crypto/sha/asm/sha1-ppc.pl

   1 #!/usr/bin/env perl
   2
   3 # ====================================================================
   4 # Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
   5 # project. The module is, however, dual licensed under OpenSSL and
   6 # CRYPTOGAMS licenses depending on where you obtain it. For further
   7 # details see http://www.openssl.org/~appro/cryptogams/.
   8 # ====================================================================
   9
  10 # I let hardware handle unaligned input(*), except on page boundaries
  11 # (see below for details). Otherwise straightforward implementation
  12 # with X vector in register bank. The module is big-endian [which is
  13 # not big deal as there're no little-endian targets left around].
  14 #
  15 # (*) this means that this module is inappropriate for PPC403? Does
  16 #     anybody know if pre-POWER3 can sustain unaligned load?
  17
  18 #                       -m64    -m32
  19 # ----------------------------------
  20 # PPC970,gcc-4.0.0      +76%    +59%
  21 # Power6,xlc-7          +68%    +33%
  22
  23 $output = shift;
  24
  25 if ($output =~ /64\.s/) {
  26         $SIZE_T =8;
  27         $UCMP   ="cmpld";
  28         $STU    ="stdu";
  29         $POP    ="ld";
  30         $PUSH   ="std";
  31 } elsif ($output =~ /32\.s/) {
  32         $SIZE_T =4;
  33         $UCMP   ="cmplw";
  34         $STU    ="stwu";
  35         $POP    ="lwz";
  36         $PUSH   ="stw";
  37 } else { die "nonsense $output"; }
  38
  39 $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
  40 ( $xlate="${dir}ppc-xlate.pl" and -f $xlate ) or
  41 ( $xlate="${dir}../../perlasm/ppc-xlate.pl" and -f $xlate) or
  42 die "can't locate ppc-xlate.pl";
  43
  44 ( defined shift || open STDOUT,"| $^X $xlate $output" ) ||
  45         die "can't call $xlate: $!";
  46
  47 $FRAME=24*$SIZE_T;
  48
  49 $K  ="r0";
  50 $sp ="r1";
  51 $toc="r2";
  52 $ctx="r3";
  53 $inp="r4";
  54 $num="r5";
  55 $t0 ="r15";
  56 $t1 ="r6";
  57
  58 $A  ="r7";
  59 $B  ="r8";
  60 $C  ="r9";
  61 $D  ="r10";
  62 $E  ="r11";
  63 $T  ="r12";
  64
  65 @V=($A,$B,$C,$D,$E,$T);
  66 @X=("r16","r17","r18","r19","r20","r21","r22","r23",
  67     "r24","r25","r26","r27","r28","r29","r30","r31");
  68
  69 sub BODY_00_19 {
  70 my ($i,$a,$b,$c,$d,$e,$f)=@_;
  71 my $j=$i+1;
  72 $code.=<<___ if ($i==0);
  73         lwz     @X[$i],`$i*4`($inp)
  74 ___
  75 $code.=<<___ if ($i<15);
  76         lwz     @X[$j],`$j*4`($inp)
  77         add     $f,$K,$e
  78         rotlwi  $e,$a,5
  79         add     $f,$f,@X[$i]
  80         and     $t0,$c,$b
  81         add     $f,$f,$e
  82         andc    $t1,$d,$b
  83         rotlwi  $b,$b,30
  84         or      $t0,$t0,$t1
  85         add     $f,$f,$t0
  86 ___
  87 $code.=<<___ if ($i>=15);
  88         add     $f,$K,$e
  89         rotlwi  $e,$a,5
  90         xor     @X[$j%16],@X[$j%16],@X[($j+2)%16]
  91         add     $f,$f,@X[$i%16]
  92         and     $t0,$c,$b
  93         xor     @X[$j%16],@X[$j%16],@X[($j+8)%16]
  94         add     $f,$f,$e
  95         andc    $t1,$d,$b
  96         rotlwi  $b,$b,30
  97         or      $t0,$t0,$t1
  98         xor     @X[$j%16],@X[$j%16],@X[($j+13)%16]
  99         add     $f,$f,$t0
 100         rotlwi  @X[$j%16],@X[$j%16],1
 101 ___
 102 }
 103
 104 sub BODY_20_39 {
 105 my ($i,$a,$b,$c,$d,$e,$f)=@_;
 106 my $j=$i+1;
 107 $code.=<<___ if ($i<79);
 108         add     $f,$K,$e
 109         rotlwi  $e,$a,5
 110         xor     @X[$j%16],@X[$j%16],@X[($j+2)%16]
 111         add     $f,$f,@X[$i%16]
 112         xor     $t0,$b,$c
 113         xor     @X[$j%16],@X[$j%16],@X[($j+8)%16]
 114         add     $f,$f,$e
 115         rotlwi  $b,$b,30
 116         xor     $t0,$t0,$d
 117         xor     @X[$j%16],@X[$j%16],@X[($j+13)%16]
 118         add     $f,$f,$t0
 119         rotlwi  @X[$j%16],@X[$j%16],1
 120 ___
 121 $code.=<<___ if ($i==79);
 122         add     $f,$K,$e
 123         rotlwi  $e,$a,5
 124         lwz     r16,0($ctx)
 125         add     $f,$f,@X[$i%16]
 126         xor     $t0,$b,$c
 127         lwz     r17,4($ctx)
 128         add     $f,$f,$e
 129         rotlwi  $b,$b,30
 130         lwz     r18,8($ctx)
 131         xor     $t0,$t0,$d
 132         lwz     r19,12($ctx)
 133         add     $f,$f,$t0
 134         lwz     r20,16($ctx)
 135 ___
 136 }
 137
 138 sub BODY_40_59 {
 139 my ($i,$a,$b,$c,$d,$e,$f)=@_;
 140 my $j=$i+1;
 141 $code.=<<___;
 142         add     $f,$K,$e
 143         rotlwi  $e,$a,5
 144         xor     @X[$j%16],@X[$j%16],@X[($j+2)%16]
 145         add     $f,$f,@X[$i%16]
 146         and     $t0,$b,$c
 147         xor     @X[$j%16],@X[$j%16],@X[($j+8)%16]
 148         add     $f,$f,$e
 149         or      $t1,$b,$c
 150         rotlwi  $b,$b,30
 151         xor     @X[$j%16],@X[$j%16],@X[($j+13)%16]
 152         and     $t1,$t1,$d
 153         or      $t0,$t0,$t1
 154         rotlwi  @X[$j%16],@X[$j%16],1
 155         add     $f,$f,$t0
 156 ___
 157 }
 158
 159 $code=<<___;
 160 .text
 161
 162 .globl  .sha1_block_data_order
 163 .align  4
 164 .sha1_block_data_order:
 165         mflr    r0
 166         $STU    $sp,`-($FRAME+64)`($sp)
 167         $PUSH   r0,`$FRAME-$SIZE_T*18`($sp)
 168         $PUSH   r15,`$FRAME-$SIZE_T*17`($sp)
 169         $PUSH   r16,`$FRAME-$SIZE_T*16`($sp)
 170         $PUSH   r17,`$FRAME-$SIZE_T*15`($sp)
 171         $PUSH   r18,`$FRAME-$SIZE_T*14`($sp)
 172         $PUSH   r19,`$FRAME-$SIZE_T*13`($sp)
 173         $PUSH   r20,`$FRAME-$SIZE_T*12`($sp)
 174         $PUSH   r21,`$FRAME-$SIZE_T*11`($sp)
 175         $PUSH   r22,`$FRAME-$SIZE_T*10`($sp)
 176         $PUSH   r23,`$FRAME-$SIZE_T*9`($sp)
 177         $PUSH   r24,`$FRAME-$SIZE_T*8`($sp)
 178         $PUSH   r25,`$FRAME-$SIZE_T*7`($sp)
 179         $PUSH   r26,`$FRAME-$SIZE_T*6`($sp)
 180         $PUSH   r27,`$FRAME-$SIZE_T*5`($sp)
 181         $PUSH   r28,`$FRAME-$SIZE_T*4`($sp)
 182         $PUSH   r29,`$FRAME-$SIZE_T*3`($sp)
 183         $PUSH   r30,`$FRAME-$SIZE_T*2`($sp)
 184         $PUSH   r31,`$FRAME-$SIZE_T*1`($sp)
 185         lwz     $A,0($ctx)
 186         lwz     $B,4($ctx)
 187         lwz     $C,8($ctx)
 188         lwz     $D,12($ctx)
 189         lwz     $E,16($ctx)
 190         andi.   r0,$inp,3
 191         bne     Lunaligned
 192 Laligned:
 193         mtctr   $num
 194         bl      Lsha1_block_private
 195 Ldone:
 196         $POP    r0,`$FRAME-$SIZE_T*18`($sp)
 197         $POP    r15,`$FRAME-$SIZE_T*17`($sp)
 198         $POP    r16,`$FRAME-$SIZE_T*16`($sp)
 199         $POP    r17,`$FRAME-$SIZE_T*15`($sp)
 200         $POP    r18,`$FRAME-$SIZE_T*14`($sp)
 201         $POP    r19,`$FRAME-$SIZE_T*13`($sp)
 202         $POP    r20,`$FRAME-$SIZE_T*12`($sp)
 203         $POP    r21,`$FRAME-$SIZE_T*11`($sp)
 204         $POP    r22,`$FRAME-$SIZE_T*10`($sp)
 205         $POP    r23,`$FRAME-$SIZE_T*9`($sp)
 206         $POP    r24,`$FRAME-$SIZE_T*8`($sp)
 207         $POP    r25,`$FRAME-$SIZE_T*7`($sp)
 208         $POP    r26,`$FRAME-$SIZE_T*6`($sp)
 209         $POP    r27,`$FRAME-$SIZE_T*5`($sp)
 210         $POP    r28,`$FRAME-$SIZE_T*4`($sp)
 211         $POP    r29,`$FRAME-$SIZE_T*3`($sp)
 212         $POP    r30,`$FRAME-$SIZE_T*2`($sp)
 213         $POP    r31,`$FRAME-$SIZE_T*1`($sp)
 214         mtlr    r0
 215         addi    $sp,$sp,`$FRAME+64`
 216         blr
 217 ___
 218
 219 # PowerPC specification allows an implementation to be ill-behaved
 220 # upon unaligned access which crosses page boundary. "Better safe
 221 # than sorry" principle makes me treat it specially. But I don't
 222 # look for particular offending word, but rather for 64-byte input
 223 # block which crosses the boundary. Once found that block is aligned
 224 # and hashed separately...
 225 $code.=<<___;
 226 .align  4
 227 Lunaligned:
 228         subfic  $t1,$inp,4096
 229         andi.   $t1,$t1,4095    ; distance to closest page boundary
 230         srwi.   $t1,$t1,6       ; t1/=64
 231         beq     Lcross_page
 232         $UCMP   $num,$t1
 233         ble-    Laligned        ; didn't cross the page boundary
 234         mtctr   $t1
 235         subfc   $num,$t1,$num
 236         bl      Lsha1_block_private
 237 Lcross_page:
 238         li      $t1,16
 239         mtctr   $t1
 240         addi    r20,$sp,$FRAME  ; spot below the frame
 241 Lmemcpy:
 242         lbz     r16,0($inp)
 243         lbz     r17,1($inp)
 244         lbz     r18,2($inp)
 245         lbz     r19,3($inp)
 246         addi    $inp,$inp,4
 247         stb     r16,0(r20)
 248         stb     r17,1(r20)
 249         stb     r18,2(r20)
 250         stb     r19,3(r20)
 251         addi    r20,r20,4
 252         bdnz    Lmemcpy
 253
 254         $PUSH   $inp,`$FRAME-$SIZE_T*19`($sp)
 255         li      $t1,1
 256         addi    $inp,$sp,$FRAME
 257         mtctr   $t1
 258         bl      Lsha1_block_private
 259         $POP    $inp,`$FRAME-$SIZE_T*19`($sp)
 260         addic.  $num,$num,-1
 261         bne-    Lunaligned
 262         b       Ldone
 263 ___
 264
 265 # This is private block function, which uses tailored calling
 266 # interface, namely upon entry SHA_CTX is pre-loaded to given
 267 # registers and counter register contains amount of chunks to
 268 # digest...
 269 $code.=<<___;
 270 .align  4
 271 Lsha1_block_private:
 272 ___
 273 $code.=<<___;   # load K_00_19
 274         lis     $K,0x5a82
 275         ori     $K,$K,0x7999
 276 ___
 277 for($i=0;$i<20;$i++)    { &BODY_00_19($i,@V); unshift(@V,pop(@V)); }
 278 $code.=<<___;   # load K_20_39
 279         lis     $K,0x6ed9
 280         ori     $K,$K,0xeba1
 281 ___
 282 for(;$i<40;$i++)        { &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
 283 $code.=<<___;   # load K_40_59
 284         lis     $K,0x8f1b
 285         ori     $K,$K,0xbcdc
 286 ___
 287 for(;$i<60;$i++)        { &BODY_40_59($i,@V); unshift(@V,pop(@V)); }
 288 $code.=<<___;   # load K_60_79
 289         lis     $K,0xca62
 290         ori     $K,$K,0xc1d6
 291 ___
 292 for(;$i<80;$i++)        { &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
 293 $code.=<<___;
 294         add     r16,r16,$E
 295         add     r17,r17,$T
 296         add     r18,r18,$A
 297         add     r19,r19,$B
 298         add     r20,r20,$C
 299         stw     r16,0($ctx)
 300         mr      $A,r16
 301         stw     r17,4($ctx)
 302         mr      $B,r17
 303         stw     r18,8($ctx)
 304         mr      $C,r18
 305         stw     r19,12($ctx)
 306         mr      $D,r19
 307         stw     r20,16($ctx)
 308         mr      $E,r20
 309         addi    $inp,$inp,`16*4`
 310         bdnz-   Lsha1_block_private
 311         blr
 312 ___
 313 $code.=<<___;
 314 .asciz  "SHA1 block transform for PPC, CRYPTOGAMS by <appro\@fy.chalmers.se>"
 315 ___
 316
 317 $code =~ s/\`([^\`]*)\`/eval $1/gem;
 318 print $code;
 319 close STDOUT;