3 # ====================================================================
4 # Written by Andy Polyakov <appro@openssl.org> 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 # ====================================================================
12 # Performance in cycles per processed byte and improvement coefficient
13 # over code generated with "default" compiler:
15 # hardware-assisted software(*)
16 # Apple A7 2.31 4.13 (+14%)
19 # (*) Software results are presented mostly for reference purposes.
22 open STDOUT,">".shift;
24 ($ctx,$inp,$num)=("x0","x1","x2");
25 @Xw=map("w$_",(3..17,19));
26 @Xx=map("x$_",(3..17,19));
27 @V=($A,$B,$C,$D,$E)=map("w$_",(20..24));
28 ($t0,$t1,$t2,$K)=map("w$_",(25..28));
32 my ($i,$a,$b,$c,$d,$e)=@_;
35 $code.=<<___ if ($i<15 && !($i&1));
36 lsr @Xx[$i+1],@Xx[$i],#32
38 $code.=<<___ if ($i<14 && !($i&1));
39 ldr @Xx[$i+2],[$inp,#`($i+2)*4-64`]
41 $code.=<<___ if ($i<14 && ($i&1));
43 ror @Xx[$i+1],@Xx[$i+1],#32
45 rev32 @Xx[$i+1],@Xx[$i+1]
48 $code.=<<___ if ($i<14);
52 add $d,$d,$K // future e+=K
54 add $e,$e,$t2 // e+=rot(a,5)
56 add $d,$d,@Xw[($i+1)&15] // future e+=X[i]
57 add $e,$e,$t0 // e+=F(b,c,d)
59 $code.=<<___ if ($i==19);
61 movk $K,#0x6ed9,lsl#16
63 $code.=<<___ if ($i>=14);
64 eor @Xw[$j],@Xw[$j],@Xw[($j+2)&15]
68 eor @Xw[$j],@Xw[$j],@Xw[($j+8)&15]
69 add $d,$d,$K // future e+=K
71 add $e,$e,$t2 // e+=rot(a,5)
72 eor @Xw[$j],@Xw[$j],@Xw[($j+13)&15]
74 add $d,$d,@Xw[($i+1)&15] // future e+=X[i]
75 add $e,$e,$t0 // e+=F(b,c,d)
76 ror @Xw[$j],@Xw[$j],#31
81 my ($i,$a,$b,$c,$d,$e)=@_;
84 $code.=<<___ if ($i==59);
86 movk $K,#0xca62,lsl#16
91 eor @Xw[$j],@Xw[$j],@Xw[($j+2)&15]
94 add $d,$d,$K // future e+=K
95 eor @Xw[$j],@Xw[$j],@Xw[($j+8)&15]
96 add $e,$e,$t2 // e+=rot(a,5)
99 eor @Xw[$j],@Xw[$j],@Xw[($j+13)&15]
100 add $d,$d,@Xw[($i+1)&15] // future e+=X[i]
101 add $e,$e,$t0 // e+=F(b,c,d)
102 ror @Xw[$j],@Xw[$j],#31
107 my ($i,$a,$b,$c,$d,$e)=@_;
110 $code.=<<___ if ($i==39);
112 movk $K,#0x8f1b,lsl#16
114 $code.=<<___ if ($i<78);
115 eor @Xw[$j],@Xw[$j],@Xw[($j+2)&15]
118 add $d,$d,$K // future e+=K
119 eor @Xw[$j],@Xw[$j],@Xw[($j+8)&15]
121 add $e,$e,$t2 // e+=rot(a,5)
123 eor @Xw[$j],@Xw[$j],@Xw[($j+13)&15]
124 add $d,$d,@Xw[($i+1)&15] // future e+=X[i]
125 add $e,$e,$t0 // e+=F(b,c,d)
126 ror @Xw[$j],@Xw[$j],#31
128 $code.=<<___ if ($i==78);
129 ldp @Xw[1],@Xw[2],[$ctx]
132 add $d,$d,$K // future e+=K
134 add $e,$e,$t2 // e+=rot(a,5)
136 add $d,$d,@Xw[($i+1)&15] // future e+=X[i]
137 add $e,$e,$t0 // e+=F(b,c,d)
139 $code.=<<___ if ($i==79);
140 ldp @Xw[3],@Xw[4],[$ctx,#8]
144 add $e,$e,$t2 // e+=rot(a,5)
146 ldr @Xw[5],[$ctx,#16]
147 add $e,$e,$t0 // e+=F(b,c,d)
152 #include "arm_arch.h"
156 .globl sha1_block_data_order
157 .type sha1_block_data_order,%function
159 sha1_block_data_order:
160 ldr x16,.LOPENSSL_armcap_P
161 adr x17,.LOPENSSL_armcap_P
167 stp x29,x30,[sp,#-96]!
180 ldr @Xx[0],[$inp],#64
183 movk $K,#0x5a82,lsl#16
185 ror $Xx[0],@Xx[0],#32
189 add $E,$E,$K // warm it up
192 for($i=0;$i<20;$i++) { &BODY_00_19($i,@V); unshift(@V,pop(@V)); }
193 for(;$i<40;$i++) { &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
194 for(;$i<60;$i++) { &BODY_40_59($i,@V); unshift(@V,pop(@V)); }
195 for(;$i<80;$i++) { &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
214 .size sha1_block_data_order,.-sha1_block_data_order
217 my ($ABCD,$E,$E0,$E1)=map("v$_.16b",(0..3));
218 my @MSG=map("v$_.16b",(4..7));
219 my @Kxx=map("v$_.4s",(16..19));
220 my ($W0,$W1)=("v20.4s","v21.4s");
221 my $ABCD_SAVE="v22.16b";
224 .type sha1_block_armv8,%function
228 stp x29,x30,[sp,#-16]!
233 ld1.32 {$ABCD},[$ctx],#16
234 ld1.32 {$E}[0],[$ctx]
236 ld1.32 {@Kxx[0]-@Kxx[3]},[x4]
239 ld1 {@MSG[0]-@MSG[3]},[$inp],#64
241 rev32 @MSG[0],@MSG[0]
242 rev32 @MSG[1],@MSG[1]
244 add.i32 $W0,@Kxx[0],@MSG[0]
245 rev32 @MSG[2],@MSG[2]
246 orr $ABCD_SAVE,$ABCD,$ABCD // offload
248 add.i32 $W1,@Kxx[0],@MSG[1]
249 rev32 @MSG[3],@MSG[3]
251 sha1c $ABCD,$E,$W0 // 0
252 add.i32 $W0,@Kxx[$j],@MSG[2]
253 sha1su0 @MSG[0],@MSG[1],@MSG[2]
255 for ($j=0,$i=1;$i<20-3;$i++) {
256 my $f=("c","p","m","p")[$i/5];
258 sha1h $E0,$ABCD // $i
260 add.i32 $W1,@Kxx[$j],@MSG[3]
261 sha1su1 @MSG[0],@MSG[3]
263 $code.=<<___ if ($i<20-4);
264 sha1su0 @MSG[1],@MSG[2],@MSG[3]
266 ($E0,$E1)=($E1,$E0); ($W0,$W1)=($W1,$W0);
267 push(@MSG,shift(@MSG)); $j++ if ((($i+3)%5)==0);
270 sha1h $E0,$ABCD // $i
272 add.i32 $W1,@Kxx[$j],@MSG[3]
274 sha1h $E1,$ABCD // 18
277 sha1h $E0,$ABCD // 19
281 add.i32 $ABCD,$ABCD,$ABCD_SAVE
285 st1.32 {$ABCD},[$ctx],#16
286 st1.32 {$E}[0],[$ctx]
290 .size sha1_block_armv8,.-sha1_block_armv8
293 .long 0x5a827999,0x5a827999,0x5a827999,0x5a827999 //K_00_19
294 .long 0x6ed9eba1,0x6ed9eba1,0x6ed9eba1,0x6ed9eba1 //K_20_39
295 .long 0x8f1bbcdc,0x8f1bbcdc,0x8f1bbcdc,0x8f1bbcdc //K_40_59
296 .long 0xca62c1d6,0xca62c1d6,0xca62c1d6,0xca62c1d6 //K_60_79
298 .quad OPENSSL_armcap_P-.
299 .asciz "SHA1 block transform for ARMv8, CRYPTOGAMS by <appro\@openssl.org>"
301 .comm OPENSSL_armcap_P,4,4
306 "sha1c" => 0x5e000000, "sha1p" => 0x5e001000,
307 "sha1m" => 0x5e002000, "sha1su0" => 0x5e003000,
308 "sha1h" => 0x5e280800, "sha1su1" => 0x5e281800 );
311 my ($mnemonic,$arg)=@_;
313 $arg =~ m/[qv]([0-9]+)[^,]*,\s*[qv]([0-9]+)[^,]*(?:,\s*[qv]([0-9]+))?/o
315 sprintf ".long\t0x%08x\t//%s %s",
316 $opcode{$mnemonic}|$1|($2<<5)|($3<<16),
321 foreach(split("\n",$code)) {
323 s/\`([^\`]*)\`/eval($1)/geo;
325 s/\b(sha1\w+)\s+([qv].*)/unsha1($1,$2)/geo;
327 s/\.\w?32\b//o and s/\.16b/\.4s/go;
328 m/(ld|st)1[^\[]+\[0\]/o and s/\.4s/\.s/go;