diff options
Diffstat (limited to 'openssl/crypto/aes/asm')
| -rw-r--r-- | openssl/crypto/aes/asm/aes-586.pl | 1533 | ||||
| -rw-r--r-- | openssl/crypto/aes/asm/aes-armv4.pl | 1030 | ||||
| -rw-r--r-- | openssl/crypto/aes/asm/aes-ia64.S | 1123 | ||||
| -rw-r--r-- | openssl/crypto/aes/asm/aes-ppc.pl | 1176 | ||||
| -rw-r--r-- | openssl/crypto/aes/asm/aes-s390x.pl | 1333 | ||||
| -rw-r--r-- | openssl/crypto/aes/asm/aes-sparcv9.pl | 1181 | ||||
| -rw-r--r-- | openssl/crypto/aes/asm/aes-x86_64.pl | 1579 |
7 files changed, 8955 insertions, 0 deletions
diff --git a/openssl/crypto/aes/asm/aes-586.pl b/openssl/crypto/aes/asm/aes-586.pl new file mode 100644 index 000000000..3bc46a968 --- /dev/null +++ b/openssl/crypto/aes/asm/aes-586.pl @@ -0,0 +1,1533 @@ +#!/usr/bin/env perl +# +# ==================================================================== +# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# project. Rights for redistribution and usage in source and binary +# forms are granted according to the OpenSSL license. +# ==================================================================== +# +# Version 3.6. +# +# You might fail to appreciate this module performance from the first +# try. If compared to "vanilla" linux-ia32-icc target, i.e. considered +# to be *the* best Intel C compiler without -KPIC, performance appears +# to be virtually identical... But try to re-configure with shared +# library support... Aha! Intel compiler "suddenly" lags behind by 30% +# [on P4, more on others]:-) And if compared to position-independent +# code generated by GNU C, this code performs *more* than *twice* as +# fast! Yes, all this buzz about PIC means that unlike other hand- +# coded implementations, this one was explicitly designed to be safe +# to use even in shared library context... This also means that this +# code isn't necessarily absolutely fastest "ever," because in order +# to achieve position independence an extra register has to be +# off-loaded to stack, which affects the benchmark result. +# +# Special note about instruction choice. Do you recall RC4_INT code +# performing poorly on P4? It might be the time to figure out why. +# RC4_INT code implies effective address calculations in base+offset*4 +# form. Trouble is that it seems that offset scaling turned to be +# critical path... At least eliminating scaling resulted in 2.8x RC4 +# performance improvement [as you might recall]. As AES code is hungry +# for scaling too, I [try to] avoid the latter by favoring off-by-2 +# shifts and masking the result with 0xFF<<2 instead of "boring" 0xFF. +# +# As was shown by Dean Gaudet <dean@arctic.org>, the above note turned +# void. Performance improvement with off-by-2 shifts was observed on +# intermediate implementation, which was spilling yet another register +# to stack... Final offset*4 code below runs just a tad faster on P4, +# but exhibits up to 10% improvement on other cores. +# +# Second version is "monolithic" replacement for aes_core.c, which in +# addition to AES_[de|en]crypt implements AES_set_[de|en]cryption_key. +# This made it possible to implement little-endian variant of the +# algorithm without modifying the base C code. Motivating factor for +# the undertaken effort was that it appeared that in tight IA-32 +# register window little-endian flavor could achieve slightly higher +# Instruction Level Parallelism, and it indeed resulted in up to 15% +# better performance on most recent µ-archs... +# +# Third version adds AES_cbc_encrypt implementation, which resulted in +# up to 40% performance imrovement of CBC benchmark results. 40% was +# observed on P4 core, where "overall" imrovement coefficient, i.e. if +# compared to PIC generated by GCC and in CBC mode, was observed to be +# as large as 4x:-) CBC performance is virtually identical to ECB now +# and on some platforms even better, e.g. 17.6 "small" cycles/byte on +# Opteron, because certain function prologues and epilogues are +# effectively taken out of the loop... +# +# Version 3.2 implements compressed tables and prefetch of these tables +# in CBC[!] mode. Former means that 3/4 of table references are now +# misaligned, which unfortunately has negative impact on elder IA-32 +# implementations, Pentium suffered 30% penalty, PIII - 10%. +# +# Version 3.3 avoids L1 cache aliasing between stack frame and +# S-boxes, and 3.4 - L1 cache aliasing even between key schedule. The +# latter is achieved by copying the key schedule to controlled place in +# stack. This unfortunately has rather strong impact on small block CBC +# performance, ~2x deterioration on 16-byte block if compared to 3.3. +# +# Version 3.5 checks if there is L1 cache aliasing between user-supplied +# key schedule and S-boxes and abstains from copying the former if +# there is no. This allows end-user to consciously retain small block +# performance by aligning key schedule in specific manner. +# +# Version 3.6 compresses Td4 to 256 bytes and prefetches it in ECB. +# +# Current ECB performance numbers for 128-bit key in CPU cycles per +# processed byte [measure commonly used by AES benchmarkers] are: +# +# small footprint fully unrolled +# P4 24 22 +# AMD K8 20 19 +# PIII 25 23 +# Pentium 81 78 + +push(@INC,"perlasm","../../perlasm"); +require "x86asm.pl"; + +&asm_init($ARGV[0],"aes-586.pl",$ARGV[$#ARGV] eq "386"); + +$s0="eax"; +$s1="ebx"; +$s2="ecx"; +$s3="edx"; +$key="edi"; +$acc="esi"; + +$compromise=0; # $compromise=128 abstains from copying key + # schedule to stack when encrypting inputs + # shorter than 128 bytes at the cost of + # risksing aliasing with S-boxes. In return + # you get way better, up to +70%, small block + # performance. +$small_footprint=1; # $small_footprint=1 code is ~5% slower [on + # recent µ-archs], but ~5 times smaller! + # I favor compact code to minimize cache + # contention and in hope to "collect" 5% back + # in real-life applications... +$vertical_spin=0; # shift "verticaly" defaults to 0, because of + # its proof-of-concept status... + +# Note that there is no decvert(), as well as last encryption round is +# performed with "horizontal" shifts. This is because this "vertical" +# implementation [one which groups shifts on a given $s[i] to form a +# "column," unlike "horizontal" one, which groups shifts on different +# $s[i] to form a "row"] is work in progress. It was observed to run +# few percents faster on Intel cores, but not AMD. On AMD K8 core it's +# whole 12% slower:-( So we face a trade-off... Shall it be resolved +# some day? Till then the code is considered experimental and by +# default remains dormant... + +sub encvert() +{ my ($te,@s) = @_; + my $v0 = $acc, $v1 = $key; + + &mov ($v0,$s[3]); # copy s3 + &mov (&DWP(4,"esp"),$s[2]); # save s2 + &mov ($v1,$s[0]); # copy s0 + &mov (&DWP(8,"esp"),$s[1]); # save s1 + + &movz ($s[2],&HB($s[0])); + &and ($s[0],0xFF); + &mov ($s[0],&DWP(0,$te,$s[0],8)); # s0>>0 + &shr ($v1,16); + &mov ($s[3],&DWP(3,$te,$s[2],8)); # s0>>8 + &movz ($s[1],&HB($v1)); + &and ($v1,0xFF); + &mov ($s[2],&DWP(2,$te,$v1,8)); # s0>>16 + &mov ($v1,$v0); + &mov ($s[1],&DWP(1,$te,$s[1],8)); # s0>>24 + + &and ($v0,0xFF); + &xor ($s[3],&DWP(0,$te,$v0,8)); # s3>>0 + &movz ($v0,&HB($v1)); + &shr ($v1,16); + &xor ($s[2],&DWP(3,$te,$v0,8)); # s3>>8 + &movz ($v0,&HB($v1)); + &and ($v1,0xFF); + &xor ($s[1],&DWP(2,$te,$v1,8)); # s3>>16 + &mov ($v1,&DWP(4,"esp")); # restore s2 + &xor ($s[0],&DWP(1,$te,$v0,8)); # s3>>24 + + &mov ($v0,$v1); + &and ($v1,0xFF); + &xor ($s[2],&DWP(0,$te,$v1,8)); # s2>>0 + &movz ($v1,&HB($v0)); + &shr ($v0,16); + &xor ($s[1],&DWP(3,$te,$v1,8)); # s2>>8 + &movz ($v1,&HB($v0)); + &and ($v0,0xFF); + &xor ($s[0],&DWP(2,$te,$v0,8)); # s2>>16 + &mov ($v0,&DWP(8,"esp")); # restore s1 + &xor ($s[3],&DWP(1,$te,$v1,8)); # s2>>24 + + &mov ($v1,$v0); + &and ($v0,0xFF); + &xor ($s[1],&DWP(0,$te,$v0,8)); # s1>>0 + &movz ($v0,&HB($v1)); + &shr ($v1,16); + &xor ($s[0],&DWP(3,$te,$v0,8)); # s1>>8 + &movz ($v0,&HB($v1)); + &and ($v1,0xFF); + &xor ($s[3],&DWP(2,$te,$v1,8)); # s1>>16 + &mov ($key,&DWP(12,"esp")); # reincarnate v1 as key + &xor ($s[2],&DWP(1,$te,$v0,8)); # s1>>24 +} + +sub encstep() +{ my ($i,$te,@s) = @_; + my $tmp = $key; + my $out = $i==3?$s[0]:$acc; + + # lines marked with #%e?x[i] denote "reordered" instructions... + if ($i==3) { &mov ($key,&DWP(12,"esp")); }##%edx + else { &mov ($out,$s[0]); + &and ($out,0xFF); } + if ($i==1) { &shr ($s[0],16); }#%ebx[1] + if ($i==2) { &shr ($s[0],24); }#%ecx[2] + &mov ($out,&DWP(0,$te,$out,8)); + + if ($i==3) { $tmp=$s[1]; }##%eax + &movz ($tmp,&HB($s[1])); + &xor ($out,&DWP(3,$te,$tmp,8)); + + if ($i==3) { $tmp=$s[2]; &mov ($s[1],&DWP(4,"esp")); }##%ebx + else { &mov ($tmp,$s[2]); + &shr ($tmp,16); } + if ($i==2) { &and ($s[1],0xFF); }#%edx[2] + &and ($tmp,0xFF); + &xor ($out,&DWP(2,$te,$tmp,8)); + + if ($i==3) { $tmp=$s[3]; &mov ($s[2],&DWP(8,"esp")); }##%ecx + elsif($i==2){ &movz ($tmp,&HB($s[3])); }#%ebx[2] + else { &mov ($tmp,$s[3]); + &shr ($tmp,24) } + &xor ($out,&DWP(1,$te,$tmp,8)); + if ($i<2) { &mov (&DWP(4+4*$i,"esp"),$out); } + if ($i==3) { &mov ($s[3],$acc); } + &comment(); +} + +sub enclast() +{ my ($i,$te,@s)=@_; + my $tmp = $key; + my $out = $i==3?$s[0]:$acc; + + if ($i==3) { &mov ($key,&DWP(12,"esp")); }##%edx + else { &mov ($out,$s[0]); } + &and ($out,0xFF); + if ($i==1) { &shr ($s[0],16); }#%ebx[1] + if ($i==2) { &shr ($s[0],24); }#%ecx[2] + &mov ($out,&DWP(2,$te,$out,8)); + &and ($out,0x000000ff); + + if ($i==3) { $tmp=$s[1]; }##%eax + &movz ($tmp,&HB($s[1])); + &mov ($tmp,&DWP(0,$te,$tmp,8)); + &and ($tmp,0x0000ff00); + &xor ($out,$tmp); + + if ($i==3) { $tmp=$s[2]; &mov ($s[1],&DWP(4,"esp")); }##%ebx + else { mov ($tmp,$s[2]); + &shr ($tmp,16); } + if ($i==2) { &and ($s[1],0xFF); }#%edx[2] + &and ($tmp,0xFF); + &mov ($tmp,&DWP(0,$te,$tmp,8)); + &and ($tmp,0x00ff0000); + &xor ($out,$tmp); + + if ($i==3) { $tmp=$s[3]; &mov ($s[2],&DWP(8,"esp")); }##%ecx + elsif($i==2){ &movz ($tmp,&HB($s[3])); }#%ebx[2] + else { &mov ($tmp,$s[3]); + &shr ($tmp,24); } + &mov ($tmp,&DWP(2,$te,$tmp,8)); + &and ($tmp,0xff000000); + &xor ($out,$tmp); + if ($i<2) { &mov (&DWP(4+4*$i,"esp"),$out); } + if ($i==3) { &mov ($s[3],$acc); } +} + +sub _data_word() { my $i; while(defined($i=shift)) { &data_word($i,$i); } } + +&public_label("AES_Te"); +&function_begin_B("_x86_AES_encrypt"); + if ($vertical_spin) { + # I need high parts of volatile registers to be accessible... + &exch ($s1="edi",$key="ebx"); + &mov ($s2="esi",$acc="ecx"); + } + + # note that caller is expected to allocate stack frame for me! + &mov (&DWP(12,"esp"),$key); # save key + + &xor ($s0,&DWP(0,$key)); # xor with key + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + + &mov ($acc,&DWP(240,$key)); # load key->rounds + + if ($small_footprint) { + &lea ($acc,&DWP(-2,$acc,$acc)); + &lea ($acc,&DWP(0,$key,$acc,8)); + &mov (&DWP(16,"esp"),$acc); # end of key schedule + &align (4); + &set_label("loop"); + if ($vertical_spin) { + &encvert("ebp",$s0,$s1,$s2,$s3); + } else { + &encstep(0,"ebp",$s0,$s1,$s2,$s3); + &encstep(1,"ebp",$s1,$s2,$s3,$s0); + &encstep(2,"ebp",$s2,$s3,$s0,$s1); + &encstep(3,"ebp",$s3,$s0,$s1,$s2); + } + &add ($key,16); # advance rd_key + &xor ($s0,&DWP(0,$key)); + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + &cmp ($key,&DWP(16,"esp")); + &mov (&DWP(12,"esp"),$key); + &jb (&label("loop")); + } + else { + &cmp ($acc,10); + &jle (&label("10rounds")); + &cmp ($acc,12); + &jle (&label("12rounds")); + + &set_label("14rounds"); + for ($i=1;$i<3;$i++) { + if ($vertical_spin) { + &encvert("ebp",$s0,$s1,$s2,$s3); + } else { + &encstep(0,"ebp",$s0,$s1,$s2,$s3); + &encstep(1,"ebp",$s1,$s2,$s3,$s0); + &encstep(2,"ebp",$s2,$s3,$s0,$s1); + &encstep(3,"ebp",$s3,$s0,$s1,$s2); + } + &xor ($s0,&DWP(16*$i+0,$key)); + &xor ($s1,&DWP(16*$i+4,$key)); + &xor ($s2,&DWP(16*$i+8,$key)); + &xor ($s3,&DWP(16*$i+12,$key)); + } + &add ($key,32); + &mov (&DWP(12,"esp"),$key); # advance rd_key + &set_label("12rounds"); + for ($i=1;$i<3;$i++) { + if ($vertical_spin) { + &encvert("ebp",$s0,$s1,$s2,$s3); + } else { + &encstep(0,"ebp",$s0,$s1,$s2,$s3); + &encstep(1,"ebp",$s1,$s2,$s3,$s0); + &encstep(2,"ebp",$s2,$s3,$s0,$s1); + &encstep(3,"ebp",$s3,$s0,$s1,$s2); + } + &xor ($s0,&DWP(16*$i+0,$key)); + &xor ($s1,&DWP(16*$i+4,$key)); + &xor ($s2,&DWP(16*$i+8,$key)); + &xor ($s3,&DWP(16*$i+12,$key)); + } + &add ($key,32); + &mov (&DWP(12,"esp"),$key); # advance rd_key + &set_label("10rounds"); + for ($i=1;$i<10;$i++) { + if ($vertical_spin) { + &encvert("ebp",$s0,$s1,$s2,$s3); + } else { + &encstep(0,"ebp",$s0,$s1,$s2,$s3); + &encstep(1,"ebp",$s1,$s2,$s3,$s0); + &encstep(2,"ebp",$s2,$s3,$s0,$s1); + &encstep(3,"ebp",$s3,$s0,$s1,$s2); + } + &xor ($s0,&DWP(16*$i+0,$key)); + &xor ($s1,&DWP(16*$i+4,$key)); + &xor ($s2,&DWP(16*$i+8,$key)); + &xor ($s3,&DWP(16*$i+12,$key)); + } + } + + if ($vertical_spin) { + # "reincarnate" some registers for "horizontal" spin... + &mov ($s1="ebx",$key="edi"); + &mov ($s2="ecx",$acc="esi"); + } + &enclast(0,"ebp",$s0,$s1,$s2,$s3); + &enclast(1,"ebp",$s1,$s2,$s3,$s0); + &enclast(2,"ebp",$s2,$s3,$s0,$s1); + &enclast(3,"ebp",$s3,$s0,$s1,$s2); + + &add ($key,$small_footprint?16:160); + &xor ($s0,&DWP(0,$key)); + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + + &ret (); + +&set_label("AES_Te",64); # Yes! I keep it in the code segment! + &_data_word(0xa56363c6, 0x847c7cf8, 0x997777ee, 0x8d7b7bf6); + &_data_word(0x0df2f2ff, 0xbd6b6bd6, 0xb16f6fde, 0x54c5c591); + &_data_word(0x50303060, 0x03010102, 0xa96767ce, 0x7d2b2b56); + &_data_word(0x19fefee7, 0x62d7d7b5, 0xe6abab4d, 0x9a7676ec); + &_data_word(0x45caca8f, 0x9d82821f, 0x40c9c989, 0x877d7dfa); + &_data_word(0x15fafaef, 0xeb5959b2, 0xc947478e, 0x0bf0f0fb); + &_data_word(0xecadad41, 0x67d4d4b3, 0xfda2a25f, 0xeaafaf45); + &_data_word(0xbf9c9c23, 0xf7a4a453, 0x967272e4, 0x5bc0c09b); + &_data_word(0xc2b7b775, 0x1cfdfde1, 0xae93933d, 0x6a26264c); + &_data_word(0x5a36366c, 0x413f3f7e, 0x02f7f7f5, 0x4fcccc83); + &_data_word(0x5c343468, 0xf4a5a551, 0x34e5e5d1, 0x08f1f1f9); + &_data_word(0x937171e2, 0x73d8d8ab, 0x53313162, 0x3f15152a); + &_data_word(0x0c040408, 0x52c7c795, 0x65232346, 0x5ec3c39d); + &_data_word(0x28181830, 0xa1969637, 0x0f05050a, 0xb59a9a2f); + &_data_word(0x0907070e, 0x36121224, 0x9b80801b, 0x3de2e2df); + &_data_word(0x26ebebcd, 0x6927274e, 0xcdb2b27f, 0x9f7575ea); + &_data_word(0x1b090912, 0x9e83831d, 0x742c2c58, 0x2e1a1a34); + &_data_word(0x2d1b1b36, 0xb26e6edc, 0xee5a5ab4, 0xfba0a05b); + &_data_word(0xf65252a4, 0x4d3b3b76, 0x61d6d6b7, 0xceb3b37d); + &_data_word(0x7b292952, 0x3ee3e3dd, 0x712f2f5e, 0x97848413); + &_data_word(0xf55353a6, 0x68d1d1b9, 0x00000000, 0x2cededc1); + &_data_word(0x60202040, 0x1ffcfce3, 0xc8b1b179, 0xed5b5bb6); + &_data_word(0xbe6a6ad4, 0x46cbcb8d, 0xd9bebe67, 0x4b393972); + &_data_word(0xde4a4a94, 0xd44c4c98, 0xe85858b0, 0x4acfcf85); + &_data_word(0x6bd0d0bb, 0x2aefefc5, 0xe5aaaa4f, 0x16fbfbed); + &_data_word(0xc5434386, 0xd74d4d9a, 0x55333366, 0x94858511); + &_data_word(0xcf45458a, 0x10f9f9e9, 0x06020204, 0x817f7ffe); + &_data_word(0xf05050a0, 0x443c3c78, 0xba9f9f25, 0xe3a8a84b); + &_data_word(0xf35151a2, 0xfea3a35d, 0xc0404080, 0x8a8f8f05); + &_data_word(0xad92923f, 0xbc9d9d21, 0x48383870, 0x04f5f5f1); + &_data_word(0xdfbcbc63, 0xc1b6b677, 0x75dadaaf, 0x63212142); + &_data_word(0x30101020, 0x1affffe5, 0x0ef3f3fd, 0x6dd2d2bf); + &_data_word(0x4ccdcd81, 0x140c0c18, 0x35131326, 0x2fececc3); + &_data_word(0xe15f5fbe, 0xa2979735, 0xcc444488, 0x3917172e); + &_data_word(0x57c4c493, 0xf2a7a755, 0x827e7efc, 0x473d3d7a); + &_data_word(0xac6464c8, 0xe75d5dba, 0x2b191932, 0x957373e6); + &_data_word(0xa06060c0, 0x98818119, 0xd14f4f9e, 0x7fdcdca3); + &_data_word(0x66222244, 0x7e2a2a54, 0xab90903b, 0x8388880b); + &_data_word(0xca46468c, 0x29eeeec7, 0xd3b8b86b, 0x3c141428); + &_data_word(0x79dedea7, 0xe25e5ebc, 0x1d0b0b16, 0x76dbdbad); + &_data_word(0x3be0e0db, 0x56323264, 0x4e3a3a74, 0x1e0a0a14); + &_data_word(0xdb494992, 0x0a06060c, 0x6c242448, 0xe45c5cb8); + &_data_word(0x5dc2c29f, 0x6ed3d3bd, 0xefacac43, 0xa66262c4); + &_data_word(0xa8919139, 0xa4959531, 0x37e4e4d3, 0x8b7979f2); + &_data_word(0x32e7e7d5, 0x43c8c88b, 0x5937376e, 0xb76d6dda); + &_data_word(0x8c8d8d01, 0x64d5d5b1, 0xd24e4e9c, 0xe0a9a949); + &_data_word(0xb46c6cd8, 0xfa5656ac, 0x07f4f4f3, 0x25eaeacf); + &_data_word(0xaf6565ca, 0x8e7a7af4, 0xe9aeae47, 0x18080810); + &_data_word(0xd5baba6f, 0x887878f0, 0x6f25254a, 0x722e2e5c); + &_data_word(0x241c1c38, 0xf1a6a657, 0xc7b4b473, 0x51c6c697); + &_data_word(0x23e8e8cb, 0x7cdddda1, 0x9c7474e8, 0x211f1f3e); + &_data_word(0xdd4b4b96, 0xdcbdbd61, 0x868b8b0d, 0x858a8a0f); + &_data_word(0x907070e0, 0x423e3e7c, 0xc4b5b571, 0xaa6666cc); + &_data_word(0xd8484890, 0x05030306, 0x01f6f6f7, 0x120e0e1c); + &_data_word(0xa36161c2, 0x5f35356a, 0xf95757ae, 0xd0b9b969); + &_data_word(0x91868617, 0x58c1c199, 0x271d1d3a, 0xb99e9e27); + &_data_word(0x38e1e1d9, 0x13f8f8eb, 0xb398982b, 0x33111122); + &_data_word(0xbb6969d2, 0x70d9d9a9, 0x898e8e07, 0xa7949433); + &_data_word(0xb69b9b2d, 0x221e1e3c, 0x92878715, 0x20e9e9c9); + &_data_word(0x49cece87, 0xff5555aa, 0x78282850, 0x7adfdfa5); + &_data_word(0x8f8c8c03, 0xf8a1a159, 0x80898909, 0x170d0d1a); + &_data_word(0xdabfbf65, 0x31e6e6d7, 0xc6424284, 0xb86868d0); + &_data_word(0xc3414182, 0xb0999929, 0x772d2d5a, 0x110f0f1e); + &_data_word(0xcbb0b07b, 0xfc5454a8, 0xd6bbbb6d, 0x3a16162c); +#rcon: + &data_word(0x00000001, 0x00000002, 0x00000004, 0x00000008); + &data_word(0x00000010, 0x00000020, 0x00000040, 0x00000080); + &data_word(0x0000001b, 0x00000036, 0, 0, 0, 0, 0, 0); +&function_end_B("_x86_AES_encrypt"); + +# void AES_encrypt (const void *inp,void *out,const AES_KEY *key); +&public_label("AES_Te"); +&function_begin("AES_encrypt"); + &mov ($acc,&wparam(0)); # load inp + &mov ($key,&wparam(2)); # load key + + &mov ($s0,"esp"); + &sub ("esp",24); + &and ("esp",-64); + &add ("esp",4); + &mov (&DWP(16,"esp"),$s0); + + &call (&label("pic_point")); # make it PIC! + &set_label("pic_point"); + &blindpop("ebp"); + &lea ("ebp",&DWP(&label("AES_Te")."-".&label("pic_point"),"ebp")); + + &mov ($s0,&DWP(0,$acc)); # load input data + &mov ($s1,&DWP(4,$acc)); + &mov ($s2,&DWP(8,$acc)); + &mov ($s3,&DWP(12,$acc)); + + &call ("_x86_AES_encrypt"); + + &mov ("esp",&DWP(16,"esp")); + + &mov ($acc,&wparam(1)); # load out + &mov (&DWP(0,$acc),$s0); # write output data + &mov (&DWP(4,$acc),$s1); + &mov (&DWP(8,$acc),$s2); + &mov (&DWP(12,$acc),$s3); +&function_end("AES_encrypt"); + +#------------------------------------------------------------------# + +sub decstep() +{ my ($i,$td,@s) = @_; + my $tmp = $key; + my $out = $i==3?$s[0]:$acc; + + # no instructions are reordered, as performance appears + # optimal... or rather that all attempts to reorder didn't + # result in better performance [which by the way is not a + # bit lower than ecryption]. + if($i==3) { &mov ($key,&DWP(12,"esp")); } + else { &mov ($out,$s[0]); } + &and ($out,0xFF); + &mov ($out,&DWP(0,$td,$out,8)); + + if ($i==3) { $tmp=$s[1]; } + &movz ($tmp,&HB($s[1])); + &xor ($out,&DWP(3,$td,$tmp,8)); + + if ($i==3) { $tmp=$s[2]; &mov ($s[1],$acc); } + else { &mov ($tmp,$s[2]); } + &shr ($tmp,16); + &and ($tmp,0xFF); + &xor ($out,&DWP(2,$td,$tmp,8)); + + if ($i==3) { $tmp=$s[3]; &mov ($s[2],&DWP(8,"esp")); } + else { &mov ($tmp,$s[3]); } + &shr ($tmp,24); + &xor ($out,&DWP(1,$td,$tmp,8)); + if ($i<2) { &mov (&DWP(4+4*$i,"esp"),$out); } + if ($i==3) { &mov ($s[3],&DWP(4,"esp")); } + &comment(); +} + +sub declast() +{ my ($i,$td,@s)=@_; + my $tmp = $key; + my $out = $i==3?$s[0]:$acc; + + if($i==3) { &mov ($key,&DWP(12,"esp")); } + else { &mov ($out,$s[0]); } + &and ($out,0xFF); + &movz ($out,&BP(2048,$td,$out,1)); + + if ($i==3) { $tmp=$s[1]; } + &movz ($tmp,&HB($s[1])); + &movz ($tmp,&BP(2048,$td,$tmp,1)); + &shl ($tmp,8); + &xor ($out,$tmp); + + if ($i==3) { $tmp=$s[2]; &mov ($s[1],$acc); } + else { mov ($tmp,$s[2]); } + &shr ($tmp,16); + &and ($tmp,0xFF); + &movz ($tmp,&BP(2048,$td,$tmp,1)); + &shl ($tmp,16); + &xor ($out,$tmp); + + if ($i==3) { $tmp=$s[3]; &mov ($s[2],&DWP(8,"esp")); } + else { &mov ($tmp,$s[3]); } + &shr ($tmp,24); + &movz ($tmp,&BP(2048,$td,$tmp,1)); + &shl ($tmp,24); + &xor ($out,$tmp); + if ($i<2) { &mov (&DWP(4+4*$i,"esp"),$out); } + if ($i==3) { &mov ($s[3],&DWP(4,"esp")); } +} + +&public_label("AES_Td"); +&function_begin_B("_x86_AES_decrypt"); + # note that caller is expected to allocate stack frame for me! + &mov (&DWP(12,"esp"),$key); # save key + + &xor ($s0,&DWP(0,$key)); # xor with key + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + + &mov ($acc,&DWP(240,$key)); # load key->rounds + + if ($small_footprint) { + &lea ($acc,&DWP(-2,$acc,$acc)); + &lea ($acc,&DWP(0,$key,$acc,8)); + &mov (&DWP(16,"esp"),$acc); # end of key schedule + &align (4); + &set_label("loop"); + &decstep(0,"ebp",$s0,$s3,$s2,$s1); + &decstep(1,"ebp",$s1,$s0,$s3,$s2); + &decstep(2,"ebp",$s2,$s1,$s0,$s3); + &decstep(3,"ebp",$s3,$s2,$s1,$s0); + &add ($key,16); # advance rd_key + &xor ($s0,&DWP(0,$key)); + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + &cmp ($key,&DWP(16,"esp")); + &mov (&DWP(12,"esp"),$key); + &jb (&label("loop")); + } + else { + &cmp ($acc,10); + &jle (&label("10rounds")); + &cmp ($acc,12); + &jle (&label("12rounds")); + + &set_label("14rounds"); + for ($i=1;$i<3;$i++) { + &decstep(0,"ebp",$s0,$s3,$s2,$s1); + &decstep(1,"ebp",$s1,$s0,$s3,$s2); + &decstep(2,"ebp",$s2,$s1,$s0,$s3); + &decstep(3,"ebp",$s3,$s2,$s1,$s0); + &xor ($s0,&DWP(16*$i+0,$key)); + &xor ($s1,&DWP(16*$i+4,$key)); + &xor ($s2,&DWP(16*$i+8,$key)); + &xor ($s3,&DWP(16*$i+12,$key)); + } + &add ($key,32); + &mov (&DWP(12,"esp"),$key); # advance rd_key + &set_label("12rounds"); + for ($i=1;$i<3;$i++) { + &decstep(0,"ebp",$s0,$s3,$s2,$s1); + &decstep(1,"ebp",$s1,$s0,$s3,$s2); + &decstep(2,"ebp",$s2,$s1,$s0,$s3); + &decstep(3,"ebp",$s3,$s2,$s1,$s0); + &xor ($s0,&DWP(16*$i+0,$key)); + &xor ($s1,&DWP(16*$i+4,$key)); + &xor ($s2,&DWP(16*$i+8,$key)); + &xor ($s3,&DWP(16*$i+12,$key)); + } + &add ($key,32); + &mov (&DWP(12,"esp"),$key); # advance rd_key + &set_label("10rounds"); + for ($i=1;$i<10;$i++) { + &decstep(0,"ebp",$s0,$s3,$s2,$s1); + &decstep(1,"ebp",$s1,$s0,$s3,$s2); + &decstep(2,"ebp",$s2,$s1,$s0,$s3); + &decstep(3,"ebp",$s3,$s2,$s1,$s0); + &xor ($s0,&DWP(16*$i+0,$key)); + &xor ($s1,&DWP(16*$i+4,$key)); + &xor ($s2,&DWP(16*$i+8,$key)); + &xor ($s3,&DWP(16*$i+12,$key)); + } + } + + &declast(0,"ebp",$s0,$s3,$s2,$s1); + &declast(1,"ebp",$s1,$s0,$s3,$s2); + &declast(2,"ebp",$s2,$s1,$s0,$s3); + &declast(3,"ebp",$s3,$s2,$s1,$s0); + + &add ($key,$small_footprint?16:160); + &xor ($s0,&DWP(0,$key)); + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + + &ret (); + +&set_label("AES_Td",64); # Yes! I keep it in the code segment! + &_data_word(0x50a7f451, 0x5365417e, 0xc3a4171a, 0x965e273a); + &_data_word(0xcb6bab3b, 0xf1459d1f, 0xab58faac, 0x9303e34b); + &_data_word(0x55fa3020, 0xf66d76ad, 0x9176cc88, 0x254c02f5); + &_data_word(0xfcd7e54f, 0xd7cb2ac5, 0x80443526, 0x8fa362b5); + &_data_word(0x495ab1de, 0x671bba25, 0x980eea45, 0xe1c0fe5d); + &_data_word(0x02752fc3, 0x12f04c81, 0xa397468d, 0xc6f9d36b); + &_data_word(0xe75f8f03, 0x959c9215, 0xeb7a6dbf, 0xda595295); + &_data_word(0x2d83bed4, 0xd3217458, 0x2969e049, 0x44c8c98e); + &_data_word(0x6a89c275, 0x78798ef4, 0x6b3e5899, 0xdd71b927); + &_data_word(0xb64fe1be, 0x17ad88f0, 0x66ac20c9, 0xb43ace7d); + &_data_word(0x184adf63, 0x82311ae5, 0x60335197, 0x457f5362); + &_data_word(0xe07764b1, 0x84ae6bbb, 0x1ca081fe, 0x942b08f9); + &_data_word(0x58684870, 0x19fd458f, 0x876cde94, 0xb7f87b52); + &_data_word(0x23d373ab, 0xe2024b72, 0x578f1fe3, 0x2aab5566); + &_data_word(0x0728ebb2, 0x03c2b52f, 0x9a7bc586, 0xa50837d3); + &_data_word(0xf2872830, 0xb2a5bf23, 0xba6a0302, 0x5c8216ed); + &_data_word(0x2b1ccf8a, 0x92b479a7, 0xf0f207f3, 0xa1e2694e); + &_data_word(0xcdf4da65, 0xd5be0506, 0x1f6234d1, 0x8afea6c4); + &_data_word(0x9d532e34, 0xa055f3a2, 0x32e18a05, 0x75ebf6a4); + &_data_word(0x39ec830b, 0xaaef6040, 0x069f715e, 0x51106ebd); + &_data_word(0xf98a213e, 0x3d06dd96, 0xae053edd, 0x46bde64d); + &_data_word(0xb58d5491, 0x055dc471, 0x6fd40604, 0xff155060); + &_data_word(0x24fb9819, 0x97e9bdd6, 0xcc434089, 0x779ed967); + &_data_word(0xbd42e8b0, 0x888b8907, 0x385b19e7, 0xdbeec879); + &_data_word(0x470a7ca1, 0xe90f427c, 0xc91e84f8, 0x00000000); + &_data_word(0x83868009, 0x48ed2b32, 0xac70111e, 0x4e725a6c); + &_data_word(0xfbff0efd, 0x5638850f, 0x1ed5ae3d, 0x27392d36); + &_data_word(0x64d90f0a, 0x21a65c68, 0xd1545b9b, 0x3a2e3624); + &_data_word(0xb1670a0c, 0x0fe75793, 0xd296eeb4, 0x9e919b1b); + &_data_word(0x4fc5c080, 0xa220dc61, 0x694b775a, 0x161a121c); + &_data_word(0x0aba93e2, 0xe52aa0c0, 0x43e0223c, 0x1d171b12); + &_data_word(0x0b0d090e, 0xadc78bf2, 0xb9a8b62d, 0xc8a91e14); + &_data_word(0x8519f157, 0x4c0775af, 0xbbdd99ee, 0xfd607fa3); + &_data_word(0x9f2601f7, 0xbcf5725c, 0xc53b6644, 0x347efb5b); + &_data_word(0x7629438b, 0xdcc623cb, 0x68fcedb6, 0x63f1e4b8); + &_data_word(0xcadc31d7, 0x10856342, 0x40229713, 0x2011c684); + &_data_word(0x7d244a85, 0xf83dbbd2, 0x1132f9ae, 0x6da129c7); + &_data_word(0x4b2f9e1d, 0xf330b2dc, 0xec52860d, 0xd0e3c177); + &_data_word(0x6c16b32b, 0x99b970a9, 0xfa489411, 0x2264e947); + &_data_word(0xc48cfca8, 0x1a3ff0a0, 0xd82c7d56, 0xef903322); + &_data_word(0xc74e4987, 0xc1d138d9, 0xfea2ca8c, 0x360bd498); + &_data_word(0xcf81f5a6, 0x28de7aa5, 0x268eb7da, 0xa4bfad3f); + &_data_word(0xe49d3a2c, 0x0d927850, 0x9bcc5f6a, 0x62467e54); + &_data_word(0xc2138df6, 0xe8b8d890, 0x5ef7392e, 0xf5afc382); + &_data_word(0xbe805d9f, 0x7c93d069, 0xa92dd56f, 0xb31225cf); + &_data_word(0x3b99acc8, 0xa77d1810, 0x6e639ce8, 0x7bbb3bdb); + &_data_word(0x097826cd, 0xf418596e, 0x01b79aec, 0xa89a4f83); + &_data_word(0x656e95e6, 0x7ee6ffaa, 0x08cfbc21, 0xe6e815ef); + &_data_word(0xd99be7ba, 0xce366f4a, 0xd4099fea, 0xd67cb029); + &_data_word(0xafb2a431, 0x31233f2a, 0x3094a5c6, 0xc066a235); + &_data_word(0x37bc4e74, 0xa6ca82fc, 0xb0d090e0, 0x15d8a733); + &_data_word(0x4a9804f1, 0xf7daec41, 0x0e50cd7f, 0x2ff69117); + &_data_word(0x8dd64d76, 0x4db0ef43, 0x544daacc, 0xdf0496e4); + &_data_word(0xe3b5d19e, 0x1b886a4c, 0xb81f2cc1, 0x7f516546); + &_data_word(0x04ea5e9d, 0x5d358c01, 0x737487fa, 0x2e410bfb); + &_data_word(0x5a1d67b3, 0x52d2db92, 0x335610e9, 0x1347d66d); + &_data_word(0x8c61d79a, 0x7a0ca137, 0x8e14f859, 0x893c13eb); + &_data_word(0xee27a9ce, 0x35c961b7, 0xede51ce1, 0x3cb1477a); + &_data_word(0x59dfd29c, 0x3f73f255, 0x79ce1418, 0xbf37c773); + &_data_word(0xeacdf753, 0x5baafd5f, 0x146f3ddf, 0x86db4478); + &_data_word(0x81f3afca, 0x3ec468b9, 0x2c342438, 0x5f40a3c2); + &_data_word(0x72c31d16, 0x0c25e2bc, 0x8b493c28, 0x41950dff); + &_data_word(0x7101a839, 0xdeb30c08, 0x9ce4b4d8, 0x90c15664); + &_data_word(0x6184cb7b, 0x70b632d5, 0x745c6c48, 0x4257b8d0); +#Td4: + &data_byte(0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38); + &data_byte(0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb); + &data_byte(0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87); + &data_byte(0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb); + &data_byte(0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d); + &data_byte(0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e); + &data_byte(0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2); + &data_byte(0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25); + &data_byte(0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16); + &data_byte(0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92); + &data_byte(0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda); + &data_byte(0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84); + &data_byte(0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a); + &data_byte(0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06); + &data_byte(0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02); + &data_byte(0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b); + &data_byte(0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea); + &data_byte(0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73); + &data_byte(0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85); + &data_byte(0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e); + &data_byte(0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89); + &data_byte(0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b); + &data_byte(0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20); + &data_byte(0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4); + &data_byte(0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31); + &data_byte(0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f); + &data_byte(0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d); + &data_byte(0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef); + &data_byte(0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0); + &data_byte(0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61); + &data_byte(0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26); + &data_byte(0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d); +&function_end_B("_x86_AES_decrypt"); + +# void AES_decrypt (const void *inp,void *out,const AES_KEY *key); +&public_label("AES_Td"); +&function_begin("AES_decrypt"); + &mov ($acc,&wparam(0)); # load inp + &mov ($key,&wparam(2)); # load key + + &mov ($s0,"esp"); + &sub ("esp",24); + &and ("esp",-64); + &add ("esp",4); + &mov (&DWP(16,"esp"),$s0); + + &call (&label("pic_point")); # make it PIC! + &set_label("pic_point"); + &blindpop("ebp"); + &lea ("ebp",&DWP(&label("AES_Td")."-".&label("pic_point"),"ebp")); + + # prefetch Td4 + &lea ("ebp",&DWP(2048+128,"ebp")); + &mov ($s0,&DWP(0-128,"ebp")); + &mov ($s1,&DWP(32-128,"ebp")); + &mov ($s2,&DWP(64-128,"ebp")); + &mov ($s3,&DWP(96-128,"ebp")); + &mov ($s0,&DWP(128-128,"ebp")); + &mov ($s1,&DWP(160-128,"ebp")); + &mov ($s2,&DWP(192-128,"ebp")); + &mov ($s3,&DWP(224-128,"ebp")); + &lea ("ebp",&DWP(-2048-128,"ebp")); + + &mov ($s0,&DWP(0,$acc)); # load input data + &mov ($s1,&DWP(4,$acc)); + &mov ($s2,&DWP(8,$acc)); + &mov ($s3,&DWP(12,$acc)); + + &call ("_x86_AES_decrypt"); + + &mov ("esp",&DWP(16,"esp")); + + &mov ($acc,&wparam(1)); # load out + &mov (&DWP(0,$acc),$s0); # write output data + &mov (&DWP(4,$acc),$s1); + &mov (&DWP(8,$acc),$s2); + &mov (&DWP(12,$acc),$s3); +&function_end("AES_decrypt"); + +# void AES_cbc_encrypt (const void char *inp, unsigned char *out, +# size_t length, const AES_KEY *key, +# unsigned char *ivp,const int enc); +{ +# stack frame layout +# -4(%esp) 0(%esp) return address +# 0(%esp) 4(%esp) tmp1 +# 4(%esp) 8(%esp) tmp2 +# 8(%esp) 12(%esp) key +# 12(%esp) 16(%esp) end of key schedule +my $_esp=&DWP(16,"esp"); #saved %esp +my $_inp=&DWP(20,"esp"); #copy of wparam(0) +my $_out=&DWP(24,"esp"); #copy of wparam(1) +my $_len=&DWP(28,"esp"); #copy of wparam(2) +my $_key=&DWP(32,"esp"); #copy of wparam(3) +my $_ivp=&DWP(36,"esp"); #copy of wparam(4) +my $_tmp=&DWP(40,"esp"); #volatile variable +my $ivec=&DWP(44,"esp"); #ivec[16] +my $aes_key=&DWP(60,"esp"); #copy of aes_key +my $mark=&DWP(60+240,"esp"); #copy of aes_key->rounds + +&public_label("AES_Te"); +&public_label("AES_Td"); +&function_begin("AES_cbc_encrypt"); + &mov ($s2 eq "ecx"? $s2 : "",&wparam(2)); # load len + &cmp ($s2,0); + &je (&label("enc_out")); + + &call (&label("pic_point")); # make it PIC! + &set_label("pic_point"); + &blindpop("ebp"); + + &pushf (); + &cld (); + + &cmp (&wparam(5),0); + &je (&label("DECRYPT")); + + &lea ("ebp",&DWP(&label("AES_Te")."-".&label("pic_point"),"ebp")); + + # allocate aligned stack frame... + &lea ($key,&DWP(-64-244,"esp")); + &and ($key,-64); + + # ... and make sure it doesn't alias with AES_Te modulo 4096 + &mov ($s0,"ebp"); + &lea ($s1,&DWP(2048,"ebp")); + &mov ($s3,$key); + &and ($s0,0xfff); # s = %ebp&0xfff + &and ($s1,0xfff); # e = (%ebp+2048)&0xfff + &and ($s3,0xfff); # p = %esp&0xfff + + &cmp ($s3,$s1); # if (p>=e) %esp =- (p-e); + &jb (&label("te_break_out")); + &sub ($s3,$s1); + &sub ($key,$s3); + &jmp (&label("te_ok")); + &set_label("te_break_out"); # else %esp -= (p-s)&0xfff + framesz; + &sub ($s3,$s0); + &and ($s3,0xfff); + &add ($s3,64+256); + &sub ($key,$s3); + &align (4); + &set_label("te_ok"); + + &mov ($s0,&wparam(0)); # load inp + &mov ($s1,&wparam(1)); # load out + &mov ($s3,&wparam(3)); # load key + &mov ($acc,&wparam(4)); # load ivp + + &exch ("esp",$key); + &add ("esp",4); # reserve for return address! + &mov ($_esp,$key); # save %esp + + &mov ($_inp,$s0); # save copy of inp + &mov ($_out,$s1); # save copy of out + &mov ($_len,$s2); # save copy of len + &mov ($_key,$s3); # save copy of key + &mov ($_ivp,$acc); # save copy of ivp + + &mov ($mark,0); # copy of aes_key->rounds = 0; + if ($compromise) { + &cmp ($s2,$compromise); + &jb (&label("skip_ecopy")); + } + # do we copy key schedule to stack? + &mov ($s1 eq "ebx" ? $s1 : "",$s3); + &mov ($s2 eq "ecx" ? $s2 : "",244/4); + &sub ($s1,"ebp"); + &mov ("esi",$s3); + &and ($s1,0xfff); + &lea ("edi",$aes_key); + &cmp ($s1,2048); + &jb (&label("do_ecopy")); + &cmp ($s1,4096-244); + &jb (&label("skip_ecopy")); + &align (4); + &set_label("do_ecopy"); + &mov ($_key,"edi"); + &data_word(0xA5F3F689); # rep movsd + &set_label("skip_ecopy"); + + &mov ($acc,$s0); + &mov ($key,16); + &align (4); + &set_label("prefetch_te"); + &mov ($s0,&DWP(0,"ebp")); + &mov ($s1,&DWP(32,"ebp")); + &mov ($s2,&DWP(64,"ebp")); + &mov ($s3,&DWP(96,"ebp")); + &lea ("ebp",&DWP(128,"ebp")); + &dec ($key); + &jnz (&label("prefetch_te")); + &sub ("ebp",2048); + + &mov ($s2,$_len); + &mov ($key,$_ivp); + &test ($s2,0xFFFFFFF0); + &jz (&label("enc_tail")); # short input... + + &mov ($s0,&DWP(0,$key)); # load iv + &mov ($s1,&DWP(4,$key)); + + &align (4); + &set_label("enc_loop"); + &mov ($s2,&DWP(8,$key)); + &mov ($s3,&DWP(12,$key)); + + &xor ($s0,&DWP(0,$acc)); # xor input data + &xor ($s1,&DWP(4,$acc)); + &xor ($s2,&DWP(8,$acc)); + &xor ($s3,&DWP(12,$acc)); + + &mov ($key,$_key); # load key + &call ("_x86_AES_encrypt"); + + &mov ($acc,$_inp); # load inp + &mov ($key,$_out); # load out + + &mov (&DWP(0,$key),$s0); # save output data + &mov (&DWP(4,$key),$s1); + &mov (&DWP(8,$key),$s2); + &mov (&DWP(12,$key),$s3); + + &mov ($s2,$_len); # load len + + &lea ($acc,&DWP(16,$acc)); + &mov ($_inp,$acc); # save inp + + &lea ($s3,&DWP(16,$key)); + &mov ($_out,$s3); # save out + + &sub ($s2,16); + &test ($s2,0xFFFFFFF0); + &mov ($_len,$s2); # save len + &jnz (&label("enc_loop")); + &test ($s2,15); + &jnz (&label("enc_tail")); + &mov ($acc,$_ivp); # load ivp + &mov ($s2,&DWP(8,$key)); # restore last dwords + &mov ($s3,&DWP(12,$key)); + &mov (&DWP(0,$acc),$s0); # save ivec + &mov (&DWP(4,$acc),$s1); + &mov (&DWP(8,$acc),$s2); + &mov (&DWP(12,$acc),$s3); + + &cmp ($mark,0); # was the key schedule copied? + &mov ("edi",$_key); + &je (&label("skip_ezero")); + # zero copy of key schedule + &mov ("ecx",240/4); + &xor ("eax","eax"); + &align (4); + &data_word(0xABF3F689); # rep stosd + &set_label("skip_ezero") + &mov ("esp",$_esp); + &popf (); + &set_label("enc_out"); + &function_end_A(); + &pushf (); # kludge, never executed + + &align (4); + &set_label("enc_tail"); + &mov ($s0,$key eq "edi" ? $key : ""); + &mov ($key,$_out); # load out + &push ($s0); # push ivp + &mov ($s1,16); + &sub ($s1,$s2); + &cmp ($key,$acc); # compare with inp + &je (&label("enc_in_place")); + &align (4); + &data_word(0xA4F3F689); # rep movsb # copy input + &jmp (&label("enc_skip_in_place")); + &set_label("enc_in_place"); + &lea ($key,&DWP(0,$key,$s2)); + &set_label("enc_skip_in_place"); + &mov ($s2,$s1); + &xor ($s0,$s0); + &align (4); + &data_word(0xAAF3F689); # rep stosb # zero tail + &pop ($key); # pop ivp + + &mov ($acc,$_out); # output as input + &mov ($s0,&DWP(0,$key)); + &mov ($s1,&DWP(4,$key)); + &mov ($_len,16); # len=16 + &jmp (&label("enc_loop")); # one more spin... + +#----------------------------- DECRYPT -----------------------------# +&align (4); +&set_label("DECRYPT"); + &lea ("ebp",&DWP(&label("AES_Td")."-".&label("pic_point"),"ebp")); + + # allocate aligned stack frame... + &lea ($key,&DWP(-64-244,"esp")); + &and ($key,-64); + + # ... and make sure it doesn't alias with AES_Td modulo 4096 + &mov ($s0,"ebp"); + &lea ($s1,&DWP(2048+256,"ebp")); + &mov ($s3,$key); + &and ($s0,0xfff); # s = %ebp&0xfff + &and ($s1,0xfff); # e = (%ebp+2048+256)&0xfff + &and ($s3,0xfff); # p = %esp&0xfff + + &cmp ($s3,$s1); # if (p>=e) %esp =- (p-e); + &jb (&label("td_break_out")); + &sub ($s3,$s1); + &sub ($key,$s3); + &jmp (&label("td_ok")); + &set_label("td_break_out"); # else %esp -= (p-s)&0xfff + framesz; + &sub ($s3,$s0); + &and ($s3,0xfff); + &add ($s3,64+256); + &sub ($key,$s3); + &align (4); + &set_label("td_ok"); + + &mov ($s0,&wparam(0)); # load inp + &mov ($s1,&wparam(1)); # load out + &mov ($s3,&wparam(3)); # load key + &mov ($acc,&wparam(4)); # load ivp + + &exch ("esp",$key); + &add ("esp",4); # reserve for return address! + &mov ($_esp,$key); # save %esp + + &mov ($_inp,$s0); # save copy of inp + &mov ($_out,$s1); # save copy of out + &mov ($_len,$s2); # save copy of len + &mov ($_key,$s3); # save copy of key + &mov ($_ivp,$acc); # save copy of ivp + + &mov ($mark,0); # copy of aes_key->rounds = 0; + if ($compromise) { + &cmp ($s2,$compromise); + &jb (&label("skip_dcopy")); + } + # do we copy key schedule to stack? + &mov ($s1 eq "ebx" ? $s1 : "",$s3); + &mov ($s2 eq "ecx" ? $s2 : "",244/4); + &sub ($s1,"ebp"); + &mov ("esi",$s3); + &and ($s1,0xfff); + &lea ("edi",$aes_key); + &cmp ($s1,2048+256); + &jb (&label("do_dcopy")); + &cmp ($s1,4096-244); + &jb (&label("skip_dcopy")); + &align (4); + &set_label("do_dcopy"); + &mov ($_key,"edi"); + &data_word(0xA5F3F689); # rep movsd + &set_label("skip_dcopy"); + + &mov ($acc,$s0); + &mov ($key,18); + &align (4); + &set_label("prefetch_td"); + &mov ($s0,&DWP(0,"ebp")); + &mov ($s1,&DWP(32,"ebp")); + &mov ($s2,&DWP(64,"ebp")); + &mov ($s3,&DWP(96,"ebp")); + &lea ("ebp",&DWP(128,"ebp")); + &dec ($key); + &jnz (&label("prefetch_td")); + &sub ("ebp",2048+256); + + &cmp ($acc,$_out); + &je (&label("dec_in_place")); # in-place processing... + + &mov ($key,$_ivp); # load ivp + &mov ($_tmp,$key); + + &align (4); + &set_label("dec_loop"); + &mov ($s0,&DWP(0,$acc)); # read input + &mov ($s1,&DWP(4,$acc)); + &mov ($s2,&DWP(8,$acc)); + &mov ($s3,&DWP(12,$acc)); + + &mov ($key,$_key); # load key + &call ("_x86_AES_decrypt"); + + &mov ($key,$_tmp); # load ivp + &mov ($acc,$_len); # load len + &xor ($s0,&DWP(0,$key)); # xor iv + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + + &sub ($acc,16); + &jc (&label("dec_partial")); + &mov ($_len,$acc); # save len + &mov ($acc,$_inp); # load inp + &mov ($key,$_out); # load out + + &mov (&DWP(0,$key),$s0); # write output + &mov (&DWP(4,$key),$s1); + &mov (&DWP(8,$key),$s2); + &mov (&DWP(12,$key),$s3); + + &mov ($_tmp,$acc); # save ivp + &lea ($acc,&DWP(16,$acc)); + &mov ($_inp,$acc); # save inp + + &lea ($key,&DWP(16,$key)); + &mov ($_out,$key); # save out + + &jnz (&label("dec_loop")); + &mov ($key,$_tmp); # load temp ivp + &set_label("dec_end"); + &mov ($acc,$_ivp); # load user ivp + &mov ($s0,&DWP(0,$key)); # load iv + &mov ($s1,&DWP(4,$key)); + &mov ($s2,&DWP(8,$key)); + &mov ($s3,&DWP(12,$key)); + &mov (&DWP(0,$acc),$s0); # copy back to user + &mov (&DWP(4,$acc),$s1); + &mov (&DWP(8,$acc),$s2); + &mov (&DWP(12,$acc),$s3); + &jmp (&label("dec_out")); + + &align (4); + &set_label("dec_partial"); + &lea ($key,$ivec); + &mov (&DWP(0,$key),$s0); # dump output to stack + &mov (&DWP(4,$key),$s1); + &mov (&DWP(8,$key),$s2); + &mov (&DWP(12,$key),$s3); + &lea ($s2 eq "ecx" ? $s2 : "",&DWP(16,$acc)); + &mov ($acc eq "esi" ? $acc : "",$key); + &mov ($key eq "edi" ? $key : "",$_out); # load out + &data_word(0xA4F3F689); # rep movsb # copy output + &mov ($key,$_inp); # use inp as temp ivp + &jmp (&label("dec_end")); + + &align (4); + &set_label("dec_in_place"); + &set_label("dec_in_place_loop"); + &lea ($key,$ivec); + &mov ($s0,&DWP(0,$acc)); # read input + &mov ($s1,&DWP(4,$acc)); + &mov ($s2,&DWP(8,$acc)); + &mov ($s3,&DWP(12,$acc)); + + &mov (&DWP(0,$key),$s0); # copy to temp + &mov (&DWP(4,$key),$s1); + &mov (&DWP(8,$key),$s2); + &mov (&DWP(12,$key),$s3); + + &mov ($key,$_key); # load key + &call ("_x86_AES_decrypt"); + + &mov ($key,$_ivp); # load ivp + &mov ($acc,$_out); # load out + &xor ($s0,&DWP(0,$key)); # xor iv + &xor ($s1,&DWP(4,$key)); + &xor ($s2,&DWP(8,$key)); + &xor ($s3,&DWP(12,$key)); + + &mov (&DWP(0,$acc),$s0); # write output + &mov (&DWP(4,$acc),$s1); + &mov (&DWP(8,$acc),$s2); + &mov (&DWP(12,$acc),$s3); + + &lea ($acc,&DWP(16,$acc)); + &mov ($_out,$acc); # save out + + &lea ($acc,$ivec); + &mov ($s0,&DWP(0,$acc)); # read temp + &mov ($s1,&DWP(4,$acc)); + &mov ($s2,&DWP(8,$acc)); + &mov ($s3,&DWP(12,$acc)); + + &mov (&DWP(0,$key),$s0); # copy iv + &mov (&DWP(4,$key),$s1); + &mov (&DWP(8,$key),$s2); + &mov (&DWP(12,$key),$s3); + + &mov ($acc,$_inp); # load inp + + &lea ($acc,&DWP(16,$acc)); + &mov ($_inp,$acc); # save inp + + &mov ($s2,$_len); # load len + &sub ($s2,16); + &jc (&label("dec_in_place_partial")); + &mov ($_len,$s2); # save len + &jnz (&label("dec_in_place_loop")); + &jmp (&label("dec_out")); + + &align (4); + &set_label("dec_in_place_partial"); + # one can argue if this is actually required... + &mov ($key eq "edi" ? $key : "",$_out); + &lea ($acc eq "esi" ? $acc : "",$ivec); + &lea ($key,&DWP(0,$key,$s2)); + &lea ($acc,&DWP(16,$acc,$s2)); + &neg ($s2 eq "ecx" ? $s2 : ""); + &data_word(0xA4F3F689); # rep movsb # restore tail + + &align (4); + &set_label("dec_out"); + &cmp ($mark,0); # was the key schedule copied? + &mov ("edi",$_key); + &je (&label("skip_dzero")); + # zero copy of key schedule + &mov ("ecx",240/4); + &xor ("eax","eax"); + &align (4); + &data_word(0xABF3F689); # rep stosd + &set_label("skip_dzero") + &mov ("esp",$_esp); + &popf (); +&function_end("AES_cbc_encrypt"); +} + +#------------------------------------------------------------------# + +sub enckey() +{ + &movz ("esi",&LB("edx")); # rk[i]>>0 + &mov ("ebx",&DWP(2,"ebp","esi",8)); + &movz ("esi",&HB("edx")); # rk[i]>>8 + &and ("ebx",0xFF000000); + &xor ("eax","ebx"); + + &mov ("ebx",&DWP(2,"ebp","esi",8)); + &shr ("edx",16); + &and ("ebx",0x000000FF); + &movz ("esi",&LB("edx")); # rk[i]>>16 + &xor ("eax","ebx"); + + &mov ("ebx",&DWP(0,"ebp","esi",8)); + &movz ("esi",&HB("edx")); # rk[i]>>24 + &and ("ebx",0x0000FF00); + &xor ("eax","ebx"); + + &mov ("ebx",&DWP(0,"ebp","esi",8)); + &and ("ebx",0x00FF0000); + &xor ("eax","ebx"); + + &xor ("eax",&DWP(2048,"ebp","ecx",4)); # rcon +} + +# int AES_set_encrypt_key(const unsigned char *userKey, const int bits, +# AES_KEY *key) +&public_label("AES_Te"); +&function_begin("AES_set_encrypt_key"); + &mov ("esi",&wparam(0)); # user supplied key + &mov ("edi",&wparam(2)); # private key schedule + + &test ("esi",-1); + &jz (&label("badpointer")); + &test ("edi",-1); + &jz (&label("badpointer")); + + &call (&label("pic_point")); + &set_label("pic_point"); + &blindpop("ebp"); + &lea ("ebp",&DWP(&label("AES_Te")."-".&label("pic_point"),"ebp")); + + &mov ("ecx",&wparam(1)); # number of bits in key + &cmp ("ecx",128); + &je (&label("10rounds")); + &cmp ("ecx",192); + &je (&label("12rounds")); + &cmp ("ecx",256); + &je (&label("14rounds")); + &mov ("eax",-2); # invalid number of bits + &jmp (&label("exit")); + + &set_label("10rounds"); + &mov ("eax",&DWP(0,"esi")); # copy first 4 dwords + &mov ("ebx",&DWP(4,"esi")); + &mov ("ecx",&DWP(8,"esi")); + &mov ("edx",&DWP(12,"esi")); + &mov (&DWP(0,"edi"),"eax"); + &mov (&DWP(4,"edi"),"ebx"); + &mov (&DWP(8,"edi"),"ecx"); + &mov (&DWP(12,"edi"),"edx"); + + &xor ("ecx","ecx"); + &jmp (&label("10shortcut")); + + &align (4); + &set_label("10loop"); + &mov ("eax",&DWP(0,"edi")); # rk[0] + &mov ("edx",&DWP(12,"edi")); # rk[3] + &set_label("10shortcut"); + &enckey (); + + &mov (&DWP(16,"edi"),"eax"); # rk[4] + &xor ("eax",&DWP(4,"edi")); + &mov (&DWP(20,"edi"),"eax"); # rk[5] + &xor ("eax",&DWP(8,"edi")); + &mov (&DWP(24,"edi"),"eax"); # rk[6] + &xor ("eax",&DWP(12,"edi")); + &mov (&DWP(28,"edi"),"eax"); # rk[7] + &inc ("ecx"); + &add ("edi",16); + &cmp ("ecx",10); + &jl (&label("10loop")); + + &mov (&DWP(80,"edi"),10); # setup number of rounds + &xor ("eax","eax"); + &jmp (&label("exit")); + + &set_label("12rounds"); + &mov ("eax",&DWP(0,"esi")); # copy first 6 dwords + &mov ("ebx",&DWP(4,"esi")); + &mov ("ecx",&DWP(8,"esi")); + &mov ("edx",&DWP(12,"esi")); + &mov (&DWP(0,"edi"),"eax"); + &mov (&DWP(4,"edi"),"ebx"); + &mov (&DWP(8,"edi"),"ecx"); + &mov (&DWP(12,"edi"),"edx"); + &mov ("ecx",&DWP(16,"esi")); + &mov ("edx",&DWP(20,"esi")); + &mov (&DWP(16,"edi"),"ecx"); + &mov (&DWP(20,"edi"),"edx"); + + &xor ("ecx","ecx"); + &jmp (&label("12shortcut")); + + &align (4); + &set_label("12loop"); + &mov ("eax",&DWP(0,"edi")); # rk[0] + &mov ("edx",&DWP(20,"edi")); # rk[5] + &set_label("12shortcut"); + &enckey (); + + &mov (&DWP(24,"edi"),"eax"); # rk[6] + &xor ("eax",&DWP(4,"edi")); + &mov (&DWP(28,"edi"),"eax"); # rk[7] + &xor ("eax",&DWP(8,"edi")); + &mov (&DWP(32,"edi"),"eax"); # rk[8] + &xor ("eax",&DWP(12,"edi")); + &mov (&DWP(36,"edi"),"eax"); # rk[9] + + &cmp ("ecx",7); + &je (&label("12break")); + &inc ("ecx"); + + &xor ("eax",&DWP(16,"edi")); + &mov (&DWP(40,"edi"),"eax"); # rk[10] + &xor ("eax",&DWP(20,"edi")); + &mov (&DWP(44,"edi"),"eax"); # rk[11] + + &add ("edi",24); + &jmp (&label("12loop")); + + &set_label("12break"); + &mov (&DWP(72,"edi"),12); # setup number of rounds + &xor ("eax","eax"); + &jmp (&label("exit")); + + &set_label("14rounds"); + &mov ("eax",&DWP(0,"esi")); # copy first 8 dwords + &mov ("ebx",&DWP(4,"esi")); + &mov ("ecx",&DWP(8,"esi")); + &mov ("edx",&DWP(12,"esi")); + &mov (&DWP(0,"edi"),"eax"); + &mov (&DWP(4,"edi"),"ebx"); + &mov (&DWP(8,"edi"),"ecx"); + &mov (&DWP(12,"edi"),"edx"); + &mov ("eax",&DWP(16,"esi")); + &mov ("ebx",&DWP(20,"esi")); + &mov ("ecx",&DWP(24,"esi")); + &mov ("edx",&DWP(28,"esi")); + &mov (&DWP(16,"edi"),"eax"); + &mov (&DWP(20,"edi"),"ebx"); + &mov (&DWP(24,"edi"),"ecx"); + &mov (&DWP(28,"edi"),"edx"); + + &xor ("ecx","ecx"); + &jmp (&label("14shortcut")); + + &align (4); + &set_label("14loop"); + &mov ("edx",&DWP(28,"edi")); # rk[7] + &set_label("14shortcut"); + &mov ("eax",&DWP(0,"edi")); # rk[0] + + &enckey (); + + &mov (&DWP(32,"edi"),"eax"); # rk[8] + &xor ("eax",&DWP(4,"edi")); + &mov (&DWP(36,"edi"),"eax"); # rk[9] + &xor ("eax",&DWP(8,"edi")); + &mov (&DWP(40,"edi"),"eax"); # rk[10] + &xor ("eax",&DWP(12,"edi")); + &mov (&DWP(44,"edi"),"eax"); # rk[11] + + &cmp ("ecx",6); + &je (&label("14break")); + &inc ("ecx"); + + &mov ("edx","eax"); + &mov ("eax",&DWP(16,"edi")); # rk[4] + &movz ("esi",&LB("edx")); # rk[11]>>0 + &mov ("ebx",&DWP(2,"ebp","esi",8)); + &movz ("esi",&HB("edx")); # rk[11]>>8 + &and ("ebx",0x000000FF); + &xor ("eax","ebx"); + + &mov ("ebx",&DWP(0,"ebp","esi",8)); + &shr ("edx",16); + &and ("ebx",0x0000FF00); + &movz ("esi",&LB("edx")); # rk[11]>>16 + &xor ("eax","ebx"); + + &mov ("ebx",&DWP(0,"ebp","esi",8)); + &movz ("esi",&HB("edx")); # rk[11]>>24 + &and ("ebx",0x00FF0000); + &xor ("eax","ebx"); + + &mov ("ebx",&DWP(2,"ebp","esi",8)); + &and ("ebx",0xFF000000); + &xor ("eax","ebx"); + + &mov (&DWP(48,"edi"),"eax"); # rk[12] + &xor ("eax",&DWP(20,"edi")); + &mov (&DWP(52,"edi"),"eax"); # rk[13] + &xor ("eax",&DWP(24,"edi")); + &mov (&DWP(56,"edi"),"eax"); # rk[14] + &xor ("eax",&DWP(28,"edi")); + &mov (&DWP(60,"edi"),"eax"); # rk[15] + + &add ("edi",32); + &jmp (&label("14loop")); + + &set_label("14break"); + &mov (&DWP(48,"edi"),14); # setup number of rounds + &xor ("eax","eax"); + &jmp (&label("exit")); + + &set_label("badpointer"); + &mov ("eax",-1); + &set_label("exit"); +&function_end("AES_set_encrypt_key"); + +sub deckey() +{ my ($i,$ptr,$te,$td) = @_; + + &mov ("eax",&DWP($i,$ptr)); + &mov ("edx","eax"); + &movz ("ebx",&HB("eax")); + &shr ("edx",16); + &and ("eax",0xFF); + &movz ("eax",&BP(2,$te,"eax",8)); + &movz ("ebx",&BP(2,$te,"ebx",8)); + &mov ("eax",&DWP(0,$td,"eax",8)); + &xor ("eax",&DWP(3,$td,"ebx",8)); + &movz ("ebx",&HB("edx")); + &and ("edx",0xFF); + &movz ("edx",&BP(2,$te,"edx",8)); + &movz ("ebx",&BP(2,$te,"ebx",8)); + &xor ("eax",&DWP(2,$td,"edx",8)); + &xor ("eax",&DWP(1,$td,"ebx",8)); + &mov (&DWP($i,$ptr),"eax"); +} + +# int AES_set_decrypt_key(const unsigned char *userKey, const int bits, +# AES_KEY *key) +&public_label("AES_Td"); +&public_label("AES_Te"); +&function_begin_B("AES_set_decrypt_key"); + &mov ("eax",&wparam(0)); + &mov ("ecx",&wparam(1)); + &mov ("edx",&wparam(2)); + &sub ("esp",12); + &mov (&DWP(0,"esp"),"eax"); + &mov (&DWP(4,"esp"),"ecx"); + &mov (&DWP(8,"esp"),"edx"); + &call ("AES_set_encrypt_key"); + &add ("esp",12); + &cmp ("eax",0); + &je (&label("proceed")); + &ret (); + + &set_label("proceed"); + &push ("ebp"); + &push ("ebx"); + &push ("esi"); + &push ("edi"); + + &mov ("esi",&wparam(2)); + &mov ("ecx",&DWP(240,"esi")); # pull number of rounds + &lea ("ecx",&DWP(0,"","ecx",4)); + &lea ("edi",&DWP(0,"esi","ecx",4)); # pointer to last chunk + + &align (4); + &set_label("invert"); # invert order of chunks + &mov ("eax",&DWP(0,"esi")); + &mov ("ebx",&DWP(4,"esi")); + &mov ("ecx",&DWP(0,"edi")); + &mov ("edx",&DWP(4,"edi")); + &mov (&DWP(0,"edi"),"eax"); + &mov (&DWP(4,"edi"),"ebx"); + &mov (&DWP(0,"esi"),"ecx"); + &mov (&DWP(4,"esi"),"edx"); + &mov ("eax",&DWP(8,"esi")); + &mov ("ebx",&DWP(12,"esi")); + &mov ("ecx",&DWP(8,"edi")); + &mov ("edx",&DWP(12,"edi")); + &mov (&DWP(8,"edi"),"eax"); + &mov (&DWP(12,"edi"),"ebx"); + &mov (&DWP(8,"esi"),"ecx"); + &mov (&DWP(12,"esi"),"edx"); + &add ("esi",16); + &sub ("edi",16); + &cmp ("esi","edi"); + &jne (&label("invert")); + + &call (&label("pic_point")); + &set_label("pic_point"); + blindpop("ebp"); + &lea ("edi",&DWP(&label("AES_Td")."-".&label("pic_point"),"ebp")); + &lea ("ebp",&DWP(&label("AES_Te")."-".&label("pic_point"),"ebp")); + + &mov ("esi",&wparam(2)); + &mov ("ecx",&DWP(240,"esi")); # pull number of rounds + &dec ("ecx"); + &align (4); + &set_label("permute"); # permute the key schedule + &add ("esi",16); + &deckey (0,"esi","ebp","edi"); + &deckey (4,"esi","ebp","edi"); + &deckey (8,"esi","ebp","edi"); + &deckey (12,"esi","ebp","edi"); + &dec ("ecx"); + &jnz (&label("permute")); + + &xor ("eax","eax"); # return success +&function_end("AES_set_decrypt_key"); + +&asm_finish(); diff --git a/openssl/crypto/aes/asm/aes-armv4.pl b/openssl/crypto/aes/asm/aes-armv4.pl new file mode 100644 index 000000000..15742c1ec --- /dev/null +++ b/openssl/crypto/aes/asm/aes-armv4.pl @@ -0,0 +1,1030 @@ +#!/usr/bin/env perl + +# ==================================================================== +# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# project. The module is, however, dual licensed under OpenSSL and +# CRYPTOGAMS licenses depending on where you obtain it. For further +# details see http://www.openssl.org/~appro/cryptogams/. +# ==================================================================== + +# AES for ARMv4 + +# January 2007. +# +# Code uses single 1K S-box and is >2 times faster than code generated +# by gcc-3.4.1. This is thanks to unique feature of ARMv4 ISA, which +# allows to merge logical or arithmetic operation with shift or rotate +# in one instruction and emit combined result every cycle. The module +# is endian-neutral. The performance is ~42 cycles/byte for 128-bit +# key. + +# May 2007. +# +# AES_set_[en|de]crypt_key is added. + +$s0="r0"; +$s1="r1"; +$s2="r2"; +$s3="r3"; +$t1="r4"; +$t2="r5"; +$t3="r6"; +$i1="r7"; +$i2="r8"; +$i3="r9"; + +$tbl="r10"; +$key="r11"; +$rounds="r12"; + +$code=<<___; +.text +.code 32 + +.type AES_Te,%object +.align 5 +AES_Te: +.word 0xc66363a5, 0xf87c7c84, 0xee777799, 0xf67b7b8d +.word 0xfff2f20d, 0xd66b6bbd, 0xde6f6fb1, 0x91c5c554 +.word 0x60303050, 0x02010103, 0xce6767a9, 0x562b2b7d +.word 0xe7fefe19, 0xb5d7d762, 0x4dababe6, 0xec76769a +.word 0x8fcaca45, 0x1f82829d, 0x89c9c940, 0xfa7d7d87 +.word 0xeffafa15, 0xb25959eb, 0x8e4747c9, 0xfbf0f00b +.word 0x41adadec, 0xb3d4d467, 0x5fa2a2fd, 0x45afafea +.word 0x239c9cbf, 0x53a4a4f7, 0xe4727296, 0x9bc0c05b +.word 0x75b7b7c2, 0xe1fdfd1c, 0x3d9393ae, 0x4c26266a +.word 0x6c36365a, 0x7e3f3f41, 0xf5f7f702, 0x83cccc4f +.word 0x6834345c, 0x51a5a5f4, 0xd1e5e534, 0xf9f1f108 +.word 0xe2717193, 0xabd8d873, 0x62313153, 0x2a15153f +.word 0x0804040c, 0x95c7c752, 0x46232365, 0x9dc3c35e +.word 0x30181828, 0x379696a1, 0x0a05050f, 0x2f9a9ab5 +.word 0x0e070709, 0x24121236, 0x1b80809b, 0xdfe2e23d +.word 0xcdebeb26, 0x4e272769, 0x7fb2b2cd, 0xea75759f +.word 0x1209091b, 0x1d83839e, 0x582c2c74, 0x341a1a2e +.word 0x361b1b2d, 0xdc6e6eb2, 0xb45a5aee, 0x5ba0a0fb +.word 0xa45252f6, 0x763b3b4d, 0xb7d6d661, 0x7db3b3ce +.word 0x5229297b, 0xdde3e33e, 0x5e2f2f71, 0x13848497 +.word 0xa65353f5, 0xb9d1d168, 0x00000000, 0xc1eded2c +.word 0x40202060, 0xe3fcfc1f, 0x79b1b1c8, 0xb65b5bed +.word 0xd46a6abe, 0x8dcbcb46, 0x67bebed9, 0x7239394b +.word 0x944a4ade, 0x984c4cd4, 0xb05858e8, 0x85cfcf4a +.word 0xbbd0d06b, 0xc5efef2a, 0x4faaaae5, 0xedfbfb16 +.word 0x864343c5, 0x9a4d4dd7, 0x66333355, 0x11858594 +.word 0x8a4545cf, 0xe9f9f910, 0x04020206, 0xfe7f7f81 +.word 0xa05050f0, 0x783c3c44, 0x259f9fba, 0x4ba8a8e3 +.word 0xa25151f3, 0x5da3a3fe, 0x804040c0, 0x058f8f8a +.word 0x3f9292ad, 0x219d9dbc, 0x70383848, 0xf1f5f504 +.word 0x63bcbcdf, 0x77b6b6c1, 0xafdada75, 0x42212163 +.word 0x20101030, 0xe5ffff1a, 0xfdf3f30e, 0xbfd2d26d +.word 0x81cdcd4c, 0x180c0c14, 0x26131335, 0xc3ecec2f +.word 0xbe5f5fe1, 0x359797a2, 0x884444cc, 0x2e171739 +.word 0x93c4c457, 0x55a7a7f2, 0xfc7e7e82, 0x7a3d3d47 +.word 0xc86464ac, 0xba5d5de7, 0x3219192b, 0xe6737395 +.word 0xc06060a0, 0x19818198, 0x9e4f4fd1, 0xa3dcdc7f +.word 0x44222266, 0x542a2a7e, 0x3b9090ab, 0x0b888883 +.word 0x8c4646ca, 0xc7eeee29, 0x6bb8b8d3, 0x2814143c +.word 0xa7dede79, 0xbc5e5ee2, 0x160b0b1d, 0xaddbdb76 +.word 0xdbe0e03b, 0x64323256, 0x743a3a4e, 0x140a0a1e +.word 0x924949db, 0x0c06060a, 0x4824246c, 0xb85c5ce4 +.word 0x9fc2c25d, 0xbdd3d36e, 0x43acacef, 0xc46262a6 +.word 0x399191a8, 0x319595a4, 0xd3e4e437, 0xf279798b +.word 0xd5e7e732, 0x8bc8c843, 0x6e373759, 0xda6d6db7 +.word 0x018d8d8c, 0xb1d5d564, 0x9c4e4ed2, 0x49a9a9e0 +.word 0xd86c6cb4, 0xac5656fa, 0xf3f4f407, 0xcfeaea25 +.word 0xca6565af, 0xf47a7a8e, 0x47aeaee9, 0x10080818 +.word 0x6fbabad5, 0xf0787888, 0x4a25256f, 0x5c2e2e72 +.word 0x381c1c24, 0x57a6a6f1, 0x73b4b4c7, 0x97c6c651 +.word 0xcbe8e823, 0xa1dddd7c, 0xe874749c, 0x3e1f1f21 +.word 0x964b4bdd, 0x61bdbddc, 0x0d8b8b86, 0x0f8a8a85 +.word 0xe0707090, 0x7c3e3e42, 0x71b5b5c4, 0xcc6666aa +.word 0x904848d8, 0x06030305, 0xf7f6f601, 0x1c0e0e12 +.word 0xc26161a3, 0x6a35355f, 0xae5757f9, 0x69b9b9d0 +.word 0x17868691, 0x99c1c158, 0x3a1d1d27, 0x279e9eb9 +.word 0xd9e1e138, 0xebf8f813, 0x2b9898b3, 0x22111133 +.word 0xd26969bb, 0xa9d9d970, 0x078e8e89, 0x339494a7 +.word 0x2d9b9bb6, 0x3c1e1e22, 0x15878792, 0xc9e9e920 +.word 0x87cece49, 0xaa5555ff, 0x50282878, 0xa5dfdf7a +.word 0x038c8c8f, 0x59a1a1f8, 0x09898980, 0x1a0d0d17 +.word 0x65bfbfda, 0xd7e6e631, 0x844242c6, 0xd06868b8 +.word 0x824141c3, 0x299999b0, 0x5a2d2d77, 0x1e0f0f11 +.word 0x7bb0b0cb, 0xa85454fc, 0x6dbbbbd6, 0x2c16163a +@ Te4[256] +.byte 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5 +.byte 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76 +.byte 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0 +.byte 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0 +.byte 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc +.byte 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15 +.byte 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a +.byte 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75 +.byte 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0 +.byte 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84 +.byte 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b +.byte 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf +.byte 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85 +.byte 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8 +.byte 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5 +.byte 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2 +.byte 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17 +.byte 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73 +.byte 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88 +.byte 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb +.byte 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c +.byte 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79 +.byte 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9 +.byte 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08 +.byte 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6 +.byte 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a +.byte 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e +.byte 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e +.byte 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94 +.byte 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf +.byte 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68 +.byte 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16 +@ rcon[] +.word 0x01000000, 0x02000000, 0x04000000, 0x08000000 +.word 0x10000000, 0x20000000, 0x40000000, 0x80000000 +.word 0x1B000000, 0x36000000, 0, 0, 0, 0, 0, 0 +.size AES_Te,.-AES_Te + +@ void AES_encrypt(const unsigned char *in, unsigned char *out, +@ const AES_KEY *key) { +.global AES_encrypt +.type AES_encrypt,%function +.align 5 +AES_encrypt: + sub r3,pc,#8 @ AES_encrypt + stmdb sp!,{r1,r4-r12,lr} + mov $rounds,r0 @ inp + mov $key,r2 + sub $tbl,r3,#AES_encrypt-AES_Te @ Te + + ldrb $s0,[$rounds,#3] @ load input data in endian-neutral + ldrb $t1,[$rounds,#2] @ manner... + ldrb $t2,[$rounds,#1] + ldrb $t3,[$rounds,#0] + orr $s0,$s0,$t1,lsl#8 + orr $s0,$s0,$t2,lsl#16 + orr $s0,$s0,$t3,lsl#24 + ldrb $s1,[$rounds,#7] + ldrb $t1,[$rounds,#6] + ldrb $t2,[$rounds,#5] + ldrb $t3,[$rounds,#4] + orr $s1,$s1,$t1,lsl#8 + orr $s1,$s1,$t2,lsl#16 + orr $s1,$s1,$t3,lsl#24 + ldrb $s2,[$rounds,#11] + ldrb $t1,[$rounds,#10] + ldrb $t2,[$rounds,#9] + ldrb $t3,[$rounds,#8] + orr $s2,$s2,$t1,lsl#8 + orr $s2,$s2,$t2,lsl#16 + orr $s2,$s2,$t3,lsl#24 + ldrb $s3,[$rounds,#15] + ldrb $t1,[$rounds,#14] + ldrb $t2,[$rounds,#13] + ldrb $t3,[$rounds,#12] + orr $s3,$s3,$t1,lsl#8 + orr $s3,$s3,$t2,lsl#16 + orr $s3,$s3,$t3,lsl#24 + + bl _armv4_AES_encrypt + + ldr $rounds,[sp],#4 @ pop out + mov $t1,$s0,lsr#24 @ write output in endian-neutral + mov $t2,$s0,lsr#16 @ manner... + mov $t3,$s0,lsr#8 + strb $t1,[$rounds,#0] + strb $t2,[$rounds,#1] + strb $t3,[$rounds,#2] + strb $s0,[$rounds,#3] + mov $t1,$s1,lsr#24 + mov $t2,$s1,lsr#16 + mov $t3,$s1,lsr#8 + strb $t1,[$rounds,#4] + strb $t2,[$rounds,#5] + strb $t3,[$rounds,#6] + strb $s1,[$rounds,#7] + mov $t1,$s2,lsr#24 + mov $t2,$s2,lsr#16 + mov $t3,$s2,lsr#8 + strb $t1,[$rounds,#8] + strb $t2,[$rounds,#9] + strb $t3,[$rounds,#10] + strb $s2,[$rounds,#11] + mov $t1,$s3,lsr#24 + mov $t2,$s3,lsr#16 + mov $t3,$s3,lsr#8 + strb $t1,[$rounds,#12] + strb $t2,[$rounds,#13] + strb $t3,[$rounds,#14] + strb $s3,[$rounds,#15] + + ldmia sp!,{r4-r12,lr} + tst lr,#1 + moveq pc,lr @ be binary compatible with V4, yet + bx lr @ interoperable with Thumb ISA:-) +.size AES_encrypt,.-AES_encrypt + +.type _armv4_AES_encrypt,%function +.align 2 +_armv4_AES_encrypt: + str lr,[sp,#-4]! @ push lr + ldr $t1,[$key],#16 + ldr $t2,[$key,#-12] + ldr $t3,[$key,#-8] + ldr $i1,[$key,#-4] + ldr $rounds,[$key,#240-16] + eor $s0,$s0,$t1 + eor $s1,$s1,$t2 + eor $s2,$s2,$t3 + eor $s3,$s3,$i1 + sub $rounds,$rounds,#1 + mov lr,#255 + +.Lenc_loop: + and $i2,lr,$s0,lsr#8 + and $i3,lr,$s0,lsr#16 + and $i1,lr,$s0 + mov $s0,$s0,lsr#24 + ldr $t1,[$tbl,$i1,lsl#2] @ Te3[s0>>0] + ldr $s0,[$tbl,$s0,lsl#2] @ Te0[s0>>24] + ldr $t2,[$tbl,$i2,lsl#2] @ Te2[s0>>8] + ldr $t3,[$tbl,$i3,lsl#2] @ Te1[s0>>16] + + and $i1,lr,$s1,lsr#16 @ i0 + and $i2,lr,$s1 + and $i3,lr,$s1,lsr#8 + mov $s1,$s1,lsr#24 + ldr $i1,[$tbl,$i1,lsl#2] @ Te1[s1>>16] + ldr $s1,[$tbl,$s1,lsl#2] @ Te0[s1>>24] + ldr $i2,[$tbl,$i2,lsl#2] @ Te3[s1>>0] + ldr $i3,[$tbl,$i3,lsl#2] @ Te2[s1>>8] + eor $s0,$s0,$i1,ror#8 + eor $s1,$s1,$t1,ror#24 + eor $t2,$t2,$i2,ror#8 + eor $t3,$t3,$i3,ror#8 + + and $i1,lr,$s2,lsr#8 @ i0 + and $i2,lr,$s2,lsr#16 @ i1 + and $i3,lr,$s2 + mov $s2,$s2,lsr#24 + ldr $i1,[$tbl,$i1,lsl#2] @ Te2[s2>>8] + ldr $i2,[$tbl,$i2,lsl#2] @ Te1[s2>>16] + ldr $s2,[$tbl,$s2,lsl#2] @ Te0[s2>>24] + ldr $i3,[$tbl,$i3,lsl#2] @ Te3[s2>>0] + eor $s0,$s0,$i1,ror#16 + eor $s1,$s1,$i2,ror#8 + eor $s2,$s2,$t2,ror#16 + eor $t3,$t3,$i3,ror#16 + + and $i1,lr,$s3 @ i0 + and $i2,lr,$s3,lsr#8 @ i1 + and $i3,lr,$s3,lsr#16 @ i2 + mov $s3,$s3,lsr#24 + ldr $i1,[$tbl,$i1,lsl#2] @ Te3[s3>>0] + ldr $i2,[$tbl,$i2,lsl#2] @ Te2[s3>>8] + ldr $i3,[$tbl,$i3,lsl#2] @ Te1[s3>>16] + ldr $s3,[$tbl,$s3,lsl#2] @ Te0[s3>>24] + eor $s0,$s0,$i1,ror#24 + eor $s1,$s1,$i2,ror#16 + eor $s2,$s2,$i3,ror#8 + eor $s3,$s3,$t3,ror#8 + + ldr $t1,[$key],#16 + ldr $t2,[$key,#-12] + ldr $t3,[$key,#-8] + ldr $i1,[$key,#-4] + eor $s0,$s0,$t1 + eor $s1,$s1,$t2 + eor $s2,$s2,$t3 + eor $s3,$s3,$i1 + + subs $rounds,$rounds,#1 + bne .Lenc_loop + + add $tbl,$tbl,#2 + + and $i1,lr,$s0 + and $i2,lr,$s0,lsr#8 + and $i3,lr,$s0,lsr#16 + mov $s0,$s0,lsr#24 + ldrb $t1,[$tbl,$i1,lsl#2] @ Te4[s0>>0] + ldrb $s0,[$tbl,$s0,lsl#2] @ Te4[s0>>24] + ldrb $t2,[$tbl,$i2,lsl#2] @ Te4[s0>>8] + ldrb $t3,[$tbl,$i3,lsl#2] @ Te4[s0>>16] + + and $i1,lr,$s1,lsr#16 @ i0 + and $i2,lr,$s1 + and $i3,lr,$s1,lsr#8 + mov $s1,$s1,lsr#24 + ldrb $i1,[$tbl,$i1,lsl#2] @ Te4[s1>>16] + ldrb $s1,[$tbl,$s1,lsl#2] @ Te4[s1>>24] + ldrb $i2,[$tbl,$i2,lsl#2] @ Te4[s1>>0] + ldrb $i3,[$tbl,$i3,lsl#2] @ Te4[s1>>8] + eor $s0,$i1,$s0,lsl#8 + eor $s1,$t1,$s1,lsl#24 + eor $t2,$i2,$t2,lsl#8 + eor $t3,$i3,$t3,lsl#8 + + and $i1,lr,$s2,lsr#8 @ i0 + and $i2,lr,$s2,lsr#16 @ i1 + and $i3,lr,$s2 + mov $s2,$s2,lsr#24 + ldrb $i1,[$tbl,$i1,lsl#2] @ Te4[s2>>8] + ldrb $i2,[$tbl,$i2,lsl#2] @ Te4[s2>>16] + ldrb $s2,[$tbl,$s2,lsl#2] @ Te4[s2>>24] + ldrb $i3,[$tbl,$i3,lsl#2] @ Te4[s2>>0] + eor $s0,$i1,$s0,lsl#8 + eor $s1,$s1,$i2,lsl#16 + eor $s2,$t2,$s2,lsl#24 + eor $t3,$i3,$t3,lsl#8 + + and $i1,lr,$s3 @ i0 + and $i2,lr,$s3,lsr#8 @ i1 + and $i3,lr,$s3,lsr#16 @ i2 + mov $s3,$s3,lsr#24 + ldrb $i1,[$tbl,$i1,lsl#2] @ Te4[s3>>0] + ldrb $i2,[$tbl,$i2,lsl#2] @ Te4[s3>>8] + ldrb $i3,[$tbl,$i3,lsl#2] @ Te4[s3>>16] + ldrb $s3,[$tbl,$s3,lsl#2] @ Te4[s3>>24] + eor $s0,$i1,$s0,lsl#8 + eor $s1,$s1,$i2,lsl#8 + eor $s2,$s2,$i3,lsl#16 + eor $s3,$t3,$s3,lsl#24 + + ldr lr,[sp],#4 @ pop lr + ldr $t1,[$key,#0] + ldr $t2,[$key,#4] + ldr $t3,[$key,#8] + ldr $i1,[$key,#12] + eor $s0,$s0,$t1 + eor $s1,$s1,$t2 + eor $s2,$s2,$t3 + eor $s3,$s3,$i1 + + sub $tbl,$tbl,#2 + mov pc,lr @ return +.size _armv4_AES_encrypt,.-_armv4_AES_encrypt + +.global AES_set_encrypt_key +.type AES_set_encrypt_key,%function +.align 5 +AES_set_encrypt_key: + sub r3,pc,#8 @ AES_set_encrypt_key + teq r0,#0 + moveq r0,#-1 + beq .Labrt + teq r2,#0 + moveq r0,#-1 + beq .Labrt + + teq r1,#128 + beq .Lok + teq r1,#192 + beq .Lok + teq r1,#256 + movne r0,#-1 + bne .Labrt + +.Lok: stmdb sp!,{r4-r12,lr} + sub $tbl,r3,#AES_set_encrypt_key-AES_Te-1024 @ Te4 + + mov $rounds,r0 @ inp + mov lr,r1 @ bits + mov $key,r2 @ key + + ldrb $s0,[$rounds,#3] @ load input data in endian-neutral + ldrb $t1,[$rounds,#2] @ manner... + ldrb $t2,[$rounds,#1] + ldrb $t3,[$rounds,#0] + orr $s0,$s0,$t1,lsl#8 + orr $s0,$s0,$t2,lsl#16 + orr $s0,$s0,$t3,lsl#24 + ldrb $s1,[$rounds,#7] + ldrb $t1,[$rounds,#6] + ldrb $t2,[$rounds,#5] + ldrb $t3,[$rounds,#4] + orr $s1,$s1,$t1,lsl#8 + orr $s1,$s1,$t2,lsl#16 + orr $s1,$s1,$t3,lsl#24 + ldrb $s2,[$rounds,#11] + ldrb $t1,[$rounds,#10] + ldrb $t2,[$rounds,#9] + ldrb $t3,[$rounds,#8] + orr $s2,$s2,$t1,lsl#8 + orr $s2,$s2,$t2,lsl#16 + orr $s2,$s2,$t3,lsl#24 + ldrb $s3,[$rounds,#15] + ldrb $t1,[$rounds,#14] + ldrb $t2,[$rounds,#13] + ldrb $t3,[$rounds,#12] + orr $s3,$s3,$t1,lsl#8 + orr $s3,$s3,$t2,lsl#16 + orr $s3,$s3,$t3,lsl#24 + str $s0,[$key],#16 + str $s1,[$key,#-12] + str $s2,[$key,#-8] + str $s3,[$key,#-4] + + teq lr,#128 + bne .Lnot128 + mov $rounds,#10 + str $rounds,[$key,#240-16] + add $t3,$tbl,#256 @ rcon + mov lr,#255 + +.L128_loop: + and $t2,lr,$s3,lsr#24 + and $i1,lr,$s3,lsr#16 + and $i2,lr,$s3,lsr#8 + and $i3,lr,$s3 + ldrb $t2,[$tbl,$t2] + ldrb $i1,[$tbl,$i1] + ldrb $i2,[$tbl,$i2] + ldrb $i3,[$tbl,$i3] + ldr $t1,[$t3],#4 @ rcon[i++] + orr $t2,$t2,$i1,lsl#24 + orr $t2,$t2,$i2,lsl#16 + orr $t2,$t2,$i3,lsl#8 + eor $t2,$t2,$t1 + eor $s0,$s0,$t2 @ rk[4]=rk[0]^... + eor $s1,$s1,$s0 @ rk[5]=rk[1]^rk[4] + eor $s2,$s2,$s1 @ rk[6]=rk[2]^rk[5] + eor $s3,$s3,$s2 @ rk[7]=rk[3]^rk[6] + str $s0,[$key],#16 + str $s1,[$key,#-12] + str $s2,[$key,#-8] + str $s3,[$key,#-4] + + subs $rounds,$rounds,#1 + bne .L128_loop + sub r2,$key,#176 + b .Ldone + +.Lnot128: + ldrb $i2,[$rounds,#19] + ldrb $t1,[$rounds,#18] + ldrb $t2,[$rounds,#17] + ldrb $t3,[$rounds,#16] + orr $i2,$i2,$t1,lsl#8 + orr $i2,$i2,$t2,lsl#16 + orr $i2,$i2,$t3,lsl#24 + ldrb $i3,[$rounds,#23] + ldrb $t1,[$rounds,#22] + ldrb $t2,[$rounds,#21] + ldrb $t3,[$rounds,#20] + orr $i3,$i3,$t1,lsl#8 + orr $i3,$i3,$t2,lsl#16 + orr $i3,$i3,$t3,lsl#24 + str $i2,[$key],#8 + str $i3,[$key,#-4] + + teq lr,#192 + bne .Lnot192 + mov $rounds,#12 + str $rounds,[$key,#240-24] + add $t3,$tbl,#256 @ rcon + mov lr,#255 + mov $rounds,#8 + +.L192_loop: + and $t2,lr,$i3,lsr#24 + and $i1,lr,$i3,lsr#16 + and $i2,lr,$i3,lsr#8 + and $i3,lr,$i3 + ldrb $t2,[$tbl,$t2] + ldrb $i1,[$tbl,$i1] + ldrb $i2,[$tbl,$i2] + ldrb $i3,[$tbl,$i3] + ldr $t1,[$t3],#4 @ rcon[i++] + orr $t2,$t2,$i1,lsl#24 + orr $t2,$t2,$i2,lsl#16 + orr $t2,$t2,$i3,lsl#8 + eor $i3,$t2,$t1 + eor $s0,$s0,$i3 @ rk[6]=rk[0]^... + eor $s1,$s1,$s0 @ rk[7]=rk[1]^rk[6] + eor $s2,$s2,$s1 @ rk[8]=rk[2]^rk[7] + eor $s3,$s3,$s2 @ rk[9]=rk[3]^rk[8] + str $s0,[$key],#24 + str $s1,[$key,#-20] + str $s2,[$key,#-16] + str $s3,[$key,#-12] + + subs $rounds,$rounds,#1 + subeq r2,$key,#216 + beq .Ldone + + ldr $i1,[$key,#-32] + ldr $i2,[$key,#-28] + eor $i1,$i1,$s3 @ rk[10]=rk[4]^rk[9] + eor $i3,$i2,$i1 @ rk[11]=rk[5]^rk[10] + str $i1,[$key,#-8] + str $i3,[$key,#-4] + b .L192_loop + +.Lnot192: + ldrb $i2,[$rounds,#27] + ldrb $t1,[$rounds,#26] + ldrb $t2,[$rounds,#25] + ldrb $t3,[$rounds,#24] + orr $i2,$i2,$t1,lsl#8 + orr $i2,$i2,$t2,lsl#16 + orr $i2,$i2,$t3,lsl#24 + ldrb $i3,[$rounds,#31] + ldrb $t1,[$rounds,#30] + ldrb $t2,[$rounds,#29] + ldrb $t3,[$rounds,#28] + orr $i3,$i3,$t1,lsl#8 + orr $i3,$i3,$t2,lsl#16 + orr $i3,$i3,$t3,lsl#24 + str $i2,[$key],#8 + str $i3,[$key,#-4] + + mov $rounds,#14 + str $rounds,[$key,#240-32] + add $t3,$tbl,#256 @ rcon + mov lr,#255 + mov $rounds,#7 + +.L256_loop: + and $t2,lr,$i3,lsr#24 + and $i1,lr,$i3,lsr#16 + and $i2,lr,$i3,lsr#8 + and $i3,lr,$i3 + ldrb $t2,[$tbl,$t2] + ldrb $i1,[$tbl,$i1] + ldrb $i2,[$tbl,$i2] + ldrb $i3,[$tbl,$i3] + ldr $t1,[$t3],#4 @ rcon[i++] + orr $t2,$t2,$i1,lsl#24 + orr $t2,$t2,$i2,lsl#16 + orr $t2,$t2,$i3,lsl#8 + eor $i3,$t2,$t1 + eor $s0,$s0,$i3 @ rk[8]=rk[0]^... + eor $s1,$s1,$s0 @ rk[9]=rk[1]^rk[8] + eor $s2,$s2,$s1 @ rk[10]=rk[2]^rk[9] + eor $s3,$s3,$s2 @ rk[11]=rk[3]^rk[10] + str $s0,[$key],#32 + str $s1,[$key,#-28] + str $s2,[$key,#-24] + str $s3,[$key,#-20] + + subs $rounds,$rounds,#1 + subeq r2,$key,#256 + beq .Ldone + + and $t2,lr,$s3 + and $i1,lr,$s3,lsr#8 + and $i2,lr,$s3,lsr#16 + and $i3,lr,$s3,lsr#24 + ldrb $t2,[$tbl,$t2] + ldrb $i1,[$tbl,$i1] + ldrb $i2,[$tbl,$i2] + ldrb $i3,[$tbl,$i3] + orr $t2,$t2,$i1,lsl#8 + orr $t2,$t2,$i2,lsl#16 + orr $t2,$t2,$i3,lsl#24 + + ldr $t1,[$key,#-48] + ldr $i1,[$key,#-44] + ldr $i2,[$key,#-40] + ldr $i3,[$key,#-36] + eor $t1,$t1,$t2 @ rk[12]=rk[4]^... + eor $i1,$i1,$t1 @ rk[13]=rk[5]^rk[12] + eor $i2,$i2,$i1 @ rk[14]=rk[6]^rk[13] + eor $i3,$i3,$i2 @ rk[15]=rk[7]^rk[14] + str $t1,[$key,#-16] + str $i1,[$key,#-12] + str $i2,[$key,#-8] + str $i3,[$key,#-4] + b .L256_loop + +.Ldone: mov r0,#0 + ldmia sp!,{r4-r12,lr} +.Labrt: tst lr,#1 + moveq pc,lr @ be binary compatible with V4, yet + bx lr @ interoperable with Thumb ISA:-) +.size AES_set_encrypt_key,.-AES_set_encrypt_key + +.global AES_set_decrypt_key +.type AES_set_decrypt_key,%function +.align 5 +AES_set_decrypt_key: + str lr,[sp,#-4]! @ push lr + bl AES_set_encrypt_key + teq r0,#0 + ldrne lr,[sp],#4 @ pop lr + bne .Labrt + + stmdb sp!,{r4-r12} + + ldr $rounds,[r2,#240] @ AES_set_encrypt_key preserves r2, + mov $key,r2 @ which is AES_KEY *key + mov $i1,r2 + add $i2,r2,$rounds,lsl#4 + +.Linv: ldr $s0,[$i1] + ldr $s1,[$i1,#4] + ldr $s2,[$i1,#8] + ldr $s3,[$i1,#12] + ldr $t1,[$i2] + ldr $t2,[$i2,#4] + ldr $t3,[$i2,#8] + ldr $i3,[$i2,#12] + str $s0,[$i2],#-16 + str $s1,[$i2,#16+4] + str $s2,[$i2,#16+8] + str $s3,[$i2,#16+12] + str $t1,[$i1],#16 + str $t2,[$i1,#-12] + str $t3,[$i1,#-8] + str $i3,[$i1,#-4] + teq $i1,$i2 + bne .Linv +___ +$mask80=$i1; +$mask1b=$i2; +$mask7f=$i3; +$code.=<<___; + ldr $s0,[$key,#16]! @ prefetch tp1 + mov $mask80,#0x80 + mov $mask1b,#0x1b + orr $mask80,$mask80,#0x8000 + orr $mask1b,$mask1b,#0x1b00 + orr $mask80,$mask80,$mask80,lsl#16 + orr $mask1b,$mask1b,$mask1b,lsl#16 + sub $rounds,$rounds,#1 + mvn $mask7f,$mask80 + mov $rounds,$rounds,lsl#2 @ (rounds-1)*4 + +.Lmix: and $t1,$s0,$mask80 + and $s1,$s0,$mask7f + sub $t1,$t1,$t1,lsr#7 + and $t1,$t1,$mask1b + eor $s1,$t1,$s1,lsl#1 @ tp2 + + and $t1,$s1,$mask80 + and $s2,$s1,$mask7f + sub $t1,$t1,$t1,lsr#7 + and $t1,$t1,$mask1b + eor $s2,$t1,$s2,lsl#1 @ tp4 + + and $t1,$s2,$mask80 + and $s3,$s2,$mask7f + sub $t1,$t1,$t1,lsr#7 + and $t1,$t1,$mask1b + eor $s3,$t1,$s3,lsl#1 @ tp8 + + eor $t1,$s1,$s2 + eor $t2,$s0,$s3 @ tp9 + eor $t1,$t1,$s3 @ tpe + eor $t1,$t1,$s1,ror#24 + eor $t1,$t1,$t2,ror#24 @ ^= ROTATE(tpb=tp9^tp2,8) + eor $t1,$t1,$s2,ror#16 + eor $t1,$t1,$t2,ror#16 @ ^= ROTATE(tpd=tp9^tp4,16) + eor $t1,$t1,$t2,ror#8 @ ^= ROTATE(tp9,24) + + ldr $s0,[$key,#4] @ prefetch tp1 + str $t1,[$key],#4 + subs $rounds,$rounds,#1 + bne .Lmix + + mov r0,#0 + ldmia sp!,{r4-r12,lr} + tst lr,#1 + moveq pc,lr @ be binary compatible with V4, yet + bx lr @ interoperable with Thumb ISA:-) +.size AES_set_decrypt_key,.-AES_set_decrypt_key + +.type AES_Td,%object +.align 5 +AES_Td: +.word 0x51f4a750, 0x7e416553, 0x1a17a4c3, 0x3a275e96 +.word 0x3bab6bcb, 0x1f9d45f1, 0xacfa58ab, 0x4be30393 +.word 0x2030fa55, 0xad766df6, 0x88cc7691, 0xf5024c25 +.word 0x4fe5d7fc, 0xc52acbd7, 0x26354480, 0xb562a38f +.word 0xdeb15a49, 0x25ba1b67, 0x45ea0e98, 0x5dfec0e1 +.word 0xc32f7502, 0x814cf012, 0x8d4697a3, 0x6bd3f9c6 +.word 0x038f5fe7, 0x15929c95, 0xbf6d7aeb, 0x955259da +.word 0xd4be832d, 0x587421d3, 0x49e06929, 0x8ec9c844 +.word 0x75c2896a, 0xf48e7978, 0x99583e6b, 0x27b971dd +.word 0xbee14fb6, 0xf088ad17, 0xc920ac66, 0x7dce3ab4 +.word 0x63df4a18, 0xe51a3182, 0x97513360, 0x62537f45 +.word 0xb16477e0, 0xbb6bae84, 0xfe81a01c, 0xf9082b94 +.word 0x70486858, 0x8f45fd19, 0x94de6c87, 0x527bf8b7 +.word 0xab73d323, 0x724b02e2, 0xe31f8f57, 0x6655ab2a +.word 0xb2eb2807, 0x2fb5c203, 0x86c57b9a, 0xd33708a5 +.word 0x302887f2, 0x23bfa5b2, 0x02036aba, 0xed16825c +.word 0x8acf1c2b, 0xa779b492, 0xf307f2f0, 0x4e69e2a1 +.word 0x65daf4cd, 0x0605bed5, 0xd134621f, 0xc4a6fe8a +.word 0x342e539d, 0xa2f355a0, 0x058ae132, 0xa4f6eb75 +.word 0x0b83ec39, 0x4060efaa, 0x5e719f06, 0xbd6e1051 +.word 0x3e218af9, 0x96dd063d, 0xdd3e05ae, 0x4de6bd46 +.word 0x91548db5, 0x71c45d05, 0x0406d46f, 0x605015ff +.word 0x1998fb24, 0xd6bde997, 0x894043cc, 0x67d99e77 +.word 0xb0e842bd, 0x07898b88, 0xe7195b38, 0x79c8eedb +.word 0xa17c0a47, 0x7c420fe9, 0xf8841ec9, 0x00000000 +.word 0x09808683, 0x322bed48, 0x1e1170ac, 0x6c5a724e +.word 0xfd0efffb, 0x0f853856, 0x3daed51e, 0x362d3927 +.word 0x0a0fd964, 0x685ca621, 0x9b5b54d1, 0x24362e3a +.word 0x0c0a67b1, 0x9357e70f, 0xb4ee96d2, 0x1b9b919e +.word 0x80c0c54f, 0x61dc20a2, 0x5a774b69, 0x1c121a16 +.word 0xe293ba0a, 0xc0a02ae5, 0x3c22e043, 0x121b171d +.word 0x0e090d0b, 0xf28bc7ad, 0x2db6a8b9, 0x141ea9c8 +.word 0x57f11985, 0xaf75074c, 0xee99ddbb, 0xa37f60fd +.word 0xf701269f, 0x5c72f5bc, 0x44663bc5, 0x5bfb7e34 +.word 0x8b432976, 0xcb23c6dc, 0xb6edfc68, 0xb8e4f163 +.word 0xd731dcca, 0x42638510, 0x13972240, 0x84c61120 +.word 0x854a247d, 0xd2bb3df8, 0xaef93211, 0xc729a16d +.word 0x1d9e2f4b, 0xdcb230f3, 0x0d8652ec, 0x77c1e3d0 +.word 0x2bb3166c, 0xa970b999, 0x119448fa, 0x47e96422 +.word 0xa8fc8cc4, 0xa0f03f1a, 0x567d2cd8, 0x223390ef +.word 0x87494ec7, 0xd938d1c1, 0x8ccaa2fe, 0x98d40b36 +.word 0xa6f581cf, 0xa57ade28, 0xdab78e26, 0x3fadbfa4 +.word 0x2c3a9de4, 0x5078920d, 0x6a5fcc9b, 0x547e4662 +.word 0xf68d13c2, 0x90d8b8e8, 0x2e39f75e, 0x82c3aff5 +.word 0x9f5d80be, 0x69d0937c, 0x6fd52da9, 0xcf2512b3 +.word 0xc8ac993b, 0x10187da7, 0xe89c636e, 0xdb3bbb7b +.word 0xcd267809, 0x6e5918f4, 0xec9ab701, 0x834f9aa8 +.word 0xe6956e65, 0xaaffe67e, 0x21bccf08, 0xef15e8e6 +.word 0xbae79bd9, 0x4a6f36ce, 0xea9f09d4, 0x29b07cd6 +.word 0x31a4b2af, 0x2a3f2331, 0xc6a59430, 0x35a266c0 +.word 0x744ebc37, 0xfc82caa6, 0xe090d0b0, 0x33a7d815 +.word 0xf104984a, 0x41ecdaf7, 0x7fcd500e, 0x1791f62f +.word 0x764dd68d, 0x43efb04d, 0xccaa4d54, 0xe49604df +.word 0x9ed1b5e3, 0x4c6a881b, 0xc12c1fb8, 0x4665517f +.word 0x9d5eea04, 0x018c355d, 0xfa877473, 0xfb0b412e +.word 0xb3671d5a, 0x92dbd252, 0xe9105633, 0x6dd64713 +.word 0x9ad7618c, 0x37a10c7a, 0x59f8148e, 0xeb133c89 +.word 0xcea927ee, 0xb761c935, 0xe11ce5ed, 0x7a47b13c +.word 0x9cd2df59, 0x55f2733f, 0x1814ce79, 0x73c737bf +.word 0x53f7cdea, 0x5ffdaa5b, 0xdf3d6f14, 0x7844db86 +.word 0xcaaff381, 0xb968c43e, 0x3824342c, 0xc2a3405f +.word 0x161dc372, 0xbce2250c, 0x283c498b, 0xff0d9541 +.word 0x39a80171, 0x080cb3de, 0xd8b4e49c, 0x6456c190 +.word 0x7bcb8461, 0xd532b670, 0x486c5c74, 0xd0b85742 +@ Td4[256] +.byte 0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38 +.byte 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb +.byte 0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87 +.byte 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb +.byte 0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d +.byte 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e +.byte 0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2 +.byte 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25 +.byte 0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16 +.byte 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92 +.byte 0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda +.byte 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84 +.byte 0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a +.byte 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06 +.byte 0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02 +.byte 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b +.byte 0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea +.byte 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73 +.byte 0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85 +.byte 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e +.byte 0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89 +.byte 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b +.byte 0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20 +.byte 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4 +.byte 0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31 +.byte 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f +.byte 0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d +.byte 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef +.byte 0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0 +.byte 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61 +.byte 0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26 +.byte 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d +.size AES_Td,.-AES_Td + +@ void AES_decrypt(const unsigned char *in, unsigned char *out, +@ const AES_KEY *key) { +.global AES_decrypt +.type AES_decrypt,%function +.align 5 +AES_decrypt: + sub r3,pc,#8 @ AES_decrypt + stmdb sp!,{r1,r4-r12,lr} + mov $rounds,r0 @ inp + mov $key,r2 + sub $tbl,r3,#AES_decrypt-AES_Td @ Td + + ldrb $s0,[$rounds,#3] @ load input data in endian-neutral + ldrb $t1,[$rounds,#2] @ manner... + ldrb $t2,[$rounds,#1] + ldrb $t3,[$rounds,#0] + orr $s0,$s0,$t1,lsl#8 + orr $s0,$s0,$t2,lsl#16 + orr $s0,$s0,$t3,lsl#24 + ldrb $s1,[$rounds,#7] + ldrb $t1,[$rounds,#6] + ldrb $t2,[$rounds,#5] + ldrb $t3,[$rounds,#4] + orr $s1,$s1,$t1,lsl#8 + orr $s1,$s1,$t2,lsl#16 + orr $s1,$s1,$t3,lsl#24 + ldrb $s2,[$rounds,#11] + ldrb $t1,[$rounds,#10] + ldrb $t2,[$rounds,#9] + ldrb $t3,[$rounds,#8] + orr $s2,$s2,$t1,lsl#8 + orr $s2,$s2,$t2,lsl#16 + orr $s2,$s2,$t3,lsl#24 + ldrb $s3,[$rounds,#15] + ldrb $t1,[$rounds,#14] + ldrb $t2,[$rounds,#13] + ldrb $t3,[$rounds,#12] + orr $s3,$s3,$t1,lsl#8 + orr $s3,$s3,$t2,lsl#16 + orr $s3,$s3,$t3,lsl#24 + + bl _armv4_AES_decrypt + + ldr $rounds,[sp],#4 @ pop out + mov $t1,$s0,lsr#24 @ write output in endian-neutral + mov $t2,$s0,lsr#16 @ manner... + mov $t3,$s0,lsr#8 + strb $t1,[$rounds,#0] + strb $t2,[$rounds,#1] + strb $t3,[$rounds,#2] + strb $s0,[$rounds,#3] + mov $t1,$s1,lsr#24 + mov $t2,$s1,lsr#16 + mov $t3,$s1,lsr#8 + strb $t1,[$rounds,#4] + strb $t2,[$rounds,#5] + strb $t3,[$rounds,#6] + strb $s1,[$rounds,#7] + mov $t1,$s2,lsr#24 + mov $t2,$s2,lsr#16 + mov $t3,$s2,lsr#8 + strb $t1,[$rounds,#8] + strb $t2,[$rounds,#9] + strb $t3,[$rounds,#10] + strb $s2,[$rounds,#11] + mov $t1,$s3,lsr#24 + mov $t2,$s3,lsr#16 + mov $t3,$s3,lsr#8 + strb $t1,[$rounds,#12] + strb $t2,[$rounds,#13] + strb $t3,[$rounds,#14] + strb $s3,[$rounds,#15] + + ldmia sp!,{r4-r12,lr} + tst lr,#1 + moveq pc,lr @ be binary compatible with V4, yet + bx lr @ interoperable with Thumb ISA:-) +.size AES_decrypt,.-AES_decrypt + +.type _armv4_AES_decrypt,%function +.align 2 +_armv4_AES_decrypt: + str lr,[sp,#-4]! @ push lr + ldr $t1,[$key],#16 + ldr $t2,[$key,#-12] + ldr $t3,[$key,#-8] + ldr $i1,[$key,#-4] + ldr $rounds,[$key,#240-16] + eor $s0,$s0,$t1 + eor $s1,$s1,$t2 + eor $s2,$s2,$t3 + eor $s3,$s3,$i1 + sub $rounds,$rounds,#1 + mov lr,#255 + +.Ldec_loop: + and $i1,lr,$s0,lsr#16 + and $i2,lr,$s0,lsr#8 + and $i3,lr,$s0 + mov $s0,$s0,lsr#24 + ldr $t1,[$tbl,$i1,lsl#2] @ Td1[s0>>16] + ldr $s0,[$tbl,$s0,lsl#2] @ Td0[s0>>24] + ldr $t2,[$tbl,$i2,lsl#2] @ Td2[s0>>8] + ldr $t3,[$tbl,$i3,lsl#2] @ Td3[s0>>0] + + and $i1,lr,$s1 @ i0 + and $i2,lr,$s1,lsr#16 + and $i3,lr,$s1,lsr#8 + mov $s1,$s1,lsr#24 + ldr $i1,[$tbl,$i1,lsl#2] @ Td3[s1>>0] + ldr $s1,[$tbl,$s1,lsl#2] @ Td0[s1>>24] + ldr $i2,[$tbl,$i2,lsl#2] @ Td1[s1>>16] + ldr $i3,[$tbl,$i3,lsl#2] @ Td2[s1>>8] + eor $s0,$s0,$i1,ror#24 + eor $s1,$s1,$t1,ror#8 + eor $t2,$i2,$t2,ror#8 + eor $t3,$i3,$t3,ror#8 + + and $i1,lr,$s2,lsr#8 @ i0 + and $i2,lr,$s2 @ i1 + and $i3,lr,$s2,lsr#16 + mov $s2,$s2,lsr#24 + ldr $i1,[$tbl,$i1,lsl#2] @ Td2[s2>>8] + ldr $i2,[$tbl,$i2,lsl#2] @ Td3[s2>>0] + ldr $s2,[$tbl,$s2,lsl#2] @ Td0[s2>>24] + ldr $i3,[$tbl,$i3,lsl#2] @ Td1[s2>>16] + eor $s0,$s0,$i1,ror#16 + eor $s1,$s1,$i2,ror#24 + eor $s2,$s2,$t2,ror#8 + eor $t3,$i3,$t3,ror#8 + + and $i1,lr,$s3,lsr#16 @ i0 + and $i2,lr,$s3,lsr#8 @ i1 + and $i3,lr,$s3 @ i2 + mov $s3,$s3,lsr#24 + ldr $i1,[$tbl,$i1,lsl#2] @ Td1[s3>>16] + ldr $i2,[$tbl,$i2,lsl#2] @ Td2[s3>>8] + ldr $i3,[$tbl,$i3,lsl#2] @ Td3[s3>>0] + ldr $s3,[$tbl,$s3,lsl#2] @ Td0[s3>>24] + eor $s0,$s0,$i1,ror#8 + eor $s1,$s1,$i2,ror#16 + eor $s2,$s2,$i3,ror#24 + eor $s3,$s3,$t3,ror#8 + + ldr $t1,[$key],#16 + ldr $t2,[$key,#-12] + ldr $t3,[$key,#-8] + ldr $i1,[$key,#-4] + eor $s0,$s0,$t1 + eor $s1,$s1,$t2 + eor $s2,$s2,$t3 + eor $s3,$s3,$i1 + + subs $rounds,$rounds,#1 + bne .Ldec_loop + + add $tbl,$tbl,#1024 + + ldr $t1,[$tbl,#0] @ prefetch Td4 + ldr $t2,[$tbl,#32] + ldr $t3,[$tbl,#64] + ldr $i1,[$tbl,#96] + ldr $i2,[$tbl,#128] + ldr $i3,[$tbl,#160] + ldr $t1,[$tbl,#192] + ldr $t2,[$tbl,#224] + + and $i1,lr,$s0,lsr#16 + and $i2,lr,$s0,lsr#8 + and $i3,lr,$s0 + ldrb $s0,[$tbl,$s0,lsr#24] @ Td4[s0>>24] + ldrb $t1,[$tbl,$i1] @ Td4[s0>>16] + ldrb $t2,[$tbl,$i2] @ Td4[s0>>8] + ldrb $t3,[$tbl,$i3] @ Td4[s0>>0] + + and $i1,lr,$s1 @ i0 + and $i2,lr,$s1,lsr#16 + and $i3,lr,$s1,lsr#8 + ldrb $i1,[$tbl,$i1] @ Td4[s1>>0] + ldrb $s1,[$tbl,$s1,lsr#24] @ Td4[s1>>24] + ldrb $i2,[$tbl,$i2] @ Td4[s1>>16] + ldrb $i3,[$tbl,$i3] @ Td4[s1>>8] + eor $s0,$i1,$s0,lsl#24 + eor $s1,$t1,$s1,lsl#8 + eor $t2,$t2,$i2,lsl#8 + eor $t3,$t3,$i3,lsl#8 + + and $i1,lr,$s2,lsr#8 @ i0 + and $i2,lr,$s2 @ i1 + and $i3,lr,$s2,lsr#16 + ldrb $i1,[$tbl,$i1] @ Td4[s2>>8] + ldrb $i2,[$tbl,$i2] @ Td4[s2>>0] + ldrb $s2,[$tbl,$s2,lsr#24] @ Td4[s2>>24] + ldrb $i3,[$tbl,$i3] @ Td4[s2>>16] + eor $s0,$s0,$i1,lsl#8 + eor $s1,$i2,$s1,lsl#16 + eor $s2,$t2,$s2,lsl#16 + eor $t3,$t3,$i3,lsl#16 + + and $i1,lr,$s3,lsr#16 @ i0 + and $i2,lr,$s3,lsr#8 @ i1 + and $i3,lr,$s3 @ i2 + ldrb $i1,[$tbl,$i1] @ Td4[s3>>16] + ldrb $i2,[$tbl,$i2] @ Td4[s3>>8] + ldrb $i3,[$tbl,$i3] @ Td4[s3>>0] + ldrb $s3,[$tbl,$s3,lsr#24] @ Td4[s3>>24] + eor $s0,$s0,$i1,lsl#16 + eor $s1,$s1,$i2,lsl#8 + eor $s2,$i3,$s2,lsl#8 + eor $s3,$t3,$s3,lsl#24 + + ldr lr,[sp],#4 @ pop lr + ldr $t1,[$key,#0] + ldr $t2,[$key,#4] + ldr $t3,[$key,#8] + ldr $i1,[$key,#12] + eor $s0,$s0,$t1 + eor $s1,$s1,$t2 + eor $s2,$s2,$t3 + eor $s3,$s3,$i1 + + sub $tbl,$tbl,#1024 + mov pc,lr @ return +.size _armv4_AES_decrypt,.-_armv4_AES_decrypt +.asciz "AES for ARMv4, CRYPTOGAMS by <appro\@openssl.org>" +___ + +$code =~ s/\bbx\s+lr\b/.word\t0xe12fff1e/gm; # make it possible to compile with -march=armv4 +print $code; diff --git a/openssl/crypto/aes/asm/aes-ia64.S b/openssl/crypto/aes/asm/aes-ia64.S new file mode 100644 index 000000000..7f6c4c366 --- /dev/null +++ b/openssl/crypto/aes/asm/aes-ia64.S @@ -0,0 +1,1123 @@ +// ==================================================================== +// Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +// project. Rights for redistribution and usage in source and binary +// forms are granted according to the OpenSSL license. +// ==================================================================== +// +// What's wrong with compiler generated code? Compiler never uses +// variable 'shr' which is pairable with 'extr'/'dep' instructions. +// Then it uses 'zxt' which is an I-type, but can be replaced with +// 'and' which in turn can be assigned to M-port [there're double as +// much M-ports as there're I-ports on Itanium 2]. By sacrificing few +// registers for small constants (255, 24 and 16) to be used with +// 'shr' and 'and' instructions I can achieve better ILP, Intruction +// Level Parallelism, and performance. This code outperforms GCC 3.3 +// generated code by over factor of 2 (two), GCC 3.4 - by 70% and +// HP C - by 40%. Measured best-case scenario, i.e. aligned +// big-endian input, ECB timing on Itanium 2 is (18 + 13*rounds) +// ticks per block, or 9.25 CPU cycles per byte for 128 bit key. + +// Version 1.2 mitigates the hazard of cache-timing attacks by +// a) compressing S-boxes from 8KB to 2KB+256B, b) scheduling +// references to S-boxes for L2 cache latency, c) prefetching T[ed]4 +// prior last round. As result performance dropped to (26 + 15*rounds) +// ticks per block or 11 cycles per byte processed with 128-bit key. +// This is ~16% deterioration. For reference Itanium 2 L1 cache has +// 64 bytes line size and L2 - 128 bytes... + +.ident "aes-ia64.S, version 1.2" +.ident "IA-64 ISA artwork by Andy Polyakov <appro@fy.chalmers.se>" +.explicit +.text + +rk0=r8; rk1=r9; + +pfssave=r2; +lcsave=r10; +prsave=r3; +maskff=r11; +twenty4=r14; +sixteen=r15; + +te00=r16; te11=r17; te22=r18; te33=r19; +te01=r20; te12=r21; te23=r22; te30=r23; +te02=r24; te13=r25; te20=r26; te31=r27; +te03=r28; te10=r29; te21=r30; te32=r31; + +// these are rotating... +t0=r32; s0=r33; +t1=r34; s1=r35; +t2=r36; s2=r37; +t3=r38; s3=r39; + +te0=r40; te1=r41; te2=r42; te3=r43; + +#if defined(_HPUX_SOURCE) && !defined(_LP64) +# define ADDP addp4 +#else +# define ADDP add +#endif + +// Offsets from Te0 +#define TE0 0 +#define TE2 2 +#if defined(_HPUX_SOURCE) || defined(B_ENDIAN) +#define TE1 3 +#define TE3 1 +#else +#define TE1 1 +#define TE3 3 +#endif + +// This implies that AES_KEY comprises 32-bit key schedule elements +// even on LP64 platforms. +#ifndef KSZ +# define KSZ 4 +# define LDKEY ld4 +#endif + +.proc _ia64_AES_encrypt# +// Input: rk0-rk1 +// te0 +// te3 as AES_KEY->rounds!!! +// s0-s3 +// maskff,twenty4,sixteen +// Output: r16,r20,r24,r28 as s0-s3 +// Clobber: r16-r31,rk0-rk1,r32-r43 +.align 32 +_ia64_AES_encrypt: + .prologue + .altrp b6 + .body +{ .mmi; alloc r16=ar.pfs,12,0,0,8 + LDKEY t0=[rk0],2*KSZ + mov pr.rot=1<<16 } +{ .mmi; LDKEY t1=[rk1],2*KSZ + add te1=TE1,te0 + add te3=-3,te3 };; +{ .mib; LDKEY t2=[rk0],2*KSZ + mov ar.ec=2 } +{ .mib; LDKEY t3=[rk1],2*KSZ + add te2=TE2,te0 + brp.loop.imp .Le_top,.Le_end-16 };; + +{ .mmi; xor s0=s0,t0 + xor s1=s1,t1 + mov ar.lc=te3 } +{ .mmi; xor s2=s2,t2 + xor s3=s3,t3 + add te3=TE3,te0 };; + +.align 32 +.Le_top: +{ .mmi; (p0) LDKEY t0=[rk0],2*KSZ // 0/0:rk[0] + (p0) and te33=s3,maskff // 0/0:s3&0xff + (p0) extr.u te22=s2,8,8 } // 0/0:s2>>8&0xff +{ .mmi; (p0) LDKEY t1=[rk1],2*KSZ // 0/1:rk[1] + (p0) and te30=s0,maskff // 0/1:s0&0xff + (p0) shr.u te00=s0,twenty4 };; // 0/0:s0>>24 +{ .mmi; (p0) LDKEY t2=[rk0],2*KSZ // 1/2:rk[2] + (p0) shladd te33=te33,3,te3 // 1/0:te0+s0>>24 + (p0) extr.u te23=s3,8,8 } // 1/1:s3>>8&0xff +{ .mmi; (p0) LDKEY t3=[rk1],2*KSZ // 1/3:rk[3] + (p0) shladd te30=te30,3,te3 // 1/1:te3+s0 + (p0) shr.u te01=s1,twenty4 };; // 1/1:s1>>24 +{ .mmi; (p0) ld4 te33=[te33] // 2/0:te3[s3&0xff] + (p0) shladd te22=te22,3,te2 // 2/0:te2+s2>>8&0xff + (p0) extr.u te20=s0,8,8 } // 2/2:s0>>8&0xff +{ .mmi; (p0) ld4 te30=[te30] // 2/1:te3[s0] + (p0) shladd te23=te23,3,te2 // 2/1:te2+s3>>8 + (p0) shr.u te02=s2,twenty4 };; // 2/2:s2>>24 +{ .mmi; (p0) ld4 te22=[te22] // 3/0:te2[s2>>8] + (p0) shladd te20=te20,3,te2 // 3/2:te2+s0>>8 + (p0) extr.u te21=s1,8,8 } // 3/3:s1>>8&0xff +{ .mmi; (p0) ld4 te23=[te23] // 3/1:te2[s3>>8] + (p0) shladd te00=te00,3,te0 // 3/0:te0+s0>>24 + (p0) shr.u te03=s3,twenty4 };; // 3/3:s3>>24 +{ .mmi; (p0) ld4 te20=[te20] // 4/2:te2[s0>>8] + (p0) shladd te21=te21,3,te2 // 4/3:te3+s2 + (p0) extr.u te11=s1,16,8 } // 4/0:s1>>16&0xff +{ .mmi; (p0) ld4 te00=[te00] // 4/0:te0[s0>>24] + (p0) shladd te01=te01,3,te0 // 4/1:te0+s1>>24 + (p0) shr.u te13=s3,sixteen };; // 4/2:s3>>16 +{ .mmi; (p0) ld4 te21=[te21] // 5/3:te2[s1>>8] + (p0) shladd te11=te11,3,te1 // 5/0:te1+s1>>16 + (p0) extr.u te12=s2,16,8 } // 5/1:s2>>16&0xff +{ .mmi; (p0) ld4 te01=[te01] // 5/1:te0[s1>>24] + (p0) shladd te02=te02,3,te0 // 5/2:te0+s2>>24 + (p0) and te31=s1,maskff };; // 5/2:s1&0xff +{ .mmi; (p0) ld4 te11=[te11] // 6/0:te1[s1>>16] + (p0) shladd te12=te12,3,te1 // 6/1:te1+s2>>16 + (p0) extr.u te10=s0,16,8 } // 6/3:s0>>16&0xff +{ .mmi; (p0) ld4 te02=[te02] // 6/2:te0[s2>>24] + (p0) shladd te03=te03,3,te0 // 6/3:te1+s0>>16 + (p0) and te32=s2,maskff };; // 6/3:s2&0xff + +{ .mmi; (p0) ld4 te12=[te12] // 7/1:te1[s2>>16] + (p0) shladd te31=te31,3,te3 // 7/2:te3+s1&0xff + (p0) and te13=te13,maskff} // 7/2:s3>>16&0xff +{ .mmi; (p0) ld4 te03=[te03] // 7/3:te0[s3>>24] + (p0) shladd te32=te32,3,te3 // 7/3:te3+s2 + (p0) xor t0=t0,te33 };; // 7/0: +{ .mmi; (p0) ld4 te31=[te31] // 8/2:te3[s1] + (p0) shladd te13=te13,3,te1 // 8/2:te1+s3>>16 + (p0) xor t0=t0,te22 } // 8/0: +{ .mmi; (p0) ld4 te32=[te32] // 8/3:te3[s2] + (p0) shladd te10=te10,3,te1 // 8/3:te1+s0>>16 + (p0) xor t1=t1,te30 };; // 8/1: +{ .mmi; (p0) ld4 te13=[te13] // 9/2:te1[s3>>16] + (p0) ld4 te10=[te10] // 9/3:te1[s0>>16] + (p0) xor t0=t0,te00 };; // 9/0: !L2 scheduling +{ .mmi; (p0) xor t1=t1,te23 // 10[9]/1: + (p0) xor t2=t2,te20 // 10[9]/2: + (p0) xor t3=t3,te21 };; // 10[9]/3: +{ .mmi; (p0) xor t0=t0,te11 // 11[10]/0:done! + (p0) xor t1=t1,te01 // 11[10]/1: + (p0) xor t2=t2,te02 };; // 11[10]/2: !L2 scheduling +{ .mmi; (p0) xor t3=t3,te03 // 12[10]/3: + (p16) cmp.eq p0,p17=r0,r0 };; // 12[10]/clear (p17) +{ .mmi; (p0) xor t1=t1,te12 // 13[11]/1:done! + (p0) xor t2=t2,te31 // 13[11]/2: + (p0) xor t3=t3,te32 } // 13[11]/3: +{ .mmi; (p17) add te0=2048,te0 // 13[11]/ + (p17) add te1=2048+64-TE1,te1};; // 13[11]/ +{ .mib; (p0) xor t2=t2,te13 // 14[12]/2:done! + (p17) add te2=2048+128-TE2,te2} // 14[12]/ +{ .mib; (p0) xor t3=t3,te10 // 14[12]/3:done! + (p17) add te3=2048+192-TE3,te3 // 14[12]/ + br.ctop.sptk .Le_top };; +.Le_end: + + +{ .mmi; ld8 te12=[te0] // prefetch Te4 + ld8 te31=[te1] } +{ .mmi; ld8 te10=[te2] + ld8 te32=[te3] } + +{ .mmi; LDKEY t0=[rk0],2*KSZ // 0/0:rk[0] + and te33=s3,maskff // 0/0:s3&0xff + extr.u te22=s2,8,8 } // 0/0:s2>>8&0xff +{ .mmi; LDKEY t1=[rk1],2*KSZ // 0/1:rk[1] + and te30=s0,maskff // 0/1:s0&0xff + shr.u te00=s0,twenty4 };; // 0/0:s0>>24 +{ .mmi; LDKEY t2=[rk0],2*KSZ // 1/2:rk[2] + add te33=te33,te0 // 1/0:te0+s0>>24 + extr.u te23=s3,8,8 } // 1/1:s3>>8&0xff +{ .mmi; LDKEY t3=[rk1],2*KSZ // 1/3:rk[3] + add te30=te30,te0 // 1/1:te0+s0 + shr.u te01=s1,twenty4 };; // 1/1:s1>>24 +{ .mmi; ld1 te33=[te33] // 2/0:te0[s3&0xff] + add te22=te22,te0 // 2/0:te0+s2>>8&0xff + extr.u te20=s0,8,8 } // 2/2:s0>>8&0xff +{ .mmi; ld1 te30=[te30] // 2/1:te0[s0] + add te23=te23,te0 // 2/1:te0+s3>>8 + shr.u te02=s2,twenty4 };; // 2/2:s2>>24 +{ .mmi; ld1 te22=[te22] // 3/0:te0[s2>>8] + add te20=te20,te0 // 3/2:te0+s0>>8 + extr.u te21=s1,8,8 } // 3/3:s1>>8&0xff +{ .mmi; ld1 te23=[te23] // 3/1:te0[s3>>8] + add te00=te00,te0 // 3/0:te0+s0>>24 + shr.u te03=s3,twenty4 };; // 3/3:s3>>24 +{ .mmi; ld1 te20=[te20] // 4/2:te0[s0>>8] + add te21=te21,te0 // 4/3:te0+s2 + extr.u te11=s1,16,8 } // 4/0:s1>>16&0xff +{ .mmi; ld1 te00=[te00] // 4/0:te0[s0>>24] + add te01=te01,te0 // 4/1:te0+s1>>24 + shr.u te13=s3,sixteen };; // 4/2:s3>>16 +{ .mmi; ld1 te21=[te21] // 5/3:te0[s1>>8] + add te11=te11,te0 // 5/0:te0+s1>>16 + extr.u te12=s2,16,8 } // 5/1:s2>>16&0xff +{ .mmi; ld1 te01=[te01] // 5/1:te0[s1>>24] + add te02=te02,te0 // 5/2:te0+s2>>24 + and te31=s1,maskff };; // 5/2:s1&0xff +{ .mmi; ld1 te11=[te11] // 6/0:te0[s1>>16] + add te12=te12,te0 // 6/1:te0+s2>>16 + extr.u te10=s0,16,8 } // 6/3:s0>>16&0xff +{ .mmi; ld1 te02=[te02] // 6/2:te0[s2>>24] + add te03=te03,te0 // 6/3:te0+s0>>16 + and te32=s2,maskff };; // 6/3:s2&0xff + +{ .mmi; ld1 te12=[te12] // 7/1:te0[s2>>16] + add te31=te31,te0 // 7/2:te0+s1&0xff + dep te33=te22,te33,8,8} // 7/0: +{ .mmi; ld1 te03=[te03] // 7/3:te0[s3>>24] + add te32=te32,te0 // 7/3:te0+s2 + and te13=te13,maskff};; // 7/2:s3>>16&0xff +{ .mmi; ld1 te31=[te31] // 8/2:te0[s1] + add te13=te13,te0 // 8/2:te0+s3>>16 + dep te30=te23,te30,8,8} // 8/1: +{ .mmi; ld1 te32=[te32] // 8/3:te0[s2] + add te10=te10,te0 // 8/3:te0+s0>>16 + shl te00=te00,twenty4};; // 8/0: +{ .mii; ld1 te13=[te13] // 9/2:te0[s3>>16] + dep te33=te11,te33,16,8 // 9/0: + shl te01=te01,twenty4};; // 9/1: +{ .mii; ld1 te10=[te10] // 10/3:te0[s0>>16] + dep te31=te20,te31,8,8 // 10/2: + shl te02=te02,twenty4};; // 10/2: +{ .mii; xor t0=t0,te33 // 11/0: + dep te32=te21,te32,8,8 // 11/3: + shl te12=te12,sixteen};; // 11/1: +{ .mii; xor r16=t0,te00 // 12/0:done! + dep te31=te13,te31,16,8 // 12/2: + shl te03=te03,twenty4};; // 12/3: +{ .mmi; xor t1=t1,te01 // 13/1: + xor t2=t2,te02 // 13/2: + dep te32=te10,te32,16,8};; // 13/3: +{ .mmi; xor t1=t1,te30 // 14/1: + xor r24=t2,te31 // 14/2:done! + xor t3=t3,te32 };; // 14/3: +{ .mib; xor r20=t1,te12 // 15/1:done! + xor r28=t3,te03 // 15/3:done! + br.ret.sptk b6 };; +.endp _ia64_AES_encrypt# + +// void AES_encrypt (const void *in,void *out,const AES_KEY *key); +.global AES_encrypt# +.proc AES_encrypt# +.align 32 +AES_encrypt: + .prologue + .save ar.pfs,pfssave +{ .mmi; alloc pfssave=ar.pfs,3,1,12,0 + and out0=3,in0 + mov r3=ip } +{ .mmi; ADDP in0=0,in0 + mov loc0=psr.um + ADDP out11=KSZ*60,in2 };; // &AES_KEY->rounds + +{ .mmi; ld4 out11=[out11] // AES_KEY->rounds + add out8=(AES_Te#-AES_encrypt#),r3 // Te0 + .save pr,prsave + mov prsave=pr } +{ .mmi; rum 1<<3 // clear um.ac + .save ar.lc,lcsave + mov lcsave=ar.lc };; + + .body +#if defined(_HPUX_SOURCE) // HPUX is big-endian, cut 15+15 cycles... +{ .mib; cmp.ne p6,p0=out0,r0 + add out0=4,in0 +(p6) br.dpnt.many .Le_i_unaligned };; + +{ .mmi; ld4 out1=[in0],8 // s0 + and out9=3,in1 + mov twenty4=24 } +{ .mmi; ld4 out3=[out0],8 // s1 + ADDP rk0=0,in2 + mov sixteen=16 };; +{ .mmi; ld4 out5=[in0] // s2 + cmp.ne p6,p0=out9,r0 + mov maskff=0xff } +{ .mmb; ld4 out7=[out0] // s3 + ADDP rk1=KSZ,in2 + br.call.sptk.many b6=_ia64_AES_encrypt };; + +{ .mib; ADDP in0=4,in1 + ADDP in1=0,in1 +(p6) br.spnt .Le_o_unaligned };; + +{ .mii; mov psr.um=loc0 + mov ar.pfs=pfssave + mov ar.lc=lcsave };; +{ .mmi; st4 [in1]=r16,8 // s0 + st4 [in0]=r20,8 // s1 + mov pr=prsave,0x1ffff };; +{ .mmb; st4 [in1]=r24 // s2 + st4 [in0]=r28 // s3 + br.ret.sptk.many b0 };; +#endif + +.align 32 +.Le_i_unaligned: +{ .mmi; add out0=1,in0 + add out2=2,in0 + add out4=3,in0 };; +{ .mmi; ld1 r16=[in0],4 + ld1 r17=[out0],4 }//;; +{ .mmi; ld1 r18=[out2],4 + ld1 out1=[out4],4 };; // s0 +{ .mmi; ld1 r20=[in0],4 + ld1 r21=[out0],4 }//;; +{ .mmi; ld1 r22=[out2],4 + ld1 out3=[out4],4 };; // s1 +{ .mmi; ld1 r24=[in0],4 + ld1 r25=[out0],4 }//;; +{ .mmi; ld1 r26=[out2],4 + ld1 out5=[out4],4 };; // s2 +{ .mmi; ld1 r28=[in0] + ld1 r29=[out0] }//;; +{ .mmi; ld1 r30=[out2] + ld1 out7=[out4] };; // s3 + +{ .mii; + dep out1=r16,out1,24,8 //;; + dep out3=r20,out3,24,8 }//;; +{ .mii; ADDP rk0=0,in2 + dep out5=r24,out5,24,8 //;; + dep out7=r28,out7,24,8 };; +{ .mii; ADDP rk1=KSZ,in2 + dep out1=r17,out1,16,8 //;; + dep out3=r21,out3,16,8 }//;; +{ .mii; mov twenty4=24 + dep out5=r25,out5,16,8 //;; + dep out7=r29,out7,16,8 };; +{ .mii; mov sixteen=16 + dep out1=r18,out1,8,8 //;; + dep out3=r22,out3,8,8 }//;; +{ .mii; mov maskff=0xff + dep out5=r26,out5,8,8 //;; + dep out7=r30,out7,8,8 };; + +{ .mib; br.call.sptk.many b6=_ia64_AES_encrypt };; + +.Le_o_unaligned: +{ .mii; ADDP out0=0,in1 + extr.u r17=r16,8,8 // s0 + shr.u r19=r16,twenty4 }//;; +{ .mii; ADDP out1=1,in1 + extr.u r18=r16,16,8 + shr.u r23=r20,twenty4 }//;; // s1 +{ .mii; ADDP out2=2,in1 + extr.u r21=r20,8,8 + shr.u r22=r20,sixteen }//;; +{ .mii; ADDP out3=3,in1 + extr.u r25=r24,8,8 // s2 + shr.u r27=r24,twenty4 };; +{ .mii; st1 [out3]=r16,4 + extr.u r26=r24,16,8 + shr.u r31=r28,twenty4 }//;; // s3 +{ .mii; st1 [out2]=r17,4 + extr.u r29=r28,8,8 + shr.u r30=r28,sixteen }//;; + +{ .mmi; st1 [out1]=r18,4 + st1 [out0]=r19,4 };; +{ .mmi; st1 [out3]=r20,4 + st1 [out2]=r21,4 }//;; +{ .mmi; st1 [out1]=r22,4 + st1 [out0]=r23,4 };; +{ .mmi; st1 [out3]=r24,4 + st1 [out2]=r25,4 + mov pr=prsave,0x1ffff }//;; +{ .mmi; st1 [out1]=r26,4 + st1 [out0]=r27,4 + mov ar.pfs=pfssave };; +{ .mmi; st1 [out3]=r28 + st1 [out2]=r29 + mov ar.lc=lcsave }//;; +{ .mmi; st1 [out1]=r30 + st1 [out0]=r31 } +{ .mfb; mov psr.um=loc0 // restore user mask + br.ret.sptk.many b0 };; +.endp AES_encrypt# + +// *AES_decrypt are autogenerated by the following script: +#if 0 +#!/usr/bin/env perl +print "// *AES_decrypt are autogenerated by the following script:\n#if 0\n"; +open(PROG,'<'.$0); while(<PROG>) { print; } close(PROG); +print "#endif\n"; +while(<>) { + $process=1 if (/\.proc\s+_ia64_AES_encrypt/); + next if (!$process); + + #s/te00=s0/td00=s0/; s/te00/td00/g; + s/te11=s1/td13=s3/; s/te11/td13/g; + #s/te22=s2/td22=s2/; s/te22/td22/g; + s/te33=s3/td31=s1/; s/te33/td31/g; + + #s/te01=s1/td01=s1/; s/te01/td01/g; + s/te12=s2/td10=s0/; s/te12/td10/g; + #s/te23=s3/td23=s3/; s/te23/td23/g; + s/te30=s0/td32=s2/; s/te30/td32/g; + + #s/te02=s2/td02=s2/; s/te02/td02/g; + s/te13=s3/td11=s1/; s/te13/td11/g; + #s/te20=s0/td20=s0/; s/te20/td20/g; + s/te31=s1/td33=s3/; s/te31/td33/g; + + #s/te03=s3/td03=s3/; s/te03/td03/g; + s/te10=s0/td12=s2/; s/te10/td12/g; + #s/te21=s1/td21=s1/; s/te21/td21/g; + s/te32=s2/td30=s0/; s/te32/td30/g; + + s/td/te/g; + + s/AES_encrypt/AES_decrypt/g; + s/\.Le_/.Ld_/g; + s/AES_Te#/AES_Td#/g; + + print; + + exit if (/\.endp\s+AES_decrypt/); +} +#endif +.proc _ia64_AES_decrypt# +// Input: rk0-rk1 +// te0 +// te3 as AES_KEY->rounds!!! +// s0-s3 +// maskff,twenty4,sixteen +// Output: r16,r20,r24,r28 as s0-s3 +// Clobber: r16-r31,rk0-rk1,r32-r43 +.align 32 +_ia64_AES_decrypt: + .prologue + .altrp b6 + .body +{ .mmi; alloc r16=ar.pfs,12,0,0,8 + LDKEY t0=[rk0],2*KSZ + mov pr.rot=1<<16 } +{ .mmi; LDKEY t1=[rk1],2*KSZ + add te1=TE1,te0 + add te3=-3,te3 };; +{ .mib; LDKEY t2=[rk0],2*KSZ + mov ar.ec=2 } +{ .mib; LDKEY t3=[rk1],2*KSZ + add te2=TE2,te0 + brp.loop.imp .Ld_top,.Ld_end-16 };; + +{ .mmi; xor s0=s0,t0 + xor s1=s1,t1 + mov ar.lc=te3 } +{ .mmi; xor s2=s2,t2 + xor s3=s3,t3 + add te3=TE3,te0 };; + +.align 32 +.Ld_top: +{ .mmi; (p0) LDKEY t0=[rk0],2*KSZ // 0/0:rk[0] + (p0) and te31=s1,maskff // 0/0:s3&0xff + (p0) extr.u te22=s2,8,8 } // 0/0:s2>>8&0xff +{ .mmi; (p0) LDKEY t1=[rk1],2*KSZ // 0/1:rk[1] + (p0) and te32=s2,maskff // 0/1:s0&0xff + (p0) shr.u te00=s0,twenty4 };; // 0/0:s0>>24 +{ .mmi; (p0) LDKEY t2=[rk0],2*KSZ // 1/2:rk[2] + (p0) shladd te31=te31,3,te3 // 1/0:te0+s0>>24 + (p0) extr.u te23=s3,8,8 } // 1/1:s3>>8&0xff +{ .mmi; (p0) LDKEY t3=[rk1],2*KSZ // 1/3:rk[3] + (p0) shladd te32=te32,3,te3 // 1/1:te3+s0 + (p0) shr.u te01=s1,twenty4 };; // 1/1:s1>>24 +{ .mmi; (p0) ld4 te31=[te31] // 2/0:te3[s3&0xff] + (p0) shladd te22=te22,3,te2 // 2/0:te2+s2>>8&0xff + (p0) extr.u te20=s0,8,8 } // 2/2:s0>>8&0xff +{ .mmi; (p0) ld4 te32=[te32] // 2/1:te3[s0] + (p0) shladd te23=te23,3,te2 // 2/1:te2+s3>>8 + (p0) shr.u te02=s2,twenty4 };; // 2/2:s2>>24 +{ .mmi; (p0) ld4 te22=[te22] // 3/0:te2[s2>>8] + (p0) shladd te20=te20,3,te2 // 3/2:te2+s0>>8 + (p0) extr.u te21=s1,8,8 } // 3/3:s1>>8&0xff +{ .mmi; (p0) ld4 te23=[te23] // 3/1:te2[s3>>8] + (p0) shladd te00=te00,3,te0 // 3/0:te0+s0>>24 + (p0) shr.u te03=s3,twenty4 };; // 3/3:s3>>24 +{ .mmi; (p0) ld4 te20=[te20] // 4/2:te2[s0>>8] + (p0) shladd te21=te21,3,te2 // 4/3:te3+s2 + (p0) extr.u te13=s3,16,8 } // 4/0:s1>>16&0xff +{ .mmi; (p0) ld4 te00=[te00] // 4/0:te0[s0>>24] + (p0) shladd te01=te01,3,te0 // 4/1:te0+s1>>24 + (p0) shr.u te11=s1,sixteen };; // 4/2:s3>>16 +{ .mmi; (p0) ld4 te21=[te21] // 5/3:te2[s1>>8] + (p0) shladd te13=te13,3,te1 // 5/0:te1+s1>>16 + (p0) extr.u te10=s0,16,8 } // 5/1:s2>>16&0xff +{ .mmi; (p0) ld4 te01=[te01] // 5/1:te0[s1>>24] + (p0) shladd te02=te02,3,te0 // 5/2:te0+s2>>24 + (p0) and te33=s3,maskff };; // 5/2:s1&0xff +{ .mmi; (p0) ld4 te13=[te13] // 6/0:te1[s1>>16] + (p0) shladd te10=te10,3,te1 // 6/1:te1+s2>>16 + (p0) extr.u te12=s2,16,8 } // 6/3:s0>>16&0xff +{ .mmi; (p0) ld4 te02=[te02] // 6/2:te0[s2>>24] + (p0) shladd te03=te03,3,te0 // 6/3:te1+s0>>16 + (p0) and te30=s0,maskff };; // 6/3:s2&0xff + +{ .mmi; (p0) ld4 te10=[te10] // 7/1:te1[s2>>16] + (p0) shladd te33=te33,3,te3 // 7/2:te3+s1&0xff + (p0) and te11=te11,maskff} // 7/2:s3>>16&0xff +{ .mmi; (p0) ld4 te03=[te03] // 7/3:te0[s3>>24] + (p0) shladd te30=te30,3,te3 // 7/3:te3+s2 + (p0) xor t0=t0,te31 };; // 7/0: +{ .mmi; (p0) ld4 te33=[te33] // 8/2:te3[s1] + (p0) shladd te11=te11,3,te1 // 8/2:te1+s3>>16 + (p0) xor t0=t0,te22 } // 8/0: +{ .mmi; (p0) ld4 te30=[te30] // 8/3:te3[s2] + (p0) shladd te12=te12,3,te1 // 8/3:te1+s0>>16 + (p0) xor t1=t1,te32 };; // 8/1: +{ .mmi; (p0) ld4 te11=[te11] // 9/2:te1[s3>>16] + (p0) ld4 te12=[te12] // 9/3:te1[s0>>16] + (p0) xor t0=t0,te00 };; // 9/0: !L2 scheduling +{ .mmi; (p0) xor t1=t1,te23 // 10[9]/1: + (p0) xor t2=t2,te20 // 10[9]/2: + (p0) xor t3=t3,te21 };; // 10[9]/3: +{ .mmi; (p0) xor t0=t0,te13 // 11[10]/0:done! + (p0) xor t1=t1,te01 // 11[10]/1: + (p0) xor t2=t2,te02 };; // 11[10]/2: !L2 scheduling +{ .mmi; (p0) xor t3=t3,te03 // 12[10]/3: + (p16) cmp.eq p0,p17=r0,r0 };; // 12[10]/clear (p17) +{ .mmi; (p0) xor t1=t1,te10 // 13[11]/1:done! + (p0) xor t2=t2,te33 // 13[11]/2: + (p0) xor t3=t3,te30 } // 13[11]/3: +{ .mmi; (p17) add te0=2048,te0 // 13[11]/ + (p17) add te1=2048+64-TE1,te1};; // 13[11]/ +{ .mib; (p0) xor t2=t2,te11 // 14[12]/2:done! + (p17) add te2=2048+128-TE2,te2} // 14[12]/ +{ .mib; (p0) xor t3=t3,te12 // 14[12]/3:done! + (p17) add te3=2048+192-TE3,te3 // 14[12]/ + br.ctop.sptk .Ld_top };; +.Ld_end: + + +{ .mmi; ld8 te10=[te0] // prefetch Td4 + ld8 te33=[te1] } +{ .mmi; ld8 te12=[te2] + ld8 te30=[te3] } + +{ .mmi; LDKEY t0=[rk0],2*KSZ // 0/0:rk[0] + and te31=s1,maskff // 0/0:s3&0xff + extr.u te22=s2,8,8 } // 0/0:s2>>8&0xff +{ .mmi; LDKEY t1=[rk1],2*KSZ // 0/1:rk[1] + and te32=s2,maskff // 0/1:s0&0xff + shr.u te00=s0,twenty4 };; // 0/0:s0>>24 +{ .mmi; LDKEY t2=[rk0],2*KSZ // 1/2:rk[2] + add te31=te31,te0 // 1/0:te0+s0>>24 + extr.u te23=s3,8,8 } // 1/1:s3>>8&0xff +{ .mmi; LDKEY t3=[rk1],2*KSZ // 1/3:rk[3] + add te32=te32,te0 // 1/1:te0+s0 + shr.u te01=s1,twenty4 };; // 1/1:s1>>24 +{ .mmi; ld1 te31=[te31] // 2/0:te0[s3&0xff] + add te22=te22,te0 // 2/0:te0+s2>>8&0xff + extr.u te20=s0,8,8 } // 2/2:s0>>8&0xff +{ .mmi; ld1 te32=[te32] // 2/1:te0[s0] + add te23=te23,te0 // 2/1:te0+s3>>8 + shr.u te02=s2,twenty4 };; // 2/2:s2>>24 +{ .mmi; ld1 te22=[te22] // 3/0:te0[s2>>8] + add te20=te20,te0 // 3/2:te0+s0>>8 + extr.u te21=s1,8,8 } // 3/3:s1>>8&0xff +{ .mmi; ld1 te23=[te23] // 3/1:te0[s3>>8] + add te00=te00,te0 // 3/0:te0+s0>>24 + shr.u te03=s3,twenty4 };; // 3/3:s3>>24 +{ .mmi; ld1 te20=[te20] // 4/2:te0[s0>>8] + add te21=te21,te0 // 4/3:te0+s2 + extr.u te13=s3,16,8 } // 4/0:s1>>16&0xff +{ .mmi; ld1 te00=[te00] // 4/0:te0[s0>>24] + add te01=te01,te0 // 4/1:te0+s1>>24 + shr.u te11=s1,sixteen };; // 4/2:s3>>16 +{ .mmi; ld1 te21=[te21] // 5/3:te0[s1>>8] + add te13=te13,te0 // 5/0:te0+s1>>16 + extr.u te10=s0,16,8 } // 5/1:s2>>16&0xff +{ .mmi; ld1 te01=[te01] // 5/1:te0[s1>>24] + add te02=te02,te0 // 5/2:te0+s2>>24 + and te33=s3,maskff };; // 5/2:s1&0xff +{ .mmi; ld1 te13=[te13] // 6/0:te0[s1>>16] + add te10=te10,te0 // 6/1:te0+s2>>16 + extr.u te12=s2,16,8 } // 6/3:s0>>16&0xff +{ .mmi; ld1 te02=[te02] // 6/2:te0[s2>>24] + add te03=te03,te0 // 6/3:te0+s0>>16 + and te30=s0,maskff };; // 6/3:s2&0xff + +{ .mmi; ld1 te10=[te10] // 7/1:te0[s2>>16] + add te33=te33,te0 // 7/2:te0+s1&0xff + dep te31=te22,te31,8,8} // 7/0: +{ .mmi; ld1 te03=[te03] // 7/3:te0[s3>>24] + add te30=te30,te0 // 7/3:te0+s2 + and te11=te11,maskff};; // 7/2:s3>>16&0xff +{ .mmi; ld1 te33=[te33] // 8/2:te0[s1] + add te11=te11,te0 // 8/2:te0+s3>>16 + dep te32=te23,te32,8,8} // 8/1: +{ .mmi; ld1 te30=[te30] // 8/3:te0[s2] + add te12=te12,te0 // 8/3:te0+s0>>16 + shl te00=te00,twenty4};; // 8/0: +{ .mii; ld1 te11=[te11] // 9/2:te0[s3>>16] + dep te31=te13,te31,16,8 // 9/0: + shl te01=te01,twenty4};; // 9/1: +{ .mii; ld1 te12=[te12] // 10/3:te0[s0>>16] + dep te33=te20,te33,8,8 // 10/2: + shl te02=te02,twenty4};; // 10/2: +{ .mii; xor t0=t0,te31 // 11/0: + dep te30=te21,te30,8,8 // 11/3: + shl te10=te10,sixteen};; // 11/1: +{ .mii; xor r16=t0,te00 // 12/0:done! + dep te33=te11,te33,16,8 // 12/2: + shl te03=te03,twenty4};; // 12/3: +{ .mmi; xor t1=t1,te01 // 13/1: + xor t2=t2,te02 // 13/2: + dep te30=te12,te30,16,8};; // 13/3: +{ .mmi; xor t1=t1,te32 // 14/1: + xor r24=t2,te33 // 14/2:done! + xor t3=t3,te30 };; // 14/3: +{ .mib; xor r20=t1,te10 // 15/1:done! + xor r28=t3,te03 // 15/3:done! + br.ret.sptk b6 };; +.endp _ia64_AES_decrypt# + +// void AES_decrypt (const void *in,void *out,const AES_KEY *key); +.global AES_decrypt# +.proc AES_decrypt# +.align 32 +AES_decrypt: + .prologue + .save ar.pfs,pfssave +{ .mmi; alloc pfssave=ar.pfs,3,1,12,0 + and out0=3,in0 + mov r3=ip } +{ .mmi; ADDP in0=0,in0 + mov loc0=psr.um + ADDP out11=KSZ*60,in2 };; // &AES_KEY->rounds + +{ .mmi; ld4 out11=[out11] // AES_KEY->rounds + add out8=(AES_Td#-AES_decrypt#),r3 // Te0 + .save pr,prsave + mov prsave=pr } +{ .mmi; rum 1<<3 // clear um.ac + .save ar.lc,lcsave + mov lcsave=ar.lc };; + + .body +#if defined(_HPUX_SOURCE) // HPUX is big-endian, cut 15+15 cycles... +{ .mib; cmp.ne p6,p0=out0,r0 + add out0=4,in0 +(p6) br.dpnt.many .Ld_i_unaligned };; + +{ .mmi; ld4 out1=[in0],8 // s0 + and out9=3,in1 + mov twenty4=24 } +{ .mmi; ld4 out3=[out0],8 // s1 + ADDP rk0=0,in2 + mov sixteen=16 };; +{ .mmi; ld4 out5=[in0] // s2 + cmp.ne p6,p0=out9,r0 + mov maskff=0xff } +{ .mmb; ld4 out7=[out0] // s3 + ADDP rk1=KSZ,in2 + br.call.sptk.many b6=_ia64_AES_decrypt };; + +{ .mib; ADDP in0=4,in1 + ADDP in1=0,in1 +(p6) br.spnt .Ld_o_unaligned };; + +{ .mii; mov psr.um=loc0 + mov ar.pfs=pfssave + mov ar.lc=lcsave };; +{ .mmi; st4 [in1]=r16,8 // s0 + st4 [in0]=r20,8 // s1 + mov pr=prsave,0x1ffff };; +{ .mmb; st4 [in1]=r24 // s2 + st4 [in0]=r28 // s3 + br.ret.sptk.many b0 };; +#endif + +.align 32 +.Ld_i_unaligned: +{ .mmi; add out0=1,in0 + add out2=2,in0 + add out4=3,in0 };; +{ .mmi; ld1 r16=[in0],4 + ld1 r17=[out0],4 }//;; +{ .mmi; ld1 r18=[out2],4 + ld1 out1=[out4],4 };; // s0 +{ .mmi; ld1 r20=[in0],4 + ld1 r21=[out0],4 }//;; +{ .mmi; ld1 r22=[out2],4 + ld1 out3=[out4],4 };; // s1 +{ .mmi; ld1 r24=[in0],4 + ld1 r25=[out0],4 }//;; +{ .mmi; ld1 r26=[out2],4 + ld1 out5=[out4],4 };; // s2 +{ .mmi; ld1 r28=[in0] + ld1 r29=[out0] }//;; +{ .mmi; ld1 r30=[out2] + ld1 out7=[out4] };; // s3 + +{ .mii; + dep out1=r16,out1,24,8 //;; + dep out3=r20,out3,24,8 }//;; +{ .mii; ADDP rk0=0,in2 + dep out5=r24,out5,24,8 //;; + dep out7=r28,out7,24,8 };; +{ .mii; ADDP rk1=KSZ,in2 + dep out1=r17,out1,16,8 //;; + dep out3=r21,out3,16,8 }//;; +{ .mii; mov twenty4=24 + dep out5=r25,out5,16,8 //;; + dep out7=r29,out7,16,8 };; +{ .mii; mov sixteen=16 + dep out1=r18,out1,8,8 //;; + dep out3=r22,out3,8,8 }//;; +{ .mii; mov maskff=0xff + dep out5=r26,out5,8,8 //;; + dep out7=r30,out7,8,8 };; + +{ .mib; br.call.sptk.many b6=_ia64_AES_decrypt };; + +.Ld_o_unaligned: +{ .mii; ADDP out0=0,in1 + extr.u r17=r16,8,8 // s0 + shr.u r19=r16,twenty4 }//;; +{ .mii; ADDP out1=1,in1 + extr.u r18=r16,16,8 + shr.u r23=r20,twenty4 }//;; // s1 +{ .mii; ADDP out2=2,in1 + extr.u r21=r20,8,8 + shr.u r22=r20,sixteen }//;; +{ .mii; ADDP out3=3,in1 + extr.u r25=r24,8,8 // s2 + shr.u r27=r24,twenty4 };; +{ .mii; st1 [out3]=r16,4 + extr.u r26=r24,16,8 + shr.u r31=r28,twenty4 }//;; // s3 +{ .mii; st1 [out2]=r17,4 + extr.u r29=r28,8,8 + shr.u r30=r28,sixteen }//;; + +{ .mmi; st1 [out1]=r18,4 + st1 [out0]=r19,4 };; +{ .mmi; st1 [out3]=r20,4 + st1 [out2]=r21,4 }//;; +{ .mmi; st1 [out1]=r22,4 + st1 [out0]=r23,4 };; +{ .mmi; st1 [out3]=r24,4 + st1 [out2]=r25,4 + mov pr=prsave,0x1ffff }//;; +{ .mmi; st1 [out1]=r26,4 + st1 [out0]=r27,4 + mov ar.pfs=pfssave };; +{ .mmi; st1 [out3]=r28 + st1 [out2]=r29 + mov ar.lc=lcsave }//;; +{ .mmi; st1 [out1]=r30 + st1 [out0]=r31 } +{ .mfb; mov psr.um=loc0 // restore user mask + br.ret.sptk.many b0 };; +.endp AES_decrypt# + +// leave it in .text segment... +.align 64 +.global AES_Te# +.type AES_Te#,@object +AES_Te: data4 0xc66363a5,0xc66363a5, 0xf87c7c84,0xf87c7c84 + data4 0xee777799,0xee777799, 0xf67b7b8d,0xf67b7b8d + data4 0xfff2f20d,0xfff2f20d, 0xd66b6bbd,0xd66b6bbd + data4 0xde6f6fb1,0xde6f6fb1, 0x91c5c554,0x91c5c554 + data4 0x60303050,0x60303050, 0x02010103,0x02010103 + data4 0xce6767a9,0xce6767a9, 0x562b2b7d,0x562b2b7d + data4 0xe7fefe19,0xe7fefe19, 0xb5d7d762,0xb5d7d762 + data4 0x4dababe6,0x4dababe6, 0xec76769a,0xec76769a + data4 0x8fcaca45,0x8fcaca45, 0x1f82829d,0x1f82829d + data4 0x89c9c940,0x89c9c940, 0xfa7d7d87,0xfa7d7d87 + data4 0xeffafa15,0xeffafa15, 0xb25959eb,0xb25959eb + data4 0x8e4747c9,0x8e4747c9, 0xfbf0f00b,0xfbf0f00b + data4 0x41adadec,0x41adadec, 0xb3d4d467,0xb3d4d467 + data4 0x5fa2a2fd,0x5fa2a2fd, 0x45afafea,0x45afafea + data4 0x239c9cbf,0x239c9cbf, 0x53a4a4f7,0x53a4a4f7 + data4 0xe4727296,0xe4727296, 0x9bc0c05b,0x9bc0c05b + data4 0x75b7b7c2,0x75b7b7c2, 0xe1fdfd1c,0xe1fdfd1c + data4 0x3d9393ae,0x3d9393ae, 0x4c26266a,0x4c26266a + data4 0x6c36365a,0x6c36365a, 0x7e3f3f41,0x7e3f3f41 + data4 0xf5f7f702,0xf5f7f702, 0x83cccc4f,0x83cccc4f + data4 0x6834345c,0x6834345c, 0x51a5a5f4,0x51a5a5f4 + data4 0xd1e5e534,0xd1e5e534, 0xf9f1f108,0xf9f1f108 + data4 0xe2717193,0xe2717193, 0xabd8d873,0xabd8d873 + data4 0x62313153,0x62313153, 0x2a15153f,0x2a15153f + data4 0x0804040c,0x0804040c, 0x95c7c752,0x95c7c752 + data4 0x46232365,0x46232365, 0x9dc3c35e,0x9dc3c35e + data4 0x30181828,0x30181828, 0x379696a1,0x379696a1 + data4 0x0a05050f,0x0a05050f, 0x2f9a9ab5,0x2f9a9ab5 + data4 0x0e070709,0x0e070709, 0x24121236,0x24121236 + data4 0x1b80809b,0x1b80809b, 0xdfe2e23d,0xdfe2e23d + data4 0xcdebeb26,0xcdebeb26, 0x4e272769,0x4e272769 + data4 0x7fb2b2cd,0x7fb2b2cd, 0xea75759f,0xea75759f + data4 0x1209091b,0x1209091b, 0x1d83839e,0x1d83839e + data4 0x582c2c74,0x582c2c74, 0x341a1a2e,0x341a1a2e + data4 0x361b1b2d,0x361b1b2d, 0xdc6e6eb2,0xdc6e6eb2 + data4 0xb45a5aee,0xb45a5aee, 0x5ba0a0fb,0x5ba0a0fb + data4 0xa45252f6,0xa45252f6, 0x763b3b4d,0x763b3b4d + data4 0xb7d6d661,0xb7d6d661, 0x7db3b3ce,0x7db3b3ce + data4 0x5229297b,0x5229297b, 0xdde3e33e,0xdde3e33e + data4 0x5e2f2f71,0x5e2f2f71, 0x13848497,0x13848497 + data4 0xa65353f5,0xa65353f5, 0xb9d1d168,0xb9d1d168 + data4 0x00000000,0x00000000, 0xc1eded2c,0xc1eded2c + data4 0x40202060,0x40202060, 0xe3fcfc1f,0xe3fcfc1f + data4 0x79b1b1c8,0x79b1b1c8, 0xb65b5bed,0xb65b5bed + data4 0xd46a6abe,0xd46a6abe, 0x8dcbcb46,0x8dcbcb46 + data4 0x67bebed9,0x67bebed9, 0x7239394b,0x7239394b + data4 0x944a4ade,0x944a4ade, 0x984c4cd4,0x984c4cd4 + data4 0xb05858e8,0xb05858e8, 0x85cfcf4a,0x85cfcf4a + data4 0xbbd0d06b,0xbbd0d06b, 0xc5efef2a,0xc5efef2a + data4 0x4faaaae5,0x4faaaae5, 0xedfbfb16,0xedfbfb16 + data4 0x864343c5,0x864343c5, 0x9a4d4dd7,0x9a4d4dd7 + data4 0x66333355,0x66333355, 0x11858594,0x11858594 + data4 0x8a4545cf,0x8a4545cf, 0xe9f9f910,0xe9f9f910 + data4 0x04020206,0x04020206, 0xfe7f7f81,0xfe7f7f81 + data4 0xa05050f0,0xa05050f0, 0x783c3c44,0x783c3c44 + data4 0x259f9fba,0x259f9fba, 0x4ba8a8e3,0x4ba8a8e3 + data4 0xa25151f3,0xa25151f3, 0x5da3a3fe,0x5da3a3fe + data4 0x804040c0,0x804040c0, 0x058f8f8a,0x058f8f8a + data4 0x3f9292ad,0x3f9292ad, 0x219d9dbc,0x219d9dbc + data4 0x70383848,0x70383848, 0xf1f5f504,0xf1f5f504 + data4 0x63bcbcdf,0x63bcbcdf, 0x77b6b6c1,0x77b6b6c1 + data4 0xafdada75,0xafdada75, 0x42212163,0x42212163 + data4 0x20101030,0x20101030, 0xe5ffff1a,0xe5ffff1a + data4 0xfdf3f30e,0xfdf3f30e, 0xbfd2d26d,0xbfd2d26d + data4 0x81cdcd4c,0x81cdcd4c, 0x180c0c14,0x180c0c14 + data4 0x26131335,0x26131335, 0xc3ecec2f,0xc3ecec2f + data4 0xbe5f5fe1,0xbe5f5fe1, 0x359797a2,0x359797a2 + data4 0x884444cc,0x884444cc, 0x2e171739,0x2e171739 + data4 0x93c4c457,0x93c4c457, 0x55a7a7f2,0x55a7a7f2 + data4 0xfc7e7e82,0xfc7e7e82, 0x7a3d3d47,0x7a3d3d47 + data4 0xc86464ac,0xc86464ac, 0xba5d5de7,0xba5d5de7 + data4 0x3219192b,0x3219192b, 0xe6737395,0xe6737395 + data4 0xc06060a0,0xc06060a0, 0x19818198,0x19818198 + data4 0x9e4f4fd1,0x9e4f4fd1, 0xa3dcdc7f,0xa3dcdc7f + data4 0x44222266,0x44222266, 0x542a2a7e,0x542a2a7e + data4 0x3b9090ab,0x3b9090ab, 0x0b888883,0x0b888883 + data4 0x8c4646ca,0x8c4646ca, 0xc7eeee29,0xc7eeee29 + data4 0x6bb8b8d3,0x6bb8b8d3, 0x2814143c,0x2814143c + data4 0xa7dede79,0xa7dede79, 0xbc5e5ee2,0xbc5e5ee2 + data4 0x160b0b1d,0x160b0b1d, 0xaddbdb76,0xaddbdb76 + data4 0xdbe0e03b,0xdbe0e03b, 0x64323256,0x64323256 + data4 0x743a3a4e,0x743a3a4e, 0x140a0a1e,0x140a0a1e + data4 0x924949db,0x924949db, 0x0c06060a,0x0c06060a + data4 0x4824246c,0x4824246c, 0xb85c5ce4,0xb85c5ce4 + data4 0x9fc2c25d,0x9fc2c25d, 0xbdd3d36e,0xbdd3d36e + data4 0x43acacef,0x43acacef, 0xc46262a6,0xc46262a6 + data4 0x399191a8,0x399191a8, 0x319595a4,0x319595a4 + data4 0xd3e4e437,0xd3e4e437, 0xf279798b,0xf279798b + data4 0xd5e7e732,0xd5e7e732, 0x8bc8c843,0x8bc8c843 + data4 0x6e373759,0x6e373759, 0xda6d6db7,0xda6d6db7 + data4 0x018d8d8c,0x018d8d8c, 0xb1d5d564,0xb1d5d564 + data4 0x9c4e4ed2,0x9c4e4ed2, 0x49a9a9e0,0x49a9a9e0 + data4 0xd86c6cb4,0xd86c6cb4, 0xac5656fa,0xac5656fa + data4 0xf3f4f407,0xf3f4f407, 0xcfeaea25,0xcfeaea25 + data4 0xca6565af,0xca6565af, 0xf47a7a8e,0xf47a7a8e + data4 0x47aeaee9,0x47aeaee9, 0x10080818,0x10080818 + data4 0x6fbabad5,0x6fbabad5, 0xf0787888,0xf0787888 + data4 0x4a25256f,0x4a25256f, 0x5c2e2e72,0x5c2e2e72 + data4 0x381c1c24,0x381c1c24, 0x57a6a6f1,0x57a6a6f1 + data4 0x73b4b4c7,0x73b4b4c7, 0x97c6c651,0x97c6c651 + data4 0xcbe8e823,0xcbe8e823, 0xa1dddd7c,0xa1dddd7c + data4 0xe874749c,0xe874749c, 0x3e1f1f21,0x3e1f1f21 + data4 0x964b4bdd,0x964b4bdd, 0x61bdbddc,0x61bdbddc + data4 0x0d8b8b86,0x0d8b8b86, 0x0f8a8a85,0x0f8a8a85 + data4 0xe0707090,0xe0707090, 0x7c3e3e42,0x7c3e3e42 + data4 0x71b5b5c4,0x71b5b5c4, 0xcc6666aa,0xcc6666aa + data4 0x904848d8,0x904848d8, 0x06030305,0x06030305 + data4 0xf7f6f601,0xf7f6f601, 0x1c0e0e12,0x1c0e0e12 + data4 0xc26161a3,0xc26161a3, 0x6a35355f,0x6a35355f + data4 0xae5757f9,0xae5757f9, 0x69b9b9d0,0x69b9b9d0 + data4 0x17868691,0x17868691, 0x99c1c158,0x99c1c158 + data4 0x3a1d1d27,0x3a1d1d27, 0x279e9eb9,0x279e9eb9 + data4 0xd9e1e138,0xd9e1e138, 0xebf8f813,0xebf8f813 + data4 0x2b9898b3,0x2b9898b3, 0x22111133,0x22111133 + data4 0xd26969bb,0xd26969bb, 0xa9d9d970,0xa9d9d970 + data4 0x078e8e89,0x078e8e89, 0x339494a7,0x339494a7 + data4 0x2d9b9bb6,0x2d9b9bb6, 0x3c1e1e22,0x3c1e1e22 + data4 0x15878792,0x15878792, 0xc9e9e920,0xc9e9e920 + data4 0x87cece49,0x87cece49, 0xaa5555ff,0xaa5555ff + data4 0x50282878,0x50282878, 0xa5dfdf7a,0xa5dfdf7a + data4 0x038c8c8f,0x038c8c8f, 0x59a1a1f8,0x59a1a1f8 + data4 0x09898980,0x09898980, 0x1a0d0d17,0x1a0d0d17 + data4 0x65bfbfda,0x65bfbfda, 0xd7e6e631,0xd7e6e631 + data4 0x844242c6,0x844242c6, 0xd06868b8,0xd06868b8 + data4 0x824141c3,0x824141c3, 0x299999b0,0x299999b0 + data4 0x5a2d2d77,0x5a2d2d77, 0x1e0f0f11,0x1e0f0f11 + data4 0x7bb0b0cb,0x7bb0b0cb, 0xa85454fc,0xa85454fc + data4 0x6dbbbbd6,0x6dbbbbd6, 0x2c16163a,0x2c16163a +// Te4: + data1 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5 + data1 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76 + data1 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0 + data1 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0 + data1 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc + data1 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15 + data1 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a + data1 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75 + data1 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0 + data1 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84 + data1 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b + data1 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf + data1 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85 + data1 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8 + data1 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5 + data1 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2 + data1 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17 + data1 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73 + data1 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88 + data1 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb + data1 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c + data1 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79 + data1 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9 + data1 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08 + data1 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6 + data1 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a + data1 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e + data1 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e + data1 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94 + data1 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf + data1 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68 + data1 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16 +.size AES_Te#,2048+256 // HP-UX assembler fails to ".-AES_Te#" + +.align 64 +.global AES_Td# +.type AES_Td#,@object +AES_Td: data4 0x51f4a750,0x51f4a750, 0x7e416553,0x7e416553 + data4 0x1a17a4c3,0x1a17a4c3, 0x3a275e96,0x3a275e96 + data4 0x3bab6bcb,0x3bab6bcb, 0x1f9d45f1,0x1f9d45f1 + data4 0xacfa58ab,0xacfa58ab, 0x4be30393,0x4be30393 + data4 0x2030fa55,0x2030fa55, 0xad766df6,0xad766df6 + data4 0x88cc7691,0x88cc7691, 0xf5024c25,0xf5024c25 + data4 0x4fe5d7fc,0x4fe5d7fc, 0xc52acbd7,0xc52acbd7 + data4 0x26354480,0x26354480, 0xb562a38f,0xb562a38f + data4 0xdeb15a49,0xdeb15a49, 0x25ba1b67,0x25ba1b67 + data4 0x45ea0e98,0x45ea0e98, 0x5dfec0e1,0x5dfec0e1 + data4 0xc32f7502,0xc32f7502, 0x814cf012,0x814cf012 + data4 0x8d4697a3,0x8d4697a3, 0x6bd3f9c6,0x6bd3f9c6 + data4 0x038f5fe7,0x038f5fe7, 0x15929c95,0x15929c95 + data4 0xbf6d7aeb,0xbf6d7aeb, 0x955259da,0x955259da + data4 0xd4be832d,0xd4be832d, 0x587421d3,0x587421d3 + data4 0x49e06929,0x49e06929, 0x8ec9c844,0x8ec9c844 + data4 0x75c2896a,0x75c2896a, 0xf48e7978,0xf48e7978 + data4 0x99583e6b,0x99583e6b, 0x27b971dd,0x27b971dd + data4 0xbee14fb6,0xbee14fb6, 0xf088ad17,0xf088ad17 + data4 0xc920ac66,0xc920ac66, 0x7dce3ab4,0x7dce3ab4 + data4 0x63df4a18,0x63df4a18, 0xe51a3182,0xe51a3182 + data4 0x97513360,0x97513360, 0x62537f45,0x62537f45 + data4 0xb16477e0,0xb16477e0, 0xbb6bae84,0xbb6bae84 + data4 0xfe81a01c,0xfe81a01c, 0xf9082b94,0xf9082b94 + data4 0x70486858,0x70486858, 0x8f45fd19,0x8f45fd19 + data4 0x94de6c87,0x94de6c87, 0x527bf8b7,0x527bf8b7 + data4 0xab73d323,0xab73d323, 0x724b02e2,0x724b02e2 + data4 0xe31f8f57,0xe31f8f57, 0x6655ab2a,0x6655ab2a + data4 0xb2eb2807,0xb2eb2807, 0x2fb5c203,0x2fb5c203 + data4 0x86c57b9a,0x86c57b9a, 0xd33708a5,0xd33708a5 + data4 0x302887f2,0x302887f2, 0x23bfa5b2,0x23bfa5b2 + data4 0x02036aba,0x02036aba, 0xed16825c,0xed16825c + data4 0x8acf1c2b,0x8acf1c2b, 0xa779b492,0xa779b492 + data4 0xf307f2f0,0xf307f2f0, 0x4e69e2a1,0x4e69e2a1 + data4 0x65daf4cd,0x65daf4cd, 0x0605bed5,0x0605bed5 + data4 0xd134621f,0xd134621f, 0xc4a6fe8a,0xc4a6fe8a + data4 0x342e539d,0x342e539d, 0xa2f355a0,0xa2f355a0 + data4 0x058ae132,0x058ae132, 0xa4f6eb75,0xa4f6eb75 + data4 0x0b83ec39,0x0b83ec39, 0x4060efaa,0x4060efaa + data4 0x5e719f06,0x5e719f06, 0xbd6e1051,0xbd6e1051 + data4 0x3e218af9,0x3e218af9, 0x96dd063d,0x96dd063d + data4 0xdd3e05ae,0xdd3e05ae, 0x4de6bd46,0x4de6bd46 + data4 0x91548db5,0x91548db5, 0x71c45d05,0x71c45d05 + data4 0x0406d46f,0x0406d46f, 0x605015ff,0x605015ff + data4 0x1998fb24,0x1998fb24, 0xd6bde997,0xd6bde997 + data4 0x894043cc,0x894043cc, 0x67d99e77,0x67d99e77 + data4 0xb0e842bd,0xb0e842bd, 0x07898b88,0x07898b88 + data4 0xe7195b38,0xe7195b38, 0x79c8eedb,0x79c8eedb + data4 0xa17c0a47,0xa17c0a47, 0x7c420fe9,0x7c420fe9 + data4 0xf8841ec9,0xf8841ec9, 0x00000000,0x00000000 + data4 0x09808683,0x09808683, 0x322bed48,0x322bed48 + data4 0x1e1170ac,0x1e1170ac, 0x6c5a724e,0x6c5a724e + data4 0xfd0efffb,0xfd0efffb, 0x0f853856,0x0f853856 + data4 0x3daed51e,0x3daed51e, 0x362d3927,0x362d3927 + data4 0x0a0fd964,0x0a0fd964, 0x685ca621,0x685ca621 + data4 0x9b5b54d1,0x9b5b54d1, 0x24362e3a,0x24362e3a + data4 0x0c0a67b1,0x0c0a67b1, 0x9357e70f,0x9357e70f + data4 0xb4ee96d2,0xb4ee96d2, 0x1b9b919e,0x1b9b919e + data4 0x80c0c54f,0x80c0c54f, 0x61dc20a2,0x61dc20a2 + data4 0x5a774b69,0x5a774b69, 0x1c121a16,0x1c121a16 + data4 0xe293ba0a,0xe293ba0a, 0xc0a02ae5,0xc0a02ae5 + data4 0x3c22e043,0x3c22e043, 0x121b171d,0x121b171d + data4 0x0e090d0b,0x0e090d0b, 0xf28bc7ad,0xf28bc7ad + data4 0x2db6a8b9,0x2db6a8b9, 0x141ea9c8,0x141ea9c8 + data4 0x57f11985,0x57f11985, 0xaf75074c,0xaf75074c + data4 0xee99ddbb,0xee99ddbb, 0xa37f60fd,0xa37f60fd + data4 0xf701269f,0xf701269f, 0x5c72f5bc,0x5c72f5bc + data4 0x44663bc5,0x44663bc5, 0x5bfb7e34,0x5bfb7e34 + data4 0x8b432976,0x8b432976, 0xcb23c6dc,0xcb23c6dc + data4 0xb6edfc68,0xb6edfc68, 0xb8e4f163,0xb8e4f163 + data4 0xd731dcca,0xd731dcca, 0x42638510,0x42638510 + data4 0x13972240,0x13972240, 0x84c61120,0x84c61120 + data4 0x854a247d,0x854a247d, 0xd2bb3df8,0xd2bb3df8 + data4 0xaef93211,0xaef93211, 0xc729a16d,0xc729a16d + data4 0x1d9e2f4b,0x1d9e2f4b, 0xdcb230f3,0xdcb230f3 + data4 0x0d8652ec,0x0d8652ec, 0x77c1e3d0,0x77c1e3d0 + data4 0x2bb3166c,0x2bb3166c, 0xa970b999,0xa970b999 + data4 0x119448fa,0x119448fa, 0x47e96422,0x47e96422 + data4 0xa8fc8cc4,0xa8fc8cc4, 0xa0f03f1a,0xa0f03f1a + data4 0x567d2cd8,0x567d2cd8, 0x223390ef,0x223390ef + data4 0x87494ec7,0x87494ec7, 0xd938d1c1,0xd938d1c1 + data4 0x8ccaa2fe,0x8ccaa2fe, 0x98d40b36,0x98d40b36 + data4 0xa6f581cf,0xa6f581cf, 0xa57ade28,0xa57ade28 + data4 0xdab78e26,0xdab78e26, 0x3fadbfa4,0x3fadbfa4 + data4 0x2c3a9de4,0x2c3a9de4, 0x5078920d,0x5078920d + data4 0x6a5fcc9b,0x6a5fcc9b, 0x547e4662,0x547e4662 + data4 0xf68d13c2,0xf68d13c2, 0x90d8b8e8,0x90d8b8e8 + data4 0x2e39f75e,0x2e39f75e, 0x82c3aff5,0x82c3aff5 + data4 0x9f5d80be,0x9f5d80be, 0x69d0937c,0x69d0937c + data4 0x6fd52da9,0x6fd52da9, 0xcf2512b3,0xcf2512b3 + data4 0xc8ac993b,0xc8ac993b, 0x10187da7,0x10187da7 + data4 0xe89c636e,0xe89c636e, 0xdb3bbb7b,0xdb3bbb7b + data4 0xcd267809,0xcd267809, 0x6e5918f4,0x6e5918f4 + data4 0xec9ab701,0xec9ab701, 0x834f9aa8,0x834f9aa8 + data4 0xe6956e65,0xe6956e65, 0xaaffe67e,0xaaffe67e + data4 0x21bccf08,0x21bccf08, 0xef15e8e6,0xef15e8e6 + data4 0xbae79bd9,0xbae79bd9, 0x4a6f36ce,0x4a6f36ce + data4 0xea9f09d4,0xea9f09d4, 0x29b07cd6,0x29b07cd6 + data4 0x31a4b2af,0x31a4b2af, 0x2a3f2331,0x2a3f2331 + data4 0xc6a59430,0xc6a59430, 0x35a266c0,0x35a266c0 + data4 0x744ebc37,0x744ebc37, 0xfc82caa6,0xfc82caa6 + data4 0xe090d0b0,0xe090d0b0, 0x33a7d815,0x33a7d815 + data4 0xf104984a,0xf104984a, 0x41ecdaf7,0x41ecdaf7 + data4 0x7fcd500e,0x7fcd500e, 0x1791f62f,0x1791f62f + data4 0x764dd68d,0x764dd68d, 0x43efb04d,0x43efb04d + data4 0xccaa4d54,0xccaa4d54, 0xe49604df,0xe49604df + data4 0x9ed1b5e3,0x9ed1b5e3, 0x4c6a881b,0x4c6a881b + data4 0xc12c1fb8,0xc12c1fb8, 0x4665517f,0x4665517f + data4 0x9d5eea04,0x9d5eea04, 0x018c355d,0x018c355d + data4 0xfa877473,0xfa877473, 0xfb0b412e,0xfb0b412e + data4 0xb3671d5a,0xb3671d5a, 0x92dbd252,0x92dbd252 + data4 0xe9105633,0xe9105633, 0x6dd64713,0x6dd64713 + data4 0x9ad7618c,0x9ad7618c, 0x37a10c7a,0x37a10c7a + data4 0x59f8148e,0x59f8148e, 0xeb133c89,0xeb133c89 + data4 0xcea927ee,0xcea927ee, 0xb761c935,0xb761c935 + data4 0xe11ce5ed,0xe11ce5ed, 0x7a47b13c,0x7a47b13c + data4 0x9cd2df59,0x9cd2df59, 0x55f2733f,0x55f2733f + data4 0x1814ce79,0x1814ce79, 0x73c737bf,0x73c737bf + data4 0x53f7cdea,0x53f7cdea, 0x5ffdaa5b,0x5ffdaa5b + data4 0xdf3d6f14,0xdf3d6f14, 0x7844db86,0x7844db86 + data4 0xcaaff381,0xcaaff381, 0xb968c43e,0xb968c43e + data4 0x3824342c,0x3824342c, 0xc2a3405f,0xc2a3405f + data4 0x161dc372,0x161dc372, 0xbce2250c,0xbce2250c + data4 0x283c498b,0x283c498b, 0xff0d9541,0xff0d9541 + data4 0x39a80171,0x39a80171, 0x080cb3de,0x080cb3de + data4 0xd8b4e49c,0xd8b4e49c, 0x6456c190,0x6456c190 + data4 0x7bcb8461,0x7bcb8461, 0xd532b670,0xd532b670 + data4 0x486c5c74,0x486c5c74, 0xd0b85742,0xd0b85742 +// Td4: + data1 0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38 + data1 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb + data1 0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87 + data1 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb + data1 0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d + data1 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e + data1 0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2 + data1 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25 + data1 0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16 + data1 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92 + data1 0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda + data1 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84 + data1 0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a + data1 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06 + data1 0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02 + data1 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b + data1 0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea + data1 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73 + data1 0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85 + data1 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e + data1 0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89 + data1 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b + data1 0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20 + data1 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4 + data1 0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31 + data1 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f + data1 0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d + data1 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef + data1 0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0 + data1 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61 + data1 0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26 + data1 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d +.size AES_Td#,2048+256 // HP-UX assembler fails to ".-AES_Td#" diff --git a/openssl/crypto/aes/asm/aes-ppc.pl b/openssl/crypto/aes/asm/aes-ppc.pl new file mode 100644 index 000000000..ce427655e --- /dev/null +++ b/openssl/crypto/aes/asm/aes-ppc.pl @@ -0,0 +1,1176 @@ +#!/usr/bin/env perl + +# ==================================================================== +# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# project. The module is, however, dual licensed under OpenSSL and +# CRYPTOGAMS licenses depending on where you obtain it. For further +# details see http://www.openssl.org/~appro/cryptogams/. +# ==================================================================== + +# Needs more work: key setup, page boundaries, CBC routine... +# +# ppc_AES_[en|de]crypt perform at 18 cycles per byte processed with +# 128-bit key, which is ~40% better than 64-bit code generated by gcc +# 4.0. But these are not the ones currently used! Their "compact" +# counterparts are, for security reason. ppc_AES_encrypt_compact runs +# at 1/2 of ppc_AES_encrypt speed, while ppc_AES_decrypt_compact - +# at 1/3 of ppc_AES_decrypt. + +$flavour = shift; + +if ($flavour =~ /64/) { + $SIZE_T =8; + $STU ="stdu"; + $POP ="ld"; + $PUSH ="std"; +} elsif ($flavour =~ /32/) { + $SIZE_T =4; + $STU ="stwu"; + $POP ="lwz"; + $PUSH ="stw"; +} else { die "nonsense $flavour"; } + +$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; +( $xlate="${dir}ppc-xlate.pl" and -f $xlate ) or +( $xlate="${dir}../../perlasm/ppc-xlate.pl" and -f $xlate) or +die "can't locate ppc-xlate.pl"; + +open STDOUT,"| $^X $xlate $flavour ".shift || die "can't call $xlate: $!"; + +$FRAME=32*$SIZE_T; + +sub _data_word() +{ my $i; + while(defined($i=shift)) { $code.=sprintf"\t.long\t0x%08x,0x%08x\n",$i,$i; } +} + +$sp="r1"; +$toc="r2"; +$inp="r3"; +$out="r4"; +$key="r5"; + +$Tbl0="r3"; +$Tbl1="r6"; +$Tbl2="r7"; +$Tbl3="r2"; + +$s0="r8"; +$s1="r9"; +$s2="r10"; +$s3="r11"; + +$t0="r12"; +$t1="r13"; +$t2="r14"; +$t3="r15"; + +$acc00="r16"; +$acc01="r17"; +$acc02="r18"; +$acc03="r19"; + +$acc04="r20"; +$acc05="r21"; +$acc06="r22"; +$acc07="r23"; + +$acc08="r24"; +$acc09="r25"; +$acc10="r26"; +$acc11="r27"; + +$acc12="r28"; +$acc13="r29"; +$acc14="r30"; +$acc15="r31"; + +# stay away from TLS pointer +if ($SIZE_T==8) { die if ($t1 ne "r13"); $t1="r0"; } +else { die if ($Tbl3 ne "r2"); $Tbl3=$t0; $t0="r0"; } +$mask80=$Tbl2; +$mask1b=$Tbl3; + +$code.=<<___; +.machine "any" +.text + +.align 7 +LAES_Te: + mflr r0 + bcl 20,31,\$+4 + mflr $Tbl0 ; vvvvv "distance" between . and 1st data entry + addi $Tbl0,$Tbl0,`128-8` + mtlr r0 + blr + .space `32-24` +LAES_Td: + mflr r0 + bcl 20,31,\$+4 + mflr $Tbl0 ; vvvvvvvv "distance" between . and 1st data entry + addi $Tbl0,$Tbl0,`128-8-32+2048+256` + mtlr r0 + blr + .space `128-32-24` +___ +&_data_word( + 0xc66363a5, 0xf87c7c84, 0xee777799, 0xf67b7b8d, + 0xfff2f20d, 0xd66b6bbd, 0xde6f6fb1, 0x91c5c554, + 0x60303050, 0x02010103, 0xce6767a9, 0x562b2b7d, + 0xe7fefe19, 0xb5d7d762, 0x4dababe6, 0xec76769a, + 0x8fcaca45, 0x1f82829d, 0x89c9c940, 0xfa7d7d87, + 0xeffafa15, 0xb25959eb, 0x8e4747c9, 0xfbf0f00b, + 0x41adadec, 0xb3d4d467, 0x5fa2a2fd, 0x45afafea, + 0x239c9cbf, 0x53a4a4f7, 0xe4727296, 0x9bc0c05b, + 0x75b7b7c2, 0xe1fdfd1c, 0x3d9393ae, 0x4c26266a, + 0x6c36365a, 0x7e3f3f41, 0xf5f7f702, 0x83cccc4f, + 0x6834345c, 0x51a5a5f4, 0xd1e5e534, 0xf9f1f108, + 0xe2717193, 0xabd8d873, 0x62313153, 0x2a15153f, + 0x0804040c, 0x95c7c752, 0x46232365, 0x9dc3c35e, + 0x30181828, 0x379696a1, 0x0a05050f, 0x2f9a9ab5, + 0x0e070709, 0x24121236, 0x1b80809b, 0xdfe2e23d, + 0xcdebeb26, 0x4e272769, 0x7fb2b2cd, 0xea75759f, + 0x1209091b, 0x1d83839e, 0x582c2c74, 0x341a1a2e, + 0x361b1b2d, 0xdc6e6eb2, 0xb45a5aee, 0x5ba0a0fb, + 0xa45252f6, 0x763b3b4d, 0xb7d6d661, 0x7db3b3ce, + 0x5229297b, 0xdde3e33e, 0x5e2f2f71, 0x13848497, + 0xa65353f5, 0xb9d1d168, 0x00000000, 0xc1eded2c, + 0x40202060, 0xe3fcfc1f, 0x79b1b1c8, 0xb65b5bed, + 0xd46a6abe, 0x8dcbcb46, 0x67bebed9, 0x7239394b, + 0x944a4ade, 0x984c4cd4, 0xb05858e8, 0x85cfcf4a, + 0xbbd0d06b, 0xc5efef2a, 0x4faaaae5, 0xedfbfb16, + 0x864343c5, 0x9a4d4dd7, 0x66333355, 0x11858594, + 0x8a4545cf, 0xe9f9f910, 0x04020206, 0xfe7f7f81, + 0xa05050f0, 0x783c3c44, 0x259f9fba, 0x4ba8a8e3, + 0xa25151f3, 0x5da3a3fe, 0x804040c0, 0x058f8f8a, + 0x3f9292ad, 0x219d9dbc, 0x70383848, 0xf1f5f504, + 0x63bcbcdf, 0x77b6b6c1, 0xafdada75, 0x42212163, + 0x20101030, 0xe5ffff1a, 0xfdf3f30e, 0xbfd2d26d, + 0x81cdcd4c, 0x180c0c14, 0x26131335, 0xc3ecec2f, + 0xbe5f5fe1, 0x359797a2, 0x884444cc, 0x2e171739, + 0x93c4c457, 0x55a7a7f2, 0xfc7e7e82, 0x7a3d3d47, + 0xc86464ac, 0xba5d5de7, 0x3219192b, 0xe6737395, + 0xc06060a0, 0x19818198, 0x9e4f4fd1, 0xa3dcdc7f, + 0x44222266, 0x542a2a7e, 0x3b9090ab, 0x0b888883, + 0x8c4646ca, 0xc7eeee29, 0x6bb8b8d3, 0x2814143c, + 0xa7dede79, 0xbc5e5ee2, 0x160b0b1d, 0xaddbdb76, + 0xdbe0e03b, 0x64323256, 0x743a3a4e, 0x140a0a1e, + 0x924949db, 0x0c06060a, 0x4824246c, 0xb85c5ce4, + 0x9fc2c25d, 0xbdd3d36e, 0x43acacef, 0xc46262a6, + 0x399191a8, 0x319595a4, 0xd3e4e437, 0xf279798b, + 0xd5e7e732, 0x8bc8c843, 0x6e373759, 0xda6d6db7, + 0x018d8d8c, 0xb1d5d564, 0x9c4e4ed2, 0x49a9a9e0, + 0xd86c6cb4, 0xac5656fa, 0xf3f4f407, 0xcfeaea25, + 0xca6565af, 0xf47a7a8e, 0x47aeaee9, 0x10080818, + 0x6fbabad5, 0xf0787888, 0x4a25256f, 0x5c2e2e72, + 0x381c1c24, 0x57a6a6f1, 0x73b4b4c7, 0x97c6c651, + 0xcbe8e823, 0xa1dddd7c, 0xe874749c, 0x3e1f1f21, + 0x964b4bdd, 0x61bdbddc, 0x0d8b8b86, 0x0f8a8a85, + 0xe0707090, 0x7c3e3e42, 0x71b5b5c4, 0xcc6666aa, + 0x904848d8, 0x06030305, 0xf7f6f601, 0x1c0e0e12, + 0xc26161a3, 0x6a35355f, 0xae5757f9, 0x69b9b9d0, + 0x17868691, 0x99c1c158, 0x3a1d1d27, 0x279e9eb9, + 0xd9e1e138, 0xebf8f813, 0x2b9898b3, 0x22111133, + 0xd26969bb, 0xa9d9d970, 0x078e8e89, 0x339494a7, + 0x2d9b9bb6, 0x3c1e1e22, 0x15878792, 0xc9e9e920, + 0x87cece49, 0xaa5555ff, 0x50282878, 0xa5dfdf7a, + 0x038c8c8f, 0x59a1a1f8, 0x09898980, 0x1a0d0d17, + 0x65bfbfda, 0xd7e6e631, 0x844242c6, 0xd06868b8, + 0x824141c3, 0x299999b0, 0x5a2d2d77, 0x1e0f0f11, + 0x7bb0b0cb, 0xa85454fc, 0x6dbbbbd6, 0x2c16163a); +$code.=<<___; +.byte 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5 +.byte 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76 +.byte 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0 +.byte 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0 +.byte 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc +.byte 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15 +.byte 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a +.byte 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75 +.byte 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0 +.byte 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84 +.byte 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b +.byte 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf +.byte 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85 +.byte 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8 +.byte 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5 +.byte 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2 +.byte 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17 +.byte 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73 +.byte 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88 +.byte 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb +.byte 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c +.byte 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79 +.byte 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9 +.byte 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08 +.byte 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6 +.byte 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a +.byte 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e +.byte 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e +.byte 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94 +.byte 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf +.byte 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68 +.byte 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16 +___ +&_data_word( + 0x51f4a750, 0x7e416553, 0x1a17a4c3, 0x3a275e96, + 0x3bab6bcb, 0x1f9d45f1, 0xacfa58ab, 0x4be30393, + 0x2030fa55, 0xad766df6, 0x88cc7691, 0xf5024c25, + 0x4fe5d7fc, 0xc52acbd7, 0x26354480, 0xb562a38f, + 0xdeb15a49, 0x25ba1b67, 0x45ea0e98, 0x5dfec0e1, + 0xc32f7502, 0x814cf012, 0x8d4697a3, 0x6bd3f9c6, + 0x038f5fe7, 0x15929c95, 0xbf6d7aeb, 0x955259da, + 0xd4be832d, 0x587421d3, 0x49e06929, 0x8ec9c844, + 0x75c2896a, 0xf48e7978, 0x99583e6b, 0x27b971dd, + 0xbee14fb6, 0xf088ad17, 0xc920ac66, 0x7dce3ab4, + 0x63df4a18, 0xe51a3182, 0x97513360, 0x62537f45, + 0xb16477e0, 0xbb6bae84, 0xfe81a01c, 0xf9082b94, + 0x70486858, 0x8f45fd19, 0x94de6c87, 0x527bf8b7, + 0xab73d323, 0x724b02e2, 0xe31f8f57, 0x6655ab2a, + 0xb2eb2807, 0x2fb5c203, 0x86c57b9a, 0xd33708a5, + 0x302887f2, 0x23bfa5b2, 0x02036aba, 0xed16825c, + 0x8acf1c2b, 0xa779b492, 0xf307f2f0, 0x4e69e2a1, + 0x65daf4cd, 0x0605bed5, 0xd134621f, 0xc4a6fe8a, + 0x342e539d, 0xa2f355a0, 0x058ae132, 0xa4f6eb75, + 0x0b83ec39, 0x4060efaa, 0x5e719f06, 0xbd6e1051, + 0x3e218af9, 0x96dd063d, 0xdd3e05ae, 0x4de6bd46, + 0x91548db5, 0x71c45d05, 0x0406d46f, 0x605015ff, + 0x1998fb24, 0xd6bde997, 0x894043cc, 0x67d99e77, + 0xb0e842bd, 0x07898b88, 0xe7195b38, 0x79c8eedb, + 0xa17c0a47, 0x7c420fe9, 0xf8841ec9, 0x00000000, + 0x09808683, 0x322bed48, 0x1e1170ac, 0x6c5a724e, + 0xfd0efffb, 0x0f853856, 0x3daed51e, 0x362d3927, + 0x0a0fd964, 0x685ca621, 0x9b5b54d1, 0x24362e3a, + 0x0c0a67b1, 0x9357e70f, 0xb4ee96d2, 0x1b9b919e, + 0x80c0c54f, 0x61dc20a2, 0x5a774b69, 0x1c121a16, + 0xe293ba0a, 0xc0a02ae5, 0x3c22e043, 0x121b171d, + 0x0e090d0b, 0xf28bc7ad, 0x2db6a8b9, 0x141ea9c8, + 0x57f11985, 0xaf75074c, 0xee99ddbb, 0xa37f60fd, + 0xf701269f, 0x5c72f5bc, 0x44663bc5, 0x5bfb7e34, + 0x8b432976, 0xcb23c6dc, 0xb6edfc68, 0xb8e4f163, + 0xd731dcca, 0x42638510, 0x13972240, 0x84c61120, + 0x854a247d, 0xd2bb3df8, 0xaef93211, 0xc729a16d, + 0x1d9e2f4b, 0xdcb230f3, 0x0d8652ec, 0x77c1e3d0, + 0x2bb3166c, 0xa970b999, 0x119448fa, 0x47e96422, + 0xa8fc8cc4, 0xa0f03f1a, 0x567d2cd8, 0x223390ef, + 0x87494ec7, 0xd938d1c1, 0x8ccaa2fe, 0x98d40b36, + 0xa6f581cf, 0xa57ade28, 0xdab78e26, 0x3fadbfa4, + 0x2c3a9de4, 0x5078920d, 0x6a5fcc9b, 0x547e4662, + 0xf68d13c2, 0x90d8b8e8, 0x2e39f75e, 0x82c3aff5, + 0x9f5d80be, 0x69d0937c, 0x6fd52da9, 0xcf2512b3, + 0xc8ac993b, 0x10187da7, 0xe89c636e, 0xdb3bbb7b, + 0xcd267809, 0x6e5918f4, 0xec9ab701, 0x834f9aa8, + 0xe6956e65, 0xaaffe67e, 0x21bccf08, 0xef15e8e6, + 0xbae79bd9, 0x4a6f36ce, 0xea9f09d4, 0x29b07cd6, + 0x31a4b2af, 0x2a3f2331, 0xc6a59430, 0x35a266c0, + 0x744ebc37, 0xfc82caa6, 0xe090d0b0, 0x33a7d815, + 0xf104984a, 0x41ecdaf7, 0x7fcd500e, 0x1791f62f, + 0x764dd68d, 0x43efb04d, 0xccaa4d54, 0xe49604df, + 0x9ed1b5e3, 0x4c6a881b, 0xc12c1fb8, 0x4665517f, + 0x9d5eea04, 0x018c355d, 0xfa877473, 0xfb0b412e, + 0xb3671d5a, 0x92dbd252, 0xe9105633, 0x6dd64713, + 0x9ad7618c, 0x37a10c7a, 0x59f8148e, 0xeb133c89, + 0xcea927ee, 0xb761c935, 0xe11ce5ed, 0x7a47b13c, + 0x9cd2df59, 0x55f2733f, 0x1814ce79, 0x73c737bf, + 0x53f7cdea, 0x5ffdaa5b, 0xdf3d6f14, 0x7844db86, + 0xcaaff381, 0xb968c43e, 0x3824342c, 0xc2a3405f, + 0x161dc372, 0xbce2250c, 0x283c498b, 0xff0d9541, + 0x39a80171, 0x080cb3de, 0xd8b4e49c, 0x6456c190, + 0x7bcb8461, 0xd532b670, 0x486c5c74, 0xd0b85742); +$code.=<<___; +.byte 0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38 +.byte 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb +.byte 0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87 +.byte 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb +.byte 0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d +.byte 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e +.byte 0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2 +.byte 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25 +.byte 0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16 +.byte 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92 +.byte 0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda +.byte 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84 +.byte 0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a +.byte 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06 +.byte 0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02 +.byte 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b +.byte 0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea +.byte 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73 +.byte 0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85 +.byte 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e +.byte 0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89 +.byte 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b +.byte 0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20 +.byte 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4 +.byte 0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31 +.byte 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f +.byte 0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d +.byte 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef +.byte 0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0 +.byte 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61 +.byte 0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26 +.byte 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d + + +.globl .AES_encrypt +.align 7 +.AES_encrypt: + mflr r0 + $STU $sp,-$FRAME($sp) + + $PUSH r0,`$FRAME-$SIZE_T*21`($sp) + $PUSH $toc,`$FRAME-$SIZE_T*20`($sp) + $PUSH r13,`$FRAME-$SIZE_T*19`($sp) + $PUSH r14,`$FRAME-$SIZE_T*18`($sp) + $PUSH r15,`$FRAME-$SIZE_T*17`($sp) + $PUSH r16,`$FRAME-$SIZE_T*16`($sp) + $PUSH r17,`$FRAME-$SIZE_T*15`($sp) + $PUSH r18,`$FRAME-$SIZE_T*14`($sp) + $PUSH r19,`$FRAME-$SIZE_T*13`($sp) + $PUSH r20,`$FRAME-$SIZE_T*12`($sp) + $PUSH r21,`$FRAME-$SIZE_T*11`($sp) + $PUSH r22,`$FRAME-$SIZE_T*10`($sp) + $PUSH r23,`$FRAME-$SIZE_T*9`($sp) + $PUSH r24,`$FRAME-$SIZE_T*8`($sp) + $PUSH r25,`$FRAME-$SIZE_T*7`($sp) + $PUSH r26,`$FRAME-$SIZE_T*6`($sp) + $PUSH r27,`$FRAME-$SIZE_T*5`($sp) + $PUSH r28,`$FRAME-$SIZE_T*4`($sp) + $PUSH r29,`$FRAME-$SIZE_T*3`($sp) + $PUSH r30,`$FRAME-$SIZE_T*2`($sp) + $PUSH r31,`$FRAME-$SIZE_T*1`($sp) + + lwz $s0,0($inp) + lwz $s1,4($inp) + lwz $s2,8($inp) + lwz $s3,12($inp) + bl LAES_Te + bl Lppc_AES_encrypt_compact + stw $s0,0($out) + stw $s1,4($out) + stw $s2,8($out) + stw $s3,12($out) + + $POP r0,`$FRAME-$SIZE_T*21`($sp) + $POP $toc,`$FRAME-$SIZE_T*20`($sp) + $POP r13,`$FRAME-$SIZE_T*19`($sp) + $POP r14,`$FRAME-$SIZE_T*18`($sp) + $POP r15,`$FRAME-$SIZE_T*17`($sp) + $POP r16,`$FRAME-$SIZE_T*16`($sp) + $POP r17,`$FRAME-$SIZE_T*15`($sp) + $POP r18,`$FRAME-$SIZE_T*14`($sp) + $POP r19,`$FRAME-$SIZE_T*13`($sp) + $POP r20,`$FRAME-$SIZE_T*12`($sp) + $POP r21,`$FRAME-$SIZE_T*11`($sp) + $POP r22,`$FRAME-$SIZE_T*10`($sp) + $POP r23,`$FRAME-$SIZE_T*9`($sp) + $POP r24,`$FRAME-$SIZE_T*8`($sp) + $POP r25,`$FRAME-$SIZE_T*7`($sp) + $POP r26,`$FRAME-$SIZE_T*6`($sp) + $POP r27,`$FRAME-$SIZE_T*5`($sp) + $POP r28,`$FRAME-$SIZE_T*4`($sp) + $POP r29,`$FRAME-$SIZE_T*3`($sp) + $POP r30,`$FRAME-$SIZE_T*2`($sp) + $POP r31,`$FRAME-$SIZE_T*1`($sp) + mtlr r0 + addi $sp,$sp,$FRAME + blr + +.align 4 +Lppc_AES_encrypt: + lwz $acc00,240($key) + lwz $t0,0($key) + lwz $t1,4($key) + lwz $t2,8($key) + lwz $t3,12($key) + addi $Tbl1,$Tbl0,3 + addi $Tbl2,$Tbl0,2 + addi $Tbl3,$Tbl0,1 + addi $acc00,$acc00,-1 + addi $key,$key,16 + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + mtctr $acc00 +.align 4 +Lenc_loop: + rlwinm $acc00,$s0,`32-24+3`,21,28 + rlwinm $acc01,$s1,`32-24+3`,21,28 + lwz $t0,0($key) + lwz $t1,4($key) + rlwinm $acc02,$s2,`32-24+3`,21,28 + rlwinm $acc03,$s3,`32-24+3`,21,28 + lwz $t2,8($key) + lwz $t3,12($key) + rlwinm $acc04,$s1,`32-16+3`,21,28 + rlwinm $acc05,$s2,`32-16+3`,21,28 + lwzx $acc00,$Tbl0,$acc00 + lwzx $acc01,$Tbl0,$acc01 + rlwinm $acc06,$s3,`32-16+3`,21,28 + rlwinm $acc07,$s0,`32-16+3`,21,28 + lwzx $acc02,$Tbl0,$acc02 + lwzx $acc03,$Tbl0,$acc03 + rlwinm $acc08,$s2,`32-8+3`,21,28 + rlwinm $acc09,$s3,`32-8+3`,21,28 + lwzx $acc04,$Tbl1,$acc04 + lwzx $acc05,$Tbl1,$acc05 + rlwinm $acc10,$s0,`32-8+3`,21,28 + rlwinm $acc11,$s1,`32-8+3`,21,28 + lwzx $acc06,$Tbl1,$acc06 + lwzx $acc07,$Tbl1,$acc07 + rlwinm $acc12,$s3,`0+3`,21,28 + rlwinm $acc13,$s0,`0+3`,21,28 + lwzx $acc08,$Tbl2,$acc08 + lwzx $acc09,$Tbl2,$acc09 + rlwinm $acc14,$s1,`0+3`,21,28 + rlwinm $acc15,$s2,`0+3`,21,28 + lwzx $acc10,$Tbl2,$acc10 + lwzx $acc11,$Tbl2,$acc11 + xor $t0,$t0,$acc00 + xor $t1,$t1,$acc01 + lwzx $acc12,$Tbl3,$acc12 + lwzx $acc13,$Tbl3,$acc13 + xor $t2,$t2,$acc02 + xor $t3,$t3,$acc03 + lwzx $acc14,$Tbl3,$acc14 + lwzx $acc15,$Tbl3,$acc15 + xor $t0,$t0,$acc04 + xor $t1,$t1,$acc05 + xor $t2,$t2,$acc06 + xor $t3,$t3,$acc07 + xor $t0,$t0,$acc08 + xor $t1,$t1,$acc09 + xor $t2,$t2,$acc10 + xor $t3,$t3,$acc11 + xor $s0,$t0,$acc12 + xor $s1,$t1,$acc13 + xor $s2,$t2,$acc14 + xor $s3,$t3,$acc15 + addi $key,$key,16 + bdnz- Lenc_loop + + addi $Tbl2,$Tbl0,2048 + nop + lwz $acc08,`2048+0`($Tbl0) ! prefetch Te4 + lwz $acc09,`2048+32`($Tbl0) + lwz $acc10,`2048+64`($Tbl0) + lwz $acc11,`2048+96`($Tbl0) + lwz $acc08,`2048+128`($Tbl0) + lwz $acc09,`2048+160`($Tbl0) + lwz $acc10,`2048+192`($Tbl0) + lwz $acc11,`2048+224`($Tbl0) + rlwinm $acc00,$s0,`32-24`,24,31 + rlwinm $acc01,$s1,`32-24`,24,31 + lwz $t0,0($key) + lwz $t1,4($key) + rlwinm $acc02,$s2,`32-24`,24,31 + rlwinm $acc03,$s3,`32-24`,24,31 + lwz $t2,8($key) + lwz $t3,12($key) + rlwinm $acc04,$s1,`32-16`,24,31 + rlwinm $acc05,$s2,`32-16`,24,31 + lbzx $acc00,$Tbl2,$acc00 + lbzx $acc01,$Tbl2,$acc01 + rlwinm $acc06,$s3,`32-16`,24,31 + rlwinm $acc07,$s0,`32-16`,24,31 + lbzx $acc02,$Tbl2,$acc02 + lbzx $acc03,$Tbl2,$acc03 + rlwinm $acc08,$s2,`32-8`,24,31 + rlwinm $acc09,$s3,`32-8`,24,31 + lbzx $acc04,$Tbl2,$acc04 + lbzx $acc05,$Tbl2,$acc05 + rlwinm $acc10,$s0,`32-8`,24,31 + rlwinm $acc11,$s1,`32-8`,24,31 + lbzx $acc06,$Tbl2,$acc06 + lbzx $acc07,$Tbl2,$acc07 + rlwinm $acc12,$s3,`0`,24,31 + rlwinm $acc13,$s0,`0`,24,31 + lbzx $acc08,$Tbl2,$acc08 + lbzx $acc09,$Tbl2,$acc09 + rlwinm $acc14,$s1,`0`,24,31 + rlwinm $acc15,$s2,`0`,24,31 + lbzx $acc10,$Tbl2,$acc10 + lbzx $acc11,$Tbl2,$acc11 + rlwinm $s0,$acc00,24,0,7 + rlwinm $s1,$acc01,24,0,7 + lbzx $acc12,$Tbl2,$acc12 + lbzx $acc13,$Tbl2,$acc13 + rlwinm $s2,$acc02,24,0,7 + rlwinm $s3,$acc03,24,0,7 + lbzx $acc14,$Tbl2,$acc14 + lbzx $acc15,$Tbl2,$acc15 + rlwimi $s0,$acc04,16,8,15 + rlwimi $s1,$acc05,16,8,15 + rlwimi $s2,$acc06,16,8,15 + rlwimi $s3,$acc07,16,8,15 + rlwimi $s0,$acc08,8,16,23 + rlwimi $s1,$acc09,8,16,23 + rlwimi $s2,$acc10,8,16,23 + rlwimi $s3,$acc11,8,16,23 + or $s0,$s0,$acc12 + or $s1,$s1,$acc13 + or $s2,$s2,$acc14 + or $s3,$s3,$acc15 + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + blr + +.align 4 +Lppc_AES_encrypt_compact: + lwz $acc00,240($key) + lwz $t0,0($key) + lwz $t1,4($key) + lwz $t2,8($key) + lwz $t3,12($key) + addi $Tbl1,$Tbl0,2048 + lis $mask80,0x8080 + lis $mask1b,0x1b1b + addi $key,$key,16 + ori $mask80,$mask80,0x8080 + ori $mask1b,$mask1b,0x1b1b + mtctr $acc00 +.align 4 +Lenc_compact_loop: + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + rlwinm $acc00,$s0,`32-24`,24,31 + rlwinm $acc01,$s1,`32-24`,24,31 + rlwinm $acc02,$s2,`32-24`,24,31 + rlwinm $acc03,$s3,`32-24`,24,31 + lbzx $acc00,$Tbl1,$acc00 + lbzx $acc01,$Tbl1,$acc01 + rlwinm $acc04,$s1,`32-16`,24,31 + rlwinm $acc05,$s2,`32-16`,24,31 + lbzx $acc02,$Tbl1,$acc02 + lbzx $acc03,$Tbl1,$acc03 + rlwinm $acc06,$s3,`32-16`,24,31 + rlwinm $acc07,$s0,`32-16`,24,31 + lbzx $acc04,$Tbl1,$acc04 + lbzx $acc05,$Tbl1,$acc05 + rlwinm $acc08,$s2,`32-8`,24,31 + rlwinm $acc09,$s3,`32-8`,24,31 + lbzx $acc06,$Tbl1,$acc06 + lbzx $acc07,$Tbl1,$acc07 + rlwinm $acc10,$s0,`32-8`,24,31 + rlwinm $acc11,$s1,`32-8`,24,31 + lbzx $acc08,$Tbl1,$acc08 + lbzx $acc09,$Tbl1,$acc09 + rlwinm $acc12,$s3,`0`,24,31 + rlwinm $acc13,$s0,`0`,24,31 + lbzx $acc10,$Tbl1,$acc10 + lbzx $acc11,$Tbl1,$acc11 + rlwinm $acc14,$s1,`0`,24,31 + rlwinm $acc15,$s2,`0`,24,31 + lbzx $acc12,$Tbl1,$acc12 + lbzx $acc13,$Tbl1,$acc13 + rlwinm $s0,$acc00,24,0,7 + rlwinm $s1,$acc01,24,0,7 + lbzx $acc14,$Tbl1,$acc14 + lbzx $acc15,$Tbl1,$acc15 + rlwinm $s2,$acc02,24,0,7 + rlwinm $s3,$acc03,24,0,7 + rlwimi $s0,$acc04,16,8,15 + rlwimi $s1,$acc05,16,8,15 + rlwimi $s2,$acc06,16,8,15 + rlwimi $s3,$acc07,16,8,15 + rlwimi $s0,$acc08,8,16,23 + rlwimi $s1,$acc09,8,16,23 + rlwimi $s2,$acc10,8,16,23 + rlwimi $s3,$acc11,8,16,23 + lwz $t0,0($key) + lwz $t1,4($key) + or $s0,$s0,$acc12 + or $s1,$s1,$acc13 + lwz $t2,8($key) + lwz $t3,12($key) + or $s2,$s2,$acc14 + or $s3,$s3,$acc15 + + addi $key,$key,16 + bdz Lenc_compact_done + + and $acc00,$s0,$mask80 # r1=r0&0x80808080 + and $acc01,$s1,$mask80 + and $acc02,$s2,$mask80 + and $acc03,$s3,$mask80 + srwi $acc04,$acc00,7 # r1>>7 + srwi $acc05,$acc01,7 + srwi $acc06,$acc02,7 + srwi $acc07,$acc03,7 + andc $acc08,$s0,$mask80 # r0&0x7f7f7f7f + andc $acc09,$s1,$mask80 + andc $acc10,$s2,$mask80 + andc $acc11,$s3,$mask80 + sub $acc00,$acc00,$acc04 # r1-(r1>>7) + sub $acc01,$acc01,$acc05 + sub $acc02,$acc02,$acc06 + sub $acc03,$acc03,$acc07 + add $acc08,$acc08,$acc08 # (r0&0x7f7f7f7f)<<1 + add $acc09,$acc09,$acc09 + add $acc10,$acc10,$acc10 + add $acc11,$acc11,$acc11 + and $acc00,$acc00,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc01,$acc01,$mask1b + and $acc02,$acc02,$mask1b + and $acc03,$acc03,$mask1b + xor $acc00,$acc00,$acc08 # r2 + xor $acc01,$acc01,$acc09 + xor $acc02,$acc02,$acc10 + xor $acc03,$acc03,$acc11 + + rotlwi $acc12,$s0,16 # ROTATE(r0,16) + rotlwi $acc13,$s1,16 + rotlwi $acc14,$s2,16 + rotlwi $acc15,$s3,16 + xor $s0,$s0,$acc00 # r0^r2 + xor $s1,$s1,$acc01 + xor $s2,$s2,$acc02 + xor $s3,$s3,$acc03 + rotrwi $s0,$s0,24 # ROTATE(r2^r0,24) + rotrwi $s1,$s1,24 + rotrwi $s2,$s2,24 + rotrwi $s3,$s3,24 + xor $s0,$s0,$acc00 # ROTATE(r2^r0,24)^r2 + xor $s1,$s1,$acc01 + xor $s2,$s2,$acc02 + xor $s3,$s3,$acc03 + rotlwi $acc08,$acc12,8 # ROTATE(r0,24) + rotlwi $acc09,$acc13,8 + rotlwi $acc10,$acc14,8 + rotlwi $acc11,$acc15,8 + xor $s0,$s0,$acc12 # + xor $s1,$s1,$acc13 + xor $s2,$s2,$acc14 + xor $s3,$s3,$acc15 + xor $s0,$s0,$acc08 # + xor $s1,$s1,$acc09 + xor $s2,$s2,$acc10 + xor $s3,$s3,$acc11 + + b Lenc_compact_loop +.align 4 +Lenc_compact_done: + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + blr + +.globl .AES_decrypt +.align 7 +.AES_decrypt: + mflr r0 + $STU $sp,-$FRAME($sp) + + $PUSH r0,`$FRAME-$SIZE_T*21`($sp) + $PUSH $toc,`$FRAME-$SIZE_T*20`($sp) + $PUSH r13,`$FRAME-$SIZE_T*19`($sp) + $PUSH r14,`$FRAME-$SIZE_T*18`($sp) + $PUSH r15,`$FRAME-$SIZE_T*17`($sp) + $PUSH r16,`$FRAME-$SIZE_T*16`($sp) + $PUSH r17,`$FRAME-$SIZE_T*15`($sp) + $PUSH r18,`$FRAME-$SIZE_T*14`($sp) + $PUSH r19,`$FRAME-$SIZE_T*13`($sp) + $PUSH r20,`$FRAME-$SIZE_T*12`($sp) + $PUSH r21,`$FRAME-$SIZE_T*11`($sp) + $PUSH r22,`$FRAME-$SIZE_T*10`($sp) + $PUSH r23,`$FRAME-$SIZE_T*9`($sp) + $PUSH r24,`$FRAME-$SIZE_T*8`($sp) + $PUSH r25,`$FRAME-$SIZE_T*7`($sp) + $PUSH r26,`$FRAME-$SIZE_T*6`($sp) + $PUSH r27,`$FRAME-$SIZE_T*5`($sp) + $PUSH r28,`$FRAME-$SIZE_T*4`($sp) + $PUSH r29,`$FRAME-$SIZE_T*3`($sp) + $PUSH r30,`$FRAME-$SIZE_T*2`($sp) + $PUSH r31,`$FRAME-$SIZE_T*1`($sp) + + lwz $s0,0($inp) + lwz $s1,4($inp) + lwz $s2,8($inp) + lwz $s3,12($inp) + bl LAES_Td + bl Lppc_AES_decrypt_compact + stw $s0,0($out) + stw $s1,4($out) + stw $s2,8($out) + stw $s3,12($out) + + $POP r0,`$FRAME-$SIZE_T*21`($sp) + $POP $toc,`$FRAME-$SIZE_T*20`($sp) + $POP r13,`$FRAME-$SIZE_T*19`($sp) + $POP r14,`$FRAME-$SIZE_T*18`($sp) + $POP r15,`$FRAME-$SIZE_T*17`($sp) + $POP r16,`$FRAME-$SIZE_T*16`($sp) + $POP r17,`$FRAME-$SIZE_T*15`($sp) + $POP r18,`$FRAME-$SIZE_T*14`($sp) + $POP r19,`$FRAME-$SIZE_T*13`($sp) + $POP r20,`$FRAME-$SIZE_T*12`($sp) + $POP r21,`$FRAME-$SIZE_T*11`($sp) + $POP r22,`$FRAME-$SIZE_T*10`($sp) + $POP r23,`$FRAME-$SIZE_T*9`($sp) + $POP r24,`$FRAME-$SIZE_T*8`($sp) + $POP r25,`$FRAME-$SIZE_T*7`($sp) + $POP r26,`$FRAME-$SIZE_T*6`($sp) + $POP r27,`$FRAME-$SIZE_T*5`($sp) + $POP r28,`$FRAME-$SIZE_T*4`($sp) + $POP r29,`$FRAME-$SIZE_T*3`($sp) + $POP r30,`$FRAME-$SIZE_T*2`($sp) + $POP r31,`$FRAME-$SIZE_T*1`($sp) + mtlr r0 + addi $sp,$sp,$FRAME + blr + +.align 4 +Lppc_AES_decrypt: + lwz $acc00,240($key) + lwz $t0,0($key) + lwz $t1,4($key) + lwz $t2,8($key) + lwz $t3,12($key) + addi $Tbl1,$Tbl0,3 + addi $Tbl2,$Tbl0,2 + addi $Tbl3,$Tbl0,1 + addi $acc00,$acc00,-1 + addi $key,$key,16 + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + mtctr $acc00 +.align 4 +Ldec_loop: + rlwinm $acc00,$s0,`32-24+3`,21,28 + rlwinm $acc01,$s1,`32-24+3`,21,28 + lwz $t0,0($key) + lwz $t1,4($key) + rlwinm $acc02,$s2,`32-24+3`,21,28 + rlwinm $acc03,$s3,`32-24+3`,21,28 + lwz $t2,8($key) + lwz $t3,12($key) + rlwinm $acc04,$s3,`32-16+3`,21,28 + rlwinm $acc05,$s0,`32-16+3`,21,28 + lwzx $acc00,$Tbl0,$acc00 + lwzx $acc01,$Tbl0,$acc01 + rlwinm $acc06,$s1,`32-16+3`,21,28 + rlwinm $acc07,$s2,`32-16+3`,21,28 + lwzx $acc02,$Tbl0,$acc02 + lwzx $acc03,$Tbl0,$acc03 + rlwinm $acc08,$s2,`32-8+3`,21,28 + rlwinm $acc09,$s3,`32-8+3`,21,28 + lwzx $acc04,$Tbl1,$acc04 + lwzx $acc05,$Tbl1,$acc05 + rlwinm $acc10,$s0,`32-8+3`,21,28 + rlwinm $acc11,$s1,`32-8+3`,21,28 + lwzx $acc06,$Tbl1,$acc06 + lwzx $acc07,$Tbl1,$acc07 + rlwinm $acc12,$s1,`0+3`,21,28 + rlwinm $acc13,$s2,`0+3`,21,28 + lwzx $acc08,$Tbl2,$acc08 + lwzx $acc09,$Tbl2,$acc09 + rlwinm $acc14,$s3,`0+3`,21,28 + rlwinm $acc15,$s0,`0+3`,21,28 + lwzx $acc10,$Tbl2,$acc10 + lwzx $acc11,$Tbl2,$acc11 + xor $t0,$t0,$acc00 + xor $t1,$t1,$acc01 + lwzx $acc12,$Tbl3,$acc12 + lwzx $acc13,$Tbl3,$acc13 + xor $t2,$t2,$acc02 + xor $t3,$t3,$acc03 + lwzx $acc14,$Tbl3,$acc14 + lwzx $acc15,$Tbl3,$acc15 + xor $t0,$t0,$acc04 + xor $t1,$t1,$acc05 + xor $t2,$t2,$acc06 + xor $t3,$t3,$acc07 + xor $t0,$t0,$acc08 + xor $t1,$t1,$acc09 + xor $t2,$t2,$acc10 + xor $t3,$t3,$acc11 + xor $s0,$t0,$acc12 + xor $s1,$t1,$acc13 + xor $s2,$t2,$acc14 + xor $s3,$t3,$acc15 + addi $key,$key,16 + bdnz- Ldec_loop + + addi $Tbl2,$Tbl0,2048 + nop + lwz $acc08,`2048+0`($Tbl0) ! prefetch Td4 + lwz $acc09,`2048+32`($Tbl0) + lwz $acc10,`2048+64`($Tbl0) + lwz $acc11,`2048+96`($Tbl0) + lwz $acc08,`2048+128`($Tbl0) + lwz $acc09,`2048+160`($Tbl0) + lwz $acc10,`2048+192`($Tbl0) + lwz $acc11,`2048+224`($Tbl0) + rlwinm $acc00,$s0,`32-24`,24,31 + rlwinm $acc01,$s1,`32-24`,24,31 + lwz $t0,0($key) + lwz $t1,4($key) + rlwinm $acc02,$s2,`32-24`,24,31 + rlwinm $acc03,$s3,`32-24`,24,31 + lwz $t2,8($key) + lwz $t3,12($key) + rlwinm $acc04,$s3,`32-16`,24,31 + rlwinm $acc05,$s0,`32-16`,24,31 + lbzx $acc00,$Tbl2,$acc00 + lbzx $acc01,$Tbl2,$acc01 + rlwinm $acc06,$s1,`32-16`,24,31 + rlwinm $acc07,$s2,`32-16`,24,31 + lbzx $acc02,$Tbl2,$acc02 + lbzx $acc03,$Tbl2,$acc03 + rlwinm $acc08,$s2,`32-8`,24,31 + rlwinm $acc09,$s3,`32-8`,24,31 + lbzx $acc04,$Tbl2,$acc04 + lbzx $acc05,$Tbl2,$acc05 + rlwinm $acc10,$s0,`32-8`,24,31 + rlwinm $acc11,$s1,`32-8`,24,31 + lbzx $acc06,$Tbl2,$acc06 + lbzx $acc07,$Tbl2,$acc07 + rlwinm $acc12,$s1,`0`,24,31 + rlwinm $acc13,$s2,`0`,24,31 + lbzx $acc08,$Tbl2,$acc08 + lbzx $acc09,$Tbl2,$acc09 + rlwinm $acc14,$s3,`0`,24,31 + rlwinm $acc15,$s0,`0`,24,31 + lbzx $acc10,$Tbl2,$acc10 + lbzx $acc11,$Tbl2,$acc11 + rlwinm $s0,$acc00,24,0,7 + rlwinm $s1,$acc01,24,0,7 + lbzx $acc12,$Tbl2,$acc12 + lbzx $acc13,$Tbl2,$acc13 + rlwinm $s2,$acc02,24,0,7 + rlwinm $s3,$acc03,24,0,7 + lbzx $acc14,$Tbl2,$acc14 + lbzx $acc15,$Tbl2,$acc15 + rlwimi $s0,$acc04,16,8,15 + rlwimi $s1,$acc05,16,8,15 + rlwimi $s2,$acc06,16,8,15 + rlwimi $s3,$acc07,16,8,15 + rlwimi $s0,$acc08,8,16,23 + rlwimi $s1,$acc09,8,16,23 + rlwimi $s2,$acc10,8,16,23 + rlwimi $s3,$acc11,8,16,23 + or $s0,$s0,$acc12 + or $s1,$s1,$acc13 + or $s2,$s2,$acc14 + or $s3,$s3,$acc15 + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + blr + +.align 4 +Lppc_AES_decrypt_compact: + lwz $acc00,240($key) + lwz $t0,0($key) + lwz $t1,4($key) + lwz $t2,8($key) + lwz $t3,12($key) + addi $Tbl1,$Tbl0,2048 + lis $mask80,0x8080 + lis $mask1b,0x1b1b + addi $key,$key,16 + ori $mask80,$mask80,0x8080 + ori $mask1b,$mask1b,0x1b1b +___ +$code.=<<___ if ($SIZE_T==8); + insrdi $mask80,$mask80,32,0 + insrdi $mask1b,$mask1b,32,0 +___ +$code.=<<___; + mtctr $acc00 +.align 4 +Ldec_compact_loop: + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + rlwinm $acc00,$s0,`32-24`,24,31 + rlwinm $acc01,$s1,`32-24`,24,31 + rlwinm $acc02,$s2,`32-24`,24,31 + rlwinm $acc03,$s3,`32-24`,24,31 + lbzx $acc00,$Tbl1,$acc00 + lbzx $acc01,$Tbl1,$acc01 + rlwinm $acc04,$s3,`32-16`,24,31 + rlwinm $acc05,$s0,`32-16`,24,31 + lbzx $acc02,$Tbl1,$acc02 + lbzx $acc03,$Tbl1,$acc03 + rlwinm $acc06,$s1,`32-16`,24,31 + rlwinm $acc07,$s2,`32-16`,24,31 + lbzx $acc04,$Tbl1,$acc04 + lbzx $acc05,$Tbl1,$acc05 + rlwinm $acc08,$s2,`32-8`,24,31 + rlwinm $acc09,$s3,`32-8`,24,31 + lbzx $acc06,$Tbl1,$acc06 + lbzx $acc07,$Tbl1,$acc07 + rlwinm $acc10,$s0,`32-8`,24,31 + rlwinm $acc11,$s1,`32-8`,24,31 + lbzx $acc08,$Tbl1,$acc08 + lbzx $acc09,$Tbl1,$acc09 + rlwinm $acc12,$s1,`0`,24,31 + rlwinm $acc13,$s2,`0`,24,31 + lbzx $acc10,$Tbl1,$acc10 + lbzx $acc11,$Tbl1,$acc11 + rlwinm $acc14,$s3,`0`,24,31 + rlwinm $acc15,$s0,`0`,24,31 + lbzx $acc12,$Tbl1,$acc12 + lbzx $acc13,$Tbl1,$acc13 + rlwinm $s0,$acc00,24,0,7 + rlwinm $s1,$acc01,24,0,7 + lbzx $acc14,$Tbl1,$acc14 + lbzx $acc15,$Tbl1,$acc15 + rlwinm $s2,$acc02,24,0,7 + rlwinm $s3,$acc03,24,0,7 + rlwimi $s0,$acc04,16,8,15 + rlwimi $s1,$acc05,16,8,15 + rlwimi $s2,$acc06,16,8,15 + rlwimi $s3,$acc07,16,8,15 + rlwimi $s0,$acc08,8,16,23 + rlwimi $s1,$acc09,8,16,23 + rlwimi $s2,$acc10,8,16,23 + rlwimi $s3,$acc11,8,16,23 + lwz $t0,0($key) + lwz $t1,4($key) + or $s0,$s0,$acc12 + or $s1,$s1,$acc13 + lwz $t2,8($key) + lwz $t3,12($key) + or $s2,$s2,$acc14 + or $s3,$s3,$acc15 + + addi $key,$key,16 + bdz Ldec_compact_done +___ +$code.=<<___ if ($SIZE_T==8); + # vectorized permutation improves decrypt performance by 10% + insrdi $s0,$s1,32,0 + insrdi $s2,$s3,32,0 + + and $acc00,$s0,$mask80 # r1=r0&0x80808080 + and $acc02,$s2,$mask80 + srdi $acc04,$acc00,7 # r1>>7 + srdi $acc06,$acc02,7 + andc $acc08,$s0,$mask80 # r0&0x7f7f7f7f + andc $acc10,$s2,$mask80 + sub $acc00,$acc00,$acc04 # r1-(r1>>7) + sub $acc02,$acc02,$acc06 + add $acc08,$acc08,$acc08 # (r0&0x7f7f7f7f)<<1 + add $acc10,$acc10,$acc10 + and $acc00,$acc00,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc02,$acc02,$mask1b + xor $acc00,$acc00,$acc08 # r2 + xor $acc02,$acc02,$acc10 + + and $acc04,$acc00,$mask80 # r1=r2&0x80808080 + and $acc06,$acc02,$mask80 + srdi $acc08,$acc04,7 # r1>>7 + srdi $acc10,$acc06,7 + andc $acc12,$acc00,$mask80 # r2&0x7f7f7f7f + andc $acc14,$acc02,$mask80 + sub $acc04,$acc04,$acc08 # r1-(r1>>7) + sub $acc06,$acc06,$acc10 + add $acc12,$acc12,$acc12 # (r2&0x7f7f7f7f)<<1 + add $acc14,$acc14,$acc14 + and $acc04,$acc04,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc06,$acc06,$mask1b + xor $acc04,$acc04,$acc12 # r4 + xor $acc06,$acc06,$acc14 + + and $acc08,$acc04,$mask80 # r1=r4&0x80808080 + and $acc10,$acc06,$mask80 + srdi $acc12,$acc08,7 # r1>>7 + srdi $acc14,$acc10,7 + sub $acc08,$acc08,$acc12 # r1-(r1>>7) + sub $acc10,$acc10,$acc14 + andc $acc12,$acc04,$mask80 # r4&0x7f7f7f7f + andc $acc14,$acc06,$mask80 + add $acc12,$acc12,$acc12 # (r4&0x7f7f7f7f)<<1 + add $acc14,$acc14,$acc14 + and $acc08,$acc08,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc10,$acc10,$mask1b + xor $acc08,$acc08,$acc12 # r8 + xor $acc10,$acc10,$acc14 + + xor $acc00,$acc00,$s0 # r2^r0 + xor $acc02,$acc02,$s2 + xor $acc04,$acc04,$s0 # r4^r0 + xor $acc06,$acc06,$s2 + + extrdi $acc01,$acc00,32,0 + extrdi $acc03,$acc02,32,0 + extrdi $acc05,$acc04,32,0 + extrdi $acc07,$acc06,32,0 + extrdi $acc09,$acc08,32,0 + extrdi $acc11,$acc10,32,0 +___ +$code.=<<___ if ($SIZE_T==4); + and $acc00,$s0,$mask80 # r1=r0&0x80808080 + and $acc01,$s1,$mask80 + and $acc02,$s2,$mask80 + and $acc03,$s3,$mask80 + srwi $acc04,$acc00,7 # r1>>7 + srwi $acc05,$acc01,7 + srwi $acc06,$acc02,7 + srwi $acc07,$acc03,7 + andc $acc08,$s0,$mask80 # r0&0x7f7f7f7f + andc $acc09,$s1,$mask80 + andc $acc10,$s2,$mask80 + andc $acc11,$s3,$mask80 + sub $acc00,$acc00,$acc04 # r1-(r1>>7) + sub $acc01,$acc01,$acc05 + sub $acc02,$acc02,$acc06 + sub $acc03,$acc03,$acc07 + add $acc08,$acc08,$acc08 # (r0&0x7f7f7f7f)<<1 + add $acc09,$acc09,$acc09 + add $acc10,$acc10,$acc10 + add $acc11,$acc11,$acc11 + and $acc00,$acc00,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc01,$acc01,$mask1b + and $acc02,$acc02,$mask1b + and $acc03,$acc03,$mask1b + xor $acc00,$acc00,$acc08 # r2 + xor $acc01,$acc01,$acc09 + xor $acc02,$acc02,$acc10 + xor $acc03,$acc03,$acc11 + + and $acc04,$acc00,$mask80 # r1=r2&0x80808080 + and $acc05,$acc01,$mask80 + and $acc06,$acc02,$mask80 + and $acc07,$acc03,$mask80 + srwi $acc08,$acc04,7 # r1>>7 + srwi $acc09,$acc05,7 + srwi $acc10,$acc06,7 + srwi $acc11,$acc07,7 + andc $acc12,$acc00,$mask80 # r2&0x7f7f7f7f + andc $acc13,$acc01,$mask80 + andc $acc14,$acc02,$mask80 + andc $acc15,$acc03,$mask80 + sub $acc04,$acc04,$acc08 # r1-(r1>>7) + sub $acc05,$acc05,$acc09 + sub $acc06,$acc06,$acc10 + sub $acc07,$acc07,$acc11 + add $acc12,$acc12,$acc12 # (r2&0x7f7f7f7f)<<1 + add $acc13,$acc13,$acc13 + add $acc14,$acc14,$acc14 + add $acc15,$acc15,$acc15 + and $acc04,$acc04,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc05,$acc05,$mask1b + and $acc06,$acc06,$mask1b + and $acc07,$acc07,$mask1b + xor $acc04,$acc04,$acc12 # r4 + xor $acc05,$acc05,$acc13 + xor $acc06,$acc06,$acc14 + xor $acc07,$acc07,$acc15 + + and $acc08,$acc04,$mask80 # r1=r4&0x80808080 + and $acc09,$acc05,$mask80 + and $acc10,$acc06,$mask80 + and $acc11,$acc07,$mask80 + srwi $acc12,$acc08,7 # r1>>7 + srwi $acc13,$acc09,7 + srwi $acc14,$acc10,7 + srwi $acc15,$acc11,7 + sub $acc08,$acc08,$acc12 # r1-(r1>>7) + sub $acc09,$acc09,$acc13 + sub $acc10,$acc10,$acc14 + sub $acc11,$acc11,$acc15 + andc $acc12,$acc04,$mask80 # r4&0x7f7f7f7f + andc $acc13,$acc05,$mask80 + andc $acc14,$acc06,$mask80 + andc $acc15,$acc07,$mask80 + add $acc12,$acc12,$acc12 # (r4&0x7f7f7f7f)<<1 + add $acc13,$acc13,$acc13 + add $acc14,$acc14,$acc14 + add $acc15,$acc15,$acc15 + and $acc08,$acc08,$mask1b # (r1-(r1>>7))&0x1b1b1b1b + and $acc09,$acc09,$mask1b + and $acc10,$acc10,$mask1b + and $acc11,$acc11,$mask1b + xor $acc08,$acc08,$acc12 # r8 + xor $acc09,$acc09,$acc13 + xor $acc10,$acc10,$acc14 + xor $acc11,$acc11,$acc15 + + xor $acc00,$acc00,$s0 # r2^r0 + xor $acc01,$acc01,$s1 + xor $acc02,$acc02,$s2 + xor $acc03,$acc03,$s3 + xor $acc04,$acc04,$s0 # r4^r0 + xor $acc05,$acc05,$s1 + xor $acc06,$acc06,$s2 + xor $acc07,$acc07,$s3 +___ +$code.=<<___; + rotrwi $s0,$s0,8 # = ROTATE(r0,8) + rotrwi $s1,$s1,8 + rotrwi $s2,$s2,8 + rotrwi $s3,$s3,8 + xor $s0,$s0,$acc00 # ^= r2^r0 + xor $s1,$s1,$acc01 + xor $s2,$s2,$acc02 + xor $s3,$s3,$acc03 + xor $acc00,$acc00,$acc08 + xor $acc01,$acc01,$acc09 + xor $acc02,$acc02,$acc10 + xor $acc03,$acc03,$acc11 + xor $s0,$s0,$acc04 # ^= r4^r0 + xor $s1,$s1,$acc05 + xor $s2,$s2,$acc06 + xor $s3,$s3,$acc07 + rotrwi $acc00,$acc00,24 + rotrwi $acc01,$acc01,24 + rotrwi $acc02,$acc02,24 + rotrwi $acc03,$acc03,24 + xor $acc04,$acc04,$acc08 + xor $acc05,$acc05,$acc09 + xor $acc06,$acc06,$acc10 + xor $acc07,$acc07,$acc11 + xor $s0,$s0,$acc08 # ^= r8 [^((r4^r0)^(r2^r0)=r4^r2)] + xor $s1,$s1,$acc09 + xor $s2,$s2,$acc10 + xor $s3,$s3,$acc11 + rotrwi $acc04,$acc04,16 + rotrwi $acc05,$acc05,16 + rotrwi $acc06,$acc06,16 + rotrwi $acc07,$acc07,16 + xor $s0,$s0,$acc00 # ^= ROTATE(r8^r2^r0,24) + xor $s1,$s1,$acc01 + xor $s2,$s2,$acc02 + xor $s3,$s3,$acc03 + rotrwi $acc08,$acc08,8 + rotrwi $acc09,$acc09,8 + rotrwi $acc10,$acc10,8 + rotrwi $acc11,$acc11,8 + xor $s0,$s0,$acc04 # ^= ROTATE(r8^r4^r0,16) + xor $s1,$s1,$acc05 + xor $s2,$s2,$acc06 + xor $s3,$s3,$acc07 + xor $s0,$s0,$acc08 # ^= ROTATE(r8,8) + xor $s1,$s1,$acc09 + xor $s2,$s2,$acc10 + xor $s3,$s3,$acc11 + + b Ldec_compact_loop +.align 4 +Ldec_compact_done: + xor $s0,$s0,$t0 + xor $s1,$s1,$t1 + xor $s2,$s2,$t2 + xor $s3,$s3,$t3 + blr +.long 0 +.asciz "AES for PPC, CRYPTOGAMS by <appro\@openssl.org>" +.align 7 +___ + +$code =~ s/\`([^\`]*)\`/eval $1/gem; +print $code; +close STDOUT; diff --git a/openssl/crypto/aes/asm/aes-s390x.pl b/openssl/crypto/aes/asm/aes-s390x.pl new file mode 100644 index 000000000..4b27afd92 --- /dev/null +++ b/openssl/crypto/aes/asm/aes-s390x.pl @@ -0,0 +1,1333 @@ +#!/usr/bin/env perl + +# ==================================================================== +# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# project. The module is, however, dual licensed under OpenSSL and +# CRYPTOGAMS licenses depending on where you obtain it. For further +# details see http://www.openssl.org/~appro/cryptogams/. +# ==================================================================== + +# AES for s390x. + +# April 2007. +# +# Software performance improvement over gcc-generated code is ~70% and +# in absolute terms is ~73 cycles per byte processed with 128-bit key. +# You're likely to exclaim "why so slow?" Keep in mind that z-CPUs are +# *strictly* in-order execution and issued instruction [in this case +# load value from memory is critical] has to complete before execution +# flow proceeds. S-boxes are compressed to 2KB[+256B]. +# +# As for hardware acceleration support. It's basically a "teaser," as +# it can and should be improved in several ways. Most notably support +# for CBC is not utilized, nor multiple blocks are ever processed. +# Then software key schedule can be postponed till hardware support +# detection... Performance improvement over assembler is reportedly +# ~2.5x, but can reach >8x [naturally on larger chunks] if proper +# support is implemented. + +# May 2007. +# +# Implement AES_set_[en|de]crypt_key. Key schedule setup is avoided +# for 128-bit keys, if hardware support is detected. + +# Januray 2009. +# +# Add support for hardware AES192/256 and reschedule instructions to +# minimize/avoid Address Generation Interlock hazard and to favour +# dual-issue z10 pipeline. This gave ~25% improvement on z10 and +# almost 50% on z9. The gain is smaller on z10, because being dual- +# issue z10 makes it improssible to eliminate the interlock condition: +# critial path is not long enough. Yet it spends ~24 cycles per byte +# processed with 128-bit key. +# +# Unlike previous version hardware support detection takes place only +# at the moment of key schedule setup, which is denoted in key->rounds. +# This is done, because deferred key setup can't be made MT-safe, not +# for key lengthes longer than 128 bits. +# +# Add AES_cbc_encrypt, which gives incredible performance improvement, +# it was measured to be ~6.6x. It's less than previously mentioned 8x, +# because software implementation was optimized. + +$softonly=0; # allow hardware support + +$t0="%r0"; $mask="%r0"; +$t1="%r1"; +$t2="%r2"; $inp="%r2"; +$t3="%r3"; $out="%r3"; $bits="%r3"; +$key="%r4"; +$i1="%r5"; +$i2="%r6"; +$i3="%r7"; +$s0="%r8"; +$s1="%r9"; +$s2="%r10"; +$s3="%r11"; +$tbl="%r12"; +$rounds="%r13"; +$ra="%r14"; +$sp="%r15"; + +sub _data_word() +{ my $i; + while(defined($i=shift)) { $code.=sprintf".long\t0x%08x,0x%08x\n",$i,$i; } +} + +$code=<<___; +.text + +.type AES_Te,\@object +.align 256 +AES_Te: +___ +&_data_word( + 0xc66363a5, 0xf87c7c84, 0xee777799, 0xf67b7b8d, + 0xfff2f20d, 0xd66b6bbd, 0xde6f6fb1, 0x91c5c554, + 0x60303050, 0x02010103, 0xce6767a9, 0x562b2b7d, + 0xe7fefe19, 0xb5d7d762, 0x4dababe6, 0xec76769a, + 0x8fcaca45, 0x1f82829d, 0x89c9c940, 0xfa7d7d87, + 0xeffafa15, 0xb25959eb, 0x8e4747c9, 0xfbf0f00b, + 0x41adadec, 0xb3d4d467, 0x5fa2a2fd, 0x45afafea, + 0x239c9cbf, 0x53a4a4f7, 0xe4727296, 0x9bc0c05b, + 0x75b7b7c2, 0xe1fdfd1c, 0x3d9393ae, 0x4c26266a, + 0x6c36365a, 0x7e3f3f41, 0xf5f7f702, 0x83cccc4f, + 0x6834345c, 0x51a5a5f4, 0xd1e5e534, 0xf9f1f108, + 0xe2717193, 0xabd8d873, 0x62313153, 0x2a15153f, + 0x0804040c, 0x95c7c752, 0x46232365, 0x9dc3c35e, + 0x30181828, 0x379696a1, 0x0a05050f, 0x2f9a9ab5, + 0x0e070709, 0x24121236, 0x1b80809b, 0xdfe2e23d, + 0xcdebeb26, 0x4e272769, 0x7fb2b2cd, 0xea75759f, + 0x1209091b, 0x1d83839e, 0x582c2c74, 0x341a1a2e, + 0x361b1b2d, 0xdc6e6eb2, 0xb45a5aee, 0x5ba0a0fb, + 0xa45252f6, 0x763b3b4d, 0xb7d6d661, 0x7db3b3ce, + 0x5229297b, 0xdde3e33e, 0x5e2f2f71, 0x13848497, + 0xa65353f5, 0xb9d1d168, 0x00000000, 0xc1eded2c, + 0x40202060, 0xe3fcfc1f, 0x79b1b1c8, 0xb65b5bed, + 0xd46a6abe, 0x8dcbcb46, 0x67bebed9, 0x7239394b, + 0x944a4ade, 0x984c4cd4, 0xb05858e8, 0x85cfcf4a, + 0xbbd0d06b, 0xc5efef2a, 0x4faaaae5, 0xedfbfb16, + 0x864343c5, 0x9a4d4dd7, 0x66333355, 0x11858594, + 0x8a4545cf, 0xe9f9f910, 0x04020206, 0xfe7f7f81, + 0xa05050f0, 0x783c3c44, 0x259f9fba, 0x4ba8a8e3, + 0xa25151f3, 0x5da3a3fe, 0x804040c0, 0x058f8f8a, + 0x3f9292ad, 0x219d9dbc, 0x70383848, 0xf1f5f504, + 0x63bcbcdf, 0x77b6b6c1, 0xafdada75, 0x42212163, + 0x20101030, 0xe5ffff1a, 0xfdf3f30e, 0xbfd2d26d, + 0x81cdcd4c, 0x180c0c14, 0x26131335, 0xc3ecec2f, + 0xbe5f5fe1, 0x359797a2, 0x884444cc, 0x2e171739, + 0x93c4c457, 0x55a7a7f2, 0xfc7e7e82, 0x7a3d3d47, + 0xc86464ac, 0xba5d5de7, 0x3219192b, 0xe6737395, + 0xc06060a0, 0x19818198, 0x9e4f4fd1, 0xa3dcdc7f, + 0x44222266, 0x542a2a7e, 0x3b9090ab, 0x0b888883, + 0x8c4646ca, 0xc7eeee29, 0x6bb8b8d3, 0x2814143c, + 0xa7dede79, 0xbc5e5ee2, 0x160b0b1d, 0xaddbdb76, + 0xdbe0e03b, 0x64323256, 0x743a3a4e, 0x140a0a1e, + 0x924949db, 0x0c06060a, 0x4824246c, 0xb85c5ce4, + 0x9fc2c25d, 0xbdd3d36e, 0x43acacef, 0xc46262a6, + 0x399191a8, 0x319595a4, 0xd3e4e437, 0xf279798b, + 0xd5e7e732, 0x8bc8c843, 0x6e373759, 0xda6d6db7, + 0x018d8d8c, 0xb1d5d564, 0x9c4e4ed2, 0x49a9a9e0, + 0xd86c6cb4, 0xac5656fa, 0xf3f4f407, 0xcfeaea25, + 0xca6565af, 0xf47a7a8e, 0x47aeaee9, 0x10080818, + 0x6fbabad5, 0xf0787888, 0x4a25256f, 0x5c2e2e72, + 0x381c1c24, 0x57a6a6f1, 0x73b4b4c7, 0x97c6c651, + 0xcbe8e823, 0xa1dddd7c, 0xe874749c, 0x3e1f1f21, + 0x964b4bdd, 0x61bdbddc, 0x0d8b8b86, 0x0f8a8a85, + 0xe0707090, 0x7c3e3e42, 0x71b5b5c4, 0xcc6666aa, + 0x904848d8, 0x06030305, 0xf7f6f601, 0x1c0e0e12, + 0xc26161a3, 0x6a35355f, 0xae5757f9, 0x69b9b9d0, + 0x17868691, 0x99c1c158, 0x3a1d1d27, 0x279e9eb9, + 0xd9e1e138, 0xebf8f813, 0x2b9898b3, 0x22111133, + 0xd26969bb, 0xa9d9d970, 0x078e8e89, 0x339494a7, + 0x2d9b9bb6, 0x3c1e1e22, 0x15878792, 0xc9e9e920, + 0x87cece49, 0xaa5555ff, 0x50282878, 0xa5dfdf7a, + 0x038c8c8f, 0x59a1a1f8, 0x09898980, 0x1a0d0d17, + 0x65bfbfda, 0xd7e6e631, 0x844242c6, 0xd06868b8, + 0x824141c3, 0x299999b0, 0x5a2d2d77, 0x1e0f0f11, + 0x7bb0b0cb, 0xa85454fc, 0x6dbbbbd6, 0x2c16163a); +$code.=<<___; +# Te4[256] +.byte 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5 +.byte 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76 +.byte 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0 +.byte 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0 +.byte 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc +.byte 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15 +.byte 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a +.byte 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75 +.byte 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0 +.byte 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84 +.byte 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b +.byte 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf +.byte 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85 +.byte 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8 +.byte 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5 +.byte 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2 +.byte 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17 +.byte 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73 +.byte 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88 +.byte 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb +.byte 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c +.byte 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79 +.byte 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9 +.byte 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08 +.byte 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6 +.byte 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a +.byte 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e +.byte 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e +.byte 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94 +.byte 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf +.byte 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68 +.byte 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16 +# rcon[] +.long 0x01000000, 0x02000000, 0x04000000, 0x08000000 +.long 0x10000000, 0x20000000, 0x40000000, 0x80000000 +.long 0x1B000000, 0x36000000, 0, 0, 0, 0, 0, 0 +.align 256 +.size AES_Te,.-AES_Te + +# void AES_encrypt(const unsigned char *inp, unsigned char *out, +# const AES_KEY *key) { +.globl AES_encrypt +.type AES_encrypt,\@function +AES_encrypt: +___ +$code.=<<___ if (!$softonly); + l %r0,240($key) + lhi %r1,16 + clr %r0,%r1 + jl .Lesoft + + la %r1,0($key) + #la %r2,0($inp) + la %r4,0($out) + lghi %r3,16 # single block length + .long 0xb92e0042 # km %r4,%r2 + brc 1,.-4 # can this happen? + br %r14 +.align 64 +.Lesoft: +___ +$code.=<<___; + stmg %r3,$ra,24($sp) + + llgf $s0,0($inp) + llgf $s1,4($inp) + llgf $s2,8($inp) + llgf $s3,12($inp) + + larl $tbl,AES_Te + bras $ra,_s390x_AES_encrypt + + lg $out,24($sp) + st $s0,0($out) + st $s1,4($out) + st $s2,8($out) + st $s3,12($out) + + lmg %r6,$ra,48($sp) + br $ra +.size AES_encrypt,.-AES_encrypt + +.type _s390x_AES_encrypt,\@function +.align 16 +_s390x_AES_encrypt: + stg $ra,152($sp) + x $s0,0($key) + x $s1,4($key) + x $s2,8($key) + x $s3,12($key) + l $rounds,240($key) + llill $mask,`0xff<<3` + aghi $rounds,-1 + j .Lenc_loop +.align 16 +.Lenc_loop: + sllg $t1,$s0,`0+3` + srlg $t2,$s0,`8-3` + srlg $t3,$s0,`16-3` + srl $s0,`24-3` + nr $s0,$mask + ngr $t1,$mask + nr $t2,$mask + nr $t3,$mask + + srlg $i1,$s1,`16-3` # i0 + sllg $i2,$s1,`0+3` + srlg $i3,$s1,`8-3` + srl $s1,`24-3` + nr $i1,$mask + nr $s1,$mask + ngr $i2,$mask + nr $i3,$mask + + l $s0,0($s0,$tbl) # Te0[s0>>24] + l $t1,1($t1,$tbl) # Te3[s0>>0] + l $t2,2($t2,$tbl) # Te2[s0>>8] + l $t3,3($t3,$tbl) # Te1[s0>>16] + + x $s0,3($i1,$tbl) # Te1[s1>>16] + l $s1,0($s1,$tbl) # Te0[s1>>24] + x $t2,1($i2,$tbl) # Te3[s1>>0] + x $t3,2($i3,$tbl) # Te2[s1>>8] + + srlg $i1,$s2,`8-3` # i0 + srlg $i2,$s2,`16-3` # i1 + nr $i1,$mask + nr $i2,$mask + sllg $i3,$s2,`0+3` + srl $s2,`24-3` + nr $s2,$mask + ngr $i3,$mask + + xr $s1,$t1 + srlg $ra,$s3,`8-3` # i1 + sllg $t1,$s3,`0+3` # i0 + nr $ra,$mask + la $key,16($key) + ngr $t1,$mask + + x $s0,2($i1,$tbl) # Te2[s2>>8] + x $s1,3($i2,$tbl) # Te1[s2>>16] + l $s2,0($s2,$tbl) # Te0[s2>>24] + x $t3,1($i3,$tbl) # Te3[s2>>0] + + srlg $i3,$s3,`16-3` # i2 + xr $s2,$t2 + srl $s3,`24-3` + nr $i3,$mask + nr $s3,$mask + + x $s0,0($key) + x $s1,4($key) + x $s2,8($key) + x $t3,12($key) + + x $s0,1($t1,$tbl) # Te3[s3>>0] + x $s1,2($ra,$tbl) # Te2[s3>>8] + x $s2,3($i3,$tbl) # Te1[s3>>16] + l $s3,0($s3,$tbl) # Te0[s3>>24] + xr $s3,$t3 + + brct $rounds,.Lenc_loop + .align 16 + + sllg $t1,$s0,`0+3` + srlg $t2,$s0,`8-3` + ngr $t1,$mask + srlg $t3,$s0,`16-3` + srl $s0,`24-3` + nr $s0,$mask + nr $t2,$mask + nr $t3,$mask + + srlg $i1,$s1,`16-3` # i0 + sllg $i2,$s1,`0+3` + ngr $i2,$mask + srlg $i3,$s1,`8-3` + srl $s1,`24-3` + nr $i1,$mask + nr $s1,$mask + nr $i3,$mask + + llgc $s0,2($s0,$tbl) # Te4[s0>>24] + llgc $t1,2($t1,$tbl) # Te4[s0>>0] + sll $s0,24 + llgc $t2,2($t2,$tbl) # Te4[s0>>8] + llgc $t3,2($t3,$tbl) # Te4[s0>>16] + sll $t2,8 + sll $t3,16 + + llgc $i1,2($i1,$tbl) # Te4[s1>>16] + llgc $s1,2($s1,$tbl) # Te4[s1>>24] + llgc $i2,2($i2,$tbl) # Te4[s1>>0] + llgc $i3,2($i3,$tbl) # Te4[s1>>8] + sll $i1,16 + sll $s1,24 + sll $i3,8 + or $s0,$i1 + or $s1,$t1 + or $t2,$i2 + or $t3,$i3 + + srlg $i1,$s2,`8-3` # i0 + srlg $i2,$s2,`16-3` # i1 + nr $i1,$mask + nr $i2,$mask + sllg $i3,$s2,`0+3` + srl $s2,`24-3` + ngr $i3,$mask + nr $s2,$mask + + sllg $t1,$s3,`0+3` # i0 + srlg $ra,$s3,`8-3` # i1 + ngr $t1,$mask + + llgc $i1,2($i1,$tbl) # Te4[s2>>8] + llgc $i2,2($i2,$tbl) # Te4[s2>>16] + sll $i1,8 + llgc $s2,2($s2,$tbl) # Te4[s2>>24] + llgc $i3,2($i3,$tbl) # Te4[s2>>0] + sll $i2,16 + nr $ra,$mask + sll $s2,24 + or $s0,$i1 + or $s1,$i2 + or $s2,$t2 + or $t3,$i3 + + srlg $i3,$s3,`16-3` # i2 + srl $s3,`24-3` + nr $i3,$mask + nr $s3,$mask + + l $t0,16($key) + l $t2,20($key) + + llgc $i1,2($t1,$tbl) # Te4[s3>>0] + llgc $i2,2($ra,$tbl) # Te4[s3>>8] + llgc $i3,2($i3,$tbl) # Te4[s3>>16] + llgc $s3,2($s3,$tbl) # Te4[s3>>24] + sll $i2,8 + sll $i3,16 + sll $s3,24 + or $s0,$i1 + or $s1,$i2 + or $s2,$i3 + or $s3,$t3 + + lg $ra,152($sp) + xr $s0,$t0 + xr $s1,$t2 + x $s2,24($key) + x $s3,28($key) + + br $ra +.size _s390x_AES_encrypt,.-_s390x_AES_encrypt +___ + +$code.=<<___; +.type AES_Td,\@object +.align 256 +AES_Td: +___ +&_data_word( + 0x51f4a750, 0x7e416553, 0x1a17a4c3, 0x3a275e96, + 0x3bab6bcb, 0x1f9d45f1, 0xacfa58ab, 0x4be30393, + 0x2030fa55, 0xad766df6, 0x88cc7691, 0xf5024c25, + 0x4fe5d7fc, 0xc52acbd7, 0x26354480, 0xb562a38f, + 0xdeb15a49, 0x25ba1b67, 0x45ea0e98, 0x5dfec0e1, + 0xc32f7502, 0x814cf012, 0x8d4697a3, 0x6bd3f9c6, + 0x038f5fe7, 0x15929c95, 0xbf6d7aeb, 0x955259da, + 0xd4be832d, 0x587421d3, 0x49e06929, 0x8ec9c844, + 0x75c2896a, 0xf48e7978, 0x99583e6b, 0x27b971dd, + 0xbee14fb6, 0xf088ad17, 0xc920ac66, 0x7dce3ab4, + 0x63df4a18, 0xe51a3182, 0x97513360, 0x62537f45, + 0xb16477e0, 0xbb6bae84, 0xfe81a01c, 0xf9082b94, + 0x70486858, 0x8f45fd19, 0x94de6c87, 0x527bf8b7, + 0xab73d323, 0x724b02e2, 0xe31f8f57, 0x6655ab2a, + 0xb2eb2807, 0x2fb5c203, 0x86c57b9a, 0xd33708a5, + 0x302887f2, 0x23bfa5b2, 0x02036aba, 0xed16825c, + 0x8acf1c2b, 0xa779b492, 0xf307f2f0, 0x4e69e2a1, + 0x65daf4cd, 0x0605bed5, 0xd134621f, 0xc4a6fe8a, + 0x342e539d, 0xa2f355a0, 0x058ae132, 0xa4f6eb75, + 0x0b83ec39, 0x4060efaa, 0x5e719f06, 0xbd6e1051, + 0x3e218af9, 0x96dd063d, 0xdd3e05ae, 0x4de6bd46, + 0x91548db5, 0x71c45d05, 0x0406d46f, 0x605015ff, + 0x1998fb24, 0xd6bde997, 0x894043cc, 0x67d99e77, + 0xb0e842bd, 0x07898b88, 0xe7195b38, 0x79c8eedb, + 0xa17c0a47, 0x7c420fe9, 0xf8841ec9, 0x00000000, + 0x09808683, 0x322bed48, 0x1e1170ac, 0x6c5a724e, + 0xfd0efffb, 0x0f853856, 0x3daed51e, 0x362d3927, + 0x0a0fd964, 0x685ca621, 0x9b5b54d1, 0x24362e3a, + 0x0c0a67b1, 0x9357e70f, 0xb4ee96d2, 0x1b9b919e, + 0x80c0c54f, 0x61dc20a2, 0x5a774b69, 0x1c121a16, + 0xe293ba0a, 0xc0a02ae5, 0x3c22e043, 0x121b171d, + 0x0e090d0b, 0xf28bc7ad, 0x2db6a8b9, 0x141ea9c8, + 0x57f11985, 0xaf75074c, 0xee99ddbb, 0xa37f60fd, + 0xf701269f, 0x5c72f5bc, 0x44663bc5, 0x5bfb7e34, + 0x8b432976, 0xcb23c6dc, 0xb6edfc68, 0xb8e4f163, + 0xd731dcca, 0x42638510, 0x13972240, 0x84c61120, + 0x854a247d, 0xd2bb3df8, 0xaef93211, 0xc729a16d, + 0x1d9e2f4b, 0xdcb230f3, 0x0d8652ec, 0x77c1e3d0, + 0x2bb3166c, 0xa970b999, 0x119448fa, 0x47e96422, + 0xa8fc8cc4, 0xa0f03f1a, 0x567d2cd8, 0x223390ef, + 0x87494ec7, 0xd938d1c1, 0x8ccaa2fe, 0x98d40b36, + 0xa6f581cf, 0xa57ade28, 0xdab78e26, 0x3fadbfa4, + 0x2c3a9de4, 0x5078920d, 0x6a5fcc9b, 0x547e4662, + 0xf68d13c2, 0x90d8b8e8, 0x2e39f75e, 0x82c3aff5, + 0x9f5d80be, 0x69d0937c, 0x6fd52da9, 0xcf2512b3, + 0xc8ac993b, 0x10187da7, 0xe89c636e, 0xdb3bbb7b, + 0xcd267809, 0x6e5918f4, 0xec9ab701, 0x834f9aa8, + 0xe6956e65, 0xaaffe67e, 0x21bccf08, 0xef15e8e6, + 0xbae79bd9, 0x4a6f36ce, 0xea9f09d4, 0x29b07cd6, + 0x31a4b2af, 0x2a3f2331, 0xc6a59430, 0x35a266c0, + 0x744ebc37, 0xfc82caa6, 0xe090d0b0, 0x33a7d815, + 0xf104984a, 0x41ecdaf7, 0x7fcd500e, 0x1791f62f, + 0x764dd68d, 0x43efb04d, 0xccaa4d54, 0xe49604df, + 0x9ed1b5e3, 0x4c6a881b, 0xc12c1fb8, 0x4665517f, + 0x9d5eea04, 0x018c355d, 0xfa877473, 0xfb0b412e, + 0xb3671d5a, 0x92dbd252, 0xe9105633, 0x6dd64713, + 0x9ad7618c, 0x37a10c7a, 0x59f8148e, 0xeb133c89, + 0xcea927ee, 0xb761c935, 0xe11ce5ed, 0x7a47b13c, + 0x9cd2df59, 0x55f2733f, 0x1814ce79, 0x73c737bf, + 0x53f7cdea, 0x5ffdaa5b, 0xdf3d6f14, 0x7844db86, + 0xcaaff381, 0xb968c43e, 0x3824342c, 0xc2a3405f, + 0x161dc372, 0xbce2250c, 0x283c498b, 0xff0d9541, + 0x39a80171, 0x080cb3de, 0xd8b4e49c, 0x6456c190, + 0x7bcb8461, 0xd532b670, 0x486c5c74, 0xd0b85742); +$code.=<<___; +# Td4[256] +.byte 0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38 +.byte 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb +.byte 0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87 +.byte 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb +.byte 0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d +.byte 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e +.byte 0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2 +.byte 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25 +.byte 0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16 +.byte 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92 +.byte 0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda +.byte 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84 +.byte 0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a +.byte 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06 +.byte 0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02 +.byte 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b +.byte 0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea +.byte 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73 +.byte 0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85 +.byte 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e +.byte 0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89 +.byte 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b +.byte 0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20 +.byte 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4 +.byte 0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31 +.byte 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f +.byte 0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d +.byte 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef +.byte 0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0 +.byte 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61 +.byte 0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26 +.byte 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d +.size AES_Td,.-AES_Td + +# void AES_decrypt(const unsigned char *inp, unsigned char *out, +# const AES_KEY *key) { +.globl AES_decrypt +.type AES_decrypt,\@function +AES_decrypt: +___ +$code.=<<___ if (!$softonly); + l %r0,240($key) + lhi %r1,16 + clr %r0,%r1 + jl .Ldsoft + + la %r1,0($key) + #la %r2,0($inp) + la %r4,0($out) + lghi %r3,16 # single block length + .long 0xb92e0042 # km %r4,%r2 + brc 1,.-4 # can this happen? + br %r14 +.align 64 +.Ldsoft: +___ +$code.=<<___; + stmg %r3,$ra,24($sp) + + llgf $s0,0($inp) + llgf $s1,4($inp) + llgf $s2,8($inp) + llgf $s3,12($inp) + + larl $tbl,AES_Td + bras $ra,_s390x_AES_decrypt + + lg $out,24($sp) + st $s0,0($out) + st $s1,4($out) + st $s2,8($out) + st $s3,12($out) + + lmg %r6,$ra,48($sp) + br $ra +.size AES_decrypt,.-AES_decrypt + +.type _s390x_AES_decrypt,\@function +.align 16 +_s390x_AES_decrypt: + stg $ra,152($sp) + x $s0,0($key) + x $s1,4($key) + x $s2,8($key) + x $s3,12($key) + l $rounds,240($key) + llill $mask,`0xff<<3` + aghi $rounds,-1 + j .Ldec_loop +.align 16 +.Ldec_loop: + srlg $t1,$s0,`16-3` + srlg $t2,$s0,`8-3` + sllg $t3,$s0,`0+3` + srl $s0,`24-3` + nr $s0,$mask + nr $t1,$mask + nr $t2,$mask + ngr $t3,$mask + + sllg $i1,$s1,`0+3` # i0 + srlg $i2,$s1,`16-3` + srlg $i3,$s1,`8-3` + srl $s1,`24-3` + ngr $i1,$mask + nr $s1,$mask + nr $i2,$mask + nr $i3,$mask + + l $s0,0($s0,$tbl) # Td0[s0>>24] + l $t1,3($t1,$tbl) # Td1[s0>>16] + l $t2,2($t2,$tbl) # Td2[s0>>8] + l $t3,1($t3,$tbl) # Td3[s0>>0] + + x $s0,1($i1,$tbl) # Td3[s1>>0] + l $s1,0($s1,$tbl) # Td0[s1>>24] + x $t2,3($i2,$tbl) # Td1[s1>>16] + x $t3,2($i3,$tbl) # Td2[s1>>8] + + srlg $i1,$s2,`8-3` # i0 + sllg $i2,$s2,`0+3` # i1 + srlg $i3,$s2,`16-3` + srl $s2,`24-3` + nr $i1,$mask + ngr $i2,$mask + nr $s2,$mask + nr $i3,$mask + + xr $s1,$t1 + srlg $ra,$s3,`8-3` # i1 + srlg $t1,$s3,`16-3` # i0 + nr $ra,$mask + la $key,16($key) + nr $t1,$mask + + x $s0,2($i1,$tbl) # Td2[s2>>8] + x $s1,1($i2,$tbl) # Td3[s2>>0] + l $s2,0($s2,$tbl) # Td0[s2>>24] + x $t3,3($i3,$tbl) # Td1[s2>>16] + + sllg $i3,$s3,`0+3` # i2 + srl $s3,`24-3` + ngr $i3,$mask + nr $s3,$mask + + xr $s2,$t2 + x $s0,0($key) + x $s1,4($key) + x $s2,8($key) + x $t3,12($key) + + x $s0,3($t1,$tbl) # Td1[s3>>16] + x $s1,2($ra,$tbl) # Td2[s3>>8] + x $s2,1($i3,$tbl) # Td3[s3>>0] + l $s3,0($s3,$tbl) # Td0[s3>>24] + xr $s3,$t3 + + brct $rounds,.Ldec_loop + .align 16 + + l $t1,`2048+0`($tbl) # prefetch Td4 + l $t2,`2048+64`($tbl) + l $t3,`2048+128`($tbl) + l $i1,`2048+192`($tbl) + llill $mask,0xff + + srlg $i3,$s0,24 # i0 + srlg $t1,$s0,16 + srlg $t2,$s0,8 + nr $s0,$mask # i3 + nr $t1,$mask + + srlg $i1,$s1,24 + nr $t2,$mask + srlg $i2,$s1,16 + srlg $ra,$s1,8 + nr $s1,$mask # i0 + nr $i2,$mask + nr $ra,$mask + + llgc $i3,2048($i3,$tbl) # Td4[s0>>24] + llgc $t1,2048($t1,$tbl) # Td4[s0>>16] + llgc $t2,2048($t2,$tbl) # Td4[s0>>8] + sll $t1,16 + llgc $t3,2048($s0,$tbl) # Td4[s0>>0] + sllg $s0,$i3,24 + sll $t2,8 + + llgc $s1,2048($s1,$tbl) # Td4[s1>>0] + llgc $i1,2048($i1,$tbl) # Td4[s1>>24] + llgc $i2,2048($i2,$tbl) # Td4[s1>>16] + sll $i1,24 + llgc $i3,2048($ra,$tbl) # Td4[s1>>8] + sll $i2,16 + sll $i3,8 + or $s0,$s1 + or $t1,$i1 + or $t2,$i2 + or $t3,$i3 + + srlg $i1,$s2,8 # i0 + srlg $i2,$s2,24 + srlg $i3,$s2,16 + nr $s2,$mask # i1 + nr $i1,$mask + nr $i3,$mask + llgc $i1,2048($i1,$tbl) # Td4[s2>>8] + llgc $s1,2048($s2,$tbl) # Td4[s2>>0] + llgc $i2,2048($i2,$tbl) # Td4[s2>>24] + llgc $i3,2048($i3,$tbl) # Td4[s2>>16] + sll $i1,8 + sll $i2,24 + or $s0,$i1 + sll $i3,16 + or $t2,$i2 + or $t3,$i3 + + srlg $i1,$s3,16 # i0 + srlg $i2,$s3,8 # i1 + srlg $i3,$s3,24 + nr $s3,$mask # i2 + nr $i1,$mask + nr $i2,$mask + + lg $ra,152($sp) + or $s1,$t1 + l $t0,16($key) + l $t1,20($key) + + llgc $i1,2048($i1,$tbl) # Td4[s3>>16] + llgc $i2,2048($i2,$tbl) # Td4[s3>>8] + sll $i1,16 + llgc $s2,2048($s3,$tbl) # Td4[s3>>0] + llgc $s3,2048($i3,$tbl) # Td4[s3>>24] + sll $i2,8 + sll $s3,24 + or $s0,$i1 + or $s1,$i2 + or $s2,$t2 + or $s3,$t3 + + xr $s0,$t0 + xr $s1,$t1 + x $s2,24($key) + x $s3,28($key) + + br $ra +.size _s390x_AES_decrypt,.-_s390x_AES_decrypt +___ + +$code.=<<___; +# void AES_set_encrypt_key(const unsigned char *in, int bits, +# AES_KEY *key) { +.globl AES_set_encrypt_key +.type AES_set_encrypt_key,\@function +.align 16 +AES_set_encrypt_key: + lghi $t0,0 + clgr $inp,$t0 + je .Lminus1 + clgr $key,$t0 + je .Lminus1 + + lghi $t0,128 + clr $bits,$t0 + je .Lproceed + lghi $t0,192 + clr $bits,$t0 + je .Lproceed + lghi $t0,256 + clr $bits,$t0 + je .Lproceed + lghi %r2,-2 + br %r14 + +.align 16 +.Lproceed: +___ +$code.=<<___ if (!$softonly); + # convert bits to km code, [128,192,256]->[18,19,20] + lhi %r5,-128 + lhi %r0,18 + ar %r5,$bits + srl %r5,6 + ar %r5,%r0 + + lghi %r0,0 # query capability vector + la %r1,16($sp) + .long 0xb92f0042 # kmc %r4,%r2 + + llihh %r1,0x8000 + srlg %r1,%r1,0(%r5) + ng %r1,16($sp) + jz .Lekey_internal + + lmg %r0,%r1,0($inp) # just copy 128 bits... + stmg %r0,%r1,0($key) + lhi %r0,192 + cr $bits,%r0 + jl 1f + lg %r1,16($inp) + stg %r1,16($key) + je 1f + lg %r1,24($inp) + stg %r1,24($key) +1: st $bits,236($key) # save bits + st %r5,240($key) # save km code + lghi %r2,0 + br %r14 +___ +$code.=<<___; +.align 16 +.Lekey_internal: + stmg %r6,%r13,48($sp) # all non-volatile regs + + larl $tbl,AES_Te+2048 + + llgf $s0,0($inp) + llgf $s1,4($inp) + llgf $s2,8($inp) + llgf $s3,12($inp) + st $s0,0($key) + st $s1,4($key) + st $s2,8($key) + st $s3,12($key) + lghi $t0,128 + cr $bits,$t0 + jne .Lnot128 + + llill $mask,0xff + lghi $t3,0 # i=0 + lghi $rounds,10 + st $rounds,240($key) + + llgfr $t2,$s3 # temp=rk[3] + srlg $i1,$s3,8 + srlg $i2,$s3,16 + srlg $i3,$s3,24 + nr $t2,$mask + nr $i1,$mask + nr $i2,$mask + +.align 16 +.L128_loop: + la $t2,0($t2,$tbl) + la $i1,0($i1,$tbl) + la $i2,0($i2,$tbl) + la $i3,0($i3,$tbl) + icm $t2,2,0($t2) # Te4[rk[3]>>0]<<8 + icm $t2,4,0($i1) # Te4[rk[3]>>8]<<16 + icm $t2,8,0($i2) # Te4[rk[3]>>16]<<24 + icm $t2,1,0($i3) # Te4[rk[3]>>24] + x $t2,256($t3,$tbl) # rcon[i] + xr $s0,$t2 # rk[4]=rk[0]^... + xr $s1,$s0 # rk[5]=rk[1]^rk[4] + xr $s2,$s1 # rk[6]=rk[2]^rk[5] + xr $s3,$s2 # rk[7]=rk[3]^rk[6] + + llgfr $t2,$s3 # temp=rk[3] + srlg $i1,$s3,8 + srlg $i2,$s3,16 + nr $t2,$mask + nr $i1,$mask + srlg $i3,$s3,24 + nr $i2,$mask + + st $s0,16($key) + st $s1,20($key) + st $s2,24($key) + st $s3,28($key) + la $key,16($key) # key+=4 + la $t3,4($t3) # i++ + brct $rounds,.L128_loop + lghi %r2,0 + lmg %r6,%r13,48($sp) + br $ra + +.align 16 +.Lnot128: + llgf $t0,16($inp) + llgf $t1,20($inp) + st $t0,16($key) + st $t1,20($key) + lghi $t0,192 + cr $bits,$t0 + jne .Lnot192 + + llill $mask,0xff + lghi $t3,0 # i=0 + lghi $rounds,12 + st $rounds,240($key) + lghi $rounds,8 + + srlg $i1,$t1,8 + srlg $i2,$t1,16 + srlg $i3,$t1,24 + nr $t1,$mask + nr $i1,$mask + nr $i2,$mask + +.align 16 +.L192_loop: + la $t1,0($t1,$tbl) + la $i1,0($i1,$tbl) + la $i2,0($i2,$tbl) + la $i3,0($i3,$tbl) + icm $t1,2,0($t1) # Te4[rk[5]>>0]<<8 + icm $t1,4,0($i1) # Te4[rk[5]>>8]<<16 + icm $t1,8,0($i2) # Te4[rk[5]>>16]<<24 + icm $t1,1,0($i3) # Te4[rk[5]>>24] + x $t1,256($t3,$tbl) # rcon[i] + xr $s0,$t1 # rk[6]=rk[0]^... + xr $s1,$s0 # rk[7]=rk[1]^rk[6] + xr $s2,$s1 # rk[8]=rk[2]^rk[7] + xr $s3,$s2 # rk[9]=rk[3]^rk[8] + + st $s0,24($key) + st $s1,28($key) + st $s2,32($key) + st $s3,36($key) + brct $rounds,.L192_continue + lghi %r2,0 + lmg %r6,%r13,48($sp) + br $ra + +.align 16 +.L192_continue: + lgr $t1,$s3 + x $t1,16($key) # rk[10]=rk[4]^rk[9] + st $t1,40($key) + x $t1,20($key) # rk[11]=rk[5]^rk[10] + st $t1,44($key) + + srlg $i1,$t1,8 + srlg $i2,$t1,16 + srlg $i3,$t1,24 + nr $t1,$mask + nr $i1,$mask + nr $i2,$mask + + la $key,24($key) # key+=6 + la $t3,4($t3) # i++ + j .L192_loop + +.align 16 +.Lnot192: + llgf $t0,24($inp) + llgf $t1,28($inp) + st $t0,24($key) + st $t1,28($key) + llill $mask,0xff + lghi $t3,0 # i=0 + lghi $rounds,14 + st $rounds,240($key) + lghi $rounds,7 + + srlg $i1,$t1,8 + srlg $i2,$t1,16 + srlg $i3,$t1,24 + nr $t1,$mask + nr $i1,$mask + nr $i2,$mask + +.align 16 +.L256_loop: + la $t1,0($t1,$tbl) + la $i1,0($i1,$tbl) + la $i2,0($i2,$tbl) + la $i3,0($i3,$tbl) + icm $t1,2,0($t1) # Te4[rk[7]>>0]<<8 + icm $t1,4,0($i1) # Te4[rk[7]>>8]<<16 + icm $t1,8,0($i2) # Te4[rk[7]>>16]<<24 + icm $t1,1,0($i3) # Te4[rk[7]>>24] + x $t1,256($t3,$tbl) # rcon[i] + xr $s0,$t1 # rk[8]=rk[0]^... + xr $s1,$s0 # rk[9]=rk[1]^rk[8] + xr $s2,$s1 # rk[10]=rk[2]^rk[9] + xr $s3,$s2 # rk[11]=rk[3]^rk[10] + st $s0,32($key) + st $s1,36($key) + st $s2,40($key) + st $s3,44($key) + brct $rounds,.L256_continue + lghi %r2,0 + lmg %r6,%r13,48($sp) + br $ra + +.align 16 +.L256_continue: + lgr $t1,$s3 # temp=rk[11] + srlg $i1,$s3,8 + srlg $i2,$s3,16 + srlg $i3,$s3,24 + nr $t1,$mask + nr $i1,$mask + nr $i2,$mask + la $t1,0($t1,$tbl) + la $i1,0($i1,$tbl) + la $i2,0($i2,$tbl) + la $i3,0($i3,$tbl) + llgc $t1,0($t1) # Te4[rk[11]>>0] + icm $t1,2,0($i1) # Te4[rk[11]>>8]<<8 + icm $t1,4,0($i2) # Te4[rk[11]>>16]<<16 + icm $t1,8,0($i3) # Te4[rk[11]>>24]<<24 + x $t1,16($key) # rk[12]=rk[4]^... + st $t1,48($key) + x $t1,20($key) # rk[13]=rk[5]^rk[12] + st $t1,52($key) + x $t1,24($key) # rk[14]=rk[6]^rk[13] + st $t1,56($key) + x $t1,28($key) # rk[15]=rk[7]^rk[14] + st $t1,60($key) + + srlg $i1,$t1,8 + srlg $i2,$t1,16 + srlg $i3,$t1,24 + nr $t1,$mask + nr $i1,$mask + nr $i2,$mask + + la $key,32($key) # key+=8 + la $t3,4($t3) # i++ + j .L256_loop + +.Lminus1: + lghi %r2,-1 + br $ra +.size AES_set_encrypt_key,.-AES_set_encrypt_key + +# void AES_set_decrypt_key(const unsigned char *in, int bits, +# AES_KEY *key) { +.globl AES_set_decrypt_key +.type AES_set_decrypt_key,\@function +.align 16 +AES_set_decrypt_key: + stg $key,32($sp) # I rely on AES_set_encrypt_key to + stg $ra,112($sp) # save non-volatile registers! + bras $ra,AES_set_encrypt_key + lg $key,32($sp) + lg $ra,112($sp) + ltgr %r2,%r2 + bnzr $ra +___ +$code.=<<___ if (!$softonly); + l $t0,240($key) + lhi $t1,16 + cr $t0,$t1 + jl .Lgo + oill $t0,0x80 # set "decrypt" bit + st $t0,240($key) + br $ra + +.align 16 +.Ldkey_internal: + stg $key,32($sp) + stg $ra,40($sp) + bras $ra,.Lekey_internal + lg $key,32($sp) + lg $ra,40($sp) +___ +$code.=<<___; + +.Lgo: llgf $rounds,240($key) + la $i1,0($key) + sllg $i2,$rounds,4 + la $i2,0($i2,$key) + srl $rounds,1 + lghi $t1,-16 + +.align 16 +.Linv: lmg $s0,$s1,0($i1) + lmg $s2,$s3,0($i2) + stmg $s0,$s1,0($i2) + stmg $s2,$s3,0($i1) + la $i1,16($i1) + la $i2,0($t1,$i2) + brct $rounds,.Linv +___ +$mask80=$i1; +$mask1b=$i2; +$maskfe=$i3; +$code.=<<___; + llgf $rounds,240($key) + aghi $rounds,-1 + sll $rounds,2 # (rounds-1)*4 + llilh $mask80,0x8080 + llilh $mask1b,0x1b1b + llilh $maskfe,0xfefe + oill $mask80,0x8080 + oill $mask1b,0x1b1b + oill $maskfe,0xfefe + +.align 16 +.Lmix: l $s0,16($key) # tp1 + lr $s1,$s0 + ngr $s1,$mask80 + srlg $t1,$s1,7 + slr $s1,$t1 + nr $s1,$mask1b + sllg $t1,$s0,1 + nr $t1,$maskfe + xr $s1,$t1 # tp2 + + lr $s2,$s1 + ngr $s2,$mask80 + srlg $t1,$s2,7 + slr $s2,$t1 + nr $s2,$mask1b + sllg $t1,$s1,1 + nr $t1,$maskfe + xr $s2,$t1 # tp4 + + lr $s3,$s2 + ngr $s3,$mask80 + srlg $t1,$s3,7 + slr $s3,$t1 + nr $s3,$mask1b + sllg $t1,$s2,1 + nr $t1,$maskfe + xr $s3,$t1 # tp8 + + xr $s1,$s0 # tp2^tp1 + xr $s2,$s0 # tp4^tp1 + rll $s0,$s0,24 # = ROTATE(tp1,8) + xr $s2,$s3 # ^=tp8 + xr $s0,$s1 # ^=tp2^tp1 + xr $s1,$s3 # tp2^tp1^tp8 + xr $s0,$s2 # ^=tp4^tp1^tp8 + rll $s1,$s1,8 + rll $s2,$s2,16 + xr $s0,$s1 # ^= ROTATE(tp8^tp2^tp1,24) + rll $s3,$s3,24 + xr $s0,$s2 # ^= ROTATE(tp8^tp4^tp1,16) + xr $s0,$s3 # ^= ROTATE(tp8,8) + + st $s0,16($key) + la $key,4($key) + brct $rounds,.Lmix + + lmg %r6,%r13,48($sp)# as was saved by AES_set_encrypt_key! + lghi %r2,0 + br $ra +.size AES_set_decrypt_key,.-AES_set_decrypt_key +___ + +#void AES_cbc_encrypt(const unsigned char *in, unsigned char *out, +# size_t length, const AES_KEY *key, +# unsigned char *ivec, const int enc) +{ +my $inp="%r2"; +my $out="%r4"; # length and out are swapped +my $len="%r3"; +my $key="%r5"; +my $ivp="%r6"; + +$code.=<<___; +.globl AES_cbc_encrypt +.type AES_cbc_encrypt,\@function +.align 16 +AES_cbc_encrypt: + xgr %r3,%r4 # flip %r3 and %r4, out and len + xgr %r4,%r3 + xgr %r3,%r4 +___ +$code.=<<___ if (!$softonly); + lhi %r0,16 + cl %r0,240($key) + jh .Lcbc_software + + lg %r0,0($ivp) # copy ivec + lg %r1,8($ivp) + stmg %r0,%r1,16($sp) + lmg %r0,%r1,0($key) # copy key, cover 256 bit + stmg %r0,%r1,32($sp) + lmg %r0,%r1,16($key) + stmg %r0,%r1,48($sp) + l %r0,240($key) # load kmc code + lghi $key,15 # res=len%16, len-=res; + ngr $key,$len + slgr $len,$key + la %r1,16($sp) # parameter block - ivec || key + jz .Lkmc_truncated + .long 0xb92f0042 # kmc %r4,%r2 + brc 1,.-4 # pay attention to "partial completion" + ltr $key,$key + jnz .Lkmc_truncated +.Lkmc_done: + lmg %r0,%r1,16($sp) # copy ivec to caller + stg %r0,0($ivp) + stg %r1,8($ivp) + br $ra +.align 16 +.Lkmc_truncated: + ahi $key,-1 # it's the way it's encoded in mvc + tmll %r0,0x80 + jnz .Lkmc_truncated_dec + lghi %r1,0 + stg %r1,128($sp) + stg %r1,136($sp) + bras %r1,1f + mvc 128(1,$sp),0($inp) +1: ex $key,0(%r1) + la %r1,16($sp) # restore parameter block + la $inp,128($sp) + lghi $len,16 + .long 0xb92f0042 # kmc %r4,%r2 + j .Lkmc_done +.align 16 +.Lkmc_truncated_dec: + stg $out,64($sp) + la $out,128($sp) + lghi $len,16 + .long 0xb92f0042 # kmc %r4,%r2 + lg $out,64($sp) + bras %r1,2f + mvc 0(1,$out),128($sp) +2: ex $key,0(%r1) + j .Lkmc_done +.align 16 +.Lcbc_software: +___ +$code.=<<___; + stmg $key,$ra,40($sp) + lhi %r0,0 + cl %r0,164($sp) + je .Lcbc_decrypt + + larl $tbl,AES_Te + + llgf $s0,0($ivp) + llgf $s1,4($ivp) + llgf $s2,8($ivp) + llgf $s3,12($ivp) + + lghi $t0,16 + slgr $len,$t0 + brc 4,.Lcbc_enc_tail # if borrow +.Lcbc_enc_loop: + stmg $inp,$out,16($sp) + x $s0,0($inp) + x $s1,4($inp) + x $s2,8($inp) + x $s3,12($inp) + lgr %r4,$key + + bras $ra,_s390x_AES_encrypt + + lmg $inp,$key,16($sp) + st $s0,0($out) + st $s1,4($out) + st $s2,8($out) + st $s3,12($out) + + la $inp,16($inp) + la $out,16($out) + lghi $t0,16 + ltgr $len,$len + jz .Lcbc_enc_done + slgr $len,$t0 + brc 4,.Lcbc_enc_tail # if borrow + j .Lcbc_enc_loop +.align 16 +.Lcbc_enc_done: + lg $ivp,48($sp) + st $s0,0($ivp) + st $s1,4($ivp) + st $s2,8($ivp) + st $s3,12($ivp) + + lmg %r7,$ra,56($sp) + br $ra + +.align 16 +.Lcbc_enc_tail: + aghi $len,15 + lghi $t0,0 + stg $t0,128($sp) + stg $t0,136($sp) + bras $t1,3f + mvc 128(1,$sp),0($inp) +3: ex $len,0($t1) + lghi $len,0 + la $inp,128($sp) + j .Lcbc_enc_loop + +.align 16 +.Lcbc_decrypt: + larl $tbl,AES_Td + + lg $t0,0($ivp) + lg $t1,8($ivp) + stmg $t0,$t1,128($sp) + +.Lcbc_dec_loop: + stmg $inp,$out,16($sp) + llgf $s0,0($inp) + llgf $s1,4($inp) + llgf $s2,8($inp) + llgf $s3,12($inp) + lgr %r4,$key + + bras $ra,_s390x_AES_decrypt + + lmg $inp,$key,16($sp) + sllg $s0,$s0,32 + sllg $s2,$s2,32 + lr $s0,$s1 + lr $s2,$s3 + + lg $t0,0($inp) + lg $t1,8($inp) + xg $s0,128($sp) + xg $s2,136($sp) + lghi $s1,16 + slgr $len,$s1 + brc 4,.Lcbc_dec_tail # if borrow + brc 2,.Lcbc_dec_done # if zero + stg $s0,0($out) + stg $s2,8($out) + stmg $t0,$t1,128($sp) + + la $inp,16($inp) + la $out,16($out) + j .Lcbc_dec_loop + +.Lcbc_dec_done: + stg $s0,0($out) + stg $s2,8($out) +.Lcbc_dec_exit: + lmg $ivp,$ra,48($sp) + stmg $t0,$t1,0($ivp) + + br $ra + +.align 16 +.Lcbc_dec_tail: + aghi $len,15 + stg $s0,128($sp) + stg $s2,136($sp) + bras $s1,4f + mvc 0(1,$out),128($sp) +4: ex $len,0($s1) + j .Lcbc_dec_exit +.size AES_cbc_encrypt,.-AES_cbc_encrypt +___ +} +$code.=<<___; +.string "AES for s390x, CRYPTOGAMS by <appro\@openssl.org>" +___ + +$code =~ s/\`([^\`]*)\`/eval $1/gem; +print $code; diff --git a/openssl/crypto/aes/asm/aes-sparcv9.pl b/openssl/crypto/aes/asm/aes-sparcv9.pl new file mode 100644 index 000000000..c57b3a2d6 --- /dev/null +++ b/openssl/crypto/aes/asm/aes-sparcv9.pl @@ -0,0 +1,1181 @@ +#!/usr/bin/env perl +# +# ==================================================================== +# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# project. Rights for redistribution and usage in source and binary +# forms are granted according to the OpenSSL license. +# ==================================================================== +# +# Version 1.1 +# +# The major reason for undertaken effort was to mitigate the hazard of +# cache-timing attack. This is [currently and initially!] addressed in +# two ways. 1. S-boxes are compressed from 5KB to 2KB+256B size each. +# 2. References to them are scheduled for L2 cache latency, meaning +# that the tables don't have to reside in L1 cache. Once again, this +# is an initial draft and one should expect more countermeasures to +# be implemented... +# +# Version 1.1 prefetches T[ed]4 in order to mitigate attack on last +# round. +# +# Even though performance was not the primary goal [on the contrary, +# extra shifts "induced" by compressed S-box and longer loop epilogue +# "induced" by scheduling for L2 have negative effect on performance], +# the code turned out to run in ~23 cycles per processed byte en-/ +# decrypted with 128-bit key. This is pretty good result for code +# with mentioned qualities and UltraSPARC core. Compared to Sun C +# generated code my encrypt procedure runs just few percents faster, +# while decrypt one - whole 50% faster [yes, Sun C failed to generate +# optimal decrypt procedure]. Compared to GNU C generated code both +# procedures are more than 60% faster:-) + +$bits=32; +for (@ARGV) { $bits=64 if (/\-m64/ || /\-xarch\=v9/); } +if ($bits==64) { $bias=2047; $frame=192; } +else { $bias=0; $frame=112; } +$locals=16; + +$acc0="%l0"; +$acc1="%o0"; +$acc2="%o1"; +$acc3="%o2"; + +$acc4="%l1"; +$acc5="%o3"; +$acc6="%o4"; +$acc7="%o5"; + +$acc8="%l2"; +$acc9="%o7"; +$acc10="%g1"; +$acc11="%g2"; + +$acc12="%l3"; +$acc13="%g3"; +$acc14="%g4"; +$acc15="%g5"; + +$t0="%l4"; +$t1="%l5"; +$t2="%l6"; +$t3="%l7"; + +$s0="%i0"; +$s1="%i1"; +$s2="%i2"; +$s3="%i3"; +$tbl="%i4"; +$key="%i5"; +$rounds="%i7"; # aliases with return address, which is off-loaded to stack + +sub _data_word() +{ my $i; + while(defined($i=shift)) { $code.=sprintf"\t.long\t0x%08x,0x%08x\n",$i,$i; } +} + +$code.=<<___ if ($bits==64); +.register %g2,#scratch +.register %g3,#scratch +___ +$code.=<<___; +.section ".text",#alloc,#execinstr + +.align 256 +AES_Te: +___ +&_data_word( + 0xc66363a5, 0xf87c7c84, 0xee777799, 0xf67b7b8d, + 0xfff2f20d, 0xd66b6bbd, 0xde6f6fb1, 0x91c5c554, + 0x60303050, 0x02010103, 0xce6767a9, 0x562b2b7d, + 0xe7fefe19, 0xb5d7d762, 0x4dababe6, 0xec76769a, + 0x8fcaca45, 0x1f82829d, 0x89c9c940, 0xfa7d7d87, + 0xeffafa15, 0xb25959eb, 0x8e4747c9, 0xfbf0f00b, + 0x41adadec, 0xb3d4d467, 0x5fa2a2fd, 0x45afafea, + 0x239c9cbf, 0x53a4a4f7, 0xe4727296, 0x9bc0c05b, + 0x75b7b7c2, 0xe1fdfd1c, 0x3d9393ae, 0x4c26266a, + 0x6c36365a, 0x7e3f3f41, 0xf5f7f702, 0x83cccc4f, + 0x6834345c, 0x51a5a5f4, 0xd1e5e534, 0xf9f1f108, + 0xe2717193, 0xabd8d873, 0x62313153, 0x2a15153f, + 0x0804040c, 0x95c7c752, 0x46232365, 0x9dc3c35e, + 0x30181828, 0x379696a1, 0x0a05050f, 0x2f9a9ab5, + 0x0e070709, 0x24121236, 0x1b80809b, 0xdfe2e23d, + 0xcdebeb26, 0x4e272769, 0x7fb2b2cd, 0xea75759f, + 0x1209091b, 0x1d83839e, 0x582c2c74, 0x341a1a2e, + 0x361b1b2d, 0xdc6e6eb2, 0xb45a5aee, 0x5ba0a0fb, + 0xa45252f6, 0x763b3b4d, 0xb7d6d661, 0x7db3b3ce, + 0x5229297b, 0xdde3e33e, 0x5e2f2f71, 0x13848497, + 0xa65353f5, 0xb9d1d168, 0x00000000, 0xc1eded2c, + 0x40202060, 0xe3fcfc1f, 0x79b1b1c8, 0xb65b5bed, + 0xd46a6abe, 0x8dcbcb46, 0x67bebed9, 0x7239394b, + 0x944a4ade, 0x984c4cd4, 0xb05858e8, 0x85cfcf4a, + 0xbbd0d06b, 0xc5efef2a, 0x4faaaae5, 0xedfbfb16, + 0x864343c5, 0x9a4d4dd7, 0x66333355, 0x11858594, + 0x8a4545cf, 0xe9f9f910, 0x04020206, 0xfe7f7f81, + 0xa05050f0, 0x783c3c44, 0x259f9fba, 0x4ba8a8e3, + 0xa25151f3, 0x5da3a3fe, 0x804040c0, 0x058f8f8a, + 0x3f9292ad, 0x219d9dbc, 0x70383848, 0xf1f5f504, + 0x63bcbcdf, 0x77b6b6c1, 0xafdada75, 0x42212163, + 0x20101030, 0xe5ffff1a, 0xfdf3f30e, 0xbfd2d26d, + 0x81cdcd4c, 0x180c0c14, 0x26131335, 0xc3ecec2f, + 0xbe5f5fe1, 0x359797a2, 0x884444cc, 0x2e171739, + 0x93c4c457, 0x55a7a7f2, 0xfc7e7e82, 0x7a3d3d47, + 0xc86464ac, 0xba5d5de7, 0x3219192b, 0xe6737395, + 0xc06060a0, 0x19818198, 0x9e4f4fd1, 0xa3dcdc7f, + 0x44222266, 0x542a2a7e, 0x3b9090ab, 0x0b888883, + 0x8c4646ca, 0xc7eeee29, 0x6bb8b8d3, 0x2814143c, + 0xa7dede79, 0xbc5e5ee2, 0x160b0b1d, 0xaddbdb76, + 0xdbe0e03b, 0x64323256, 0x743a3a4e, 0x140a0a1e, + 0x924949db, 0x0c06060a, 0x4824246c, 0xb85c5ce4, + 0x9fc2c25d, 0xbdd3d36e, 0x43acacef, 0xc46262a6, + 0x399191a8, 0x319595a4, 0xd3e4e437, 0xf279798b, + 0xd5e7e732, 0x8bc8c843, 0x6e373759, 0xda6d6db7, + 0x018d8d8c, 0xb1d5d564, 0x9c4e4ed2, 0x49a9a9e0, + 0xd86c6cb4, 0xac5656fa, 0xf3f4f407, 0xcfeaea25, + 0xca6565af, 0xf47a7a8e, 0x47aeaee9, 0x10080818, + 0x6fbabad5, 0xf0787888, 0x4a25256f, 0x5c2e2e72, + 0x381c1c24, 0x57a6a6f1, 0x73b4b4c7, 0x97c6c651, + 0xcbe8e823, 0xa1dddd7c, 0xe874749c, 0x3e1f1f21, + 0x964b4bdd, 0x61bdbddc, 0x0d8b8b86, 0x0f8a8a85, + 0xe0707090, 0x7c3e3e42, 0x71b5b5c4, 0xcc6666aa, + 0x904848d8, 0x06030305, 0xf7f6f601, 0x1c0e0e12, + 0xc26161a3, 0x6a35355f, 0xae5757f9, 0x69b9b9d0, + 0x17868691, 0x99c1c158, 0x3a1d1d27, 0x279e9eb9, + 0xd9e1e138, 0xebf8f813, 0x2b9898b3, 0x22111133, + 0xd26969bb, 0xa9d9d970, 0x078e8e89, 0x339494a7, + 0x2d9b9bb6, 0x3c1e1e22, 0x15878792, 0xc9e9e920, + 0x87cece49, 0xaa5555ff, 0x50282878, 0xa5dfdf7a, + 0x038c8c8f, 0x59a1a1f8, 0x09898980, 0x1a0d0d17, + 0x65bfbfda, 0xd7e6e631, 0x844242c6, 0xd06868b8, + 0x824141c3, 0x299999b0, 0x5a2d2d77, 0x1e0f0f11, + 0x7bb0b0cb, 0xa85454fc, 0x6dbbbbd6, 0x2c16163a); +$code.=<<___; + .byte 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5 + .byte 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76 + .byte 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0 + .byte 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0 + .byte 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc + .byte 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15 + .byte 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a + .byte 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75 + .byte 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0 + .byte 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84 + .byte 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b + .byte 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf + .byte 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85 + .byte 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8 + .byte 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5 + .byte 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2 + .byte 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17 + .byte 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73 + .byte 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88 + .byte 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb + .byte 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c + .byte 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79 + .byte 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9 + .byte 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08 + .byte 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6 + .byte 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a + .byte 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e + .byte 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e + .byte 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94 + .byte 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf + .byte 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68 + .byte 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16 +.type AES_Te,#object +.size AES_Te,(.-AES_Te) + +.align 64 +.skip 16 +_sparcv9_AES_encrypt: + save %sp,-$frame-$locals,%sp + stx %i7,[%sp+$bias+$frame+0] ! off-load return address + ld [$key+240],$rounds + ld [$key+0],$t0 + ld [$key+4],$t1 ! + ld [$key+8],$t2 + srl $rounds,1,$rounds + xor $t0,$s0,$s0 + ld [$key+12],$t3 + srl $s0,21,$acc0 + xor $t1,$s1,$s1 + ld [$key+16],$t0 + srl $s1,13,$acc1 ! + xor $t2,$s2,$s2 + ld [$key+20],$t1 + xor $t3,$s3,$s3 + ld [$key+24],$t2 + and $acc0,2040,$acc0 + ld [$key+28],$t3 + nop +.Lenc_loop: + srl $s2,5,$acc2 ! + and $acc1,2040,$acc1 + ldx [$tbl+$acc0],$acc0 + sll $s3,3,$acc3 + and $acc2,2040,$acc2 + ldx [$tbl+$acc1],$acc1 + srl $s1,21,$acc4 + and $acc3,2040,$acc3 + ldx [$tbl+$acc2],$acc2 ! + srl $s2,13,$acc5 + and $acc4,2040,$acc4 + ldx [$tbl+$acc3],$acc3 + srl $s3,5,$acc6 + and $acc5,2040,$acc5 + ldx [$tbl+$acc4],$acc4 + fmovs %f0,%f0 + sll $s0,3,$acc7 ! + and $acc6,2040,$acc6 + ldx [$tbl+$acc5],$acc5 + srl $s2,21,$acc8 + and $acc7,2040,$acc7 + ldx [$tbl+$acc6],$acc6 + srl $s3,13,$acc9 + and $acc8,2040,$acc8 + ldx [$tbl+$acc7],$acc7 ! + srl $s0,5,$acc10 + and $acc9,2040,$acc9 + ldx [$tbl+$acc8],$acc8 + sll $s1,3,$acc11 + and $acc10,2040,$acc10 + ldx [$tbl+$acc9],$acc9 + fmovs %f0,%f0 + srl $s3,21,$acc12 ! + and $acc11,2040,$acc11 + ldx [$tbl+$acc10],$acc10 + srl $s0,13,$acc13 + and $acc12,2040,$acc12 + ldx [$tbl+$acc11],$acc11 + srl $s1,5,$acc14 + and $acc13,2040,$acc13 + ldx [$tbl+$acc12],$acc12 ! + sll $s2,3,$acc15 + and $acc14,2040,$acc14 + ldx [$tbl+$acc13],$acc13 + and $acc15,2040,$acc15 + add $key,32,$key + ldx [$tbl+$acc14],$acc14 + fmovs %f0,%f0 + subcc $rounds,1,$rounds ! + ldx [$tbl+$acc15],$acc15 + bz,a,pn %icc,.Lenc_last + add $tbl,2048,$rounds + + srlx $acc1,8,$acc1 + xor $acc0,$t0,$t0 + ld [$key+0],$s0 + fmovs %f0,%f0 + srlx $acc2,16,$acc2 ! + xor $acc1,$t0,$t0 + ld [$key+4],$s1 + srlx $acc3,24,$acc3 + xor $acc2,$t0,$t0 + ld [$key+8],$s2 + srlx $acc5,8,$acc5 + xor $acc3,$t0,$t0 + ld [$key+12],$s3 ! + srlx $acc6,16,$acc6 + xor $acc4,$t1,$t1 + fmovs %f0,%f0 + srlx $acc7,24,$acc7 + xor $acc5,$t1,$t1 + srlx $acc9,8,$acc9 + xor $acc6,$t1,$t1 + srlx $acc10,16,$acc10 ! + xor $acc7,$t1,$t1 + srlx $acc11,24,$acc11 + xor $acc8,$t2,$t2 + srlx $acc13,8,$acc13 + xor $acc9,$t2,$t2 + srlx $acc14,16,$acc14 + xor $acc10,$t2,$t2 + srlx $acc15,24,$acc15 ! + xor $acc11,$t2,$t2 + xor $acc12,$acc14,$acc14 + xor $acc13,$t3,$t3 + srl $t0,21,$acc0 + xor $acc14,$t3,$t3 + srl $t1,13,$acc1 + xor $acc15,$t3,$t3 + + and $acc0,2040,$acc0 ! + srl $t2,5,$acc2 + and $acc1,2040,$acc1 + ldx [$tbl+$acc0],$acc0 + sll $t3,3,$acc3 + and $acc2,2040,$acc2 + ldx [$tbl+$acc1],$acc1 + fmovs %f0,%f0 + srl $t1,21,$acc4 ! + and $acc3,2040,$acc3 + ldx [$tbl+$acc2],$acc2 + srl $t2,13,$acc5 + and $acc4,2040,$acc4 + ldx [$tbl+$acc3],$acc3 + srl $t3,5,$acc6 + and $acc5,2040,$acc5 + ldx [$tbl+$acc4],$acc4 ! + sll $t0,3,$acc7 + and $acc6,2040,$acc6 + ldx [$tbl+$acc5],$acc5 + srl $t2,21,$acc8 + and $acc7,2040,$acc7 + ldx [$tbl+$acc6],$acc6 + fmovs %f0,%f0 + srl $t3,13,$acc9 ! + and $acc8,2040,$acc8 + ldx [$tbl+$acc7],$acc7 + srl $t0,5,$acc10 + and $acc9,2040,$acc9 + ldx [$tbl+$acc8],$acc8 + sll $t1,3,$acc11 + and $acc10,2040,$acc10 + ldx [$tbl+$acc9],$acc9 ! + srl $t3,21,$acc12 + and $acc11,2040,$acc11 + ldx [$tbl+$acc10],$acc10 + srl $t0,13,$acc13 + and $acc12,2040,$acc12 + ldx [$tbl+$acc11],$acc11 + fmovs %f0,%f0 + srl $t1,5,$acc14 ! + and $acc13,2040,$acc13 + ldx [$tbl+$acc12],$acc12 + sll $t2,3,$acc15 + and $acc14,2040,$acc14 + ldx [$tbl+$acc13],$acc13 + srlx $acc1,8,$acc1 + and $acc15,2040,$acc15 + ldx [$tbl+$acc14],$acc14 ! + + srlx $acc2,16,$acc2 + xor $acc0,$s0,$s0 + ldx [$tbl+$acc15],$acc15 + srlx $acc3,24,$acc3 + xor $acc1,$s0,$s0 + ld [$key+16],$t0 + fmovs %f0,%f0 + srlx $acc5,8,$acc5 ! + xor $acc2,$s0,$s0 + ld [$key+20],$t1 + srlx $acc6,16,$acc6 + xor $acc3,$s0,$s0 + ld [$key+24],$t2 + srlx $acc7,24,$acc7 + xor $acc4,$s1,$s1 + ld [$key+28],$t3 ! + srlx $acc9,8,$acc9 + xor $acc5,$s1,$s1 + ldx [$tbl+2048+0],%g0 ! prefetch te4 + srlx $acc10,16,$acc10 + xor $acc6,$s1,$s1 + ldx [$tbl+2048+32],%g0 ! prefetch te4 + srlx $acc11,24,$acc11 + xor $acc7,$s1,$s1 + ldx [$tbl+2048+64],%g0 ! prefetch te4 + srlx $acc13,8,$acc13 + xor $acc8,$s2,$s2 + ldx [$tbl+2048+96],%g0 ! prefetch te4 + srlx $acc14,16,$acc14 ! + xor $acc9,$s2,$s2 + ldx [$tbl+2048+128],%g0 ! prefetch te4 + srlx $acc15,24,$acc15 + xor $acc10,$s2,$s2 + ldx [$tbl+2048+160],%g0 ! prefetch te4 + srl $s0,21,$acc0 + xor $acc11,$s2,$s2 + ldx [$tbl+2048+192],%g0 ! prefetch te4 + xor $acc12,$acc14,$acc14 + xor $acc13,$s3,$s3 + ldx [$tbl+2048+224],%g0 ! prefetch te4 + srl $s1,13,$acc1 ! + xor $acc14,$s3,$s3 + xor $acc15,$s3,$s3 + ba .Lenc_loop + and $acc0,2040,$acc0 + +.align 32 +.Lenc_last: + srlx $acc1,8,$acc1 ! + xor $acc0,$t0,$t0 + ld [$key+0],$s0 + srlx $acc2,16,$acc2 + xor $acc1,$t0,$t0 + ld [$key+4],$s1 + srlx $acc3,24,$acc3 + xor $acc2,$t0,$t0 + ld [$key+8],$s2 ! + srlx $acc5,8,$acc5 + xor $acc3,$t0,$t0 + ld [$key+12],$s3 + srlx $acc6,16,$acc6 + xor $acc4,$t1,$t1 + srlx $acc7,24,$acc7 + xor $acc5,$t1,$t1 + srlx $acc9,8,$acc9 ! + xor $acc6,$t1,$t1 + srlx $acc10,16,$acc10 + xor $acc7,$t1,$t1 + srlx $acc11,24,$acc11 + xor $acc8,$t2,$t2 + srlx $acc13,8,$acc13 + xor $acc9,$t2,$t2 + srlx $acc14,16,$acc14 ! + xor $acc10,$t2,$t2 + srlx $acc15,24,$acc15 + xor $acc11,$t2,$t2 + xor $acc12,$acc14,$acc14 + xor $acc13,$t3,$t3 + srl $t0,24,$acc0 + xor $acc14,$t3,$t3 + srl $t1,16,$acc1 ! + xor $acc15,$t3,$t3 + + srl $t2,8,$acc2 + and $acc1,255,$acc1 + ldub [$rounds+$acc0],$acc0 + srl $t1,24,$acc4 + and $acc2,255,$acc2 + ldub [$rounds+$acc1],$acc1 + srl $t2,16,$acc5 ! + and $t3,255,$acc3 + ldub [$rounds+$acc2],$acc2 + ldub [$rounds+$acc3],$acc3 + srl $t3,8,$acc6 + and $acc5,255,$acc5 + ldub [$rounds+$acc4],$acc4 + fmovs %f0,%f0 + srl $t2,24,$acc8 ! + and $acc6,255,$acc6 + ldub [$rounds+$acc5],$acc5 + srl $t3,16,$acc9 + and $t0,255,$acc7 + ldub [$rounds+$acc6],$acc6 + ldub [$rounds+$acc7],$acc7 + fmovs %f0,%f0 + srl $t0,8,$acc10 ! + and $acc9,255,$acc9 + ldub [$rounds+$acc8],$acc8 + srl $t3,24,$acc12 + and $acc10,255,$acc10 + ldub [$rounds+$acc9],$acc9 + srl $t0,16,$acc13 + and $t1,255,$acc11 + ldub [$rounds+$acc10],$acc10 ! + srl $t1,8,$acc14 + and $acc13,255,$acc13 + ldub [$rounds+$acc11],$acc11 + ldub [$rounds+$acc12],$acc12 + and $acc14,255,$acc14 + ldub [$rounds+$acc13],$acc13 + and $t2,255,$acc15 + ldub [$rounds+$acc14],$acc14 ! + + sll $acc0,24,$acc0 + xor $acc3,$s0,$s0 + ldub [$rounds+$acc15],$acc15 + sll $acc1,16,$acc1 + xor $acc0,$s0,$s0 + ldx [%sp+$bias+$frame+0],%i7 ! restore return address + fmovs %f0,%f0 + sll $acc2,8,$acc2 ! + xor $acc1,$s0,$s0 + sll $acc4,24,$acc4 + xor $acc2,$s0,$s0 + sll $acc5,16,$acc5 + xor $acc7,$s1,$s1 + sll $acc6,8,$acc6 + xor $acc4,$s1,$s1 + sll $acc8,24,$acc8 ! + xor $acc5,$s1,$s1 + sll $acc9,16,$acc9 + xor $acc11,$s2,$s2 + sll $acc10,8,$acc10 + xor $acc6,$s1,$s1 + sll $acc12,24,$acc12 + xor $acc8,$s2,$s2 + sll $acc13,16,$acc13 ! + xor $acc9,$s2,$s2 + sll $acc14,8,$acc14 + xor $acc10,$s2,$s2 + xor $acc12,$acc14,$acc14 + xor $acc13,$s3,$s3 + xor $acc14,$s3,$s3 + xor $acc15,$s3,$s3 + + ret + restore +.type _sparcv9_AES_encrypt,#function +.size _sparcv9_AES_encrypt,(.-_sparcv9_AES_encrypt) + +.align 32 +.globl AES_encrypt +AES_encrypt: + or %o0,%o1,%g1 + andcc %g1,3,%g0 + bnz,pn %xcc,.Lunaligned_enc + save %sp,-$frame,%sp + + ld [%i0+0],%o0 + ld [%i0+4],%o1 + ld [%i0+8],%o2 + ld [%i0+12],%o3 + +1: call .+8 + add %o7,AES_Te-1b,%o4 + call _sparcv9_AES_encrypt + mov %i2,%o5 + + st %o0,[%i1+0] + st %o1,[%i1+4] + st %o2,[%i1+8] + st %o3,[%i1+12] + + ret + restore + +.align 32 +.Lunaligned_enc: + ldub [%i0+0],%l0 + ldub [%i0+1],%l1 + ldub [%i0+2],%l2 + + sll %l0,24,%l0 + ldub [%i0+3],%l3 + sll %l1,16,%l1 + ldub [%i0+4],%l4 + sll %l2,8,%l2 + or %l1,%l0,%l0 + ldub [%i0+5],%l5 + sll %l4,24,%l4 + or %l3,%l2,%l2 + ldub [%i0+6],%l6 + sll %l5,16,%l5 + or %l0,%l2,%o0 + ldub [%i0+7],%l7 + + sll %l6,8,%l6 + or %l5,%l4,%l4 + ldub [%i0+8],%l0 + or %l7,%l6,%l6 + ldub [%i0+9],%l1 + or %l4,%l6,%o1 + ldub [%i0+10],%l2 + + sll %l0,24,%l0 + ldub [%i0+11],%l3 + sll %l1,16,%l1 + ldub [%i0+12],%l4 + sll %l2,8,%l2 + or %l1,%l0,%l0 + ldub [%i0+13],%l5 + sll %l4,24,%l4 + or %l3,%l2,%l2 + ldub [%i0+14],%l6 + sll %l5,16,%l5 + or %l0,%l2,%o2 + ldub [%i0+15],%l7 + + sll %l6,8,%l6 + or %l5,%l4,%l4 + or %l7,%l6,%l6 + or %l4,%l6,%o3 + +1: call .+8 + add %o7,AES_Te-1b,%o4 + call _sparcv9_AES_encrypt + mov %i2,%o5 + + srl %o0,24,%l0 + srl %o0,16,%l1 + stb %l0,[%i1+0] + srl %o0,8,%l2 + stb %l1,[%i1+1] + stb %l2,[%i1+2] + srl %o1,24,%l4 + stb %o0,[%i1+3] + + srl %o1,16,%l5 + stb %l4,[%i1+4] + srl %o1,8,%l6 + stb %l5,[%i1+5] + stb %l6,[%i1+6] + srl %o2,24,%l0 + stb %o1,[%i1+7] + + srl %o2,16,%l1 + stb %l0,[%i1+8] + srl %o2,8,%l2 + stb %l1,[%i1+9] + stb %l2,[%i1+10] + srl %o3,24,%l4 + stb %o2,[%i1+11] + + srl %o3,16,%l5 + stb %l4,[%i1+12] + srl %o3,8,%l6 + stb %l5,[%i1+13] + stb %l6,[%i1+14] + stb %o3,[%i1+15] + + ret + restore +.type AES_encrypt,#function +.size AES_encrypt,(.-AES_encrypt) + +___ + +$code.=<<___; +.align 256 +AES_Td: +___ +&_data_word( + 0x51f4a750, 0x7e416553, 0x1a17a4c3, 0x3a275e96, + 0x3bab6bcb, 0x1f9d45f1, 0xacfa58ab, 0x4be30393, + 0x2030fa55, 0xad766df6, 0x88cc7691, 0xf5024c25, + 0x4fe5d7fc, 0xc52acbd7, 0x26354480, 0xb562a38f, + 0xdeb15a49, 0x25ba1b67, 0x45ea0e98, 0x5dfec0e1, + 0xc32f7502, 0x814cf012, 0x8d4697a3, 0x6bd3f9c6, + 0x038f5fe7, 0x15929c95, 0xbf6d7aeb, 0x955259da, + 0xd4be832d, 0x587421d3, 0x49e06929, 0x8ec9c844, + 0x75c2896a, 0xf48e7978, 0x99583e6b, 0x27b971dd, + 0xbee14fb6, 0xf088ad17, 0xc920ac66, 0x7dce3ab4, + 0x63df4a18, 0xe51a3182, 0x97513360, 0x62537f45, + 0xb16477e0, 0xbb6bae84, 0xfe81a01c, 0xf9082b94, + 0x70486858, 0x8f45fd19, 0x94de6c87, 0x527bf8b7, + 0xab73d323, 0x724b02e2, 0xe31f8f57, 0x6655ab2a, + 0xb2eb2807, 0x2fb5c203, 0x86c57b9a, 0xd33708a5, + 0x302887f2, 0x23bfa5b2, 0x02036aba, 0xed16825c, + 0x8acf1c2b, 0xa779b492, 0xf307f2f0, 0x4e69e2a1, + 0x65daf4cd, 0x0605bed5, 0xd134621f, 0xc4a6fe8a, + 0x342e539d, 0xa2f355a0, 0x058ae132, 0xa4f6eb75, + 0x0b83ec39, 0x4060efaa, 0x5e719f06, 0xbd6e1051, + 0x3e218af9, 0x96dd063d, 0xdd3e05ae, 0x4de6bd46, + 0x91548db5, 0x71c45d05, 0x0406d46f, 0x605015ff, + 0x1998fb24, 0xd6bde997, 0x894043cc, 0x67d99e77, + 0xb0e842bd, 0x07898b88, 0xe7195b38, 0x79c8eedb, + 0xa17c0a47, 0x7c420fe9, 0xf8841ec9, 0x00000000, + 0x09808683, 0x322bed48, 0x1e1170ac, 0x6c5a724e, + 0xfd0efffb, 0x0f853856, 0x3daed51e, 0x362d3927, + 0x0a0fd964, 0x685ca621, 0x9b5b54d1, 0x24362e3a, + 0x0c0a67b1, 0x9357e70f, 0xb4ee96d2, 0x1b9b919e, + 0x80c0c54f, 0x61dc20a2, 0x5a774b69, 0x1c121a16, + 0xe293ba0a, 0xc0a02ae5, 0x3c22e043, 0x121b171d, + 0x0e090d0b, 0xf28bc7ad, 0x2db6a8b9, 0x141ea9c8, + 0x57f11985, 0xaf75074c, 0xee99ddbb, 0xa37f60fd, + 0xf701269f, 0x5c72f5bc, 0x44663bc5, 0x5bfb7e34, + 0x8b432976, 0xcb23c6dc, 0xb6edfc68, 0xb8e4f163, + 0xd731dcca, 0x42638510, 0x13972240, 0x84c61120, + 0x854a247d, 0xd2bb3df8, 0xaef93211, 0xc729a16d, + 0x1d9e2f4b, 0xdcb230f3, 0x0d8652ec, 0x77c1e3d0, + 0x2bb3166c, 0xa970b999, 0x119448fa, 0x47e96422, + 0xa8fc8cc4, 0xa0f03f1a, 0x567d2cd8, 0x223390ef, + 0x87494ec7, 0xd938d1c1, 0x8ccaa2fe, 0x98d40b36, + 0xa6f581cf, 0xa57ade28, 0xdab78e26, 0x3fadbfa4, + 0x2c3a9de4, 0x5078920d, 0x6a5fcc9b, 0x547e4662, + 0xf68d13c2, 0x90d8b8e8, 0x2e39f75e, 0x82c3aff5, + 0x9f5d80be, 0x69d0937c, 0x6fd52da9, 0xcf2512b3, + 0xc8ac993b, 0x10187da7, 0xe89c636e, 0xdb3bbb7b, + 0xcd267809, 0x6e5918f4, 0xec9ab701, 0x834f9aa8, + 0xe6956e65, 0xaaffe67e, 0x21bccf08, 0xef15e8e6, + 0xbae79bd9, 0x4a6f36ce, 0xea9f09d4, 0x29b07cd6, + 0x31a4b2af, 0x2a3f2331, 0xc6a59430, 0x35a266c0, + 0x744ebc37, 0xfc82caa6, 0xe090d0b0, 0x33a7d815, + 0xf104984a, 0x41ecdaf7, 0x7fcd500e, 0x1791f62f, + 0x764dd68d, 0x43efb04d, 0xccaa4d54, 0xe49604df, + 0x9ed1b5e3, 0x4c6a881b, 0xc12c1fb8, 0x4665517f, + 0x9d5eea04, 0x018c355d, 0xfa877473, 0xfb0b412e, + 0xb3671d5a, 0x92dbd252, 0xe9105633, 0x6dd64713, + 0x9ad7618c, 0x37a10c7a, 0x59f8148e, 0xeb133c89, + 0xcea927ee, 0xb761c935, 0xe11ce5ed, 0x7a47b13c, + 0x9cd2df59, 0x55f2733f, 0x1814ce79, 0x73c737bf, + 0x53f7cdea, 0x5ffdaa5b, 0xdf3d6f14, 0x7844db86, + 0xcaaff381, 0xb968c43e, 0x3824342c, 0xc2a3405f, + 0x161dc372, 0xbce2250c, 0x283c498b, 0xff0d9541, + 0x39a80171, 0x080cb3de, 0xd8b4e49c, 0x6456c190, + 0x7bcb8461, 0xd532b670, 0x486c5c74, 0xd0b85742); +$code.=<<___; + .byte 0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38 + .byte 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb + .byte 0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87 + .byte 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb + .byte 0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d + .byte 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e + .byte 0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2 + .byte 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25 + .byte 0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16 + .byte 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92 + .byte 0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda + .byte 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84 + .byte 0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a + .byte 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06 + .byte 0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02 + .byte 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b + .byte 0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea + .byte 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73 + .byte 0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85 + .byte 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e + .byte 0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89 + .byte 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b + .byte 0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20 + .byte 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4 + .byte 0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31 + .byte 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f + .byte 0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d + .byte 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef + .byte 0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0 + .byte 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61 + .byte 0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26 + .byte 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d +.type AES_Td,#object +.size AES_Td,(.-AES_Td) + +.align 64 +.skip 16 +_sparcv9_AES_decrypt: + save %sp,-$frame-$locals,%sp + stx %i7,[%sp+$bias+$frame+0] ! off-load return address + ld [$key+240],$rounds + ld [$key+0],$t0 + ld [$key+4],$t1 ! + ld [$key+8],$t2 + ld [$key+12],$t3 + srl $rounds,1,$rounds + xor $t0,$s0,$s0 + ld [$key+16],$t0 + xor $t1,$s1,$s1 + ld [$key+20],$t1 + srl $s0,21,$acc0 ! + xor $t2,$s2,$s2 + ld [$key+24],$t2 + xor $t3,$s3,$s3 + and $acc0,2040,$acc0 + ld [$key+28],$t3 + srl $s3,13,$acc1 + nop +.Ldec_loop: + srl $s2,5,$acc2 ! + and $acc1,2040,$acc1 + ldx [$tbl+$acc0],$acc0 + sll $s1,3,$acc3 + and $acc2,2040,$acc2 + ldx [$tbl+$acc1],$acc1 + srl $s1,21,$acc4 + and $acc3,2040,$acc3 + ldx [$tbl+$acc2],$acc2 ! + srl $s0,13,$acc5 + and $acc4,2040,$acc4 + ldx [$tbl+$acc3],$acc3 + srl $s3,5,$acc6 + and $acc5,2040,$acc5 + ldx [$tbl+$acc4],$acc4 + fmovs %f0,%f0 + sll $s2,3,$acc7 ! + and $acc6,2040,$acc6 + ldx [$tbl+$acc5],$acc5 + srl $s2,21,$acc8 + and $acc7,2040,$acc7 + ldx [$tbl+$acc6],$acc6 + srl $s1,13,$acc9 + and $acc8,2040,$acc8 + ldx [$tbl+$acc7],$acc7 ! + srl $s0,5,$acc10 + and $acc9,2040,$acc9 + ldx [$tbl+$acc8],$acc8 + sll $s3,3,$acc11 + and $acc10,2040,$acc10 + ldx [$tbl+$acc9],$acc9 + fmovs %f0,%f0 + srl $s3,21,$acc12 ! + and $acc11,2040,$acc11 + ldx [$tbl+$acc10],$acc10 + srl $s2,13,$acc13 + and $acc12,2040,$acc12 + ldx [$tbl+$acc11],$acc11 + srl $s1,5,$acc14 + and $acc13,2040,$acc13 + ldx [$tbl+$acc12],$acc12 ! + sll $s0,3,$acc15 + and $acc14,2040,$acc14 + ldx [$tbl+$acc13],$acc13 + and $acc15,2040,$acc15 + add $key,32,$key + ldx [$tbl+$acc14],$acc14 + fmovs %f0,%f0 + subcc $rounds,1,$rounds ! + ldx [$tbl+$acc15],$acc15 + bz,a,pn %icc,.Ldec_last + add $tbl,2048,$rounds + + srlx $acc1,8,$acc1 + xor $acc0,$t0,$t0 + ld [$key+0],$s0 + fmovs %f0,%f0 + srlx $acc2,16,$acc2 ! + xor $acc1,$t0,$t0 + ld [$key+4],$s1 + srlx $acc3,24,$acc3 + xor $acc2,$t0,$t0 + ld [$key+8],$s2 + srlx $acc5,8,$acc5 + xor $acc3,$t0,$t0 + ld [$key+12],$s3 ! + srlx $acc6,16,$acc6 + xor $acc4,$t1,$t1 + fmovs %f0,%f0 + srlx $acc7,24,$acc7 + xor $acc5,$t1,$t1 + srlx $acc9,8,$acc9 + xor $acc6,$t1,$t1 + srlx $acc10,16,$acc10 ! + xor $acc7,$t1,$t1 + srlx $acc11,24,$acc11 + xor $acc8,$t2,$t2 + srlx $acc13,8,$acc13 + xor $acc9,$t2,$t2 + srlx $acc14,16,$acc14 + xor $acc10,$t2,$t2 + srlx $acc15,24,$acc15 ! + xor $acc11,$t2,$t2 + xor $acc12,$acc14,$acc14 + xor $acc13,$t3,$t3 + srl $t0,21,$acc0 + xor $acc14,$t3,$t3 + xor $acc15,$t3,$t3 + srl $t3,13,$acc1 + + and $acc0,2040,$acc0 ! + srl $t2,5,$acc2 + and $acc1,2040,$acc1 + ldx [$tbl+$acc0],$acc0 + sll $t1,3,$acc3 + and $acc2,2040,$acc2 + ldx [$tbl+$acc1],$acc1 + fmovs %f0,%f0 + srl $t1,21,$acc4 ! + and $acc3,2040,$acc3 + ldx [$tbl+$acc2],$acc2 + srl $t0,13,$acc5 + and $acc4,2040,$acc4 + ldx [$tbl+$acc3],$acc3 + srl $t3,5,$acc6 + and $acc5,2040,$acc5 + ldx [$tbl+$acc4],$acc4 ! + sll $t2,3,$acc7 + and $acc6,2040,$acc6 + ldx [$tbl+$acc5],$acc5 + srl $t2,21,$acc8 + and $acc7,2040,$acc7 + ldx [$tbl+$acc6],$acc6 + fmovs %f0,%f0 + srl $t1,13,$acc9 ! + and $acc8,2040,$acc8 + ldx [$tbl+$acc7],$acc7 + srl $t0,5,$acc10 + and $acc9,2040,$acc9 + ldx [$tbl+$acc8],$acc8 + sll $t3,3,$acc11 + and $acc10,2040,$acc10 + ldx [$tbl+$acc9],$acc9 ! + srl $t3,21,$acc12 + and $acc11,2040,$acc11 + ldx [$tbl+$acc10],$acc10 + srl $t2,13,$acc13 + and $acc12,2040,$acc12 + ldx [$tbl+$acc11],$acc11 + fmovs %f0,%f0 + srl $t1,5,$acc14 ! + and $acc13,2040,$acc13 + ldx [$tbl+$acc12],$acc12 + sll $t0,3,$acc15 + and $acc14,2040,$acc14 + ldx [$tbl+$acc13],$acc13 + srlx $acc1,8,$acc1 + and $acc15,2040,$acc15 + ldx [$tbl+$acc14],$acc14 ! + + srlx $acc2,16,$acc2 + xor $acc0,$s0,$s0 + ldx [$tbl+$acc15],$acc15 + srlx $acc3,24,$acc3 + xor $acc1,$s0,$s0 + ld [$key+16],$t0 + fmovs %f0,%f0 + srlx $acc5,8,$acc5 ! + xor $acc2,$s0,$s0 + ld [$key+20],$t1 + srlx $acc6,16,$acc6 + xor $acc3,$s0,$s0 + ld [$key+24],$t2 + srlx $acc7,24,$acc7 + xor $acc4,$s1,$s1 + ld [$key+28],$t3 ! + srlx $acc9,8,$acc9 + xor $acc5,$s1,$s1 + ldx [$tbl+2048+0],%g0 ! prefetch td4 + srlx $acc10,16,$acc10 + xor $acc6,$s1,$s1 + ldx [$tbl+2048+32],%g0 ! prefetch td4 + srlx $acc11,24,$acc11 + xor $acc7,$s1,$s1 + ldx [$tbl+2048+64],%g0 ! prefetch td4 + srlx $acc13,8,$acc13 + xor $acc8,$s2,$s2 + ldx [$tbl+2048+96],%g0 ! prefetch td4 + srlx $acc14,16,$acc14 ! + xor $acc9,$s2,$s2 + ldx [$tbl+2048+128],%g0 ! prefetch td4 + srlx $acc15,24,$acc15 + xor $acc10,$s2,$s2 + ldx [$tbl+2048+160],%g0 ! prefetch td4 + srl $s0,21,$acc0 + xor $acc11,$s2,$s2 + ldx [$tbl+2048+192],%g0 ! prefetch td4 + xor $acc12,$acc14,$acc14 + xor $acc13,$s3,$s3 + ldx [$tbl+2048+224],%g0 ! prefetch td4 + and $acc0,2040,$acc0 ! + xor $acc14,$s3,$s3 + xor $acc15,$s3,$s3 + ba .Ldec_loop + srl $s3,13,$acc1 + +.align 32 +.Ldec_last: + srlx $acc1,8,$acc1 ! + xor $acc0,$t0,$t0 + ld [$key+0],$s0 + srlx $acc2,16,$acc2 + xor $acc1,$t0,$t0 + ld [$key+4],$s1 + srlx $acc3,24,$acc3 + xor $acc2,$t0,$t0 + ld [$key+8],$s2 ! + srlx $acc5,8,$acc5 + xor $acc3,$t0,$t0 + ld [$key+12],$s3 + srlx $acc6,16,$acc6 + xor $acc4,$t1,$t1 + srlx $acc7,24,$acc7 + xor $acc5,$t1,$t1 + srlx $acc9,8,$acc9 ! + xor $acc6,$t1,$t1 + srlx $acc10,16,$acc10 + xor $acc7,$t1,$t1 + srlx $acc11,24,$acc11 + xor $acc8,$t2,$t2 + srlx $acc13,8,$acc13 + xor $acc9,$t2,$t2 + srlx $acc14,16,$acc14 ! + xor $acc10,$t2,$t2 + srlx $acc15,24,$acc15 + xor $acc11,$t2,$t2 + xor $acc12,$acc14,$acc14 + xor $acc13,$t3,$t3 + srl $t0,24,$acc0 + xor $acc14,$t3,$t3 + xor $acc15,$t3,$t3 ! + srl $t3,16,$acc1 + + srl $t2,8,$acc2 + and $acc1,255,$acc1 + ldub [$rounds+$acc0],$acc0 + srl $t1,24,$acc4 + and $acc2,255,$acc2 + ldub [$rounds+$acc1],$acc1 + srl $t0,16,$acc5 ! + and $t1,255,$acc3 + ldub [$rounds+$acc2],$acc2 + ldub [$rounds+$acc3],$acc3 + srl $t3,8,$acc6 + and $acc5,255,$acc5 + ldub [$rounds+$acc4],$acc4 + fmovs %f0,%f0 + srl $t2,24,$acc8 ! + and $acc6,255,$acc6 + ldub [$rounds+$acc5],$acc5 + srl $t1,16,$acc9 + and $t2,255,$acc7 + ldub [$rounds+$acc6],$acc6 + ldub [$rounds+$acc7],$acc7 + fmovs %f0,%f0 + srl $t0,8,$acc10 ! + and $acc9,255,$acc9 + ldub [$rounds+$acc8],$acc8 + srl $t3,24,$acc12 + and $acc10,255,$acc10 + ldub [$rounds+$acc9],$acc9 + srl $t2,16,$acc13 + and $t3,255,$acc11 + ldub [$rounds+$acc10],$acc10 ! + srl $t1,8,$acc14 + and $acc13,255,$acc13 + ldub [$rounds+$acc11],$acc11 + ldub [$rounds+$acc12],$acc12 + and $acc14,255,$acc14 + ldub [$rounds+$acc13],$acc13 + and $t0,255,$acc15 + ldub [$rounds+$acc14],$acc14 ! + + sll $acc0,24,$acc0 + xor $acc3,$s0,$s0 + ldub [$rounds+$acc15],$acc15 + sll $acc1,16,$acc1 + xor $acc0,$s0,$s0 + ldx [%sp+$bias+$frame+0],%i7 ! restore return address + fmovs %f0,%f0 + sll $acc2,8,$acc2 ! + xor $acc1,$s0,$s0 + sll $acc4,24,$acc4 + xor $acc2,$s0,$s0 + sll $acc5,16,$acc5 + xor $acc7,$s1,$s1 + sll $acc6,8,$acc6 + xor $acc4,$s1,$s1 + sll $acc8,24,$acc8 ! + xor $acc5,$s1,$s1 + sll $acc9,16,$acc9 + xor $acc11,$s2,$s2 + sll $acc10,8,$acc10 + xor $acc6,$s1,$s1 + sll $acc12,24,$acc12 + xor $acc8,$s2,$s2 + sll $acc13,16,$acc13 ! + xor $acc9,$s2,$s2 + sll $acc14,8,$acc14 + xor $acc10,$s2,$s2 + xor $acc12,$acc14,$acc14 + xor $acc13,$s3,$s3 + xor $acc14,$s3,$s3 + xor $acc15,$s3,$s3 + + ret + restore +.type _sparcv9_AES_decrypt,#function +.size _sparcv9_AES_decrypt,(.-_sparcv9_AES_decrypt) + +.align 32 +.globl AES_decrypt +AES_decrypt: + or %o0,%o1,%g1 + andcc %g1,3,%g0 + bnz,pn %xcc,.Lunaligned_dec + save %sp,-$frame,%sp + + ld [%i0+0],%o0 + ld [%i0+4],%o1 + ld [%i0+8],%o2 + ld [%i0+12],%o3 + +1: call .+8 + add %o7,AES_Td-1b,%o4 + call _sparcv9_AES_decrypt + mov %i2,%o5 + + st %o0,[%i1+0] + st %o1,[%i1+4] + st %o2,[%i1+8] + st %o3,[%i1+12] + + ret + restore + +.align 32 +.Lunaligned_dec: + ldub [%i0+0],%l0 + ldub [%i0+1],%l1 + ldub [%i0+2],%l2 + + sll %l0,24,%l0 + ldub [%i0+3],%l3 + sll %l1,16,%l1 + ldub [%i0+4],%l4 + sll %l2,8,%l2 + or %l1,%l0,%l0 + ldub [%i0+5],%l5 + sll %l4,24,%l4 + or %l3,%l2,%l2 + ldub [%i0+6],%l6 + sll %l5,16,%l5 + or %l0,%l2,%o0 + ldub [%i0+7],%l7 + + sll %l6,8,%l6 + or %l5,%l4,%l4 + ldub [%i0+8],%l0 + or %l7,%l6,%l6 + ldub [%i0+9],%l1 + or %l4,%l6,%o1 + ldub [%i0+10],%l2 + + sll %l0,24,%l0 + ldub [%i0+11],%l3 + sll %l1,16,%l1 + ldub [%i0+12],%l4 + sll %l2,8,%l2 + or %l1,%l0,%l0 + ldub [%i0+13],%l5 + sll %l4,24,%l4 + or %l3,%l2,%l2 + ldub [%i0+14],%l6 + sll %l5,16,%l5 + or %l0,%l2,%o2 + ldub [%i0+15],%l7 + + sll %l6,8,%l6 + or %l5,%l4,%l4 + or %l7,%l6,%l6 + or %l4,%l6,%o3 + +1: call .+8 + add %o7,AES_Td-1b,%o4 + call _sparcv9_AES_decrypt + mov %i2,%o5 + + srl %o0,24,%l0 + srl %o0,16,%l1 + stb %l0,[%i1+0] + srl %o0,8,%l2 + stb %l1,[%i1+1] + stb %l2,[%i1+2] + srl %o1,24,%l4 + stb %o0,[%i1+3] + + srl %o1,16,%l5 + stb %l4,[%i1+4] + srl %o1,8,%l6 + stb %l5,[%i1+5] + stb %l6,[%i1+6] + srl %o2,24,%l0 + stb %o1,[%i1+7] + + srl %o2,16,%l1 + stb %l0,[%i1+8] + srl %o2,8,%l2 + stb %l1,[%i1+9] + stb %l2,[%i1+10] + srl %o3,24,%l4 + stb %o2,[%i1+11] + + srl %o3,16,%l5 + stb %l4,[%i1+12] + srl %o3,8,%l6 + stb %l5,[%i1+13] + stb %l6,[%i1+14] + stb %o3,[%i1+15] + + ret + restore +.type AES_decrypt,#function +.size AES_decrypt,(.-AES_decrypt) +___ + +# fmovs instructions substituting for FP nops were originally added +# to meet specific instruction alignment requirements to maximize ILP. +# As UltraSPARC T1, a.k.a. Niagara, has shared FPU, FP nops can have +# undesired effect, so just omit them and sacrifice some portion of +# percent in performance... +$code =~ s/fmovs.*$//gem; + +print $code; diff --git a/openssl/crypto/aes/asm/aes-x86_64.pl b/openssl/crypto/aes/asm/aes-x86_64.pl new file mode 100644 index 000000000..f616f1751 --- /dev/null +++ b/openssl/crypto/aes/asm/aes-x86_64.pl @@ -0,0 +1,1579 @@ +#!/usr/bin/env perl +# +# ==================================================================== +# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# project. Rights for redistribution and usage in source and binary +# forms are granted according to the OpenSSL license. +# ==================================================================== +# +# Version 1.2. +# +# aes-*-cbc benchmarks are improved by >70% [compared to gcc 3.3.2 on +# Opteron 240 CPU] plus all the bells-n-whistles from 32-bit version +# [you'll notice a lot of resemblance], such as compressed S-boxes +# in little-endian byte order, prefetch of these tables in CBC mode, +# as well as avoiding L1 cache aliasing between stack frame and key +# schedule and already mentioned tables, compressed Td4... +# +# Performance in number of cycles per processed byte for 128-bit key: +# +# ECB CBC encrypt +# AMD64 13.7 13.0(*) +# EM64T 20.2 18.6(*) +# +# (*) CBC benchmarks are better than ECB thanks to custom ABI used +# by the private block encryption function. + +$verticalspin=1; # unlike 32-bit version $verticalspin performs + # ~15% better on both AMD and Intel cores +$output=shift; +open STDOUT,"| $^X ../perlasm/x86_64-xlate.pl $output"; + +$code=".text\n"; + +$s0="%eax"; +$s1="%ebx"; +$s2="%ecx"; +$s3="%edx"; +$acc0="%esi"; +$acc1="%edi"; +$acc2="%ebp"; +$inp="%r8"; +$out="%r9"; +$t0="%r10d"; +$t1="%r11d"; +$t2="%r12d"; +$rnds="%r13d"; +$sbox="%r14"; +$key="%r15"; + +sub hi() { my $r=shift; $r =~ s/%[er]([a-d])x/%\1h/; $r; } +sub lo() { my $r=shift; $r =~ s/%[er]([a-d])x/%\1l/; + $r =~ s/%[er]([sd]i)/%\1l/; + $r =~ s/%(r[0-9]+)[d]?/%\1b/; $r; } +sub _data_word() +{ my $i; + while(defined($i=shift)) { $code.=sprintf".long\t0x%08x,0x%08x\n",$i,$i; } +} +sub data_word() +{ my $i; + my $last=pop(@_); + $code.=".long\t"; + while(defined($i=shift)) { $code.=sprintf"0x%08x,",$i; } + $code.=sprintf"0x%08x\n",$last; +} + +sub data_byte() +{ my $i; + my $last=pop(@_); + $code.=".byte\t"; + while(defined($i=shift)) { $code.=sprintf"0x%02x,",$i&0xff; } + $code.=sprintf"0x%02x\n",$last&0xff; +} + +sub encvert() +{ my $t3="%r8d"; # zaps $inp! + +$code.=<<___; + # favor 3-way issue Opteron pipeline... + movzb `&lo("$s0")`,$acc0 + movzb `&lo("$s1")`,$acc1 + movzb `&lo("$s2")`,$acc2 + mov 0($sbox,$acc0,8),$t0 + mov 0($sbox,$acc1,8),$t1 + mov 0($sbox,$acc2,8),$t2 + + movzb `&hi("$s1")`,$acc0 + movzb `&hi("$s2")`,$acc1 + movzb `&lo("$s3")`,$acc2 + xor 3($sbox,$acc0,8),$t0 + xor 3($sbox,$acc1,8),$t1 + mov 0($sbox,$acc2,8),$t3 + + movzb `&hi("$s3")`,$acc0 + shr \$16,$s2 + movzb `&hi("$s0")`,$acc2 + xor 3($sbox,$acc0,8),$t2 + shr \$16,$s3 + xor 3($sbox,$acc2,8),$t3 + + shr \$16,$s1 + lea 16($key),$key + shr \$16,$s0 + + movzb `&lo("$s2")`,$acc0 + movzb `&lo("$s3")`,$acc1 + movzb `&lo("$s0")`,$acc2 + xor 2($sbox,$acc0,8),$t0 + xor 2($sbox,$acc1,8),$t1 + xor 2($sbox,$acc2,8),$t2 + + movzb `&hi("$s3")`,$acc0 + movzb `&hi("$s0")`,$acc1 + movzb `&lo("$s1")`,$acc2 + xor 1($sbox,$acc0,8),$t0 + xor 1($sbox,$acc1,8),$t1 + xor 2($sbox,$acc2,8),$t3 + + mov 12($key),$s3 + movzb `&hi("$s1")`,$acc1 + movzb `&hi("$s2")`,$acc2 + mov 0($key),$s0 + xor 1($sbox,$acc1,8),$t2 + xor 1($sbox,$acc2,8),$t3 + + mov 4($key),$s1 + mov 8($key),$s2 + xor $t0,$s0 + xor $t1,$s1 + xor $t2,$s2 + xor $t3,$s3 +___ +} + +sub enclastvert() +{ my $t3="%r8d"; # zaps $inp! + +$code.=<<___; + movzb `&lo("$s0")`,$acc0 + movzb `&lo("$s1")`,$acc1 + movzb `&lo("$s2")`,$acc2 + mov 2($sbox,$acc0,8),$t0 + mov 2($sbox,$acc1,8),$t1 + mov 2($sbox,$acc2,8),$t2 + + and \$0x000000ff,$t0 + and \$0x000000ff,$t1 + and \$0x000000ff,$t2 + + movzb `&lo("$s3")`,$acc0 + movzb `&hi("$s1")`,$acc1 + movzb `&hi("$s2")`,$acc2 + mov 2($sbox,$acc0,8),$t3 + mov 0($sbox,$acc1,8),$acc1 #$t0 + mov 0($sbox,$acc2,8),$acc2 #$t1 + + and \$0x000000ff,$t3 + and \$0x0000ff00,$acc1 + and \$0x0000ff00,$acc2 + + xor $acc1,$t0 + xor $acc2,$t1 + shr \$16,$s2 + + movzb `&hi("$s3")`,$acc0 + movzb `&hi("$s0")`,$acc1 + shr \$16,$s3 + mov 0($sbox,$acc0,8),$acc0 #$t2 + mov 0($sbox,$acc1,8),$acc1 #$t3 + + and \$0x0000ff00,$acc0 + and \$0x0000ff00,$acc1 + shr \$16,$s1 + xor $acc0,$t2 + xor $acc1,$t3 + shr \$16,$s0 + + movzb `&lo("$s2")`,$acc0 + movzb `&lo("$s3")`,$acc1 + movzb `&lo("$s0")`,$acc2 + mov 0($sbox,$acc0,8),$acc0 #$t0 + mov 0($sbox,$acc1,8),$acc1 #$t1 + mov 0($sbox,$acc2,8),$acc2 #$t2 + + and \$0x00ff0000,$acc0 + and \$0x00ff0000,$acc1 + and \$0x00ff0000,$acc2 + + xor $acc0,$t0 + xor $acc1,$t1 + xor $acc2,$t2 + + movzb `&lo("$s1")`,$acc0 + movzb `&hi("$s3")`,$acc1 + movzb `&hi("$s0")`,$acc2 + mov 0($sbox,$acc0,8),$acc0 #$t3 + mov 2($sbox,$acc1,8),$acc1 #$t0 + mov 2($sbox,$acc2,8),$acc2 #$t1 + + and \$0x00ff0000,$acc0 + and \$0xff000000,$acc1 + and \$0xff000000,$acc2 + + xor $acc0,$t3 + xor $acc1,$t0 + xor $acc2,$t1 + + movzb `&hi("$s1")`,$acc0 + movzb `&hi("$s2")`,$acc1 + mov 16+12($key),$s3 + mov 2($sbox,$acc0,8),$acc0 #$t2 + mov 2($sbox,$acc1,8),$acc1 #$t3 + mov 16+0($key),$s0 + + and \$0xff000000,$acc0 + and \$0xff000000,$acc1 + + xor $acc0,$t2 + xor $acc1,$t3 + + mov 16+4($key),$s1 + mov 16+8($key),$s2 + xor $t0,$s0 + xor $t1,$s1 + xor $t2,$s2 + xor $t3,$s3 +___ +} + +sub encstep() +{ my ($i,@s) = @_; + my $tmp0=$acc0; + my $tmp1=$acc1; + my $tmp2=$acc2; + my $out=($t0,$t1,$t2,$s[0])[$i]; + + if ($i==3) { + $tmp0=$s[1]; + $tmp1=$s[2]; + $tmp2=$s[3]; + } + $code.=" movzb ".&lo($s[0]).",$out\n"; + $code.=" mov $s[2],$tmp1\n" if ($i!=3); + $code.=" lea 16($key),$key\n" if ($i==0); + + $code.=" movzb ".&hi($s[1]).",$tmp0\n"; + $code.=" mov 0($sbox,$out,8),$out\n"; + + $code.=" shr \$16,$tmp1\n"; + $code.=" mov $s[3],$tmp2\n" if ($i!=3); + $code.=" xor 3($sbox,$tmp0,8),$out\n"; + + $code.=" movzb ".&lo($tmp1).",$tmp1\n"; + $code.=" shr \$24,$tmp2\n"; + $code.=" xor 4*$i($key),$out\n"; + + $code.=" xor 2($sbox,$tmp1,8),$out\n"; + $code.=" xor 1($sbox,$tmp2,8),$out\n"; + + $code.=" mov $t0,$s[1]\n" if ($i==3); + $code.=" mov $t1,$s[2]\n" if ($i==3); + $code.=" mov $t2,$s[3]\n" if ($i==3); + $code.="\n"; +} + +sub enclast() +{ my ($i,@s)=@_; + my $tmp0=$acc0; + my $tmp1=$acc1; + my $tmp2=$acc2; + my $out=($t0,$t1,$t2,$s[0])[$i]; + + if ($i==3) { + $tmp0=$s[1]; + $tmp1=$s[2]; + $tmp2=$s[3]; + } + $code.=" movzb ".&lo($s[0]).",$out\n"; + $code.=" mov $s[2],$tmp1\n" if ($i!=3); + + $code.=" mov 2($sbox,$out,8),$out\n"; + $code.=" shr \$16,$tmp1\n"; + $code.=" mov $s[3],$tmp2\n" if ($i!=3); + + $code.=" and \$0x000000ff,$out\n"; + $code.=" movzb ".&hi($s[1]).",$tmp0\n"; + $code.=" movzb ".&lo($tmp1).",$tmp1\n"; + $code.=" shr \$24,$tmp2\n"; + + $code.=" mov 0($sbox,$tmp0,8),$tmp0\n"; + $code.=" mov 0($sbox,$tmp1,8),$tmp1\n"; + $code.=" mov 2($sbox,$tmp2,8),$tmp2\n"; + + $code.=" and \$0x0000ff00,$tmp0\n"; + $code.=" and \$0x00ff0000,$tmp1\n"; + $code.=" and \$0xff000000,$tmp2\n"; + + $code.=" xor $tmp0,$out\n"; + $code.=" mov $t0,$s[1]\n" if ($i==3); + $code.=" xor $tmp1,$out\n"; + $code.=" mov $t1,$s[2]\n" if ($i==3); + $code.=" xor $tmp2,$out\n"; + $code.=" mov $t2,$s[3]\n" if ($i==3); + $code.="\n"; +} + +$code.=<<___; +.type _x86_64_AES_encrypt,\@abi-omnipotent +.align 16 +_x86_64_AES_encrypt: + xor 0($key),$s0 # xor with key + xor 4($key),$s1 + xor 8($key),$s2 + xor 12($key),$s3 + + mov 240($key),$rnds # load key->rounds + sub \$1,$rnds + jmp .Lenc_loop +.align 16 +.Lenc_loop: +___ + if ($verticalspin) { &encvert(); } + else { &encstep(0,$s0,$s1,$s2,$s3); + &encstep(1,$s1,$s2,$s3,$s0); + &encstep(2,$s2,$s3,$s0,$s1); + &encstep(3,$s3,$s0,$s1,$s2); + } +$code.=<<___; + sub \$1,$rnds + jnz .Lenc_loop +___ + if ($verticalspin) { &enclastvert(); } + else { &enclast(0,$s0,$s1,$s2,$s3); + &enclast(1,$s1,$s2,$s3,$s0); + &enclast(2,$s2,$s3,$s0,$s1); + &enclast(3,$s3,$s0,$s1,$s2); + $code.=<<___; + xor 16+0($key),$s0 # xor with key + xor 16+4($key),$s1 + xor 16+8($key),$s2 + xor 16+12($key),$s3 +___ + } +$code.=<<___; + .byte 0xf3,0xc3 # rep ret +.size _x86_64_AES_encrypt,.-_x86_64_AES_encrypt +___ + +# void AES_encrypt (const void *inp,void *out,const AES_KEY *key); +$code.=<<___; +.globl AES_encrypt +.type AES_encrypt,\@function,3 +.align 16 +AES_encrypt: + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + mov %rdx,$key + mov %rdi,$inp + mov %rsi,$out + + .picmeup $sbox + lea AES_Te-.($sbox),$sbox + + mov 0($inp),$s0 + mov 4($inp),$s1 + mov 8($inp),$s2 + mov 12($inp),$s3 + + call _x86_64_AES_encrypt + + mov $s0,0($out) + mov $s1,4($out) + mov $s2,8($out) + mov $s3,12($out) + + pop %r15 + pop %r14 + pop %r13 + pop %r12 + pop %rbp + pop %rbx + ret +.size AES_encrypt,.-AES_encrypt +___ + +#------------------------------------------------------------------# + +sub decvert() +{ my $t3="%r8d"; # zaps $inp! + +$code.=<<___; + # favor 3-way issue Opteron pipeline... + movzb `&lo("$s0")`,$acc0 + movzb `&lo("$s1")`,$acc1 + movzb `&lo("$s2")`,$acc2 + mov 0($sbox,$acc0,8),$t0 + mov 0($sbox,$acc1,8),$t1 + mov 0($sbox,$acc2,8),$t2 + + movzb `&hi("$s3")`,$acc0 + movzb `&hi("$s0")`,$acc1 + movzb `&lo("$s3")`,$acc2 + xor 3($sbox,$acc0,8),$t0 + xor 3($sbox,$acc1,8),$t1 + mov 0($sbox,$acc2,8),$t3 + + movzb `&hi("$s1")`,$acc0 + shr \$16,$s0 + movzb `&hi("$s2")`,$acc2 + xor 3($sbox,$acc0,8),$t2 + shr \$16,$s3 + xor 3($sbox,$acc2,8),$t3 + + shr \$16,$s1 + lea 16($key),$key + shr \$16,$s2 + + movzb `&lo("$s2")`,$acc0 + movzb `&lo("$s3")`,$acc1 + movzb `&lo("$s0")`,$acc2 + xor 2($sbox,$acc0,8),$t0 + xor 2($sbox,$acc1,8),$t1 + xor 2($sbox,$acc2,8),$t2 + + movzb `&hi("$s1")`,$acc0 + movzb `&hi("$s2")`,$acc1 + movzb `&lo("$s1")`,$acc2 + xor 1($sbox,$acc0,8),$t0 + xor 1($sbox,$acc1,8),$t1 + xor 2($sbox,$acc2,8),$t3 + + movzb `&hi("$s3")`,$acc0 + mov 12($key),$s3 + movzb `&hi("$s0")`,$acc2 + xor 1($sbox,$acc0,8),$t2 + mov 0($key),$s0 + xor 1($sbox,$acc2,8),$t3 + + xor $t0,$s0 + mov 4($key),$s1 + mov 8($key),$s2 + xor $t2,$s2 + xor $t1,$s1 + xor $t3,$s3 +___ +} + +sub declastvert() +{ my $t3="%r8d"; # zaps $inp! + +$code.=<<___; + movzb `&lo("$s0")`,$acc0 + movzb `&lo("$s1")`,$acc1 + movzb `&lo("$s2")`,$acc2 + movzb 2048($sbox,$acc0,1),$t0 + movzb 2048($sbox,$acc1,1),$t1 + movzb 2048($sbox,$acc2,1),$t2 + + movzb `&lo("$s3")`,$acc0 + movzb `&hi("$s3")`,$acc1 + movzb `&hi("$s0")`,$acc2 + movzb 2048($sbox,$acc0,1),$t3 + movzb 2048($sbox,$acc1,1),$acc1 #$t0 + movzb 2048($sbox,$acc2,1),$acc2 #$t1 + + shl \$8,$acc1 + shl \$8,$acc2 + + xor $acc1,$t0 + xor $acc2,$t1 + shr \$16,$s3 + + movzb `&hi("$s1")`,$acc0 + movzb `&hi("$s2")`,$acc1 + shr \$16,$s0 + movzb 2048($sbox,$acc0,1),$acc0 #$t2 + movzb 2048($sbox,$acc1,1),$acc1 #$t3 + + shl \$8,$acc0 + shl \$8,$acc1 + shr \$16,$s1 + xor $acc0,$t2 + xor $acc1,$t3 + shr \$16,$s2 + + movzb `&lo("$s2")`,$acc0 + movzb `&lo("$s3")`,$acc1 + movzb `&lo("$s0")`,$acc2 + movzb 2048($sbox,$acc0,1),$acc0 #$t0 + movzb 2048($sbox,$acc1,1),$acc1 #$t1 + movzb 2048($sbox,$acc2,1),$acc2 #$t2 + + shl \$16,$acc0 + shl \$16,$acc1 + shl \$16,$acc2 + + xor $acc0,$t0 + xor $acc1,$t1 + xor $acc2,$t2 + + movzb `&lo("$s1")`,$acc0 + movzb `&hi("$s1")`,$acc1 + movzb `&hi("$s2")`,$acc2 + movzb 2048($sbox,$acc0,1),$acc0 #$t3 + movzb 2048($sbox,$acc1,1),$acc1 #$t0 + movzb 2048($sbox,$acc2,1),$acc2 #$t1 + + shl \$16,$acc0 + shl \$24,$acc1 + shl \$24,$acc2 + + xor $acc0,$t3 + xor $acc1,$t0 + xor $acc2,$t1 + + movzb `&hi("$s3")`,$acc0 + movzb `&hi("$s0")`,$acc1 + mov 16+12($key),$s3 + movzb 2048($sbox,$acc0,1),$acc0 #$t2 + movzb 2048($sbox,$acc1,1),$acc1 #$t3 + mov 16+0($key),$s0 + + shl \$24,$acc0 + shl \$24,$acc1 + + xor $acc0,$t2 + xor $acc1,$t3 + + mov 16+4($key),$s1 + mov 16+8($key),$s2 + xor $t0,$s0 + xor $t1,$s1 + xor $t2,$s2 + xor $t3,$s3 +___ +} + +sub decstep() +{ my ($i,@s) = @_; + my $tmp0=$acc0; + my $tmp1=$acc1; + my $tmp2=$acc2; + my $out=($t0,$t1,$t2,$s[0])[$i]; + + $code.=" mov $s[0],$out\n" if ($i!=3); + $tmp1=$s[2] if ($i==3); + $code.=" mov $s[2],$tmp1\n" if ($i!=3); + $code.=" and \$0xFF,$out\n"; + + $code.=" mov 0($sbox,$out,8),$out\n"; + $code.=" shr \$16,$tmp1\n"; + $tmp2=$s[3] if ($i==3); + $code.=" mov $s[3],$tmp2\n" if ($i!=3); + + $tmp0=$s[1] if ($i==3); + $code.=" movzb ".&hi($s[1]).",$tmp0\n"; + $code.=" and \$0xFF,$tmp1\n"; + $code.=" shr \$24,$tmp2\n"; + + $code.=" xor 3($sbox,$tmp0,8),$out\n"; + $code.=" xor 2($sbox,$tmp1,8),$out\n"; + $code.=" xor 1($sbox,$tmp2,8),$out\n"; + + $code.=" mov $t2,$s[1]\n" if ($i==3); + $code.=" mov $t1,$s[2]\n" if ($i==3); + $code.=" mov $t0,$s[3]\n" if ($i==3); + $code.="\n"; +} + +sub declast() +{ my ($i,@s)=@_; + my $tmp0=$acc0; + my $tmp1=$acc1; + my $tmp2=$acc2; + my $out=($t0,$t1,$t2,$s[0])[$i]; + + $code.=" mov $s[0],$out\n" if ($i!=3); + $tmp1=$s[2] if ($i==3); + $code.=" mov $s[2],$tmp1\n" if ($i!=3); + $code.=" and \$0xFF,$out\n"; + + $code.=" movzb 2048($sbox,$out,1),$out\n"; + $code.=" shr \$16,$tmp1\n"; + $tmp2=$s[3] if ($i==3); + $code.=" mov $s[3],$tmp2\n" if ($i!=3); + + $tmp0=$s[1] if ($i==3); + $code.=" movzb ".&hi($s[1]).",$tmp0\n"; + $code.=" and \$0xFF,$tmp1\n"; + $code.=" shr \$24,$tmp2\n"; + + $code.=" movzb 2048($sbox,$tmp0,1),$tmp0\n"; + $code.=" movzb 2048($sbox,$tmp1,1),$tmp1\n"; + $code.=" movzb 2048($sbox,$tmp2,1),$tmp2\n"; + + $code.=" shl \$8,$tmp0\n"; + $code.=" shl \$16,$tmp1\n"; + $code.=" shl \$24,$tmp2\n"; + + $code.=" xor $tmp0,$out\n"; + $code.=" mov $t2,$s[1]\n" if ($i==3); + $code.=" xor $tmp1,$out\n"; + $code.=" mov $t1,$s[2]\n" if ($i==3); + $code.=" xor $tmp2,$out\n"; + $code.=" mov $t0,$s[3]\n" if ($i==3); + $code.="\n"; +} + +$code.=<<___; +.type _x86_64_AES_decrypt,\@abi-omnipotent +.align 16 +_x86_64_AES_decrypt: + xor 0($key),$s0 # xor with key + xor 4($key),$s1 + xor 8($key),$s2 + xor 12($key),$s3 + + mov 240($key),$rnds # load key->rounds + sub \$1,$rnds + jmp .Ldec_loop +.align 16 +.Ldec_loop: +___ + if ($verticalspin) { &decvert(); } + else { &decstep(0,$s0,$s3,$s2,$s1); + &decstep(1,$s1,$s0,$s3,$s2); + &decstep(2,$s2,$s1,$s0,$s3); + &decstep(3,$s3,$s2,$s1,$s0); + $code.=<<___; + lea 16($key),$key + xor 0($key),$s0 # xor with key + xor 4($key),$s1 + xor 8($key),$s2 + xor 12($key),$s3 +___ + } +$code.=<<___; + sub \$1,$rnds + jnz .Ldec_loop +___ + if ($verticalspin) { &declastvert(); } + else { &declast(0,$s0,$s3,$s2,$s1); + &declast(1,$s1,$s0,$s3,$s2); + &declast(2,$s2,$s1,$s0,$s3); + &declast(3,$s3,$s2,$s1,$s0); + $code.=<<___; + xor 16+0($key),$s0 # xor with key + xor 16+4($key),$s1 + xor 16+8($key),$s2 + xor 16+12($key),$s3 +___ + } +$code.=<<___; + .byte 0xf3,0xc3 # rep ret +.size _x86_64_AES_decrypt,.-_x86_64_AES_decrypt +___ + +# void AES_decrypt (const void *inp,void *out,const AES_KEY *key); +$code.=<<___; +.globl AES_decrypt +.type AES_decrypt,\@function,3 +.align 16 +AES_decrypt: + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + mov %rdx,$key + mov %rdi,$inp + mov %rsi,$out + + .picmeup $sbox + lea AES_Td-.($sbox),$sbox + + # prefetch Td4 + lea 2048+128($sbox),$sbox; + mov 0-128($sbox),$s0 + mov 32-128($sbox),$s1 + mov 64-128($sbox),$s2 + mov 96-128($sbox),$s3 + mov 128-128($sbox),$s0 + mov 160-128($sbox),$s1 + mov 192-128($sbox),$s2 + mov 224-128($sbox),$s3 + lea -2048-128($sbox),$sbox; + + mov 0($inp),$s0 + mov 4($inp),$s1 + mov 8($inp),$s2 + mov 12($inp),$s3 + + call _x86_64_AES_decrypt + + mov $s0,0($out) + mov $s1,4($out) + mov $s2,8($out) + mov $s3,12($out) + + pop %r15 + pop %r14 + pop %r13 + pop %r12 + pop %rbp + pop %rbx + ret +.size AES_decrypt,.-AES_decrypt +___ +#------------------------------------------------------------------# + +sub enckey() +{ +$code.=<<___; + movz %dl,%esi # rk[i]>>0 + mov 2(%rbp,%rsi,8),%ebx + movz %dh,%esi # rk[i]>>8 + and \$0xFF000000,%ebx + xor %ebx,%eax + + mov 2(%rbp,%rsi,8),%ebx + shr \$16,%edx + and \$0x000000FF,%ebx + movz %dl,%esi # rk[i]>>16 + xor %ebx,%eax + + mov 0(%rbp,%rsi,8),%ebx + movz %dh,%esi # rk[i]>>24 + and \$0x0000FF00,%ebx + xor %ebx,%eax + + mov 0(%rbp,%rsi,8),%ebx + and \$0x00FF0000,%ebx + xor %ebx,%eax + + xor 2048(%rbp,%rcx,4),%eax # rcon +___ +} + +# int AES_set_encrypt_key(const unsigned char *userKey, const int bits, +# AES_KEY *key) +$code.=<<___; +.globl AES_set_encrypt_key +.type AES_set_encrypt_key,\@function,3 +.align 16 +AES_set_encrypt_key: + push %rbx + push %rbp + + mov %esi,%ecx # %ecx=bits + mov %rdi,%rsi # %rsi=userKey + mov %rdx,%rdi # %rdi=key + + test \$-1,%rsi + jz .Lbadpointer + test \$-1,%rdi + jz .Lbadpointer + + .picmeup %rbp + lea AES_Te-.(%rbp),%rbp + + cmp \$128,%ecx + je .L10rounds + cmp \$192,%ecx + je .L12rounds + cmp \$256,%ecx + je .L14rounds + mov \$-2,%rax # invalid number of bits + jmp .Lexit + +.L10rounds: + mov 0(%rsi),%eax # copy first 4 dwords + mov 4(%rsi),%ebx + mov 8(%rsi),%ecx + mov 12(%rsi),%edx + mov %eax,0(%rdi) + mov %ebx,4(%rdi) + mov %ecx,8(%rdi) + mov %edx,12(%rdi) + + xor %ecx,%ecx + jmp .L10shortcut +.align 4 +.L10loop: + mov 0(%rdi),%eax # rk[0] + mov 12(%rdi),%edx # rk[3] +.L10shortcut: +___ + &enckey (); +$code.=<<___; + mov %eax,16(%rdi) # rk[4] + xor 4(%rdi),%eax + mov %eax,20(%rdi) # rk[5] + xor 8(%rdi),%eax + mov %eax,24(%rdi) # rk[6] + xor 12(%rdi),%eax + mov %eax,28(%rdi) # rk[7] + add \$1,%ecx + lea 16(%rdi),%rdi + cmp \$10,%ecx + jl .L10loop + + movl \$10,80(%rdi) # setup number of rounds + xor %rax,%rax + jmp .Lexit + +.L12rounds: + mov 0(%rsi),%eax # copy first 6 dwords + mov 4(%rsi),%ebx + mov 8(%rsi),%ecx + mov 12(%rsi),%edx + mov %eax,0(%rdi) + mov %ebx,4(%rdi) + mov %ecx,8(%rdi) + mov %edx,12(%rdi) + mov 16(%rsi),%ecx + mov 20(%rsi),%edx + mov %ecx,16(%rdi) + mov %edx,20(%rdi) + + xor %ecx,%ecx + jmp .L12shortcut +.align 4 +.L12loop: + mov 0(%rdi),%eax # rk[0] + mov 20(%rdi),%edx # rk[5] +.L12shortcut: +___ + &enckey (); +$code.=<<___; + mov %eax,24(%rdi) # rk[6] + xor 4(%rdi),%eax + mov %eax,28(%rdi) # rk[7] + xor 8(%rdi),%eax + mov %eax,32(%rdi) # rk[8] + xor 12(%rdi),%eax + mov %eax,36(%rdi) # rk[9] + + cmp \$7,%ecx + je .L12break + add \$1,%ecx + + xor 16(%rdi),%eax + mov %eax,40(%rdi) # rk[10] + xor 20(%rdi),%eax + mov %eax,44(%rdi) # rk[11] + + lea 24(%rdi),%rdi + jmp .L12loop +.L12break: + movl \$12,72(%rdi) # setup number of rounds + xor %rax,%rax + jmp .Lexit + +.L14rounds: + mov 0(%rsi),%eax # copy first 8 dwords + mov 4(%rsi),%ebx + mov 8(%rsi),%ecx + mov 12(%rsi),%edx + mov %eax,0(%rdi) + mov %ebx,4(%rdi) + mov %ecx,8(%rdi) + mov %edx,12(%rdi) + mov 16(%rsi),%eax + mov 20(%rsi),%ebx + mov 24(%rsi),%ecx + mov 28(%rsi),%edx + mov %eax,16(%rdi) + mov %ebx,20(%rdi) + mov %ecx,24(%rdi) + mov %edx,28(%rdi) + + xor %ecx,%ecx + jmp .L14shortcut +.align 4 +.L14loop: + mov 28(%rdi),%edx # rk[4] +.L14shortcut: + mov 0(%rdi),%eax # rk[0] +___ + &enckey (); +$code.=<<___; + mov %eax,32(%rdi) # rk[8] + xor 4(%rdi),%eax + mov %eax,36(%rdi) # rk[9] + xor 8(%rdi),%eax + mov %eax,40(%rdi) # rk[10] + xor 12(%rdi),%eax + mov %eax,44(%rdi) # rk[11] + + cmp \$6,%ecx + je .L14break + add \$1,%ecx + + mov %eax,%edx + mov 16(%rdi),%eax # rk[4] + movz %dl,%esi # rk[11]>>0 + mov 2(%rbp,%rsi,8),%ebx + movz %dh,%esi # rk[11]>>8 + and \$0x000000FF,%ebx + xor %ebx,%eax + + mov 0(%rbp,%rsi,8),%ebx + shr \$16,%edx + and \$0x0000FF00,%ebx + movz %dl,%esi # rk[11]>>16 + xor %ebx,%eax + + mov 0(%rbp,%rsi,8),%ebx + movz %dh,%esi # rk[11]>>24 + and \$0x00FF0000,%ebx + xor %ebx,%eax + + mov 2(%rbp,%rsi,8),%ebx + and \$0xFF000000,%ebx + xor %ebx,%eax + + mov %eax,48(%rdi) # rk[12] + xor 20(%rdi),%eax + mov %eax,52(%rdi) # rk[13] + xor 24(%rdi),%eax + mov %eax,56(%rdi) # rk[14] + xor 28(%rdi),%eax + mov %eax,60(%rdi) # rk[15] + + lea 32(%rdi),%rdi + jmp .L14loop +.L14break: + movl \$14,48(%rdi) # setup number of rounds + xor %rax,%rax + jmp .Lexit + +.Lbadpointer: + mov \$-1,%rax +.Lexit: + pop %rbp + pop %rbx + ret +.size AES_set_encrypt_key,.-AES_set_encrypt_key +___ + +sub deckey() +{ my ($i,$ptr,$te,$td) = @_; +$code.=<<___; + mov $i($ptr),%eax + mov %eax,%edx + movz %ah,%ebx + shr \$16,%edx + and \$0xFF,%eax + movzb 2($te,%rax,8),%rax + movzb 2($te,%rbx,8),%rbx + mov 0($td,%rax,8),%eax + xor 3($td,%rbx,8),%eax + movzb %dh,%ebx + and \$0xFF,%edx + movzb 2($te,%rdx,8),%rdx + movzb 2($te,%rbx,8),%rbx + xor 2($td,%rdx,8),%eax + xor 1($td,%rbx,8),%eax + mov %eax,$i($ptr) +___ +} + +# int AES_set_decrypt_key(const unsigned char *userKey, const int bits, +# AES_KEY *key) +$code.=<<___; +.globl AES_set_decrypt_key +.type AES_set_decrypt_key,\@function,3 +.align 16 +AES_set_decrypt_key: + push %rdx + call AES_set_encrypt_key + cmp \$0,%eax + je .Lproceed + lea 24(%rsp),%rsp + ret +.Lproceed: + mov (%rsp),%r8 # restore key schedule + mov %rbx,(%rsp) + + mov 240(%r8),%ecx # pull number of rounds + xor %rdi,%rdi + lea (%rdi,%rcx,4),%rcx + mov %r8,%rsi + lea (%r8,%rcx,4),%rdi # pointer to last chunk +.align 4 +.Linvert: + mov 0(%rsi),%rax + mov 8(%rsi),%rbx + mov 0(%rdi),%rcx + mov 8(%rdi),%rdx + mov %rax,0(%rdi) + mov %rbx,8(%rdi) + mov %rcx,0(%rsi) + mov %rdx,8(%rsi) + lea 16(%rsi),%rsi + lea -16(%rdi),%rdi + cmp %rsi,%rdi + jne .Linvert + + .picmeup %r9 + lea AES_Td-.(%r9),%rdi + lea AES_Te-AES_Td(%rdi),%r9 + + mov %r8,%rsi + mov 240(%r8),%ecx # pull number of rounds + sub \$1,%ecx +.align 4 +.Lpermute: + lea 16(%rsi),%rsi +___ + &deckey (0,"%rsi","%r9","%rdi"); + &deckey (4,"%rsi","%r9","%rdi"); + &deckey (8,"%rsi","%r9","%rdi"); + &deckey (12,"%rsi","%r9","%rdi"); +$code.=<<___; + sub \$1,%ecx + jnz .Lpermute + + xor %rax,%rax + pop %rbx + ret +.size AES_set_decrypt_key,.-AES_set_decrypt_key +___ + +# void AES_cbc_encrypt (const void char *inp, unsigned char *out, +# size_t length, const AES_KEY *key, +# unsigned char *ivp,const int enc); +{ +# stack frame layout +# -8(%rsp) return address +my $_rsp="0(%rsp)"; # saved %rsp +my $_len="8(%rsp)"; # copy of 3rd parameter, length +my $_key="16(%rsp)"; # copy of 4th parameter, key +my $_ivp="24(%rsp)"; # copy of 5th parameter, ivp +my $keyp="32(%rsp)"; # one to pass as $key +my $ivec="40(%rsp)"; # ivec[16] +my $aes_key="56(%rsp)"; # copy of aes_key +my $mark="56+240(%rsp)"; # copy of aes_key->rounds + +$code.=<<___; +.globl AES_cbc_encrypt +.type AES_cbc_encrypt,\@function,6 +.align 16 +AES_cbc_encrypt: + cmp \$0,%rdx # check length + je .Lcbc_just_ret + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + pushfq + cld + mov %r9d,%r9d # clear upper half of enc + + .picmeup $sbox +.Lcbc_pic_point: + + cmp \$0,%r9 + je .LDECRYPT + + lea AES_Te-.Lcbc_pic_point($sbox),$sbox + + # allocate aligned stack frame... + lea -64-248(%rsp),$key + and \$-64,$key + + # ... and make it doesn't alias with AES_Te modulo 4096 + mov $sbox,%r10 + lea 2048($sbox),%r11 + mov $key,%r12 + and \$0xFFF,%r10 # s = $sbox&0xfff + and \$0xFFF,%r11 # e = ($sbox+2048)&0xfff + and \$0xFFF,%r12 # p = %rsp&0xfff + + cmp %r11,%r12 # if (p=>e) %rsp =- (p-e); + jb .Lcbc_te_break_out + sub %r11,%r12 + sub %r12,$key + jmp .Lcbc_te_ok +.Lcbc_te_break_out: # else %rsp -= (p-s)&0xfff + framesz + sub %r10,%r12 + and \$0xFFF,%r12 + add \$320,%r12 + sub %r12,$key +.align 4 +.Lcbc_te_ok: + + xchg %rsp,$key + add \$8,%rsp # reserve for return address! + mov $key,$_rsp # save %rsp + mov %rdx,$_len # save copy of len + mov %rcx,$_key # save copy of key + mov %r8,$_ivp # save copy of ivp + movl \$0,$mark # copy of aes_key->rounds = 0; + mov %r8,%rbp # rearrange input arguments + mov %rsi,$out + mov %rdi,$inp + mov %rcx,$key + + # do we copy key schedule to stack? + mov $key,%r10 + sub $sbox,%r10 + and \$0xfff,%r10 + cmp \$2048,%r10 + jb .Lcbc_do_ecopy + cmp \$4096-248,%r10 + jb .Lcbc_skip_ecopy +.align 4 +.Lcbc_do_ecopy: + mov $key,%rsi + lea $aes_key,%rdi + lea $aes_key,$key + mov \$240/8,%ecx + .long 0x90A548F3 # rep movsq + mov (%rsi),%eax # copy aes_key->rounds + mov %eax,(%rdi) +.Lcbc_skip_ecopy: + mov $key,$keyp # save key pointer + + mov \$16,%ecx +.align 4 +.Lcbc_prefetch_te: + mov 0($sbox),%r10 + mov 32($sbox),%r11 + mov 64($sbox),%r12 + mov 96($sbox),%r13 + lea 128($sbox),$sbox + sub \$1,%ecx + jnz .Lcbc_prefetch_te + sub \$2048,$sbox + + test \$-16,%rdx # check upon length + mov %rdx,%r10 + mov 0(%rbp),$s0 # load iv + mov 4(%rbp),$s1 + mov 8(%rbp),$s2 + mov 12(%rbp),$s3 + jz .Lcbc_enc_tail # short input... + +.align 4 +.Lcbc_enc_loop: + xor 0($inp),$s0 + xor 4($inp),$s1 + xor 8($inp),$s2 + xor 12($inp),$s3 + mov $inp,$ivec # if ($verticalspin) save inp + + mov $keyp,$key # restore key + call _x86_64_AES_encrypt + + mov $ivec,$inp # if ($verticalspin) restore inp + mov $s0,0($out) + mov $s1,4($out) + mov $s2,8($out) + mov $s3,12($out) + + mov $_len,%r10 + lea 16($inp),$inp + lea 16($out),$out + sub \$16,%r10 + test \$-16,%r10 + mov %r10,$_len + jnz .Lcbc_enc_loop + test \$15,%r10 + jnz .Lcbc_enc_tail + mov $_ivp,%rbp # restore ivp + mov $s0,0(%rbp) # save ivec + mov $s1,4(%rbp) + mov $s2,8(%rbp) + mov $s3,12(%rbp) + +.align 4 +.Lcbc_cleanup: + cmpl \$0,$mark # was the key schedule copied? + lea $aes_key,%rdi + mov $_rsp,%rsp + je .Lcbc_exit + mov \$240/8,%ecx + xor %rax,%rax + .long 0x90AB48F3 # rep stosq +.Lcbc_exit: + popfq + pop %r15 + pop %r14 + pop %r13 + pop %r12 + pop %rbp + pop %rbx +.Lcbc_just_ret: + ret +.align 4 +.Lcbc_enc_tail: + mov %rax,%r11 + mov %rcx,%r12 + mov %r10,%rcx + mov $inp,%rsi + mov $out,%rdi + .long 0xF689A4F3 # rep movsb + mov \$16,%rcx # zero tail + sub %r10,%rcx + xor %rax,%rax + .long 0xF689AAF3 # rep stosb + mov $out,$inp # this is not a mistake! + movq \$16,$_len # len=16 + mov %r11,%rax + mov %r12,%rcx + jmp .Lcbc_enc_loop # one more spin... +#----------------------------- DECRYPT -----------------------------# +.align 16 +.LDECRYPT: + lea AES_Td-.Lcbc_pic_point($sbox),$sbox + + # allocate aligned stack frame... + lea -64-248(%rsp),$key + and \$-64,$key + + # ... and make it doesn't alias with AES_Td modulo 4096 + mov $sbox,%r10 + lea 2304($sbox),%r11 + mov $key,%r12 + and \$0xFFF,%r10 # s = $sbox&0xfff + and \$0xFFF,%r11 # e = ($sbox+2048+256)&0xfff + and \$0xFFF,%r12 # p = %rsp&0xfff + + cmp %r11,%r12 # if (p=>e) %rsp =- (p-e); + jb .Lcbc_td_break_out + sub %r11,%r12 + sub %r12,$key + jmp .Lcbc_td_ok +.Lcbc_td_break_out: # else %rsp -= (p-s)&0xfff + framesz + sub %r10,%r12 + and \$0xFFF,%r12 + add \$320,%r12 + sub %r12,$key +.align 4 +.Lcbc_td_ok: + + xchg %rsp,$key + add \$8,%rsp # reserve for return address! + mov $key,$_rsp # save %rsp + mov %rdx,$_len # save copy of len + mov %rcx,$_key # save copy of key + mov %r8,$_ivp # save copy of ivp + movl \$0,$mark # copy of aes_key->rounds = 0; + mov %r8,%rbp # rearrange input arguments + mov %rsi,$out + mov %rdi,$inp + mov %rcx,$key + + # do we copy key schedule to stack? + mov $key,%r10 + sub $sbox,%r10 + and \$0xfff,%r10 + cmp \$2304,%r10 + jb .Lcbc_do_dcopy + cmp \$4096-248,%r10 + jb .Lcbc_skip_dcopy +.align 4 +.Lcbc_do_dcopy: + mov $key,%rsi + lea $aes_key,%rdi + lea $aes_key,$key + mov \$240/8,%ecx + .long 0x90A548F3 # rep movsq + mov (%rsi),%eax # copy aes_key->rounds + mov %eax,(%rdi) +.Lcbc_skip_dcopy: + mov $key,$keyp # save key pointer + + mov \$18,%ecx +.align 4 +.Lcbc_prefetch_td: + mov 0($sbox),%r10 + mov 32($sbox),%r11 + mov 64($sbox),%r12 + mov 96($sbox),%r13 + lea 128($sbox),$sbox + sub \$1,%ecx + jnz .Lcbc_prefetch_td + sub \$2304,$sbox + + cmp $inp,$out + je .Lcbc_dec_in_place + + mov %rbp,$ivec +.align 4 +.Lcbc_dec_loop: + mov 0($inp),$s0 # read input + mov 4($inp),$s1 + mov 8($inp),$s2 + mov 12($inp),$s3 + mov $inp,8+$ivec # if ($verticalspin) save inp + + mov $keyp,$key # restore key + call _x86_64_AES_decrypt + + mov $ivec,%rbp # load ivp + mov 8+$ivec,$inp # if ($verticalspin) restore inp + xor 0(%rbp),$s0 # xor iv + xor 4(%rbp),$s1 + xor 8(%rbp),$s2 + xor 12(%rbp),$s3 + mov $inp,%rbp # current input, next iv + + mov $_len,%r10 # load len + sub \$16,%r10 + jc .Lcbc_dec_partial + mov %r10,$_len # update len + mov %rbp,$ivec # update ivp + + mov $s0,0($out) # write output + mov $s1,4($out) + mov $s2,8($out) + mov $s3,12($out) + + lea 16($inp),$inp + lea 16($out),$out + jnz .Lcbc_dec_loop +.Lcbc_dec_end: + mov $_ivp,%r12 # load user ivp + mov 0(%rbp),%r10 # load iv + mov 8(%rbp),%r11 + mov %r10,0(%r12) # copy back to user + mov %r11,8(%r12) + jmp .Lcbc_cleanup + +.align 4 +.Lcbc_dec_partial: + mov $s0,0+$ivec # dump output to stack + mov $s1,4+$ivec + mov $s2,8+$ivec + mov $s3,12+$ivec + mov $out,%rdi + lea $ivec,%rsi + mov \$16,%rcx + add %r10,%rcx # number of bytes to copy + .long 0xF689A4F3 # rep movsb + jmp .Lcbc_dec_end + +.align 16 +.Lcbc_dec_in_place: + mov 0($inp),$s0 # load input + mov 4($inp),$s1 + mov 8($inp),$s2 + mov 12($inp),$s3 + + mov $inp,$ivec # if ($verticalspin) save inp + mov $keyp,$key + call _x86_64_AES_decrypt + + mov $ivec,$inp # if ($verticalspin) restore inp + mov $_ivp,%rbp + xor 0(%rbp),$s0 + xor 4(%rbp),$s1 + xor 8(%rbp),$s2 + xor 12(%rbp),$s3 + + mov 0($inp),%r10 # copy input to iv + mov 8($inp),%r11 + mov %r10,0(%rbp) + mov %r11,8(%rbp) + + mov $s0,0($out) # save output [zaps input] + mov $s1,4($out) + mov $s2,8($out) + mov $s3,12($out) + + mov $_len,%rcx + lea 16($inp),$inp + lea 16($out),$out + sub \$16,%rcx + jc .Lcbc_dec_in_place_partial + mov %rcx,$_len + jnz .Lcbc_dec_in_place + jmp .Lcbc_cleanup + +.align 4 +.Lcbc_dec_in_place_partial: + # one can argue if this is actually required + lea ($out,%rcx),%rdi + lea (%rbp,%rcx),%rsi + neg %rcx + .long 0xF689A4F3 # rep movsb # restore tail + jmp .Lcbc_cleanup +.size AES_cbc_encrypt,.-AES_cbc_encrypt +___ +} + +$code.=<<___; +.globl AES_Te +.align 64 +AES_Te: +___ + &_data_word(0xa56363c6, 0x847c7cf8, 0x997777ee, 0x8d7b7bf6); + &_data_word(0x0df2f2ff, 0xbd6b6bd6, 0xb16f6fde, 0x54c5c591); + &_data_word(0x50303060, 0x03010102, 0xa96767ce, 0x7d2b2b56); + &_data_word(0x19fefee7, 0x62d7d7b5, 0xe6abab4d, 0x9a7676ec); + &_data_word(0x45caca8f, 0x9d82821f, 0x40c9c989, 0x877d7dfa); + &_data_word(0x15fafaef, 0xeb5959b2, 0xc947478e, 0x0bf0f0fb); + &_data_word(0xecadad41, 0x67d4d4b3, 0xfda2a25f, 0xeaafaf45); + &_data_word(0xbf9c9c23, 0xf7a4a453, 0x967272e4, 0x5bc0c09b); + &_data_word(0xc2b7b775, 0x1cfdfde1, 0xae93933d, 0x6a26264c); + &_data_word(0x5a36366c, 0x413f3f7e, 0x02f7f7f5, 0x4fcccc83); + &_data_word(0x5c343468, 0xf4a5a551, 0x34e5e5d1, 0x08f1f1f9); + &_data_word(0x937171e2, 0x73d8d8ab, 0x53313162, 0x3f15152a); + &_data_word(0x0c040408, 0x52c7c795, 0x65232346, 0x5ec3c39d); + &_data_word(0x28181830, 0xa1969637, 0x0f05050a, 0xb59a9a2f); + &_data_word(0x0907070e, 0x36121224, 0x9b80801b, 0x3de2e2df); + &_data_word(0x26ebebcd, 0x6927274e, 0xcdb2b27f, 0x9f7575ea); + &_data_word(0x1b090912, 0x9e83831d, 0x742c2c58, 0x2e1a1a34); + &_data_word(0x2d1b1b36, 0xb26e6edc, 0xee5a5ab4, 0xfba0a05b); + &_data_word(0xf65252a4, 0x4d3b3b76, 0x61d6d6b7, 0xceb3b37d); + &_data_word(0x7b292952, 0x3ee3e3dd, 0x712f2f5e, 0x97848413); + &_data_word(0xf55353a6, 0x68d1d1b9, 0x00000000, 0x2cededc1); + &_data_word(0x60202040, 0x1ffcfce3, 0xc8b1b179, 0xed5b5bb6); + &_data_word(0xbe6a6ad4, 0x46cbcb8d, 0xd9bebe67, 0x4b393972); + &_data_word(0xde4a4a94, 0xd44c4c98, 0xe85858b0, 0x4acfcf85); + &_data_word(0x6bd0d0bb, 0x2aefefc5, 0xe5aaaa4f, 0x16fbfbed); + &_data_word(0xc5434386, 0xd74d4d9a, 0x55333366, 0x94858511); + &_data_word(0xcf45458a, 0x10f9f9e9, 0x06020204, 0x817f7ffe); + &_data_word(0xf05050a0, 0x443c3c78, 0xba9f9f25, 0xe3a8a84b); + &_data_word(0xf35151a2, 0xfea3a35d, 0xc0404080, 0x8a8f8f05); + &_data_word(0xad92923f, 0xbc9d9d21, 0x48383870, 0x04f5f5f1); + &_data_word(0xdfbcbc63, 0xc1b6b677, 0x75dadaaf, 0x63212142); + &_data_word(0x30101020, 0x1affffe5, 0x0ef3f3fd, 0x6dd2d2bf); + &_data_word(0x4ccdcd81, 0x140c0c18, 0x35131326, 0x2fececc3); + &_data_word(0xe15f5fbe, 0xa2979735, 0xcc444488, 0x3917172e); + &_data_word(0x57c4c493, 0xf2a7a755, 0x827e7efc, 0x473d3d7a); + &_data_word(0xac6464c8, 0xe75d5dba, 0x2b191932, 0x957373e6); + &_data_word(0xa06060c0, 0x98818119, 0xd14f4f9e, 0x7fdcdca3); + &_data_word(0x66222244, 0x7e2a2a54, 0xab90903b, 0x8388880b); + &_data_word(0xca46468c, 0x29eeeec7, 0xd3b8b86b, 0x3c141428); + &_data_word(0x79dedea7, 0xe25e5ebc, 0x1d0b0b16, 0x76dbdbad); + &_data_word(0x3be0e0db, 0x56323264, 0x4e3a3a74, 0x1e0a0a14); + &_data_word(0xdb494992, 0x0a06060c, 0x6c242448, 0xe45c5cb8); + &_data_word(0x5dc2c29f, 0x6ed3d3bd, 0xefacac43, 0xa66262c4); + &_data_word(0xa8919139, 0xa4959531, 0x37e4e4d3, 0x8b7979f2); + &_data_word(0x32e7e7d5, 0x43c8c88b, 0x5937376e, 0xb76d6dda); + &_data_word(0x8c8d8d01, 0x64d5d5b1, 0xd24e4e9c, 0xe0a9a949); + &_data_word(0xb46c6cd8, 0xfa5656ac, 0x07f4f4f3, 0x25eaeacf); + &_data_word(0xaf6565ca, 0x8e7a7af4, 0xe9aeae47, 0x18080810); + &_data_word(0xd5baba6f, 0x887878f0, 0x6f25254a, 0x722e2e5c); + &_data_word(0x241c1c38, 0xf1a6a657, 0xc7b4b473, 0x51c6c697); + &_data_word(0x23e8e8cb, 0x7cdddda1, 0x9c7474e8, 0x211f1f3e); + &_data_word(0xdd4b4b96, 0xdcbdbd61, 0x868b8b0d, 0x858a8a0f); + &_data_word(0x907070e0, 0x423e3e7c, 0xc4b5b571, 0xaa6666cc); + &_data_word(0xd8484890, 0x05030306, 0x01f6f6f7, 0x120e0e1c); + &_data_word(0xa36161c2, 0x5f35356a, 0xf95757ae, 0xd0b9b969); + &_data_word(0x91868617, 0x58c1c199, 0x271d1d3a, 0xb99e9e27); + &_data_word(0x38e1e1d9, 0x13f8f8eb, 0xb398982b, 0x33111122); + &_data_word(0xbb6969d2, 0x70d9d9a9, 0x898e8e07, 0xa7949433); + &_data_word(0xb69b9b2d, 0x221e1e3c, 0x92878715, 0x20e9e9c9); + &_data_word(0x49cece87, 0xff5555aa, 0x78282850, 0x7adfdfa5); + &_data_word(0x8f8c8c03, 0xf8a1a159, 0x80898909, 0x170d0d1a); + &_data_word(0xdabfbf65, 0x31e6e6d7, 0xc6424284, 0xb86868d0); + &_data_word(0xc3414182, 0xb0999929, 0x772d2d5a, 0x110f0f1e); + &_data_word(0xcbb0b07b, 0xfc5454a8, 0xd6bbbb6d, 0x3a16162c); +#rcon: +$code.=<<___; + .long 0x00000001, 0x00000002, 0x00000004, 0x00000008 + .long 0x00000010, 0x00000020, 0x00000040, 0x00000080 + .long 0x0000001b, 0x00000036, 0, 0, 0, 0, 0, 0 +___ +$code.=<<___; +.globl AES_Td +.align 64 +AES_Td: +___ + &_data_word(0x50a7f451, 0x5365417e, 0xc3a4171a, 0x965e273a); + &_data_word(0xcb6bab3b, 0xf1459d1f, 0xab58faac, 0x9303e34b); + &_data_word(0x55fa3020, 0xf66d76ad, 0x9176cc88, 0x254c02f5); + &_data_word(0xfcd7e54f, 0xd7cb2ac5, 0x80443526, 0x8fa362b5); + &_data_word(0x495ab1de, 0x671bba25, 0x980eea45, 0xe1c0fe5d); + &_data_word(0x02752fc3, 0x12f04c81, 0xa397468d, 0xc6f9d36b); + &_data_word(0xe75f8f03, 0x959c9215, 0xeb7a6dbf, 0xda595295); + &_data_word(0x2d83bed4, 0xd3217458, 0x2969e049, 0x44c8c98e); + &_data_word(0x6a89c275, 0x78798ef4, 0x6b3e5899, 0xdd71b927); + &_data_word(0xb64fe1be, 0x17ad88f0, 0x66ac20c9, 0xb43ace7d); + &_data_word(0x184adf63, 0x82311ae5, 0x60335197, 0x457f5362); + &_data_word(0xe07764b1, 0x84ae6bbb, 0x1ca081fe, 0x942b08f9); + &_data_word(0x58684870, 0x19fd458f, 0x876cde94, 0xb7f87b52); + &_data_word(0x23d373ab, 0xe2024b72, 0x578f1fe3, 0x2aab5566); + &_data_word(0x0728ebb2, 0x03c2b52f, 0x9a7bc586, 0xa50837d3); + &_data_word(0xf2872830, 0xb2a5bf23, 0xba6a0302, 0x5c8216ed); + &_data_word(0x2b1ccf8a, 0x92b479a7, 0xf0f207f3, 0xa1e2694e); + &_data_word(0xcdf4da65, 0xd5be0506, 0x1f6234d1, 0x8afea6c4); + &_data_word(0x9d532e34, 0xa055f3a2, 0x32e18a05, 0x75ebf6a4); + &_data_word(0x39ec830b, 0xaaef6040, 0x069f715e, 0x51106ebd); + &_data_word(0xf98a213e, 0x3d06dd96, 0xae053edd, 0x46bde64d); + &_data_word(0xb58d5491, 0x055dc471, 0x6fd40604, 0xff155060); + &_data_word(0x24fb9819, 0x97e9bdd6, 0xcc434089, 0x779ed967); + &_data_word(0xbd42e8b0, 0x888b8907, 0x385b19e7, 0xdbeec879); + &_data_word(0x470a7ca1, 0xe90f427c, 0xc91e84f8, 0x00000000); + &_data_word(0x83868009, 0x48ed2b32, 0xac70111e, 0x4e725a6c); + &_data_word(0xfbff0efd, 0x5638850f, 0x1ed5ae3d, 0x27392d36); + &_data_word(0x64d90f0a, 0x21a65c68, 0xd1545b9b, 0x3a2e3624); + &_data_word(0xb1670a0c, 0x0fe75793, 0xd296eeb4, 0x9e919b1b); + &_data_word(0x4fc5c080, 0xa220dc61, 0x694b775a, 0x161a121c); + &_data_word(0x0aba93e2, 0xe52aa0c0, 0x43e0223c, 0x1d171b12); + &_data_word(0x0b0d090e, 0xadc78bf2, 0xb9a8b62d, 0xc8a91e14); + &_data_word(0x8519f157, 0x4c0775af, 0xbbdd99ee, 0xfd607fa3); + &_data_word(0x9f2601f7, 0xbcf5725c, 0xc53b6644, 0x347efb5b); + &_data_word(0x7629438b, 0xdcc623cb, 0x68fcedb6, 0x63f1e4b8); + &_data_word(0xcadc31d7, 0x10856342, 0x40229713, 0x2011c684); + &_data_word(0x7d244a85, 0xf83dbbd2, 0x1132f9ae, 0x6da129c7); + &_data_word(0x4b2f9e1d, 0xf330b2dc, 0xec52860d, 0xd0e3c177); + &_data_word(0x6c16b32b, 0x99b970a9, 0xfa489411, 0x2264e947); + &_data_word(0xc48cfca8, 0x1a3ff0a0, 0xd82c7d56, 0xef903322); + &_data_word(0xc74e4987, 0xc1d138d9, 0xfea2ca8c, 0x360bd498); + &_data_word(0xcf81f5a6, 0x28de7aa5, 0x268eb7da, 0xa4bfad3f); + &_data_word(0xe49d3a2c, 0x0d927850, 0x9bcc5f6a, 0x62467e54); + &_data_word(0xc2138df6, 0xe8b8d890, 0x5ef7392e, 0xf5afc382); + &_data_word(0xbe805d9f, 0x7c93d069, 0xa92dd56f, 0xb31225cf); + &_data_word(0x3b99acc8, 0xa77d1810, 0x6e639ce8, 0x7bbb3bdb); + &_data_word(0x097826cd, 0xf418596e, 0x01b79aec, 0xa89a4f83); + &_data_word(0x656e95e6, 0x7ee6ffaa, 0x08cfbc21, 0xe6e815ef); + &_data_word(0xd99be7ba, 0xce366f4a, 0xd4099fea, 0xd67cb029); + &_data_word(0xafb2a431, 0x31233f2a, 0x3094a5c6, 0xc066a235); + &_data_word(0x37bc4e74, 0xa6ca82fc, 0xb0d090e0, 0x15d8a733); + &_data_word(0x4a9804f1, 0xf7daec41, 0x0e50cd7f, 0x2ff69117); + &_data_word(0x8dd64d76, 0x4db0ef43, 0x544daacc, 0xdf0496e4); + &_data_word(0xe3b5d19e, 0x1b886a4c, 0xb81f2cc1, 0x7f516546); + &_data_word(0x04ea5e9d, 0x5d358c01, 0x737487fa, 0x2e410bfb); + &_data_word(0x5a1d67b3, 0x52d2db92, 0x335610e9, 0x1347d66d); + &_data_word(0x8c61d79a, 0x7a0ca137, 0x8e14f859, 0x893c13eb); + &_data_word(0xee27a9ce, 0x35c961b7, 0xede51ce1, 0x3cb1477a); + &_data_word(0x59dfd29c, 0x3f73f255, 0x79ce1418, 0xbf37c773); + &_data_word(0xeacdf753, 0x5baafd5f, 0x146f3ddf, 0x86db4478); + &_data_word(0x81f3afca, 0x3ec468b9, 0x2c342438, 0x5f40a3c2); + &_data_word(0x72c31d16, 0x0c25e2bc, 0x8b493c28, 0x41950dff); + &_data_word(0x7101a839, 0xdeb30c08, 0x9ce4b4d8, 0x90c15664); + &_data_word(0x6184cb7b, 0x70b632d5, 0x745c6c48, 0x4257b8d0); +#Td4: + &data_byte(0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38); + &data_byte(0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb); + &data_byte(0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87); + &data_byte(0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb); + &data_byte(0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d); + &data_byte(0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e); + &data_byte(0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2); + &data_byte(0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25); + &data_byte(0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16); + &data_byte(0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92); + &data_byte(0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda); + &data_byte(0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84); + &data_byte(0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a); + &data_byte(0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06); + &data_byte(0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02); + &data_byte(0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b); + &data_byte(0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea); + &data_byte(0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73); + &data_byte(0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85); + &data_byte(0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e); + &data_byte(0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89); + &data_byte(0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b); + &data_byte(0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20); + &data_byte(0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4); + &data_byte(0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31); + &data_byte(0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f); + &data_byte(0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d); + &data_byte(0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef); + &data_byte(0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0); + &data_byte(0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61); + &data_byte(0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26); + &data_byte(0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d); + +$code =~ s/\`([^\`]*)\`/eval($1)/gem; + +print $code; + +close STDOUT; |