crypto/sha/asm/sha1-alpha.pl

   1 #! /usr/bin/env perl
   2 # Copyright 2009-2016 The OpenSSL Project Authors. All Rights Reserved.
   3 #
   4 # Licensed under the Apache License 2.0 (the "License").  You may not use
   5 # this file except in compliance with the License.  You can obtain a copy
   6 # in the file LICENSE in the source distribution or at
   7 # https://www.openssl.org/source/license.html
   8
   9
  10 # ====================================================================
  11 # Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
  12 # project. The module is, however, dual licensed under OpenSSL and
  13 # CRYPTOGAMS licenses depending on where you obtain it. For further
  14 # details see http://www.openssl.org/~appro/cryptogams/.
  15 # ====================================================================
  16
  17 # SHA1 block procedure for Alpha.
  18
  19 # On 21264 performance is 33% better than code generated by vendor
  20 # compiler, and 75% better than GCC [3.4], and in absolute terms is
  21 # 8.7 cycles per processed byte. Implementation features vectorized
  22 # byte swap, but not Xupdate.
  23
  24 @X=(    "\$0",  "\$1",  "\$2",  "\$3",  "\$4",  "\$5",  "\$6",  "\$7",
  25         "\$8",  "\$9",  "\$10", "\$11", "\$12", "\$13", "\$14", "\$15");
  26 $ctx="a0";      # $16
  27 $inp="a1";
  28 $num="a2";
  29 $A="a3";
  30 $B="a4";        # 20
  31 $C="a5";
  32 $D="t8";
  33 $E="t9";        @V=($A,$B,$C,$D,$E);
  34 $t0="t10";      # 24
  35 $t1="t11";
  36 $t2="ra";
  37 $t3="t12";
  38 $K="AT";        # 28
  39
  40 sub BODY_00_19 {
  41 my ($i,$a,$b,$c,$d,$e)=@_;
  42 my $j=$i+1;
  43 $code.=<<___ if ($i==0);
  44         ldq_u   @X[0],0+0($inp)
  45         ldq_u   @X[1],0+7($inp)
  46 ___
  47 $code.=<<___ if (!($i&1) && $i<14);
  48         ldq_u   @X[$i+2],($i+2)*4+0($inp)
  49         ldq_u   @X[$i+3],($i+2)*4+7($inp)
  50 ___
  51 $code.=<<___ if (!($i&1) && $i<15);
  52         extql   @X[$i],$inp,@X[$i]
  53         extqh   @X[$i+1],$inp,@X[$i+1]
  54
  55         or      @X[$i+1],@X[$i],@X[$i]  # pair of 32-bit values are fetched
  56
  57         srl     @X[$i],24,$t0           # vectorized byte swap
  58         srl     @X[$i],8,$t2
  59
  60         sll     @X[$i],8,$t3
  61         sll     @X[$i],24,@X[$i]
  62         zapnot  $t0,0x11,$t0
  63         zapnot  $t2,0x22,$t2
  64
  65         zapnot  @X[$i],0x88,@X[$i]
  66         or      $t0,$t2,$t0
  67         zapnot  $t3,0x44,$t3
  68         sll     $a,5,$t1
  69
  70         or      @X[$i],$t0,@X[$i]
  71         addl    $K,$e,$e
  72         and     $b,$c,$t2
  73         zapnot  $a,0xf,$a
  74
  75         or      @X[$i],$t3,@X[$i]
  76         srl     $a,27,$t0
  77         bic     $d,$b,$t3
  78         sll     $b,30,$b
  79
  80         extll   @X[$i],4,@X[$i+1]       # extract upper half
  81         or      $t2,$t3,$t2
  82         addl    @X[$i],$e,$e
  83
  84         addl    $t1,$e,$e
  85         srl     $b,32,$t3
  86         zapnot  @X[$i],0xf,@X[$i]
  87
  88         addl    $t0,$e,$e
  89         addl    $t2,$e,$e
  90         or      $t3,$b,$b
  91 ___
  92 $code.=<<___ if (($i&1) && $i<15);
  93         sll     $a,5,$t1
  94         addl    $K,$e,$e
  95         and     $b,$c,$t2
  96         zapnot  $a,0xf,$a
  97
  98         srl     $a,27,$t0
  99         addl    @X[$i%16],$e,$e
 100         bic     $d,$b,$t3
 101         sll     $b,30,$b
 102
 103         or      $t2,$t3,$t2
 104         addl    $t1,$e,$e
 105         srl     $b,32,$t3
 106         zapnot  @X[$i],0xf,@X[$i]
 107
 108         addl    $t0,$e,$e
 109         addl    $t2,$e,$e
 110         or      $t3,$b,$b
 111 ___
 112 $code.=<<___ if ($i>=15);       # with forward Xupdate
 113         sll     $a,5,$t1
 114         addl    $K,$e,$e
 115         and     $b,$c,$t2
 116         xor     @X[($j+2)%16],@X[$j%16],@X[$j%16]
 117
 118         zapnot  $a,0xf,$a
 119         addl    @X[$i%16],$e,$e
 120         bic     $d,$b,$t3
 121         xor     @X[($j+8)%16],@X[$j%16],@X[$j%16]
 122
 123         srl     $a,27,$t0
 124         addl    $t1,$e,$e
 125         or      $t2,$t3,$t2
 126         xor     @X[($j+13)%16],@X[$j%16],@X[$j%16]
 127
 128         sll     $b,30,$b
 129         addl    $t0,$e,$e
 130         srl     @X[$j%16],31,$t1
 131
 132         addl    $t2,$e,$e
 133         srl     $b,32,$t3
 134         addl    @X[$j%16],@X[$j%16],@X[$j%16]
 135
 136         or      $t3,$b,$b
 137         zapnot  @X[$i%16],0xf,@X[$i%16]
 138         or      $t1,@X[$j%16],@X[$j%16]
 139 ___
 140 }
 141
 142 sub BODY_20_39 {
 143 my ($i,$a,$b,$c,$d,$e)=@_;
 144 my $j=$i+1;
 145 $code.=<<___ if ($i<79);        # with forward Xupdate
 146         sll     $a,5,$t1
 147         addl    $K,$e,$e
 148         zapnot  $a,0xf,$a
 149         xor     @X[($j+2)%16],@X[$j%16],@X[$j%16]
 150
 151         sll     $b,30,$t3
 152         addl    $t1,$e,$e
 153         xor     $b,$c,$t2
 154         xor     @X[($j+8)%16],@X[$j%16],@X[$j%16]
 155
 156         srl     $b,2,$b
 157         addl    @X[$i%16],$e,$e
 158         xor     $d,$t2,$t2
 159         xor     @X[($j+13)%16],@X[$j%16],@X[$j%16]
 160
 161         srl     @X[$j%16],31,$t1
 162         addl    $t2,$e,$e
 163         srl     $a,27,$t0
 164         addl    @X[$j%16],@X[$j%16],@X[$j%16]
 165
 166         or      $t3,$b,$b
 167         addl    $t0,$e,$e
 168         or      $t1,@X[$j%16],@X[$j%16]
 169 ___
 170 $code.=<<___ if ($i<77);
 171         zapnot  @X[$i%16],0xf,@X[$i%16]
 172 ___
 173 $code.=<<___ if ($i==79);       # with context fetch
 174         sll     $a,5,$t1
 175         addl    $K,$e,$e
 176         zapnot  $a,0xf,$a
 177         ldl     @X[0],0($ctx)
 178
 179         sll     $b,30,$t3
 180         addl    $t1,$e,$e
 181         xor     $b,$c,$t2
 182         ldl     @X[1],4($ctx)
 183
 184         srl     $b,2,$b
 185         addl    @X[$i%16],$e,$e
 186         xor     $d,$t2,$t2
 187         ldl     @X[2],8($ctx)
 188
 189         srl     $a,27,$t0
 190         addl    $t2,$e,$e
 191         ldl     @X[3],12($ctx)
 192
 193         or      $t3,$b,$b
 194         addl    $t0,$e,$e
 195         ldl     @X[4],16($ctx)
 196 ___
 197 }
 198
 199 sub BODY_40_59 {
 200 my ($i,$a,$b,$c,$d,$e)=@_;
 201 my $j=$i+1;
 202 $code.=<<___;   # with forward Xupdate
 203         sll     $a,5,$t1
 204         addl    $K,$e,$e
 205         zapnot  $a,0xf,$a
 206         xor     @X[($j+2)%16],@X[$j%16],@X[$j%16]
 207
 208         srl     $a,27,$t0
 209         and     $b,$c,$t2
 210         and     $b,$d,$t3
 211         xor     @X[($j+8)%16],@X[$j%16],@X[$j%16]
 212
 213         sll     $b,30,$b
 214         addl    $t1,$e,$e
 215         xor     @X[($j+13)%16],@X[$j%16],@X[$j%16]
 216
 217         srl     @X[$j%16],31,$t1
 218         addl    $t0,$e,$e
 219         or      $t2,$t3,$t2
 220         and     $c,$d,$t3
 221
 222         or      $t2,$t3,$t2
 223         srl     $b,32,$t3
 224         addl    @X[$i%16],$e,$e
 225         addl    @X[$j%16],@X[$j%16],@X[$j%16]
 226
 227         or      $t3,$b,$b
 228         addl    $t2,$e,$e
 229         or      $t1,@X[$j%16],@X[$j%16]
 230         zapnot  @X[$i%16],0xf,@X[$i%16]
 231 ___
 232 }
 233
 234 $code=<<___;
 235 #ifdef __linux__
 236 #include <asm/regdef.h>
 237 #else
 238 #include <asm.h>
 239 #include <regdef.h>
 240 #endif
 241
 242 .text
 243
 244 .set    noat
 245 .set    noreorder
 246 .globl  sha1_block_data_order
 247 .align  5
 248 .ent    sha1_block_data_order
 249 sha1_block_data_order:
 250         lda     sp,-64(sp)
 251         stq     ra,0(sp)
 252         stq     s0,8(sp)
 253         stq     s1,16(sp)
 254         stq     s2,24(sp)
 255         stq     s3,32(sp)
 256         stq     s4,40(sp)
 257         stq     s5,48(sp)
 258         stq     fp,56(sp)
 259         .mask   0x0400fe00,-64
 260         .frame  sp,64,ra
 261         .prologue 0
 262
 263         ldl     $A,0($ctx)
 264         ldl     $B,4($ctx)
 265         sll     $num,6,$num
 266         ldl     $C,8($ctx)
 267         ldl     $D,12($ctx)
 268         ldl     $E,16($ctx)
 269         addq    $inp,$num,$num
 270
 271 .Lloop:
 272         .set    noreorder
 273         ldah    $K,23170(zero)
 274         zapnot  $B,0xf,$B
 275         lda     $K,31129($K)    # K_00_19
 276 ___
 277 for ($i=0;$i<20;$i++) { &BODY_00_19($i,@V); unshift(@V,pop(@V)); }
 278
 279 $code.=<<___;
 280         ldah    $K,28378(zero)
 281         lda     $K,-5215($K)    # K_20_39
 282 ___
 283 for (;$i<40;$i++) { &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
 284
 285 $code.=<<___;
 286         ldah    $K,-28900(zero)
 287         lda     $K,-17188($K)   # K_40_59
 288 ___
 289 for (;$i<60;$i++) { &BODY_40_59($i,@V); unshift(@V,pop(@V)); }
 290
 291 $code.=<<___;
 292         ldah    $K,-13725(zero)
 293         lda     $K,-15914($K)   # K_60_79
 294 ___
 295 for (;$i<80;$i++) { &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
 296
 297 $code.=<<___;
 298         addl    @X[0],$A,$A
 299         addl    @X[1],$B,$B
 300         addl    @X[2],$C,$C
 301         addl    @X[3],$D,$D
 302         addl    @X[4],$E,$E
 303         stl     $A,0($ctx)
 304         stl     $B,4($ctx)
 305         addq    $inp,64,$inp
 306         stl     $C,8($ctx)
 307         stl     $D,12($ctx)
 308         stl     $E,16($ctx)
 309         cmpult  $inp,$num,$t1
 310         bne     $t1,.Lloop
 311
 312         .set    noreorder
 313         ldq     ra,0(sp)
 314         ldq     s0,8(sp)
 315         ldq     s1,16(sp)
 316         ldq     s2,24(sp)
 317         ldq     s3,32(sp)
 318         ldq     s4,40(sp)
 319         ldq     s5,48(sp)
 320         ldq     fp,56(sp)
 321         lda     sp,64(sp)
 322         ret     (ra)
 323 .end    sha1_block_data_order
 324 .ascii  "SHA1 block transform for Alpha, CRYPTOGAMS by <appro\@openssl.org>"
 325 .align  2
 326 ___
 327 $output=pop and open STDOUT,">$output";
 328 print $code;
 329 close STDOUT;