdfe2e23d, |
| 147 | 0xcdebeb26, 0x4e272769, 0x7fb2b2cd, 0xea75759f, |
| 148 | 0x1209091b, 0x1d83839e, 0x582c2c74, 0x341a1a2e, |
| 149 | 0x361b1b2d, 0xdc6e6eb2, 0xb45a5aee, 0x5ba0a0fb, |
| 150 | 0xa45252f6, 0x763b3b4d, 0xb7d6d661, 0x7db3b3ce, |
| 151 | 0x5229297b, 0xdde3e33e, 0x5e2f2f71, 0x13848497, |
| 152 | 0xa65353f5, 0xb9d1d168, 0x00000000, 0xc1eded2c, |
| 153 | 0x40202060, 0xe3fcfc1f, 0x79b1b1c8, 0xb65b5bed, |
| 154 | 0xd46a6abe, 0x8dcbcb46, 0x67bebed9, 0x7239394b, |
| 155 | 0x944a4ade, 0x984c4cd4, 0xb05858e8, 0x85cfcf4a, |
| 156 | 0xbbd0d06b, 0xc5efef2a, 0x4faaaae5, 0xedfbfb16, |
| 157 | 0x864343c5, 0x9a4d4dd7, 0x66333355, 0x11858594, |
| 158 | 0x8a4545cf, 0xe9f9f910, 0x04020206, 0xfe7f7f81, |
| 159 | 0xa05050f0, 0x783c3c44, 0x259f9fba, 0x4ba8a8e3, |
| 160 | 0xa25151f3, 0x5da3a3fe, 0x804040c0, 0x058f8f8a, |
| 161 | 0x3f9292ad, 0x219d9dbc, 0x70383848, 0xf1f5f504, |
| 162 | 0x63bcbcdf, 0x77b6b6c1, 0xafdada75, 0x42212163, |
| 163 | 0x20101030, 0xe5ffff1a, 0xfdf3f30e, 0xbfd2d26d, |
| 164 | 0x81cdcd4c, 0x180c0c14, 0x26131335, 0xc3ecec2f, |
| 165 | 0xbe5f5fe1, 0x359797a2, 0x884444cc, 0x2e171739, |
| 166 | 0x93c4c457, 0x55a7a7f2, 0xfc7e7e82, 0x7a3d3d47, |
| 167 | 0xc86464ac, 0xba5d5de7, 0x3219192b, 0xe6737395, |
| 168 | 0xc06060a0, 0x19818198, 0x9e4f4fd1, 0xa3dcdc7f, |
| 169 | 0x44222266, 0x542a2a7e, 0x3b9090ab, 0x0b888883, |
| 170 | 0x8c4646ca, 0xc7eeee29, 0x6bb8b8d3, 0x2814143c, |
| 171 | 0xa7dede79, 0xbc5e5ee2, 0x160b0b1d, 0xaddbdb76, |
| 172 | 0xdbe0e03b, 0x64323256, 0x743a3a4e, 0x140a0a1e, |
| 173 | 0x924949db, 0x0c06060a, 0x4824246c, 0xb85c5ce4, |
| 174 | 0x9fc2c25d, 0xbdd3d36e, 0x43acacef, 0xc46262a6, |
| 175 | 0x399191a8, 0x319595a4, 0xd3e4e437, 0xf279798b, |
| 176 | 0xd5e7e732, 0x8bc8c843, 0x6e373759, 0xda6d6db7, |
| 177 | 0x018d8d8c, 0xb1d5d564, 0x9c4e4ed2, 0x49a9a9e0, |
| 178 | 0xd86c6cb4, 0xac5656fa, 0xf3f4f407, 0xcfeaea25, |
| 179 | 0xca6565af, 0xf47a7a8e, 0x47aeaee9, 0x10080818, |
| 180 | 0x6fbabad5, 0xf0787888, 0x4a25256f, 0x5c2e2e72, |
| 181 | 0x381c1c24, 0x57a6a6f1, 0x73b4b4c7, 0x97c6c651, |
| 182 | 0xcbe8e823, 0xa1dddd7c, 0xe874749c, 0x3e1f1f21, |
| 183 | 0x964b4bdd, 0x61bdbddc, 0x0d8b8b86, 0x0f8a8a85, |
| 184 | 0xe0707090, 0x7c3e3e42, 0x71b5b5c4, 0xcc6666aa, |
| 185 | 0x904848d8, 0x06030305, 0xf7f6f601, 0x1c0e0e12, |
| 186 | 0xc26161a3, 0x6a35355f, 0xae5757f9, 0x69b9b9d0, |
| 187 | 0x17868691, 0x99c1c158, 0x3a1d1d27, 0x279e9eb9, |
| 188 | 0xd9e1e138, 0xebf8f813, 0x2b9898b3, 0x22111133, |
| 189 | 0xd26969bb, 0xa9d9d970, 0x078e8e89, 0x339494a7, |
| 190 | 0x2d9b9bb6, 0x3c1e1e22, 0x15878792, 0xc9e9e920, |
| 191 | 0x87cece49, 0xaa5555ff, 0x50282878, 0xa5dfdf7a, |
| 192 | 0x038c8c8f, 0x59a1a1f8, 0x09898980, 0x1a0d0d17, |
| 193 | 0x65bfbfda, 0xd7e6e631, 0x844242c6, 0xd06868b8, |
| 194 | 0x824141c3, 0x299999b0, 0x5a2d2d77, 0x1e0f0f11, |
| 195 | 0x7bb0b0cb, 0xa85454fc, 0x6dbbbbd6, 0x2c16163a); |
| 196 | $code.=<<___; |
| 197 | # Te4[256] |
| 198 | .byte 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5 |
| 199 | .byte 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76 |
| 200 | .byte 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0 |
| 201 | .byte 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0 |
| 202 | .byte 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc |
| 203 | .byte 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15 |
| 204 | .byte 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a |
| 205 | .byte 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75 |
| 206 | .byte 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0 |
| 207 | .byte 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84 |
| 208 | .byte 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b |
| 209 | .byte 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf |
| 210 | .byte 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85 |
| 211 | .byte 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8 |
| 212 | .byte 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5 |
| 213 | .byte 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2 |
| 214 | .byte 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17 |
| 215 | .byte 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73 |
| 216 | .byte 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88 |
| 217 | .byte 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb |
| 218 | .byte 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c |
| 219 | .byte 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79 |
| 220 | .byte 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9 |
| 221 | .byte 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08 |
| 222 | .byte 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6 |
| 223 | .byte 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a |
| 224 | .byte 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e |
| 225 | .byte 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e |
| 226 | .byte 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94 |
| 227 | .byte 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf |
| 228 | .byte 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68 |
| 229 | .byte 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16 |
| 230 | # rcon[] |
| 231 | .long 0x01000000, 0x02000000, 0x04000000, 0x08000000 |
| 232 | .long 0x10000000, 0x20000000, 0x40000000, 0x80000000 |
| 233 | .long 0x1B000000, 0x36000000, 0, 0, 0, 0, 0, 0 |
| 234 | .align 256 |
| 235 | .size AES_Te,.-AES_Te |
| 236 | |
| 237 | # void AES_encrypt(const unsigned char *inp, unsigned char *out, |
| 238 | # const AES_KEY *key) { |
| 239 | .globl AES_encrypt |
| 240 | .type AES_encrypt,\@function |
| 241 | AES_encrypt: |
| 242 | ___ |
| 243 | $code.=<<___ if (!$softonly); |
| 244 | l %r0,240($key) |
| 245 | lhi %r1,16 |
| 246 | clr %r0,%r1 |
| 247 | jl .Lesoft |
| 248 | |
| 249 | la %r1,0($key) |
| 250 | #la %r2,0($inp) |
| 251 | la %r4,0($out) |
| 252 | lghi %r3,16 # single block length |
| 253 | .long 0xb92e0042 # km %r4,%r2 |
| 254 | brc 1,.-4 # can this happen? |
| 255 | br %r14 |
| 256 | .align 64 |
| 257 | .Lesoft: |
| 258 | ___ |
| 259 | $code.=<<___; |
| 260 | stm${g} %r3,$ra,3*$SIZE_T($sp) |
| 261 | |
| 262 | llgf $s0,0($inp) |
| 263 | llgf $s1,4($inp) |
| 264 | llgf $s2,8($inp) |
| 265 | llgf $s3,12($inp) |
| 266 | |
| 267 | larl $tbl,AES_Te |
| 268 | bras $ra,_s390x_AES_encrypt |
| 269 | |
| 270 | l${g} $out,3*$SIZE_T($sp) |
| 271 | st $s0,0($out) |
| 272 | st $s1,4($out) |
| 273 | st $s2,8($out) |
| 274 | st $s3,12($out) |
| 275 | |
| 276 | lm${g} %r6,$ra,6*$SIZE_T($sp) |
| 277 | br $ra |
| 278 | .size AES_encrypt,.-AES_encrypt |
| 279 | |
| 280 | .type _s390x_AES_encrypt,\@function |
| 281 | .align 16 |
| 282 | _s390x_AES_encrypt: |
| 283 | st${g} $ra,15*$SIZE_T($sp) |
| 284 | x $s0,0($key) |
| 285 | x $s1,4($key) |
| 286 | x $s2,8($key) |
| 287 | x $s3,12($key) |
| 288 | l $rounds,240($key) |
| 289 | llill $mask,`0xff<<3` |
| 290 | aghi $rounds,-1 |
| 291 | j .Lenc_loop |
| 292 | .align 16 |
| 293 | .Lenc_loop: |
| 294 | sllg $t1,$s0,`0+3` |
| 295 | srlg $t2,$s0,`8-3` |
| 296 | srlg $t3,$s0,`16-3` |
| 297 | srl $s0,`24-3` |
| 298 | nr $s0,$mask |
| 299 | ngr $t1,$mask |
| 300 | nr $t2,$mask |
| 301 | nr $t3,$mask |
| 302 | |
| 303 | srlg $i1,$s1,`16-3` # i0 |
| 304 | sllg $i2,$s1,`0+3` |
| 305 | srlg $i3,$s1,`8-3` |
| 306 | srl $s1,`24-3` |
| 307 | nr $i1,$mask |
| 308 | nr $s1,$mask |
| 309 | ngr $i2,$mask |
| 310 | nr $i3,$mask |
| 311 | |
| 312 | l $s0,0($s0,$tbl) # Te0[s0>>24] |
| 313 | l $t1,1($t1,$tbl) # Te3[s0>>0] |
| 314 | l $t2,2($t2,$tbl) # Te2[s0>>8] |
| 315 | l $t3,3($t3,$tbl) # Te1[s0>>16] |
| 316 | |
| 317 | x $s0,3($i1,$tbl) # Te1[s1>>16] |
| 318 | l $s1,0($s1,$tbl) # Te0[s1>>24] |
| 319 | x $t2,1($i2,$tbl) # Te3[s1>>0] |
| 320 | x $t3,2($i3,$tbl) # Te2[s1>>8] |
| 321 | |
| 322 | srlg $i1,$s2,`8-3` # i0 |
| 323 | srlg $i2,$s2,`16-3` # i1 |
| 324 | nr $i1,$mask |
| 325 | nr $i2,$mask |
| 326 | sllg $i3,$s2,`0+3` |
| 327 | srl $s2,`24-3` |
| 328 | nr $s2,$mask |
| 329 | ngr $i3,$mask |
| 330 | |
| 331 | xr $s1,$t1 |
| 332 | srlg $ra,$s3,`8-3` # i1 |
| 333 | sllg $t1,$s3,`0+3` # i0 |
| 334 | nr $ra,$mask |
| 335 | la $key,16($key) |
| 336 | ngr $t1,$mask |
| 337 | |
| 338 | x $s0,2($i1,$tbl) # Te2[s2>>8] |
| 339 | x $s1,3($i2,$tbl) # Te1[s2>>16] |
| 340 | l $s2,0($s2,$tbl) # Te0[s2>>24] |
| 341 | x $t3,1($i3,$tbl) # Te3[s2>>0] |
| 342 | |
| 343 | srlg $i3,$s3,`16-3` # i2 |
| 344 | xr $s2,$t2 |
| 345 | srl $s3,`24-3` |
| 346 | nr $i3,$mask |
| 347 | nr $s3,$mask |
| 348 | |
| 349 | x $s0,0($key) |
| 350 | x $s1,4($key) |
| 351 | x $s2,8($key) |
| 352 | x $t3,12($key) |
| 353 | |
| 354 | x $s0,1($t1,$tbl) # Te3[s3>>0] |
| 355 | x $s1,2($ra,$tbl) # Te2[s3>>8] |
| 356 | x $s2,3($i3,$tbl) # Te1[s3>>16] |
| 357 | l $s3,0($s3,$tbl) # Te0[s3>>24] |
| 358 | xr $s3,$t3 |
| 359 | |
| 360 | brct $rounds,.Lenc_loop |
| 361 | .align 16 |
| 362 | |
| 363 | sllg $t1,$s0,`0+3` |
| 364 | srlg $t2,$s0,`8-3` |
| 365 | ngr $t1,$mask |
| 366 | srlg $t3,$s0,`16-3` |
| 367 | srl $s0,`24-3` |
| 368 | nr $s0,$mask |
| 369 | nr $t2,$mask |
| 370 | nr $t3,$mask |
| 371 | |
| 372 | srlg $i1,$s1,`16-3` # i0 |
| 373 | sllg $i2,$s1,`0+3` |
| 374 | ngr $i2,$mask |
| 375 | srlg $i3,$s1,`8-3` |
| 376 | srl $s1,`24-3` |
| 377 | nr $i1,$mask |
| 378 | nr $s1,$mask |
| 379 | nr $i3,$mask |
| 380 | |
| 381 | llgc $s0,2($s0,$tbl) # Te4[s0>>24] |
| 382 | llgc $t1,2($t1,$tbl) # Te4[s0>>0] |
| 383 | sll $s0,24 |
| 384 | llgc $t2,2($t2,$tbl) # Te4[s0>>8] |
| 385 | llgc $t3,2($t3,$tbl) # Te4[s0>>16] |
| 386 | sll $t2,8 |
| 387 | sll $t3,16 |
| 388 | |
| 389 | llgc $i1,2($i1,$tbl) # Te4[s1>>16] |
| 390 | llgc $s1,2($s1,$tbl) # Te4[s1>>24] |
| 391 | llgc $i2,2($i2,$tbl) # Te4[s1>>0] |
| 392 | llgc $i3,2($i3,$tbl) # Te4[s1>>8] |
| 393 | sll $i1,16 |
| 394 | sll $s1,24 |
| 395 | sll $i3,8 |
| 396 | or $s0,$i1 |
| 397 | or $s1,$t1 |
| 398 | or $t2,$i2 |
| 399 | or $t3,$i3 |
| 400 | |
| 401 | srlg $i1,$s2,`8-3` # i0 |
| 402 | srlg $i2,$s2,`16-3` # i1 |
| 403 | nr $i1,$mask |
| 404 | nr $i2,$mask |
| 405 | sllg $i3,$s2,`0+3` |
| 406 | srl $s2,`24-3` |
| 407 | ngr $i3,$mask |
| 408 | nr $s2,$mask |
| 409 | |
| 410 | sllg $t1,$s3,`0+3` # i0 |
| 411 | srlg $ra,$s3,`8-3` # i1 |
| 412 | ngr $t1,$mask |
| 413 | |
| 414 | llgc $i1,2($i1,$tbl) # Te4[s2>>8] |
| 415 | llgc $i2,2($i2,$tbl) # Te4[s2>>16] |
| 416 | sll $i1,8 |
| 417 | llgc $s2,2($s2,$tbl) # Te4[s2>>24] |
| 418 | llgc $i3,2($i3,$tbl) # Te4[s2>>0] |
| 419 | sll $i2,16 |
| 420 | nr $ra,$mask |
| 421 | sll $s2,24 |
| 422 | or $s0,$i1 |
| 423 | or $s1,$i2 |
| 424 | or $s2,$t2 |
| 425 | or $t3,$i3 |
| 426 | |
| 427 | srlg $i3,$s3,`16-3` # i2 |
| 428 | srl $s3,`24-3` |
| 429 | nr $i3,$mask |
| 430 | nr $s3,$mask |
| 431 | |
| 432 | l $t0,16($key) |
| 433 | l $t2,20($key) |
| 434 | |
| 435 | llgc $i1,2($t1,$tbl) # Te4[s3>>0] |
| 436 | llgc $i2,2($ra,$tbl) # Te4[s3>>8] |
| 437 | llgc $i3,2($i3,$tbl) # Te4[s3>>16] |
| 438 | llgc $s3,2($s3,$tbl) # Te4[s3>>24] |
| 439 | sll $i2,8 |
| 440 | sll $i3,16 |
| 441 | sll $s3,24 |
| 442 | or $s0,$i1 |
| 443 | or $s1,$i2 |
| 444 | or $s2,$i3 |
| 445 | or $s3,$t3 |
| 446 | |
| 447 | l${g} $ra,15*$SIZE_T($sp) |
| 448 | xr $s0,$t0 |
| 449 | xr $s1,$t2 |
| 450 | x $s2,24($key) |
| 451 | x $s3,28($key) |
| 452 | |
| 453 | br $ra |
| 454 | .size _s390x_AES_encrypt,.-_s390x_AES_encrypt |
| 455 | ___ |
| 456 | |
| 457 | $code.=<<___; |
| 458 | .type AES_Td,\@object |
| 459 | .align 256 |
| 460 | AES_Td: |
| 461 | ___ |
| 462 | &_data_word( |
| 463 | 0x51f4a750, 0x7e416553, 0x1a17a4c3, 0x3a275e96, |
| 464 | 0x3bab6bcb, 0x1f9d45f1, 0xacfa58ab, 0x4be30393, |
| 465 | 0x2030fa55, 0xad766df6, 0x88cc7691, 0xf5024c25, |
| 466 | 0x4fe5d7fc, 0xc52acbd7, 0x26354480, 0xb562a38f, |
| 467 | 0xdeb15a49, 0x25ba1b67, 0x45ea0e98, 0x5dfec0e1, |
| 468 | 0xc32f7502, 0x814cf012, 0x8d4697a3, 0x6bd3f9c6, |
| 469 | 0x038f5fe7, 0x15929c95, 0xbf6d7aeb, 0x955259da, |
| 470 | 0xd4be832d, 0x587421d3, 0x49e06929, 0x8ec9c844, |
| 471 | 0x75c2896a, 0xf48e7978, 0x99583e6b, 0x27b971dd, |
| 472 | 0xbee14fb6, 0xf088ad17, 0xc920ac66, 0x7dce3ab4, |
| 473 | 0x63df4a18, 0xe51a3182, 0x97513360, 0x62537f45, |
| 474 | 0xb16477e0, 0xbb6bae84, 0xfe81a01c, 0xf9082b94, |
| 475 | 0x70486858, 0x8f45fd19, 0x94de6c87, 0x527bf8b7, |
| 476 | 0xab73d323, 0x724b02e2, 0xe31f8f57, 0x6655ab2a, |
| 477 | 0xb2eb2807, 0x2fb5c203, 0x86c57b9a, 0xd33708a5, |
| 478 | 0x302887f2, 0x23bfa5b2, 0x02036aba, 0xed16825c, |
| 479 | 0x8acf1c2b, 0xa779b492, 0xf307f2f0, 0x4e69e2a1, |
| 480 | 0x65daf4cd, 0x0605bed5, 0xd134621f, 0xc4a6fe8a, |
| 481 | 0x342e539d, 0xa2f355a0, 0x058ae132, 0xa4f6eb75, |
| 482 | 0x0b83ec39, 0x4060efaa, 0x5e719f06, 0xbd6e1051, |
| 483 | 0x3e218af9, 0x96dd063d, 0xdd3e05ae, 0x4de6bd46, |
| 484 | 0x91548db5, 0x71c45d05, 0x0406d46f, 0x605015ff, |
| 485 | 0x1998fb24, 0xd6bde997, 0x894043cc, 0x67d99e77, |
| 486 | 0xb0e842bd, 0x07898b88, 0xe7195b38, 0x79c8eedb, |
| 487 | 0xa17c0a47, 0x7c420fe9, 0xf8841ec9, 0x00000000, |
| 488 | 0x09808683, 0x322bed48, 0x1e1170ac, 0x6c5a724e, |
| 489 | 0xfd0efffb, 0x0f853856, 0x3daed51e, 0x362d3927, |
| 490 | 0x0a0fd964, 0x685ca621, 0x9b5b54d1, 0x24362e3a, |
| 491 | 0x0c0a67b1, 0x9357e70f, 0xb4ee96d2, 0x1b9b919e, |
| 492 | 0x80c0c54f, 0x61dc20a2, 0x5a774b69, 0x1c121a16, |
| 493 | 0xe293ba0a, 0xc0a02ae5, 0x3c22e043, 0x121b171d, |
| 494 | 0x0e090d0b, 0xf28bc7ad, 0x2db6a8b9, 0x141ea9c8, |
| 495 | 0x57f11985, 0xaf75074c, 0xee99ddbb, 0xa37f60fd, |
| 496 | 0xf701269f, 0x5c72f5bc, 0x44663bc5, 0x5bfb7e34, |
| 497 | 0x8b432976, 0xcb23c6dc, 0xb6edfc68, 0xb8e4f163, |
| 498 | 0xd731dcca, 0x42638510, 0x13972240, 0x84c61120, |
| 499 | 0x854a247d, 0xd2bb3df8, 0xaef93211, 0xc729a16d, |
| 500 | 0x1d9e2f4b, 0xdcb230f3, 0x0d8652ec, 0x77c1e3d0, |
| 501 | 0x2bb3166c, 0xa970b999, 0x119448fa, 0x47e96422, |
| 502 | 0xa8fc8cc4, 0xa0f03f1a, 0x567d2cd8, 0x223390ef, |
| 503 | 0x87494ec7, 0xd938d1c1, 0x8ccaa2fe, 0x98d40b36, |
| 504 | 0xa6f581cf, 0xa57ade28, 0xdab78e26, 0x3fadbfa4, |
| 505 | 0x2c3a9de4, 0x5078920d, 0x6a5fcc9b, 0x547e4662, |
| 506 | 0xf68d13c2, 0x90d8b8e8, 0x2e39f75e, 0x82c3aff5, |
| 507 | 0x9f5d80be, 0x69d0937c, 0x6fd52da9, 0xcf2512b3, |
| 508 | 0xc8ac993b, 0x10187da7, 0xe89c636e, 0xdb3bbb7b, |
| 509 | 0xcd267809, 0x6e5918f4, 0xec9ab701, 0x834f9aa8, |
| 510 | 0xe6956e65, 0xaaffe67e, 0x21bccf08, 0xef15e8e6, |
| 511 | 0xbae79bd9, 0x4a6f36ce, 0xea9f09d4, 0x29b07cd6, |
| 512 | 0x31a4b2af, 0x2a3f2331, 0xc6a59430, 0x35a266c0, |
| 513 | 0x744ebc37, 0xfc82caa6, 0xe090d0b0, 0x33a7d815, |
| 514 | 0xf104984a, 0x41ecdaf7, 0x7fcd500e, 0x1791f62f, |
| 515 | 0x764dd68d, 0x43efb04d, 0xccaa4d54, 0xe49604df, |
| 516 | 0x9ed1b5e3, 0x4c6a881b, 0xc12c1fb8, 0x4665517f, |
| 517 | 0x9d5eea04, 0x018c355d, 0xfa877473, 0xfb0b412e, |
| 518 | 0xb3671d5a, 0x92dbd252, 0xe9105633, 0x6dd64713, |
| 519 | 0x9ad7618c, 0x37a10c7a, 0x59f8148e, 0xeb133c89, |
| 520 | 0xcea927ee, 0xb761c935, 0xe11ce5ed, 0x7a47b13c, |
| 521 | 0x9cd2df59, 0x55f2733f, 0x1814ce79, 0x73c737bf, |
| 522 | 0x53f7cdea, 0x5ffdaa5b, 0xdf3d6f14, 0x7844db86, |
| 523 | 0xcaaff381, 0xb968c43e, 0x3824342c, 0xc2a3405f, |
| 524 | 0x161dc372, 0xbce2250c, 0x283c498b, 0xff0d9541, |
| 525 | 0x39a80171, 0x080cb3de, 0xd8b4e49c, 0x6456c190, |
| 526 | 0x7bcb8461, 0xd532b670, 0x486c5c74, 0xd0b85742); |
| 527 | $code.=<<___; |
| 528 | # Td4[256] |
| 529 | .byte 0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38 |
| 530 | .byte 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb |
| 531 | .byte 0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87 |
| 532 | .byte 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb |
| 533 | .byte 0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d |
| 534 | .byte 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e |
| 535 | .byte 0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2 |
| 536 | .byte 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25 |
| 537 | .byte 0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16 |
| 538 | .byte 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92 |
| 539 | .byte 0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda |
| 540 | .byte 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84 |
| 541 | .byte 0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a |
| 542 | .byte 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06 |
| 543 | .byte 0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02 |
| 544 | .byte 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b |
| 545 | .byte 0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea |
| 546 | .byte 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73 |
| 547 | .byte 0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85 |
| 548 | .byte 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e |
| 549 | .byte 0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89 |
| 550 | .byte 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b |
| 551 | .byte 0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20 |
| 552 | .byte 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4 |
| 553 | .byte 0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31 |
| 554 | .byte 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f |
| 555 | .byte 0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d |
| 556 | .byte 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef |
| 557 | .byte 0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0 |
| 558 | .byte 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61 |
| 559 | .byte 0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26 |
| 560 | .byte 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d |
| 561 | .size AES_Td,.-AES_Td |
| 562 | |
| 563 | # void AES_decrypt(const unsigned char *inp, unsigned char *out, |
| 564 | # const AES_KEY *key) { |
| 565 | .globl AES_decrypt |
| 566 | .type AES_decrypt,\@function |
| 567 | AES_decrypt: |
| 568 | ___ |
| 569 | $code.=<<___ if (!$softonly); |
| 570 | l %r0,240($key) |
| 571 | lhi %r1,16 |
| 572 | clr %r0,%r1 |
| 573 | jl .Ldsoft |
| 574 | |
| 575 | la %r1,0($key) |
| 576 | #la %r2,0($inp) |
| 577 | la %r4,0($out) |
| 578 | lghi %r3,16 # single block length |
| 579 | .long 0xb92e0042 # km %r4,%r2 |
| 580 | brc 1,.-4 # can this happen? |
| 581 | br %r14 |
| 582 | .align 64 |
| 583 | .Ldsoft: |
| 584 | ___ |
| 585 | $code.=<<___; |
| 586 | stm${g} %r3,$ra,3*$SIZE_T($sp) |
| 587 | |
| 588 | llgf $s0,0($inp) |
| 589 | llgf $s1,4($inp) |
| 590 | llgf $s2,8($inp) |
| 591 | llgf $s3,12($inp) |
| 592 | |
| 593 | larl $tbl,AES_Td |
| 594 | bras $ra,_s390x_AES_decrypt |
| 595 | |
| 596 | l${g} $out,3*$SIZE_T($sp) |
| 597 | st $s0,0($out) |
| 598 | st $s1,4($out) |
| 599 | st $s2,8($out) |
| 600 | st $s3,12($out) |
| 601 | |
| 602 | lm${g} %r6,$ra,6*$SIZE_T($sp) |
| 603 | br $ra |
| 604 | .size AES_decrypt,.-AES_decrypt |
| 605 | |
| 606 | .type _s390x_AES_decrypt,\@function |
| 607 | .align 16 |
| 608 | _s390x_AES_decrypt: |
| 609 | st${g} $ra,15*$SIZE_T($sp) |
| 610 | x $s0,0($key) |
| 611 | x $s1,4($key) |
| 612 | x $s2,8($key) |
| 613 | x $s3,12($key) |
| 614 | l $rounds,240($key) |
| 615 | llill $mask,`0xff<<3` |
| 616 | aghi $rounds,-1 |
| 617 | j .Ldec_loop |
| 618 | .align 16 |
| 619 | .Ldec_loop: |
| 620 | srlg $t1,$s0,`16-3` |
| 621 | srlg $t2,$s0,`8-3` |
| 622 | sllg $t3,$s0,`0+3` |
| 623 | srl $s0,`24-3` |
| 624 | nr $s0,$mask |
| 625 | nr $t1,$mask |
| 626 | nr $t2,$mask |
| 627 | ngr $t3,$mask |
| 628 | |
| 629 | sllg $i1,$s1,`0+3` # i0 |
| 630 | srlg $i2,$s1,`16-3` |
| 631 | srlg $i3,$s1,`8-3` |
| 632 | srl $s1,`24-3` |
| 633 | ngr $i1,$mask |
| 634 | nr $s1,$mask |
| 635 | nr $i2,$mask |
| 636 | nr $i3,$mask |
| 637 | |
| 638 | l $s0,0($s0,$tbl) # Td0[s0>>24] |
| 639 | l $t1,3($t1,$tbl) # Td1[s0>>16] |
| 640 | l $t2,2($t2,$tbl) # Td2[s0>>8] |
| 641 | l $t3,1($t3,$tbl) # Td3[s0>>0] |
| 642 | |
| 643 | x $s0,1($i1,$tbl) # Td3[s1>>0] |
| 644 | l $s1,0($s1,$tbl) # Td0[s1>>24] |
| 645 | x $t2,3($i2,$tbl) # Td1[s1>>16] |
| 646 | x $t3,2($i3,$tbl) # Td2[s1>>8] |
| 647 | |
| 648 | srlg $i1,$s2,`8-3` # i0 |
| 649 | sllg $i2,$s2,`0+3` # i1 |
| 650 | srlg $i3,$s2,`16-3` |
| 651 | srl $s2,`24-3` |
| 652 | nr $i1,$mask |
| 653 | ngr $i2,$mask |
| 654 | nr $s2,$mask |
| 655 | nr $i3,$mask |
| 656 | |
| 657 | xr $s1,$t1 |
| 658 | srlg $ra,$s3,`8-3` # i1 |
| 659 | srlg $t1,$s3,`16-3` # i0 |
| 660 | nr $ra,$mask |
| 661 | la $key,16($key) |
| 662 | nr $t1,$mask |
| 663 | |
| 664 | x $s0,2($i1,$tbl) # Td2[s2>>8] |
| 665 | x $s1,1($i2,$tbl) # Td3[s2>>0] |
| 666 | l $s2,0($s2,$tbl) # Td0[s2>>24] |
| 667 | x $t3,3($i3,$tbl) # Td1[s2>>16] |
| 668 | |
| 669 | sllg $i3,$s3,`0+3` # i2 |
| 670 | srl $s3,`24-3` |
| 671 | ngr $i3,$mask |
| 672 | nr $s3,$mask |
| 673 | |
| 674 | xr $s2,$t2 |
| 675 | x $s0,0($key) |
| 676 | x $s1,4($key) |
| 677 | x $s2,8($key) |
| 678 | x $t3,12($key) |
| 679 | |
| 680 | x $s0,3($t1,$tbl) # Td1[s3>>16] |
| 681 | x $s1,2($ra,$tbl) # Td2[s3>>8] |
| 682 | x $s2,1($i3,$tbl) # Td3[s3>>0] |
| 683 | l $s3,0($s3,$tbl) # Td0[s3>>24] |
| 684 | xr $s3,$t3 |
| 685 | |
| 686 | brct $rounds,.Ldec_loop |
| 687 | .align 16 |
| 688 | |
| 689 | l $t1,`2048+0`($tbl) # prefetch Td4 |
| 690 | l $t2,`2048+64`($tbl) |
| 691 | l $t3,`2048+128`($tbl) |
| 692 | l $i1,`2048+192`($tbl) |
| 693 | llill $mask,0xff |
| 694 | |
| 695 | srlg $i3,$s0,24 # i0 |
| 696 | srlg $t1,$s0,16 |
| 697 | srlg $t2,$s0,8 |
| 698 | nr $s0,$mask # i3 |
| 699 | nr $t1,$mask |
| 700 | |
| 701 | srlg $i1,$s1,24 |
| 702 | nr $t2,$mask |
| 703 | srlg $i2,$s1,16 |
| 704 | srlg $ra,$s1,8 |
| 705 | nr $s1,$mask # i0 |
| 706 | nr $i2,$mask |
| 707 | nr $ra,$mask |
| 708 | |
| 709 | llgc $i3,2048($i3,$tbl) # Td4[s0>>24] |
| 710 | llgc $t1,2048($t1,$tbl) # Td4[s0>>16] |
| 711 | llgc $t2,2048($t2,$tbl) # Td4[s0>>8] |
| 712 | sll $t1,16 |
| 713 | llgc $t3,2048($s0,$tbl) # Td4[s0>>0] |
| 714 | sllg $s0,$i3,24 |
| 715 | sll $t2,8 |
| 716 | |
| 717 | llgc $s1,2048($s1,$tbl) # Td4[s1>>0] |
| 718 | llgc $i1,2048($i1,$tbl) # Td4[s1>>24] |
| 719 | llgc $i2,2048($i2,$tbl) # Td4[s1>>16] |
| 720 | sll $i1,24 |
| 721 | llgc $i3,2048($ra,$tbl) # Td4[s1>>8] |
| 722 | sll $i2,16 |
| 723 | sll $i3,8 |
| 724 | or $s0,$s1 |
| 725 | or $t1,$i1 |
| 726 | or $t2,$i2 |
| 727 | or $t3,$i3 |
| 728 | |
| 729 | srlg $i1,$s2,8 # i0 |
| 730 | srlg $i2,$s2,24 |
| 731 | srlg $i3,$s2,16 |
| 732 | nr $s2,$mask # i1 |
| 733 | nr $i1,$mask |
| 734 | nr $i3,$mask |
| 735 | llgc $i1,2048($i1,$tbl) # Td4[s2>>8] |
| 736 | llgc $s1,2048($s2,$tbl) # Td4[s2>>0] |
| 737 | llgc $i2,2048($i2,$tbl) # Td4[s2>>24] |
| 738 | llgc $i3,2048($i3,$tbl) # Td4[s2>>16] |
| 739 | sll $i1,8 |
| 740 | sll $i2,24 |
| 741 | or $s0,$i1 |
| 742 | sll $i3,16 |
| 743 | or $t2,$i2 |
| 744 | or $t3,$i3 |
| 745 | |
| 746 | srlg $i1,$s3,16 # i0 |
| 747 | srlg $i2,$s3,8 # i1 |
| 748 | srlg $i3,$s3,24 |
| 749 | nr $s3,$mask # i2 |
| 750 | nr $i1,$mask |
| 751 | nr $i2,$mask |
| 752 | |
| 753 | l${g} $ra,15*$SIZE_T($sp) |
| 754 | or $s1,$t1 |
| 755 | l $t0,16($key) |
| 756 | l $t1,20($key) |
| 757 | |
| 758 | llgc $i1,2048($i1,$tbl) # Td4[s3>>16] |
| 759 | llgc $i2,2048($i2,$tbl) # Td4[s3>>8] |
| 760 | sll $i1,16 |
| 761 | llgc $s2,2048($s3,$tbl) # Td4[s3>>0] |
| 762 | llgc $s3,2048($i3,$tbl) # Td4[s3>>24] |
| 763 | sll $i2,8 |
| 764 | sll $s3,24 |
| 765 | or $s0,$i1 |
| 766 | or $s1,$i2 |
| 767 | or $s2,$t2 |
| 768 | or $s3,$t3 |
| 769 | |
| 770 | xr $s0,$t0 |
| 771 | xr $s1,$t1 |
| 772 | x $s2,24($key) |
| 773 | x $s3,28($key) |
| 774 | |
| 775 | br $ra |
| 776 | .size _s390x_AES_decrypt,.-_s390x_AES_decrypt |
| 777 | ___ |
| 778 | |
| 779 | $code.=<<___; |
| 780 | # void AES_set_encrypt_key(const unsigned char *in, int bits, |
| 781 | # AES_KEY *key) { |
| 782 | .globl private_AES_set_encrypt_key |
| 783 | .type private_AES_set_encrypt_key,\@function |
| 784 | .align 16 |
| 785 | private_AES_set_encrypt_key: |
| 786 | _s390x_AES_set_encrypt_key: |
| 787 | lghi $t0,0 |
| 788 | cl${g}r $inp,$t0 |
| 789 | je .Lminus1 |
| 790 | cl${g}r $key,$t0 |
| 791 | je .Lminus1 |
| 792 | |
| 793 | lghi $t0,128 |
| 794 | clr $bits,$t0 |
| 795 | je .Lproceed |
| 796 | lghi $t0,192 |
| 797 | clr $bits,$t0 |
| 798 | je .Lproceed |
| 799 | lghi $t0,256 |
| 800 | clr $bits,$t0 |
| 801 | je .Lproceed |
| 802 | lghi %r2,-2 |
| 803 | br %r14 |
| 804 | |
| 805 | .align 16 |
| 806 | .Lproceed: |
| 807 | ___ |
| 808 | $code.=<<___ if (!$softonly); |
| 809 | # convert bits to km code, [128,192,256]->[18,19,20] |
| 810 | lhi %r5,-128 |
| 811 | lhi %r0,18 |
| 812 | ar %r5,$bits |
| 813 | srl %r5,6 |
| 814 | ar %r5,%r0 |
| 815 | |
| 816 | larl %r1,OPENSSL_s390xcap_P |
| 817 | lg %r0,0(%r1) |
| 818 | tmhl %r0,0x4000 # check for message-security assist |
| 819 | jz .Lekey_internal |
| 820 | |
| 821 | lghi %r0,0 # query capability vector |
| 822 | la %r1,16($sp) |
| 823 | .long 0xb92f0042 # kmc %r4,%r2 |
| 824 | |
| 825 | llihh %r1,0x8000 |
| 826 | srlg %r1,%r1,0(%r5) |
| 827 | ng %r1,16($sp) |
| 828 | jz .Lekey_internal |
| 829 | |
| 830 | lmg %r0,%r1,0($inp) # just copy 128 bits... |
| 831 | stmg %r0,%r1,0($key) |
| 832 | lhi %r0,192 |
| 833 | cr $bits,%r0 |
| 834 | jl 1f |
| 835 | lg %r1,16($inp) |
| 836 | stg %r1,16($key) |
| 837 | je 1f |
| 838 | lg %r1,24($inp) |
| 839 | stg %r1,24($key) |
| 840 | 1: st $bits,236($key) # save bits [for debugging purposes] |
| 841 | lgr $t0,%r5 |
| 842 | st %r5,240($key) # save km code |
| 843 | lghi %r2,0 |
| 844 | br %r14 |
| 845 | ___ |
| 846 | $code.=<<___; |
| 847 | .align 16 |
| 848 | .Lekey_internal: |
| 849 | stm${g} %r4,%r13,4*$SIZE_T($sp) # all non-volatile regs and $key |
| 850 | |
| 851 | larl $tbl,AES_Te+2048 |
| 852 | |
| 853 | llgf $s0,0($inp) |
| 854 | llgf $s1,4($inp) |
| 855 | llgf $s2,8($inp) |
| 856 | llgf $s3,12($inp) |
| 857 | st $s0,0($key) |
| 858 | st $s1,4($key) |
| 859 | st $s2,8($key) |
| 860 | st $s3,12($key) |
| 861 | lghi $t0,128 |
| 862 | cr $bits,$t0 |
| 863 | jne .Lnot128 |
| 864 | |
| 865 | llill $mask,0xff |
| 866 | lghi $t3,0 # i=0 |
| 867 | lghi $rounds,10 |
| 868 | st $rounds,240($key) |
| 869 | |
| 870 | llgfr $t2,$s3 # temp=rk[3] |
| 871 | srlg $i1,$s3,8 |
| 872 | srlg $i2,$s3,16 |
| 873 | srlg $i3,$s3,24 |
| 874 | nr $t2,$mask |
| 875 | nr $i1,$mask |
| 876 | nr $i2,$mask |
| 877 | |
| 878 | .align 16 |
| 879 | .L128_loop: |
| 880 | la $t2,0($t2,$tbl) |
| 881 | la $i1,0($i1,$tbl) |
| 882 | la $i2,0($i2,$tbl) |
| 883 | la $i3,0($i3,$tbl) |
| 884 | icm $t2,2,0($t2) # Te4[rk[3]>>0]<<8 |
| 885 | icm $t2,4,0($i1) # Te4[rk[3]>>8]<<16 |
| 886 | icm $t2,8,0($i2) # Te4[rk[3]>>16]<<24 |
| 887 | icm $t2,1,0($i3) # Te4[rk[3]>>24] |
| 888 | x $t2,256($t3,$tbl) # rcon[i] |
| 889 | xr $s0,$t2 # rk[4]=rk[0]^... |
| 890 | xr $s1,$s0 # rk[5]=rk[1]^rk[4] |
| 891 | xr $s2,$s1 # rk[6]=rk[2]^rk[5] |
| 892 | xr $s3,$s2 # rk[7]=rk[3]^rk[6] |
| 893 | |
| 894 | llgfr $t2,$s3 # temp=rk[3] |
| 895 | srlg $i1,$s3,8 |
| 896 | srlg $i2,$s3,16 |
| 897 | nr $t2,$mask |
| 898 | nr $i1,$mask |
| 899 | srlg $i3,$s3,24 |
| 900 | nr $i2,$mask |
| 901 | |
| 902 | st $s0,16($key) |
| 903 | st $s1,20($key) |
| 904 | st $s2,24($key) |
| 905 | st $s3,28($key) |
| 906 | la $key,16($key) # key+=4 |
| 907 | la $t3,4($t3) # i++ |
| 908 | brct $rounds,.L128_loop |
| 909 | lghi $t0,10 |
| 910 | lghi %r2,0 |
| 911 | lm${g} %r4,%r13,4*$SIZE_T($sp) |
| 912 | br $ra |
| 913 | |
| 914 | .align 16 |
| 915 | .Lnot128: |
| 916 | llgf $t0,16($inp) |
| 917 | llgf $t1,20($inp) |
| 918 | st $t0,16($key) |
| 919 | st $t1,20($key) |
| 920 | lghi $t0,192 |
| 921 | cr $bits,$t0 |
| 922 | jne .Lnot192 |
| 923 | |
| 924 | llill $mask,0xff |
| 925 | lghi $t3,0 # i=0 |
| 926 | lghi $rounds,12 |
| 927 | st $rounds,240($key) |
| 928 | lghi $rounds,8 |
| 929 | |
| 930 | srlg $i1,$t1,8 |
| 931 | srlg $i2,$t1,16 |
| 932 | srlg $i3,$t1,24 |
| 933 | nr $t1,$mask |
| 934 | nr $i1,$mask |
| 935 | nr $i2,$mask |
| 936 | |
| 937 | .align 16 |
| 938 | .L192_loop: |
| 939 | la $t1,0($t1,$tbl) |
| 940 | la $i1,0($i1,$tbl) |
| 941 | la $i2,0($i2,$tbl) |
| 942 | la $i3,0($i3,$tbl) |
| 943 | icm $t1,2,0($t1) # Te4[rk[5]>>0]<<8 |
| 944 | icm $t1,4,0($i1) # Te4[rk[5]>>8]<<16 |
| 945 | icm $t1,8,0($i2) # Te4[rk[5]>>16]<<24 |
| 946 | icm $t1,1,0($i3) # Te4[rk[5]>>24] |
| 947 | x $t1,256($t3,$tbl) # rcon[i] |
| 948 | xr $s0,$t1 # rk[6]=rk[0]^... |
| 949 | xr $s1,$s0 # rk[7]=rk[1]^rk[6] |
| 950 | xr $s2,$s1 # rk[8]=rk[2]^rk[7] |
| 951 | xr $s3,$s2 # rk[9]=rk[3]^rk[8] |
| 952 | |
| 953 | st $s0,24($key) |
| 954 | st $s1,28($key) |
| 955 | st $s2,32($key) |
| 956 | st $s3,36($key) |
| 957 | brct $rounds,.L192_continue |
| 958 | lghi $t0,12 |
| 959 | lghi %r2,0 |
| 960 | lm${g} %r4,%r13,4*$SIZE_T($sp) |
| 961 | br $ra |
| 962 | |
| 963 | .align 16 |
| 964 | .L192_continue: |
| 965 | lgr $t1,$s3 |
| 966 | x $t1,16($key) # rk[10]=rk[4]^rk[9] |
| 967 | st $t1,40($key) |
| 968 | x $t1,20($key) # rk[11]=rk[5]^rk[10] |
| 969 | st $t1,44($key) |
| 970 | |
| 971 | srlg $i1,$t1,8 |
| 972 | srlg $i2,$t1,16 |
| 973 | srlg $i3,$t1,24 |
| 974 | nr $t1,$mask |
| 975 | nr $i1,$mask |
| 976 | nr $i2,$mask |
| 977 | |
| 978 | la $key,24($key) # key+=6 |
| 979 | la $t3,4($t3) # i++ |
| 980 | j .L192_loop |
| 981 | |
| 982 | .align 16 |
| 983 | .Lnot192: |
| 984 | llgf $t0,24($inp) |
| 985 | llgf $t1,28($inp) |
| 986 | st $t0,24($key) |
| 987 | st $t1,28($key) |
| 988 | llill $mask,0xff |
| 989 | lghi $t3,0 # i=0 |
| 990 | lghi $rounds,14 |
| 991 | st $rounds,240($key) |
| 992 | lghi $rounds,7 |
| 993 | |
| 994 | srlg $i1,$t1,8 |
| 995 | srlg $i2,$t1,16 |
| 996 | srlg $i3,$t1,24 |
| 997 | nr $t1,$mask |
| 998 | nr $i1,$mask |
| 999 | nr $i2,$mask |
| 1000 | |
| 1001 | .align 16 |
| 1002 | .L256_loop: |
| 1003 | la $t1,0($t1,$tbl) |
| 1004 | la $i1,0($i1,$tbl) |
| 1005 | la $i2,0($i2,$tbl) |
| 1006 | la $i3,0($i3,$tbl) |
| 1007 | icm $t1,2,0($t1) # Te4[rk[7]>>0]<<8 |
| 1008 | icm $t1,4,0($i1) # Te4[rk[7]>>8]<<16 |
| 1009 | icm $t1,8,0($i2) # Te4[rk[7]>>16]<<24 |
| 1010 | icm $t1,1,0($i3) # Te4[rk[7]>>24] |
| 1011 | x $t1,256($t3,$tbl) # rcon[i] |
| 1012 | xr $s0,$t1 # rk[8]=rk[0]^... |
| 1013 | xr $s1,$s0 # rk[9]=rk[1]^rk[8] |
| 1014 | xr $s2,$s1 # rk[10]=rk[2]^rk[9] |
| 1015 | xr $s3,$s2 # rk[11]=rk[3]^rk[10] |
| 1016 | st $s0,32($key) |
| 1017 | st $s1,36($key) |
| 1018 | st $s2,40($key) |
| 1019 | st $s3,44($key) |
| 1020 | brct $rounds,.L256_continue |
| 1021 | lghi $t0,14 |
| 1022 | lghi %r2,0 |
| 1023 | lm${g} %r4,%r13,4*$SIZE_T($sp) |
| 1024 | br $ra |
| 1025 | |
| 1026 | .align 16 |
| 1027 | .L256_continue: |
| 1028 | lgr $t1,$s3 # temp=rk[11] |
| 1029 | srlg $i1,$s3,8 |
| 1030 | srlg $i2,$s3,16 |
| 1031 | srlg $i3,$s3,24 |
| 1032 | nr $t1,$mask |
| 1033 | nr $i1,$mask |
| 1034 | nr $i2,$mask |
| 1035 | la $t1,0($t1,$tbl) |
| 1036 | la $i1,0($i1,$tbl) |
| 1037 | la $i2,0($i2,$tbl) |
| 1038 | la $i3,0($i3,$tbl) |
| 1039 | llgc $t1,0($t1) # Te4[rk[11]>>0] |
| 1040 | icm $t1,2,0($i1) # Te4[rk[11]>>8]<<8 |
| 1041 | icm $t1,4,0($i2) # Te4[rk[11]>>16]<<16 |
| 1042 | icm $t1,8,0($i3) # Te4[rk[11]>>24]<<24 |
| 1043 | x $t1,16($key) # rk[12]=rk[4]^... |
| 1044 | st $t1,48($key) |
| 1045 | x $t1,20($key) # rk[13]=rk[5]^rk[12] |
| 1046 | st $t1,52($key) |
| 1047 | x $t1,24($key) # rk[14]=rk[6]^rk[13] |
| 1048 | st $t1,56($key) |
| 1049 | x $t1,28($key) # rk[15]=rk[7]^rk[14] |
| 1050 | st $t1,60($key) |
| 1051 | |
| 1052 | srlg $i1,$t1,8 |
| 1053 | srlg $i2,$t1,16 |
| 1054 | srlg $i3,$t1,24 |
| 1055 | nr $t1,$mask |
| 1056 | nr $i1,$mask |
| 1057 | nr $i2,$mask |
| 1058 | |
| 1059 | la $key,32($key) # key+=8 |
| 1060 | la $t3,4($t3) # i++ |
| 1061 | j .L256_loop |
| 1062 | |
| 1063 | .Lminus1: |
| 1064 | lghi %r2,-1 |
| 1065 | br $ra |
| 1066 | .size private_AES_set_encrypt_key,.-private_AES_set_encrypt_key |
| 1067 | |
| 1068 | # void AES_set_decrypt_key(const unsigned char *in, int bits, |
| 1069 | # AES_KEY *key) { |
| 1070 | .globl private_AES_set_decrypt_key |
| 1071 | .type private_AES_set_decrypt_key,\@function |
| 1072 | .align 16 |
| 1073 | private_AES_set_decrypt_key: |
| 1074 | #st${g} $key,4*$SIZE_T($sp) # I rely on AES_set_encrypt_key to |
| 1075 | st${g} $ra,14*$SIZE_T($sp) # save non-volatile registers and $key! |
| 1076 | bras $ra,_s390x_AES_set_encrypt_key |
| 1077 | #l${g} $key,4*$SIZE_T($sp) |
| 1078 | l${g} $ra,14*$SIZE_T($sp) |
| 1079 | ltgr %r2,%r2 |
| 1080 | bnzr $ra |
| 1081 | ___ |
| 1082 | $code.=<<___ if (!$softonly); |
| 1083 | #l $t0,240($key) |
| 1084 | lhi $t1,16 |
| 1085 | cr $t0,$t1 |
| 1086 | jl .Lgo |
| 1087 | oill $t0,0x80 # set "decrypt" bit |
| 1088 | st $t0,240($key) |
| 1089 | br $ra |
| 1090 | ___ |
| 1091 | $code.=<<___; |
| 1092 | .align 16 |
| 1093 | .Lgo: lgr $rounds,$t0 #llgf $rounds,240($key) |
| 1094 | la $i1,0($key) |
| 1095 | sllg $i2,$rounds,4 |
| 1096 | la $i2,0($i2,$key) |
| 1097 | srl $rounds,1 |
| 1098 | lghi $t1,-16 |
| 1099 | |
| 1100 | .align 16 |
| 1101 | .Linv: lmg $s0,$s1,0($i1) |
| 1102 | lmg $s2,$s3,0($i2) |
| 1103 | stmg $s0,$s1,0($i2) |
| 1104 | stmg $s2,$s3,0($i1) |
| 1105 | la $i1,16($i1) |
| 1106 | la $i2,0($t1,$i2) |
| 1107 | brct $rounds,.Linv |
| 1108 | ___ |
| 1109 | $mask80=$i1; |
| 1110 | $mask1b=$i2; |
| 1111 | $maskfe=$i3; |
| 1112 | $code.=<<___; |
| 1113 | llgf $rounds,240($key) |
| 1114 | aghi $rounds,-1 |
| 1115 | sll $rounds,2 # (rounds-1)*4 |
| 1116 | llilh $mask80,0x8080 |
| 1117 | llilh $mask1b,0x1b1b |
| 1118 | llilh $maskfe,0xfefe |
| 1119 | oill $mask80,0x8080 |
| 1120 | oill $mask1b,0x1b1b |
| 1121 | oill $maskfe,0xfefe |
| 1122 | |
| 1123 | .align 16 |
| 1124 | .Lmix: l $s0,16($key) # tp1 |
| 1125 | lr $s1,$s0 |
| 1126 | ngr $s1,$mask80 |
| 1127 | srlg $t1,$s1,7 |
| 1128 | slr $s1,$t1 |
| 1129 | nr $s1,$mask1b |
| 1130 | sllg $t1,$s0,1 |
| 1131 | nr $t1,$maskfe |
| 1132 | xr $s1,$t1 # tp2 |
| 1133 | |
| 1134 | lr $s2,$s1 |
| 1135 | ngr $s2,$mask80 |
| 1136 | srlg $t1,$s2,7 |
| 1137 | slr $s2,$t1 |
| 1138 | nr $s2,$mask1b |
| 1139 | sllg $t1,$s1,1 |
| 1140 | nr $t1,$maskfe |
| 1141 | xr $s2,$t1 # tp4 |
| 1142 | |
| 1143 | lr $s3,$s2 |
| 1144 | ngr $s3,$mask80 |
| 1145 | srlg $t1,$s3,7 |
| 1146 | slr $s3,$t1 |
| 1147 | nr $s3,$mask1b |
| 1148 | sllg $t1,$s2,1 |
| 1149 | nr $t1,$maskfe |
| 1150 | xr $s3,$t1 # tp8 |
| 1151 | |
| 1152 | xr $s1,$s0 # tp2^tp1 |
| 1153 | xr $s2,$s0 # tp4^tp1 |
| 1154 | rll $s0,$s0,24 # = ROTATE(tp1,8) |
| 1155 | xr $s2,$s3 # ^=tp8 |
| 1156 | xr $s0,$s1 # ^=tp2^tp1 |
| 1157 | xr $s1,$s3 # tp2^tp1^tp8 |
| 1158 | xr $s0,$s2 # ^=tp4^tp1^tp8 |
| 1159 | rll $s1,$s1,8 |
| 1160 | rll $s2,$s2,16 |
| 1161 | xr $s0,$s1 # ^= ROTATE(tp8^tp2^tp1,24) |
| 1162 | rll $s3,$s3,24 |
| 1163 | xr $s0,$s2 # ^= ROTATE(tp8^tp4^tp1,16) |
| 1164 | xr $s0,$s3 # ^= ROTATE(tp8,8) |
| 1165 | |
| 1166 | st $s0,16($key) |
| 1167 | la $key,4($key) |
| 1168 | brct $rounds,.Lmix |
| 1169 | |
| 1170 | lm${g} %r6,%r13,6*$SIZE_T($sp)# as was saved by AES_set_encrypt_key! |
| 1171 | lghi %r2,0 |
| 1172 | br $ra |
| 1173 | .size private_AES_set_decrypt_key,.-private_AES_set_decrypt_key |
| 1174 | ___ |
| 1175 | |
| 1176 | ######################################################################## |
| 1177 | # void AES_cbc_encrypt(const unsigned char *in, unsigned char *out, |
| 1178 | # size_t length, const AES_KEY *key, |
| 1179 | # unsigned char *ivec, const int enc) |
| 1180 | { |
| 1181 | my $inp="%r2"; |
| 1182 | my $out="%r4"; # length and out are swapped |
| 1183 | my $len="%r3"; |
| 1184 | my $key="%r5"; |
| 1185 | my $ivp="%r6"; |
| 1186 | |
| 1187 | $code.=<<___; |
| 1188 | .globl AES_cbc_encrypt |
| 1189 | .type AES_cbc_encrypt,\@function |
| 1190 | .align 16 |
| 1191 | AES_cbc_encrypt: |
| 1192 | xgr %r3,%r4 # flip %r3 and %r4, out and len |
| 1193 | xgr %r4,%r3 |
| 1194 | xgr %r3,%r4 |
| 1195 | ___ |
| 1196 | $code.=<<___ if (!$softonly); |
| 1197 | lhi %r0,16 |
| 1198 | cl %r0,240($key) |
| 1199 | jh .Lcbc_software |
| 1200 | |
| 1201 | lg %r0,0($ivp) # copy ivec |
| 1202 | lg %r1,8($ivp) |
| 1203 | stmg %r0,%r1,16($sp) |
| 1204 | lmg %r0,%r1,0($key) # copy key, cover 256 bit |
| 1205 | stmg %r0,%r1,32($sp) |
| 1206 | lmg %r0,%r1,16($key) |
| 1207 | stmg %r0,%r1,48($sp) |
| 1208 | l %r0,240($key) # load kmc code |
| 1209 | lghi $key,15 # res=len%16, len-=res; |
| 1210 | ngr $key,$len |
| 1211 | sl${g}r $len,$key |
| 1212 | la %r1,16($sp) # parameter block - ivec || key |
| 1213 | jz .Lkmc_truncated |
| 1214 | .long 0xb92f0042 # kmc %r4,%r2 |
| 1215 | brc 1,.-4 # pay attention to "partial completion" |
| 1216 | ltr $key,$key |
| 1217 | jnz .Lkmc_truncated |
| 1218 | .Lkmc_done: |
| 1219 | lmg %r0,%r1,16($sp) # copy ivec to caller |
| 1220 | stg %r0,0($ivp) |
| 1221 | stg %r1,8($ivp) |
| 1222 | br $ra |
| 1223 | .align 16 |
| 1224 | .Lkmc_truncated: |
| 1225 | ahi $key,-1 # it's the way it's encoded in mvc |
| 1226 | tmll %r0,0x80 |
| 1227 | jnz .Lkmc_truncated_dec |
| 1228 | lghi %r1,0 |
| 1229 | stg %r1,16*$SIZE_T($sp) |
| 1230 | stg %r1,16*$SIZE_T+8($sp) |
| 1231 | bras %r1,1f |
| 1232 | mvc 16*$SIZE_T(1,$sp),0($inp) |
| 1233 | 1: ex $key,0(%r1) |
| 1234 | la %r1,16($sp) # restore parameter block |
| 1235 | la $inp,16*$SIZE_T($sp) |
| 1236 | lghi $len,16 |
| 1237 | .long 0xb92f0042 # kmc %r4,%r2 |
| 1238 | j .Lkmc_done |
| 1239 | .align 16 |
| 1240 | .Lkmc_truncated_dec: |
| 1241 | st${g} $out,4*$SIZE_T($sp) |
| 1242 | la $out,16*$SIZE_T($sp) |
| 1243 | lghi $len,16 |
| 1244 | .long 0xb92f0042 # kmc %r4,%r2 |
| 1245 | l${g} $out,4*$SIZE_T($sp) |
| 1246 | bras %r1,2f |
| 1247 | mvc 0(1,$out),16*$SIZE_T($sp) |
| 1248 | 2: ex $key,0(%r1) |
| 1249 | j .Lkmc_done |
| 1250 | .align 16 |
| 1251 | .Lcbc_software: |
| 1252 | ___ |
| 1253 | $code.=<<___; |
| 1254 | stm${g} $key,$ra,5*$SIZE_T($sp) |
| 1255 | lhi %r0,0 |
| 1256 | cl %r0,`$stdframe+$SIZE_T-4`($sp) |
| 1257 | je .Lcbc_decrypt |
| 1258 | |
| 1259 | larl $tbl,AES_Te |
| 1260 | |
| 1261 | llgf $s0,0($ivp) |
| 1262 | llgf $s1,4($ivp) |
| 1263 | llgf $s2,8($ivp) |
| 1264 | llgf $s3,12($ivp) |
| 1265 | |
| 1266 | lghi $t0,16 |
| 1267 | sl${g}r $len,$t0 |
| 1268 | brc 4,.Lcbc_enc_tail # if borrow |
| 1269 | .Lcbc_enc_loop: |
| 1270 | stm${g} $inp,$out,2*$SIZE_T($sp) |
| 1271 | x $s0,0($inp) |
| 1272 | x $s1,4($inp) |
| 1273 | x $s2,8($inp) |
| 1274 | x $s3,12($inp) |
| 1275 | lgr %r4,$key |
| 1276 | |
| 1277 | bras $ra,_s390x_AES_encrypt |
| 1278 | |
| 1279 | lm${g} $inp,$key,2*$SIZE_T($sp) |
| 1280 | st $s0,0($out) |
| 1281 | st $s1,4($out) |
| 1282 | st $s2,8($out) |
| 1283 | st $s3,12($out) |
| 1284 | |
| 1285 | la $inp,16($inp) |
| 1286 | la $out,16($out) |
| 1287 | lghi $t0,16 |
| 1288 | lt${g}r $len,$len |
| 1289 | jz .Lcbc_enc_done |
| 1290 | sl${g}r $len,$t0 |
| 1291 | brc 4,.Lcbc_enc_tail # if borrow |
| 1292 | j .Lcbc_enc_loop |
| 1293 | .align 16 |
| 1294 | .Lcbc_enc_done: |
| 1295 | l${g} $ivp,6*$SIZE_T($sp) |
| 1296 | st $s0,0($ivp) |
| 1297 | st $s1,4($ivp) |
| 1298 | st $s2,8($ivp) |
| 1299 | st $s3,12($ivp) |
| 1300 | |
| 1301 | lm${g} %r7,$ra,7*$SIZE_T($sp) |
| 1302 | br $ra |
| 1303 | |
| 1304 | .align 16 |
| 1305 | .Lcbc_enc_tail: |
| 1306 | aghi $len,15 |
| 1307 | lghi $t0,0 |
| 1308 | stg $t0,16*$SIZE_T($sp) |
| 1309 | stg $t0,16*$SIZE_T+8($sp) |
| 1310 | bras $t1,3f |
| 1311 | mvc 16*$SIZE_T(1,$sp),0($inp) |
| 1312 | 3: ex $len,0($t1) |
| 1313 | lghi $len,0 |
| 1314 | la $inp,16*$SIZE_T($sp) |
| 1315 | j .Lcbc_enc_loop |
| 1316 | |
| 1317 | .align 16 |
| 1318 | .Lcbc_decrypt: |
| 1319 | larl $tbl,AES_Td |
| 1320 | |
| 1321 | lg $t0,0($ivp) |
| 1322 | lg $t1,8($ivp) |
| 1323 | stmg $t0,$t1,16*$SIZE_T($sp) |
| 1324 | |
| 1325 | .Lcbc_dec_loop: |
| 1326 | stm${g} $inp,$out,2*$SIZE_T($sp) |
| 1327 | llgf $s0,0($inp) |
| 1328 | llgf $s1,4($inp) |
| 1329 | llgf $s2,8($inp) |
| 1330 | llgf $s3,12($inp) |
| 1331 | lgr %r4,$key |
| 1332 | |
| 1333 | bras $ra,_s390x_AES_decrypt |
| 1334 | |
| 1335 | lm${g} $inp,$key,2*$SIZE_T($sp) |
| 1336 | sllg $s0,$s0,32 |
| 1337 | sllg $s2,$s2,32 |
| 1338 | lr $s0,$s1 |
| 1339 | lr $s2,$s3 |
| 1340 | |
| 1341 | lg $t0,0($inp) |
| 1342 | lg $t1,8($inp) |
| 1343 | xg $s0,16*$SIZE_T($sp) |
| 1344 | xg $s2,16*$SIZE_T+8($sp) |
| 1345 | lghi $s1,16 |
| 1346 | sl${g}r $len,$s1 |
| 1347 | brc 4,.Lcbc_dec_tail # if borrow |
| 1348 | brc 2,.Lcbc_dec_done # if zero |
| 1349 | stg $s0,0($out) |
| 1350 | stg $s2,8($out) |
| 1351 | stmg $t0,$t1,16*$SIZE_T($sp) |
| 1352 | |
| 1353 | la $inp,16($inp) |
| 1354 | la $out,16($out) |
| 1355 | j .Lcbc_dec_loop |
| 1356 | |
| 1357 | .Lcbc_dec_done: |
| 1358 | stg $s0,0($out) |
| 1359 | stg $s2,8($out) |
| 1360 | .Lcbc_dec_exit: |
| 1361 | lm${g} %r6,$ra,6*$SIZE_T($sp) |
| 1362 | stmg $t0,$t1,0($ivp) |
| 1363 | |
| 1364 | br $ra |
| 1365 | |
| 1366 | .align 16 |
| 1367 | .Lcbc_dec_tail: |
| 1368 | aghi $len,15 |
| 1369 | stg $s0,16*$SIZE_T($sp) |
| 1370 | stg $s2,16*$SIZE_T+8($sp) |
| 1371 | bras $s1,4f |
| 1372 | mvc 0(1,$out),16*$SIZE_T($sp) |
| 1373 | 4: ex $len,0($s1) |
| 1374 | j .Lcbc_dec_exit |
| 1375 | .size AES_cbc_encrypt,.-AES_cbc_encrypt |
| 1376 | ___ |
| 1377 | } |
| 1378 | ######################################################################## |
| 1379 | # void AES_ctr32_encrypt(const unsigned char *in, unsigned char *out, |
| 1380 | # size_t blocks, const AES_KEY *key, |
| 1381 | # const unsigned char *ivec) |
| 1382 | { |
| 1383 | my $inp="%r2"; |
| 1384 | my $out="%r4"; # blocks and out are swapped |
| 1385 | my $len="%r3"; |
| 1386 | my $key="%r5"; my $iv0="%r5"; |
| 1387 | my $ivp="%r6"; |
| 1388 | my $fp ="%r7"; |
| 1389 | |
| 1390 | $code.=<<___; |
| 1391 | .globl AES_ctr32_encrypt |
| 1392 | .type AES_ctr32_encrypt,\@function |
| 1393 | .align 16 |
| 1394 | AES_ctr32_encrypt: |
| 1395 | xgr %r3,%r4 # flip %r3 and %r4, $out and $len |
| 1396 | xgr %r4,%r3 |
| 1397 | xgr %r3,%r4 |
| 1398 | llgfr $len,$len # safe in ctr32 subroutine even in 64-bit case |
| 1399 | ___ |
| 1400 | $code.=<<___ if (!$softonly); |
| 1401 | l %r0,240($key) |
| 1402 | lhi %r1,16 |
| 1403 | clr %r0,%r1 |
| 1404 | jl .Lctr32_software |
| 1405 | |
| 1406 | stm${g} %r6,$s3,6*$SIZE_T($sp) |
| 1407 | |
| 1408 | slgr $out,$inp |
| 1409 | la %r1,0($key) # %r1 is permanent copy of $key |
| 1410 | lg $iv0,0($ivp) # load ivec |
| 1411 | lg $ivp,8($ivp) |
| 1412 | |
| 1413 | # prepare and allocate stack frame at the top of 4K page |
| 1414 | # with 1K reserved for eventual signal handling |
| 1415 | lghi $s0,-1024-256-16# guarantee at least 256-bytes buffer |
| 1416 | lghi $s1,-4096 |
| 1417 | algr $s0,$sp |
| 1418 | lgr $fp,$sp |
| 1419 | ngr $s0,$s1 # align at page boundary |
| 1420 | slgr $fp,$s0 # total buffer size |
| 1421 | lgr $s2,$sp |
| 1422 | lghi $s1,1024+16 # sl[g]fi is extended-immediate facility |
| 1423 | slgr $fp,$s1 # deduct reservation to get usable buffer size |
| 1424 | # buffer size is at lest 256 and at most 3072+256-16 |
| 1425 | |
| 1426 | la $sp,1024($s0) # alloca |
| 1427 | srlg $fp,$fp,4 # convert bytes to blocks, minimum 16 |
| 1428 | st${g} $s2,0($sp) # back-chain |
| 1429 | st${g} $fp,$SIZE_T($sp) |
| 1430 | |
| 1431 | slgr $len,$fp |
| 1432 | brc 1,.Lctr32_hw_switch # not zero, no borrow |
| 1433 | algr $fp,$len # input is shorter than allocated buffer |
| 1434 | lghi $len,0 |
| 1435 | st${g} $fp,$SIZE_T($sp) |
| 1436 | |
| 1437 | .Lctr32_hw_switch: |
| 1438 | ___ |
| 1439 | $code.=<<___ if (0); ######### kmctr code was measured to be ~12% slower |
| 1440 | larl $s0,OPENSSL_s390xcap_P |
| 1441 | lg $s0,8($s0) |
| 1442 | tmhh $s0,0x0004 # check for message_security-assist-4 |
| 1443 | jz .Lctr32_km_loop |
| 1444 | |
| 1445 | llgfr $s0,%r0 |
| 1446 | lgr $s1,%r1 |
| 1447 | lghi %r0,0 |
| 1448 | la %r1,16($sp) |
| 1449 | .long 0xb92d2042 # kmctr %r4,%r2,%r2 |
| 1450 | |
| 1451 | llihh %r0,0x8000 # check if kmctr supports the function code |
| 1452 | srlg %r0,%r0,0($s0) |
| 1453 | ng %r0,16($sp) |
| 1454 | lgr %r0,$s0 |
| 1455 | lgr %r1,$s1 |
| 1456 | jz .Lctr32_km_loop |
| 1457 | |
| 1458 | ####### kmctr code |
| 1459 | algr $out,$inp # restore $out |
| 1460 | lgr $s1,$len # $s1 undertakes $len |
| 1461 | j .Lctr32_kmctr_loop |
| 1462 | .align 16 |
| 1463 | .Lctr32_kmctr_loop: |
| 1464 | la $s2,16($sp) |
| 1465 | lgr $s3,$fp |
| 1466 | .Lctr32_kmctr_prepare: |
| 1467 | stg $iv0,0($s2) |
| 1468 | stg $ivp,8($s2) |
| 1469 | la $s2,16($s2) |
| 1470 | ahi $ivp,1 # 32-bit increment, preserves upper half |
| 1471 | brct $s3,.Lctr32_kmctr_prepare |
| 1472 | |
| 1473 | #la $inp,0($inp) # inp |
| 1474 | sllg $len,$fp,4 # len |
| 1475 | #la $out,0($out) # out |
| 1476 | la $s2,16($sp) # iv |
| 1477 | .long 0xb92da042 # kmctr $out,$s2,$inp |
| 1478 | brc 1,.-4 # pay attention to "partial completion" |
| 1479 | |
| 1480 | slgr $s1,$fp |
| 1481 | brc 1,.Lctr32_kmctr_loop # not zero, no borrow |
| 1482 | algr $fp,$s1 |
| 1483 | lghi $s1,0 |
| 1484 | brc 4+1,.Lctr32_kmctr_loop # not zero |
| 1485 | |
| 1486 | l${g} $sp,0($sp) |
| 1487 | lm${g} %r6,$s3,6*$SIZE_T($sp) |
| 1488 | br $ra |
| 1489 | .align 16 |
| 1490 | ___ |
| 1491 | $code.=<<___; |
| 1492 | .Lctr32_km_loop: |
| 1493 | la $s2,16($sp) |
| 1494 | lgr $s3,$fp |
| 1495 | .Lctr32_km_prepare: |
| 1496 | stg $iv0,0($s2) |
| 1497 | stg $ivp,8($s2) |
| 1498 | la $s2,16($s2) |
| 1499 | ahi $ivp,1 # 32-bit increment, preserves upper half |
| 1500 | brct $s3,.Lctr32_km_prepare |
| 1501 | |
| 1502 | la $s0,16($sp) # inp |
| 1503 | sllg $s1,$fp,4 # len |
| 1504 | la $s2,16($sp) # out |
| 1505 | .long 0xb92e00a8 # km %r10,%r8 |
| 1506 | brc 1,.-4 # pay attention to "partial completion" |
| 1507 | |
| 1508 | la $s2,16($sp) |
| 1509 | lgr $s3,$fp |
| 1510 | slgr $s2,$inp |
| 1511 | .Lctr32_km_xor: |
| 1512 | lg $s0,0($inp) |
| 1513 | lg $s1,8($inp) |
| 1514 | xg $s0,0($s2,$inp) |
| 1515 | xg $s1,8($s2,$inp) |
| 1516 | stg $s0,0($out,$inp) |
| 1517 | stg $s1,8($out,$inp) |
| 1518 | la $inp,16($inp) |
| 1519 | brct $s3,.Lctr32_km_xor |
| 1520 | |
| 1521 | slgr $len,$fp |
| 1522 | brc 1,.Lctr32_km_loop # not zero, no borrow |
| 1523 | algr $fp,$len |
| 1524 | lghi $len,0 |
| 1525 | brc 4+1,.Lctr32_km_loop # not zero |
| 1526 | |
| 1527 | l${g} $s0,0($sp) |
| 1528 | l${g} $s1,$SIZE_T($sp) |
| 1529 | la $s2,16($sp) |
| 1530 | .Lctr32_km_zap: |
| 1531 | stg $s0,0($s2) |
| 1532 | stg $s0,8($s2) |
| 1533 | la $s2,16($s2) |
| 1534 | brct $s1,.Lctr32_km_zap |
| 1535 | |
| 1536 | la $sp,0($s0) |
| 1537 | lm${g} %r6,$s3,6*$SIZE_T($sp) |
| 1538 | br $ra |
| 1539 | .align 16 |
| 1540 | .Lctr32_software: |
| 1541 | ___ |
| 1542 | $code.=<<___; |
| 1543 | stm${g} $key,$ra,5*$SIZE_T($sp) |
| 1544 | sl${g}r $inp,$out |
| 1545 | larl $tbl,AES_Te |
| 1546 | llgf $t1,12($ivp) |
| 1547 | |
| 1548 | .Lctr32_loop: |
| 1549 | stm${g} $inp,$out,2*$SIZE_T($sp) |
| 1550 | llgf $s0,0($ivp) |
| 1551 | llgf $s1,4($ivp) |
| 1552 | llgf $s2,8($ivp) |
| 1553 | lgr $s3,$t1 |
| 1554 | st $t1,16*$SIZE_T($sp) |
| 1555 | lgr %r4,$key |
| 1556 | |
| 1557 | bras $ra,_s390x_AES_encrypt |
| 1558 | |
| 1559 | lm${g} $inp,$ivp,2*$SIZE_T($sp) |
| 1560 | llgf $t1,16*$SIZE_T($sp) |
| 1561 | x $s0,0($inp,$out) |
| 1562 | x $s1,4($inp,$out) |
| 1563 | x $s2,8($inp,$out) |
| 1564 | x $s3,12($inp,$out) |
| 1565 | stm $s0,$s3,0($out) |
| 1566 | |
| 1567 | la $out,16($out) |
| 1568 | ahi $t1,1 # 32-bit increment |
| 1569 | brct $len,.Lctr32_loop |
| 1570 | |
| 1571 | lm${g} %r6,$ra,6*$SIZE_T($sp) |
| 1572 | br $ra |
| 1573 | .size AES_ctr32_encrypt,.-AES_ctr32_encrypt |
| 1574 | ___ |
| 1575 | } |
| 1576 | |
| 1577 | ######################################################################## |
| 1578 | # void AES_xts_encrypt(const char *inp,char *out,size_t len, |
| 1579 | # const AES_KEY *key1, const AES_KEY *key2, |
| 1580 | # const unsigned char iv[16]); |
| 1581 | # |
| 1582 | { |
| 1583 | my $inp="%r2"; |
| 1584 | my $out="%r4"; # len and out are swapped |
| 1585 | my $len="%r3"; |
| 1586 | my $key1="%r5"; # $i1 |
| 1587 | my $key2="%r6"; # $i2 |
| 1588 | my $fp="%r7"; # $i3 |
| 1589 | my $tweak=16*$SIZE_T+16; # or $stdframe-16, bottom of the frame... |
| 1590 | |
| 1591 | $code.=<<___; |
| 1592 | .type _s390x_xts_km,\@function |
| 1593 | .align 16 |
| 1594 | _s390x_xts_km: |
| 1595 | ___ |
| 1596 | $code.=<<___ if(1); |
| 1597 | llgfr $s0,%r0 # put aside the function code |
| 1598 | lghi $s1,0x7f |
| 1599 | nr $s1,%r0 |
| 1600 | lghi %r0,0 # query capability vector |
| 1601 | la %r1,2*$SIZE_T($sp) |
| 1602 | .long 0xb92e0042 # km %r4,%r2 |
| 1603 | llihh %r1,0x8000 |
| 1604 | srlg %r1,%r1,32($s1) # check for 32+function code |
| 1605 | ng %r1,2*$SIZE_T($sp) |
| 1606 | lgr %r0,$s0 # restore the function code |
| 1607 | la %r1,0($key1) # restore $key1 |
| 1608 | jz .Lxts_km_vanilla |
| 1609 | |
| 1610 | lmg $i2,$i3,$tweak($sp) # put aside the tweak value |
| 1611 | algr $out,$inp |
| 1612 | |
| 1613 | oill %r0,32 # switch to xts function code |
| 1614 | aghi $s1,-18 # |
| 1615 | sllg $s1,$s1,3 # (function code - 18)*8, 0 or 16 |
| 1616 | la %r1,$tweak-16($sp) |
| 1617 | slgr %r1,$s1 # parameter block position |
| 1618 | lmg $s0,$s3,0($key1) # load 256 bits of key material, |
| 1619 | stmg $s0,$s3,0(%r1) # and copy it to parameter block. |
| 1620 | # yes, it contains junk and overlaps |
| 1621 | # with the tweak in 128-bit case. |
| 1622 | # it's done to avoid conditional |
| 1623 | # branch. |
| 1624 | stmg $i2,$i3,$tweak($sp) # "re-seat" the tweak value |
| 1625 | |
| 1626 | .long 0xb92e0042 # km %r4,%r2 |
| 1627 | brc 1,.-4 # pay attention to "partial completion" |
| 1628 | |
| 1629 | lrvg $s0,$tweak+0($sp) # load the last tweak |
| 1630 | lrvg $s1,$tweak+8($sp) |
| 1631 | stmg %r0,%r3,$tweak-32(%r1) # wipe copy of the key |
| 1632 | |
| 1633 | nill %r0,0xffdf # switch back to original function code |
| 1634 | la %r1,0($key1) # restore pointer to $key1 |
| 1635 | slgr $out,$inp |
| 1636 | |
| 1637 | llgc $len,2*$SIZE_T-1($sp) |
| 1638 | nill $len,0x0f # $len%=16 |
| 1639 | br $ra |
| 1640 | |
| 1641 | .align 16 |
| 1642 | .Lxts_km_vanilla: |
| 1643 | ___ |
| 1644 | $code.=<<___; |
| 1645 | # prepare and allocate stack frame at the top of 4K page |
| 1646 | # with 1K reserved for eventual signal handling |
| 1647 | lghi $s0,-1024-256-16# guarantee at least 256-bytes buffer |
| 1648 | lghi $s1,-4096 |
| 1649 | algr $s0,$sp |
| 1650 | lgr $fp,$sp |
| 1651 | ngr $s0,$s1 # align at page boundary |
| 1652 | slgr $fp,$s0 # total buffer size |
| 1653 | lgr $s2,$sp |
| 1654 | lghi $s1,1024+16 # sl[g]fi is extended-immediate facility |
| 1655 | slgr $fp,$s1 # deduct reservation to get usable buffer size |
| 1656 | # buffer size is at lest 256 and at most 3072+256-16 |
| 1657 | |
| 1658 | la $sp,1024($s0) # alloca |
| 1659 | nill $fp,0xfff0 # round to 16*n |
| 1660 | st${g} $s2,0($sp) # back-chain |
| 1661 | nill $len,0xfff0 # redundant |
| 1662 | st${g} $fp,$SIZE_T($sp) |
| 1663 | |
| 1664 | slgr $len,$fp |
| 1665 | brc 1,.Lxts_km_go # not zero, no borrow |
| 1666 | algr $fp,$len # input is shorter than allocated buffer |
| 1667 | lghi $len,0 |
| 1668 | st${g} $fp,$SIZE_T($sp) |
| 1669 | |
| 1670 | .Lxts_km_go: |
| 1671 | lrvg $s0,$tweak+0($s2) # load the tweak value in little-endian |
| 1672 | lrvg $s1,$tweak+8($s2) |
| 1673 | |
| 1674 | la $s2,16($sp) # vector of ascending tweak values |
| 1675 | slgr $s2,$inp |
| 1676 | srlg $s3,$fp,4 |
| 1677 | j .Lxts_km_start |
| 1678 | |
| 1679 | .Lxts_km_loop: |
| 1680 | la $s2,16($sp) |
| 1681 | slgr $s2,$inp |
| 1682 | srlg $s3,$fp,4 |
| 1683 | .Lxts_km_prepare: |
| 1684 | lghi $i1,0x87 |
| 1685 | srag $i2,$s1,63 # broadcast upper bit |
| 1686 | ngr $i1,$i2 # rem |
| 1687 | srlg $i2,$s0,63 # carry bit from lower half |
| 1688 | sllg $s0,$s0,1 |
| 1689 | sllg $s1,$s1,1 |
| 1690 | xgr $s0,$i1 |
| 1691 | ogr $s1,$i2 |
| 1692 | .Lxts_km_start: |
| 1693 | lrvgr $i1,$s0 # flip byte order |
| 1694 | lrvgr $i2,$s1 |
| 1695 | stg $i1,0($s2,$inp) |
| 1696 | stg $i2,8($s2,$inp) |
| 1697 | xg $i1,0($inp) |
| 1698 | xg $i2,8($inp) |
| 1699 | stg $i1,0($out,$inp) |
| 1700 | stg $i2,8($out,$inp) |
| 1701 | la $inp,16($inp) |
| 1702 | brct $s3,.Lxts_km_prepare |
| 1703 | |
| 1704 | slgr $inp,$fp # rewind $inp |
| 1705 | la $s2,0($out,$inp) |
| 1706 | lgr $s3,$fp |
| 1707 | .long 0xb92e00aa # km $s2,$s2 |
| 1708 | brc 1,.-4 # pay attention to "partial completion" |
| 1709 | |
| 1710 | la $s2,16($sp) |
| 1711 | slgr $s2,$inp |
| 1712 | srlg $s3,$fp,4 |
| 1713 | .Lxts_km_xor: |
| 1714 | lg $i1,0($out,$inp) |
| 1715 | lg $i2,8($out,$inp) |
| 1716 | xg $i1,0($s2,$inp) |
| 1717 | xg $i2,8($s2,$inp) |
| 1718 | stg $i1,0($out,$inp) |
| 1719 | stg $i2,8($out,$inp) |
| 1720 | la $inp,16($inp) |
| 1721 | brct $s3,.Lxts_km_xor |
| 1722 | |
| 1723 | slgr $len,$fp |
| 1724 | brc 1,.Lxts_km_loop # not zero, no borrow |
| 1725 | algr $fp,$len |
| 1726 | lghi $len,0 |
| 1727 | brc 4+1,.Lxts_km_loop # not zero |
| 1728 | |
| 1729 | l${g} $i1,0($sp) # back-chain |
| 1730 | llgf $fp,`2*$SIZE_T-4`($sp) # bytes used |
| 1731 | la $i2,16($sp) |
| 1732 | srlg $fp,$fp,4 |
| 1733 | .Lxts_km_zap: |
| 1734 | stg $i1,0($i2) |
| 1735 | stg $i1,8($i2) |
| 1736 | la $i2,16($i2) |
| 1737 | brct $fp,.Lxts_km_zap |
| 1738 | |
| 1739 | la $sp,0($i1) |
| 1740 | llgc $len,2*$SIZE_T-1($i1) |
| 1741 | nill $len,0x0f # $len%=16 |
| 1742 | bzr $ra |
| 1743 | |
| 1744 | # generate one more tweak... |
| 1745 | lghi $i1,0x87 |
| 1746 | srag $i2,$s1,63 # broadcast upper bit |
| 1747 | ngr $i1,$i2 # rem |
| 1748 | srlg $i2,$s0,63 # carry bit from lower half |
| 1749 | sllg $s0,$s0,1 |
| 1750 | sllg $s1,$s1,1 |
| 1751 | xgr $s0,$i1 |
| 1752 | ogr $s1,$i2 |
| 1753 | |
| 1754 | ltr $len,$len # clear zero flag |
| 1755 | br $ra |
| 1756 | .size _s390x_xts_km,.-_s390x_xts_km |
| 1757 | |
| 1758 | .globl AES_xts_encrypt |
| 1759 | .type AES_xts_encrypt,\@function |
| 1760 | .align 16 |
| 1761 | AES_xts_encrypt: |
| 1762 | xgr %r3,%r4 # flip %r3 and %r4, $out and $len |
| 1763 | xgr %r4,%r3 |
| 1764 | xgr %r3,%r4 |
| 1765 | ___ |
| 1766 | $code.=<<___ if ($SIZE_T==4); |
| 1767 | llgfr $len,$len |
| 1768 | ___ |
| 1769 | $code.=<<___; |
| 1770 | st${g} $len,1*$SIZE_T($sp) # save copy of $len |
| 1771 | srag $len,$len,4 # formally wrong, because it expands |
| 1772 | # sign byte, but who can afford asking |
| 1773 | # to process more than 2^63-1 bytes? |
| 1774 | # I use it, because it sets condition |
| 1775 | # code... |
| 1776 | bcr 8,$ra # abort if zero (i.e. less than 16) |
| 1777 | ___ |
| 1778 | $code.=<<___ if (!$softonly); |
| 1779 | llgf %r0,240($key2) |
| 1780 | lhi %r1,16 |
| 1781 | clr %r0,%r1 |
| 1782 | jl .Lxts_enc_software |
| 1783 | |
| 1784 | stm${g} %r6,$s3,6*$SIZE_T($sp) |
| 1785 | st${g} $ra,14*$SIZE_T($sp) |
| 1786 | |
| 1787 | sllg $len,$len,4 # $len&=~15 |
| 1788 | slgr $out,$inp |
| 1789 | |
| 1790 | # generate the tweak value |
| 1791 | l${g} $s3,$stdframe($sp) # pointer to iv |
| 1792 | la $s2,$tweak($sp) |
| 1793 | lmg $s0,$s1,0($s3) |
| 1794 | lghi $s3,16 |
| 1795 | stmg $s0,$s1,0($s2) |
| 1796 | la %r1,0($key2) # $key2 is not needed anymore |
| 1797 | .long 0xb92e00aa # km $s2,$s2, generate the tweak |
| 1798 | brc 1,.-4 # can this happen? |
| 1799 | |
| 1800 | l %r0,240($key1) |
| 1801 | la %r1,0($key1) # $key1 is not needed anymore |
| 1802 | bras $ra,_s390x_xts_km |
| 1803 | jz .Lxts_enc_km_done |
| 1804 | |
| 1805 | aghi $inp,-16 # take one step back |
| 1806 | la $i3,0($out,$inp) # put aside real $out |
| 1807 | .Lxts_enc_km_steal: |
| 1808 | llgc $i1,16($inp) |
| 1809 | llgc $i2,0($out,$inp) |
| 1810 | stc $i1,0($out,$inp) |
| 1811 | stc $i2,16($out,$inp) |
| 1812 | la $inp,1($inp) |
| 1813 | brct $len,.Lxts_enc_km_steal |
| 1814 | |
| 1815 | la $s2,0($i3) |
| 1816 | lghi $s3,16 |
| 1817 | lrvgr $i1,$s0 # flip byte order |
| 1818 | lrvgr $i2,$s1 |
| 1819 | xg $i1,0($s2) |
| 1820 | xg $i2,8($s2) |
| 1821 | stg $i1,0($s2) |
| 1822 | stg $i2,8($s2) |
| 1823 | .long 0xb92e00aa # km $s2,$s2 |
| 1824 | brc 1,.-4 # can this happen? |
| 1825 | lrvgr $i1,$s0 # flip byte order |
| 1826 | lrvgr $i2,$s1 |
| 1827 | xg $i1,0($i3) |
| 1828 | xg $i2,8($i3) |
| 1829 | stg $i1,0($i3) |
| 1830 | stg $i2,8($i3) |
| 1831 | |
| 1832 | .Lxts_enc_km_done: |
| 1833 | l${g} $ra,14*$SIZE_T($sp) |
| 1834 | st${g} $sp,$tweak($sp) # wipe tweak |
| 1835 | st${g} $sp,$tweak($sp) |
| 1836 | lm${g} %r6,$s3,6*$SIZE_T($sp) |
| 1837 | br $ra |
| 1838 | .align 16 |
| 1839 | .Lxts_enc_software: |
| 1840 | ___ |
| 1841 | $code.=<<___; |
| 1842 | stm${g} %r6,$ra,6*$SIZE_T($sp) |
| 1843 | |
| 1844 | slgr $out,$inp |
| 1845 | |
| 1846 | xgr $s0,$s0 # clear upper half |
| 1847 | xgr $s1,$s1 |
| 1848 | lrv $s0,$stdframe+4($sp) # load secno |
| 1849 | lrv $s1,$stdframe+0($sp) |
| 1850 | xgr $s2,$s2 |
| 1851 | xgr $s3,$s3 |
| 1852 | stm${g} %r2,%r5,2*$SIZE_T($sp) |
| 1853 | la $key,0($key2) |
| 1854 | larl $tbl,AES_Te |
| 1855 | bras $ra,_s390x_AES_encrypt # generate the tweak |
| 1856 | lm${g} %r2,%r5,2*$SIZE_T($sp) |
| 1857 | stm $s0,$s3,$tweak($sp) # save the tweak |
| 1858 | j .Lxts_enc_enter |
| 1859 | |
| 1860 | .align 16 |
| 1861 | .Lxts_enc_loop: |
| 1862 | lrvg $s1,$tweak+0($sp) # load the tweak in little-endian |
| 1863 | lrvg $s3,$tweak+8($sp) |
| 1864 | lghi %r1,0x87 |
| 1865 | srag %r0,$s3,63 # broadcast upper bit |
| 1866 | ngr %r1,%r0 # rem |
| 1867 | srlg %r0,$s1,63 # carry bit from lower half |
| 1868 | sllg $s1,$s1,1 |
| 1869 | sllg $s3,$s3,1 |
| 1870 | xgr $s1,%r1 |
| 1871 | ogr $s3,%r0 |
| 1872 | lrvgr $s1,$s1 # flip byte order |
| 1873 | lrvgr $s3,$s3 |
| 1874 | srlg $s0,$s1,32 # smash the tweak to 4x32-bits |
| 1875 | stg $s1,$tweak+0($sp) # save the tweak |
| 1876 | llgfr $s1,$s1 |
| 1877 | srlg $s2,$s3,32 |
| 1878 | stg $s3,$tweak+8($sp) |
| 1879 | llgfr $s3,$s3 |
| 1880 | la $inp,16($inp) # $inp+=16 |
| 1881 | .Lxts_enc_enter: |
| 1882 | x $s0,0($inp) # ^=*($inp) |
| 1883 | x $s1,4($inp) |
| 1884 | x $s2,8($inp) |
| 1885 | x $s3,12($inp) |
| 1886 | stm${g} %r2,%r3,2*$SIZE_T($sp) # only two registers are changing |
| 1887 | la $key,0($key1) |
| 1888 | bras $ra,_s390x_AES_encrypt |
| 1889 | lm${g} %r2,%r5,2*$SIZE_T($sp) |
| 1890 | x $s0,$tweak+0($sp) # ^=tweak |
| 1891 | x $s1,$tweak+4($sp) |
| 1892 | x $s2,$tweak+8($sp) |
| 1893 | x $s3,$tweak+12($sp) |
| 1894 | st $s0,0($out,$inp) |
| 1895 | st $s1,4($out,$inp) |
| 1896 | st $s2,8($out,$inp) |
| 1897 | st $s3,12($out,$inp) |
| 1898 | brct${g} $len,.Lxts_enc_loop |
| 1899 | |
| 1900 | llgc $len,`2*$SIZE_T-1`($sp) |
| 1901 | nill $len,0x0f # $len%16 |
| 1902 | jz .Lxts_enc_done |
| 1903 | |
| 1904 | la $i3,0($inp,$out) # put aside real $out |
| 1905 | .Lxts_enc_steal: |
| 1906 | llgc %r0,16($inp) |
| 1907 | llgc %r1,0($out,$inp) |
| 1908 | stc %r0,0($out,$inp) |
| 1909 | stc %r1,16($out,$inp) |
| 1910 | la $inp,1($inp) |
| 1911 | brct $len,.Lxts_enc_steal |
| 1912 | la $out,0($i3) # restore real $out |
| 1913 | |
| 1914 | # generate last tweak... |
| 1915 | lrvg $s1,$tweak+0($sp) # load the tweak in little-endian |
| 1916 | lrvg $s3,$tweak+8($sp) |
| 1917 | lghi %r1,0x87 |
| 1918 | srag %r0,$s3,63 # broadcast upper bit |
| 1919 | ngr %r1,%r0 # rem |
| 1920 | srlg %r0,$s1,63 # carry bit from lower half |
| 1921 | sllg $s1,$s1,1 |
| 1922 | sllg $s3,$s3,1 |
| 1923 | xgr $s1,%r1 |
| 1924 | ogr $s3,%r0 |
| 1925 | lrvgr $s1,$s1 # flip byte order |
| 1926 | lrvgr $s3,$s3 |
| 1927 | srlg $s0,$s1,32 # smash the tweak to 4x32-bits |
| 1928 | stg $s1,$tweak+0($sp) # save the tweak |
| 1929 | llgfr $s1,$s1 |
| 1930 | srlg $s2,$s3,32 |
| 1931 | stg $s3,$tweak+8($sp) |
| 1932 | llgfr $s3,$s3 |
| 1933 | |
| 1934 | x $s0,0($out) # ^=*(inp)|stolen cipther-text |
| 1935 | x $s1,4($out) |
| 1936 | x $s2,8($out) |
| 1937 | x $s3,12($out) |
| 1938 | st${g} $out,4*$SIZE_T($sp) |
| 1939 | la $key,0($key1) |
| 1940 | bras $ra,_s390x_AES_encrypt |
| 1941 | l${g} $out,4*$SIZE_T($sp) |
| 1942 | x $s0,`$tweak+0`($sp) # ^=tweak |
| 1943 | x $s1,`$tweak+4`($sp) |
| 1944 | x $s2,`$tweak+8`($sp) |
| 1945 | x $s3,`$tweak+12`($sp) |
| 1946 | st $s0,0($out) |
| 1947 | st $s1,4($out) |
| 1948 | st $s2,8($out) |
| 1949 | st $s3,12($out) |
| 1950 | |
| 1951 | .Lxts_enc_done: |
| 1952 | stg $sp,$tweak+0($sp) # wipe tweak |
| 1953 | stg $sp,$twesk+8($sp) |
| 1954 | lm${g} %r6,$ra,6*$SIZE_T($sp) |
| 1955 | br $ra |
| 1956 | .size AES_xts_encrypt,.-AES_xts_encrypt |
| 1957 | ___ |
| 1958 | # void AES_xts_decrypt(const char *inp,char *out,size_t len, |
| 1959 | # const AES_KEY *key1, const AES_KEY *key2,u64 secno); |
| 1960 | # |
| 1961 | $code.=<<___; |
| 1962 | .globl AES_xts_decrypt |
| 1963 | .type AES_xts_decrypt,\@function |
| 1964 | .align 16 |
| 1965 | AES_xts_decrypt: |
| 1966 | xgr %r3,%r4 # flip %r3 and %r4, $out and $len |
| 1967 | xgr %r4,%r3 |
| 1968 | xgr %r3,%r4 |
| 1969 | ___ |
| 1970 | $code.=<<___ if ($SIZE_T==4); |
| 1971 | llgfr $len,$len |
| 1972 | ___ |
| 1973 | $code.=<<___; |
| 1974 | st${g} $len,1*$SIZE_T($sp) # save copy of $len |
| 1975 | aghi $len,-16 |
| 1976 | bcr 4,$ra # abort if less than zero. formally |
| 1977 | # wrong, because $len is unsigned, |
| 1978 | # but who can afford asking to |
| 1979 | # process more than 2^63-1 bytes? |
| 1980 | tmll $len,0x0f |
| 1981 | jnz .Lxts_dec_proceed |
| 1982 | aghi $len,16 |
| 1983 | .Lxts_dec_proceed: |
| 1984 | ___ |
| 1985 | $code.=<<___ if (!$softonly); |
| 1986 | llgf %r0,240($key2) |
| 1987 | lhi %r1,16 |
| 1988 | clr %r0,%r1 |
| 1989 | jl .Lxts_dec_software |
| 1990 | |
| 1991 | stm${g} %r6,$s3,6*$SIZE_T($sp) |
| 1992 | st${g} $ra,14*$SIZE_T($sp) |
| 1993 | |
| 1994 | nill $len,0xfff0 # $len&=~15 |
| 1995 | slgr $out,$inp |
| 1996 | |
| 1997 | # generate the tweak value |
| 1998 | l${g} $s3,$stdframe($sp) # pointer to iv |
| 1999 | la $s2,$tweak($sp) |
| 2000 | lmg $s0,$s1,0($s3) |
| 2001 | lghi $s3,16 |
| 2002 | stmg $s0,$s1,0($s2) |
| 2003 | la %r1,0($key2) # $key2 is not needed past this point |
| 2004 | .long 0xb92e00aa # km $s2,$s2, generate the tweak |
| 2005 | brc 1,.-4 # can this happen? |
| 2006 | |
| 2007 | l %r0,240($key1) |
| 2008 | la %r1,0($key1) # $key1 is not needed anymore |
| 2009 | |
| 2010 | ltgr $len,$len |
| 2011 | jz .Lxts_dec_km_short |
| 2012 | bras $ra,_s390x_xts_km |
| 2013 | jz .Lxts_dec_km_done |
| 2014 | |
| 2015 | lrvgr $s2,$s0 # make copy in reverse byte order |
| 2016 | lrvgr $s3,$s1 |
| 2017 | j .Lxts_dec_km_2ndtweak |
| 2018 | |
| 2019 | .Lxts_dec_km_short: |
| 2020 | llgc $len,`2*$SIZE_T-1`($sp) |
| 2021 | nill $len,0x0f # $len%=16 |
| 2022 | lrvg $s0,$tweak+0($sp) # load the tweak |
| 2023 | lrvg $s1,$tweak+8($sp) |
| 2024 | lrvgr $s2,$s0 # make copy in reverse byte order |
| 2025 | lrvgr $s3,$s1 |
| 2026 | |
| 2027 | .Lxts_dec_km_2ndtweak: |
| 2028 | lghi $i1,0x87 |
| 2029 | srag $i2,$s1,63 # broadcast upper bit |
| 2030 | ngr $i1,$i2 # rem |
| 2031 | srlg $i2,$s0,63 # carry bit from lower half |
| 2032 | sllg $s0,$s0,1 |
| 2033 | sllg $s1,$s1,1 |
| 2034 | xgr $s0,$i1 |
| 2035 | ogr $s1,$i2 |
| 2036 | lrvgr $i1,$s0 # flip byte order |
| 2037 | lrvgr $i2,$s1 |
| 2038 | |
| 2039 | xg $i1,0($inp) |
| 2040 | xg $i2,8($inp) |
| 2041 | stg $i1,0($out,$inp) |
| 2042 | stg $i2,8($out,$inp) |
| 2043 | la $i2,0($out,$inp) |
| 2044 | lghi $i3,16 |
| 2045 | .long 0xb92e0066 # km $i2,$i2 |
| 2046 | brc 1,.-4 # can this happen? |
| 2047 | lrvgr $i1,$s0 |
| 2048 | lrvgr $i2,$s1 |
| 2049 | xg $i1,0($out,$inp) |
| 2050 | xg $i2,8($out,$inp) |
| 2051 | stg $i1,0($out,$inp) |
| 2052 | stg $i2,8($out,$inp) |
| 2053 | |
| 2054 | la $i3,0($out,$inp) # put aside real $out |
| 2055 | .Lxts_dec_km_steal: |
| 2056 | llgc $i1,16($inp) |
| 2057 | llgc $i2,0($out,$inp) |
| 2058 | stc $i1,0($out,$inp) |
| 2059 | stc $i2,16($out,$inp) |
| 2060 | la $inp,1($inp) |
| 2061 | brct $len,.Lxts_dec_km_steal |
| 2062 | |
| 2063 | lgr $s0,$s2 |
| 2064 | lgr $s1,$s3 |
| 2065 | xg $s0,0($i3) |
| 2066 | xg $s1,8($i3) |
| 2067 | stg $s0,0($i3) |
| 2068 | stg $s1,8($i3) |
| 2069 | la $s0,0($i3) |
| 2070 | lghi $s1,16 |
| 2071 | .long 0xb92e0088 # km $s0,$s0 |
| 2072 | brc 1,.-4 # can this happen? |
| 2073 | xg $s2,0($i3) |
| 2074 | xg $s3,8($i3) |
| 2075 | stg $s2,0($i3) |
| 2076 | stg $s3,8($i3) |
| 2077 | .Lxts_dec_km_done: |
| 2078 | l${g} $ra,14*$SIZE_T($sp) |
| 2079 | st${g} $sp,$tweak($sp) # wipe tweak |
| 2080 | st${g} $sp,$tweak($sp) |
| 2081 | lm${g} %r6,$s3,6*$SIZE_T($sp) |
| 2082 | br $ra |
| 2083 | .align 16 |
| 2084 | .Lxts_dec_software: |
| 2085 | ___ |
| 2086 | $code.=<<___; |
| 2087 | stm${g} %r6,$ra,6*$SIZE_T($sp) |
| 2088 | |
| 2089 | srlg $len,$len,4 |
| 2090 | slgr $out,$inp |
| 2091 | |
| 2092 | xgr $s0,$s0 # clear upper half |
| 2093 | xgr $s1,$s1 |
| 2094 | lrv $s0,$stdframe+4($sp) # load secno |
| 2095 | lrv $s1,$stdframe+0($sp) |
| 2096 | xgr $s2,$s2 |
| 2097 | xgr $s3,$s3 |
| 2098 | stm${g} %r2,%r5,2*$SIZE_T($sp) |
| 2099 | la $key,0($key2) |
| 2100 | larl $tbl,AES_Te |
| 2101 | bras $ra,_s390x_AES_encrypt # generate the tweak |
| 2102 | lm${g} %r2,%r5,2*$SIZE_T($sp) |
| 2103 | larl $tbl,AES_Td |
| 2104 | lt${g}r $len,$len |
| 2105 | stm $s0,$s3,$tweak($sp) # save the tweak |
| 2106 | jz .Lxts_dec_short |
| 2107 | j .Lxts_dec_enter |
| 2108 | |
| 2109 | .align 16 |
| 2110 | .Lxts_dec_loop: |
| 2111 | lrvg $s1,$tweak+0($sp) # load the tweak in little-endian |
| 2112 | lrvg $s3,$tweak+8($sp) |
| 2113 | lghi %r1,0x87 |
| 2114 | srag %r0,$s3,63 # broadcast upper bit |
| 2115 | ngr %r1,%r0 # rem |
| 2116 | srlg %r0,$s1,63 # carry bit from lower half |
| 2117 | sllg $s1,$s1,1 |
| 2118 | sllg $s3,$s3,1 |
| 2119 | xgr $s1,%r1 |
| 2120 | ogr $s3,%r0 |
| 2121 | lrvgr $s1,$s1 # flip byte order |
| 2122 | lrvgr $s3,$s3 |
| 2123 | srlg $s0,$s1,32 # smash the tweak to 4x32-bits |
| 2124 | stg $s1,$tweak+0($sp) # save the tweak |
| 2125 | llgfr $s1,$s1 |
| 2126 | srlg $s2,$s3,32 |
| 2127 | stg $s3,$tweak+8($sp) |
| 2128 | llgfr $s3,$s3 |
| 2129 | .Lxts_dec_enter: |
| 2130 | x $s0,0($inp) # tweak^=*(inp) |
| 2131 | x $s1,4($inp) |
| 2132 | x $s2,8($inp) |
| 2133 | x $s3,12($inp) |
| 2134 | stm${g} %r2,%r3,2*$SIZE_T($sp) # only two registers are changing |
| 2135 | la $key,0($key1) |
| 2136 | bras $ra,_s390x_AES_decrypt |
| 2137 | lm${g} %r2,%r5,2*$SIZE_T($sp) |
| 2138 | x $s0,$tweak+0($sp) # ^=tweak |
| 2139 | x $s1,$tweak+4($sp) |
| 2140 | x $s2,$tweak+8($sp) |
| 2141 | x $s3,$tweak+12($sp) |
| 2142 | st $s0,0($out,$inp) |
| 2143 | st $s1,4($out,$inp) |
| 2144 | st $s2,8($out,$inp) |
| 2145 | st $s3,12($out,$inp) |
| 2146 | la $inp,16($inp) |
| 2147 | brct${g} $len,.Lxts_dec_loop |
| 2148 | |
| 2149 | llgc $len,`2*$SIZE_T-1`($sp) |
| 2150 | nill $len,0x0f # $len%16 |
| 2151 | jz .Lxts_dec_done |
| 2152 | |
| 2153 | # generate pair of tweaks... |
| 2154 | lrvg $s1,$tweak+0($sp) # load the tweak in little-endian |
| 2155 | lrvg $s3,$tweak+8($sp) |
| 2156 | lghi %r1,0x87 |
| 2157 | srag %r0,$s3,63 # broadcast upper bit |
| 2158 | ngr %r1,%r0 # rem |
| 2159 | srlg %r0,$s1,63 # carry bit from lower half |
| 2160 | sllg $s1,$s1,1 |
| 2161 | sllg $s3,$s3,1 |
| 2162 | xgr $s1,%r1 |
| 2163 | ogr $s3,%r0 |
| 2164 | lrvgr $i2,$s1 # flip byte order |
| 2165 | lrvgr $i3,$s3 |
| 2166 | stmg $i2,$i3,$tweak($sp) # save the 1st tweak |
| 2167 | j .Lxts_dec_2ndtweak |
| 2168 | |
| 2169 | .align 16 |
| 2170 | .Lxts_dec_short: |
| 2171 | llgc $len,`2*$SIZE_T-1`($sp) |
| 2172 | nill $len,0x0f # $len%16 |
| 2173 | lrvg $s1,$tweak+0($sp) # load the tweak in little-endian |
| 2174 | lrvg $s3,$tweak+8($sp) |
| 2175 | .Lxts_dec_2ndtweak: |
| 2176 | lghi %r1,0x87 |
| 2177 | srag %r0,$s3,63 # broadcast upper bit |
| 2178 | ngr %r1,%r0 # rem |
| 2179 | srlg %r0,$s1,63 # carry bit from lower half |
| 2180 | sllg $s1,$s1,1 |
| 2181 | sllg $s3,$s3,1 |
| 2182 | xgr $s1,%r1 |
| 2183 | ogr $s3,%r0 |
| 2184 | lrvgr $s1,$s1 # flip byte order |
| 2185 | lrvgr $s3,$s3 |
| 2186 | srlg $s0,$s1,32 # smash the tweak to 4x32-bits |
| 2187 | stg $s1,$tweak-16+0($sp) # save the 2nd tweak |
| 2188 | llgfr $s1,$s1 |
| 2189 | srlg $s2,$s3,32 |
| 2190 | stg $s3,$tweak-16+8($sp) |
| 2191 | llgfr $s3,$s3 |
| 2192 | |
| 2193 | x $s0,0($inp) # tweak_the_2nd^=*(inp) |
| 2194 | x $s1,4($inp) |
| 2195 | x $s2,8($inp) |
| 2196 | x $s3,12($inp) |
| 2197 | stm${g} %r2,%r3,2*$SIZE_T($sp) |
| 2198 | la $key,0($key1) |
| 2199 | bras $ra,_s390x_AES_decrypt |
| 2200 | lm${g} %r2,%r5,2*$SIZE_T($sp) |
| 2201 | x $s0,$tweak-16+0($sp) # ^=tweak_the_2nd |
| 2202 | x $s1,$tweak-16+4($sp) |
| 2203 | x $s2,$tweak-16+8($sp) |
| 2204 | x $s3,$tweak-16+12($sp) |
| 2205 | st $s0,0($out,$inp) |
| 2206 | st $s1,4($out,$inp) |
| 2207 | st $s2,8($out,$inp) |
| 2208 | st $s3,12($out,$inp) |
| 2209 | |
| 2210 | la $i3,0($out,$inp) # put aside real $out |
| 2211 | .Lxts_dec_steal: |
| 2212 | llgc %r0,16($inp) |
| 2213 | llgc %r1,0($out,$inp) |
| 2214 | stc %r0,0($out,$inp) |
| 2215 | stc %r1,16($out,$inp) |
| 2216 | la $inp,1($inp) |
| 2217 | brct $len,.Lxts_dec_steal |
| 2218 | la $out,0($i3) # restore real $out |
| 2219 | |
| 2220 | lm $s0,$s3,$tweak($sp) # load the 1st tweak |
| 2221 | x $s0,0($out) # tweak^=*(inp)|stolen cipher-text |
| 2222 | x $s1,4($out) |
| 2223 | x $s2,8($out) |
| 2224 | x $s3,12($out) |
| 2225 | st${g} $out,4*$SIZE_T($sp) |
| 2226 | la $key,0($key1) |
| 2227 | bras $ra,_s390x_AES_decrypt |
| 2228 | l${g} $out,4*$SIZE_T($sp) |
| 2229 | x $s0,$tweak+0($sp) # ^=tweak |
| 2230 | x $s1,$tweak+4($sp) |
| 2231 | x $s2,$tweak+8($sp) |
| 2232 | x $s3,$tweak+12($sp) |
| 2233 | st $s0,0($out) |
| 2234 | st $s1,4($out) |
| 2235 | st $s2,8($out) |
| 2236 | st $s3,12($out) |
| 2237 | stg $sp,$tweak-16+0($sp) # wipe 2nd tweak |
| 2238 | stg $sp,$tweak-16+8($sp) |
| 2239 | .Lxts_dec_done: |
| 2240 | stg $sp,$tweak+0($sp) # wipe tweak |
| 2241 | stg $sp,$twesk+8($sp) |
| 2242 | lm${g} %r6,$ra,6*$SIZE_T($sp) |
| 2243 | br $ra |
| 2244 | .size AES_xts_decrypt,.-AES_xts_decrypt |
| 2245 | ___ |
| 2246 | } |
| 2247 | $code.=<<___; |
| 2248 | .string "AES for s390x, CRYPTOGAMS by <appro\@openssl.org>" |
| 2249 | .comm OPENSSL_s390xcap_P,16,8 |
| 2250 | ___ |
| 2251 | |
| 2252 | $code =~ s/\`([^\`]*)\`/eval $1/gem; |
| 2253 | print $code; |
| 2254 | close STDOUT; # force flush |