summaryrefslogtreecommitdiff
path: root/src/lib/libcrypto/md5/asm
diff options
context:
space:
mode:
Diffstat (limited to 'src/lib/libcrypto/md5/asm')
-rw-r--r--src/lib/libcrypto/md5/asm/md5-586.pl306
-rwxr-xr-xsrc/lib/libcrypto/md5/asm/md5-x86_64.pl245
2 files changed, 0 insertions, 551 deletions
diff --git a/src/lib/libcrypto/md5/asm/md5-586.pl b/src/lib/libcrypto/md5/asm/md5-586.pl
deleted file mode 100644
index 76ac235f7d..0000000000
--- a/src/lib/libcrypto/md5/asm/md5-586.pl
+++ /dev/null
@@ -1,306 +0,0 @@
1#!/usr/local/bin/perl
2
3# Normal is the
4# md5_block_x86(MD5_CTX *c, ULONG *X);
5# version, non-normal is the
6# md5_block_x86(MD5_CTX *c, ULONG *X,int blocks);
7
8$normal=0;
9
10push(@INC,"perlasm","../../perlasm");
11require "x86asm.pl";
12
13&asm_init($ARGV[0],$0);
14
15$A="eax";
16$B="ebx";
17$C="ecx";
18$D="edx";
19$tmp1="edi";
20$tmp2="ebp";
21$X="esi";
22
23# What we need to load into $tmp for the next round
24%Ltmp1=("R0",&Np($C), "R1",&Np($C), "R2",&Np($C), "R3",&Np($D));
25@xo=(
26 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, # R0
27 1, 6, 11, 0, 5, 10, 15, 4, 9, 14, 3, 8, 13, 2, 7, 12, # R1
28 5, 8, 11, 14, 1, 4, 7, 10, 13, 0, 3, 6, 9, 12, 15, 2, # R2
29 0, 7, 14, 5, 12, 3, 10, 1, 8, 15, 6, 13, 4, 11, 2, 9, # R3
30 );
31
32&md5_block("md5_block_asm_data_order");
33&asm_finish();
34
35sub Np
36 {
37 local($p)=@_;
38 local(%n)=($A,$D,$B,$A,$C,$B,$D,$C);
39 return($n{$p});
40 }
41
42sub R0
43 {
44 local($pos,$a,$b,$c,$d,$K,$ki,$s,$t)=@_;
45
46 &mov($tmp1,$C) if $pos < 0;
47 &mov($tmp2,&DWP($xo[$ki]*4,$K,"",0)) if $pos < 0; # very first one
48
49 # body proper
50
51 &comment("R0 $ki");
52 &xor($tmp1,$d); # F function - part 2
53
54 &and($tmp1,$b); # F function - part 3
55 &lea($a,&DWP($t,$a,$tmp2,1));
56
57 &xor($tmp1,$d); # F function - part 4
58
59 &add($a,$tmp1);
60 &mov($tmp1,&Np($c)) if $pos < 1; # next tmp1 for R0
61 &mov($tmp1,&Np($c)) if $pos == 1; # next tmp1 for R1
62
63 &rotl($a,$s);
64
65 &mov($tmp2,&DWP($xo[$ki+1]*4,$K,"",0)) if ($pos != 2);
66
67 &add($a,$b);
68 }
69
70sub R1
71 {
72 local($pos,$a,$b,$c,$d,$K,$ki,$s,$t)=@_;
73
74 &comment("R1 $ki");
75
76 &lea($a,&DWP($t,$a,$tmp2,1));
77
78 &xor($tmp1,$b); # G function - part 2
79 &and($tmp1,$d); # G function - part 3
80
81 &mov($tmp2,&DWP($xo[$ki+1]*4,$K,"",0)) if ($pos != 2);
82 &xor($tmp1,$c); # G function - part 4
83
84 &add($a,$tmp1);
85 &mov($tmp1,&Np($c)) if $pos < 1; # G function - part 1
86 &mov($tmp1,&Np($c)) if $pos == 1; # G function - part 1
87
88 &rotl($a,$s);
89
90 &add($a,$b);
91 }
92
93sub R2
94 {
95 local($n,$pos,$a,$b,$c,$d,$K,$ki,$s,$t)=@_;
96 # This one is different, only 3 logical operations
97
98if (($n & 1) == 0)
99 {
100 &comment("R2 $ki");
101 # make sure to do 'D' first, not 'B', else we clash with
102 # the last add from the previous round.
103
104 &xor($tmp1,$d); # H function - part 2
105
106 &xor($tmp1,$b); # H function - part 3
107 &lea($a,&DWP($t,$a,$tmp2,1));
108
109 &add($a,$tmp1);
110
111 &rotl($a,$s);
112
113 &mov($tmp2,&DWP($xo[$ki+1]*4,$K,"",0));
114 &mov($tmp1,&Np($c));
115 }
116else
117 {
118 &comment("R2 $ki");
119 # make sure to do 'D' first, not 'B', else we clash with
120 # the last add from the previous round.
121
122 &lea($a,&DWP($t,$a,$tmp2,1));
123
124 &add($b,$c); # MOVED FORWARD
125 &xor($tmp1,$d); # H function - part 2
126
127 &xor($tmp1,$b); # H function - part 3
128 &mov($tmp2,&DWP($xo[$ki+1]*4,$K,"",0)) if ($pos != 2);
129
130 &add($a,$tmp1);
131 &mov($tmp1,&Np($c)) if $pos < 1; # H function - part 1
132 &mov($tmp1,-1) if $pos == 1; # I function - part 1
133
134 &rotl($a,$s);
135
136 &add($a,$b);
137 }
138 }
139
140sub R3
141 {
142 local($pos,$a,$b,$c,$d,$K,$ki,$s,$t)=@_;
143
144 &comment("R3 $ki");
145
146 # &not($tmp1)
147 &xor($tmp1,$d) if $pos < 0; # I function - part 2
148
149 &or($tmp1,$b); # I function - part 3
150 &lea($a,&DWP($t,$a,$tmp2,1));
151
152 &xor($tmp1,$c); # I function - part 4
153 &mov($tmp2,&DWP($xo[$ki+1]*4,$K,"",0)) if $pos != 2; # load X/k value
154 &mov($tmp2,&wparam(0)) if $pos == 2;
155
156 &add($a,$tmp1);
157 &mov($tmp1,-1) if $pos < 1; # H function - part 1
158 &add($K,64) if $pos >=1 && !$normal;
159
160 &rotl($a,$s);
161
162 &xor($tmp1,&Np($d)) if $pos <= 0; # I function - part = first time
163 &mov($tmp1,&DWP( 0,$tmp2,"",0)) if $pos > 0;
164 &add($a,$b);
165 }
166
167
168sub md5_block
169 {
170 local($name)=@_;
171
172 &function_begin_B($name,"",3);
173
174 # parameter 1 is the MD5_CTX structure.
175 # A 0
176 # B 4
177 # C 8
178 # D 12
179
180 &push("esi");
181 &push("edi");
182 &mov($tmp1, &wparam(0)); # edi
183 &mov($X, &wparam(1)); # esi
184 &mov($C, &wparam(2));
185 &push("ebp");
186 &shl($C, 6);
187 &push("ebx");
188 &add($C, $X); # offset we end at
189 &sub($C, 64);
190 &mov($A, &DWP( 0,$tmp1,"",0));
191 &push($C); # Put on the TOS
192 &mov($B, &DWP( 4,$tmp1,"",0));
193 &mov($C, &DWP( 8,$tmp1,"",0));
194 &mov($D, &DWP(12,$tmp1,"",0));
195
196 &set_label("start") unless $normal;
197 &comment("");
198 &comment("R0 section");
199
200 &R0(-2,$A,$B,$C,$D,$X, 0, 7,0xd76aa478);
201 &R0( 0,$D,$A,$B,$C,$X, 1,12,0xe8c7b756);
202 &R0( 0,$C,$D,$A,$B,$X, 2,17,0x242070db);
203 &R0( 0,$B,$C,$D,$A,$X, 3,22,0xc1bdceee);
204 &R0( 0,$A,$B,$C,$D,$X, 4, 7,0xf57c0faf);
205 &R0( 0,$D,$A,$B,$C,$X, 5,12,0x4787c62a);
206 &R0( 0,$C,$D,$A,$B,$X, 6,17,0xa8304613);
207 &R0( 0,$B,$C,$D,$A,$X, 7,22,0xfd469501);
208 &R0( 0,$A,$B,$C,$D,$X, 8, 7,0x698098d8);
209 &R0( 0,$D,$A,$B,$C,$X, 9,12,0x8b44f7af);
210 &R0( 0,$C,$D,$A,$B,$X,10,17,0xffff5bb1);
211 &R0( 0,$B,$C,$D,$A,$X,11,22,0x895cd7be);
212 &R0( 0,$A,$B,$C,$D,$X,12, 7,0x6b901122);
213 &R0( 0,$D,$A,$B,$C,$X,13,12,0xfd987193);
214 &R0( 0,$C,$D,$A,$B,$X,14,17,0xa679438e);
215 &R0( 1,$B,$C,$D,$A,$X,15,22,0x49b40821);
216
217 &comment("");
218 &comment("R1 section");
219 &R1(-1,$A,$B,$C,$D,$X,16, 5,0xf61e2562);
220 &R1( 0,$D,$A,$B,$C,$X,17, 9,0xc040b340);
221 &R1( 0,$C,$D,$A,$B,$X,18,14,0x265e5a51);
222 &R1( 0,$B,$C,$D,$A,$X,19,20,0xe9b6c7aa);
223 &R1( 0,$A,$B,$C,$D,$X,20, 5,0xd62f105d);
224 &R1( 0,$D,$A,$B,$C,$X,21, 9,0x02441453);
225 &R1( 0,$C,$D,$A,$B,$X,22,14,0xd8a1e681);
226 &R1( 0,$B,$C,$D,$A,$X,23,20,0xe7d3fbc8);
227 &R1( 0,$A,$B,$C,$D,$X,24, 5,0x21e1cde6);
228 &R1( 0,$D,$A,$B,$C,$X,25, 9,0xc33707d6);
229 &R1( 0,$C,$D,$A,$B,$X,26,14,0xf4d50d87);
230 &R1( 0,$B,$C,$D,$A,$X,27,20,0x455a14ed);
231 &R1( 0,$A,$B,$C,$D,$X,28, 5,0xa9e3e905);
232 &R1( 0,$D,$A,$B,$C,$X,29, 9,0xfcefa3f8);
233 &R1( 0,$C,$D,$A,$B,$X,30,14,0x676f02d9);
234 &R1( 1,$B,$C,$D,$A,$X,31,20,0x8d2a4c8a);
235
236 &comment("");
237 &comment("R2 section");
238 &R2( 0,-1,$A,$B,$C,$D,$X,32, 4,0xfffa3942);
239 &R2( 1, 0,$D,$A,$B,$C,$X,33,11,0x8771f681);
240 &R2( 2, 0,$C,$D,$A,$B,$X,34,16,0x6d9d6122);
241 &R2( 3, 0,$B,$C,$D,$A,$X,35,23,0xfde5380c);
242 &R2( 4, 0,$A,$B,$C,$D,$X,36, 4,0xa4beea44);
243 &R2( 5, 0,$D,$A,$B,$C,$X,37,11,0x4bdecfa9);
244 &R2( 6, 0,$C,$D,$A,$B,$X,38,16,0xf6bb4b60);
245 &R2( 7, 0,$B,$C,$D,$A,$X,39,23,0xbebfbc70);
246 &R2( 8, 0,$A,$B,$C,$D,$X,40, 4,0x289b7ec6);
247 &R2( 9, 0,$D,$A,$B,$C,$X,41,11,0xeaa127fa);
248 &R2(10, 0,$C,$D,$A,$B,$X,42,16,0xd4ef3085);
249 &R2(11, 0,$B,$C,$D,$A,$X,43,23,0x04881d05);
250 &R2(12, 0,$A,$B,$C,$D,$X,44, 4,0xd9d4d039);
251 &R2(13, 0,$D,$A,$B,$C,$X,45,11,0xe6db99e5);
252 &R2(14, 0,$C,$D,$A,$B,$X,46,16,0x1fa27cf8);
253 &R2(15, 1,$B,$C,$D,$A,$X,47,23,0xc4ac5665);
254
255 &comment("");
256 &comment("R3 section");
257 &R3(-1,$A,$B,$C,$D,$X,48, 6,0xf4292244);
258 &R3( 0,$D,$A,$B,$C,$X,49,10,0x432aff97);
259 &R3( 0,$C,$D,$A,$B,$X,50,15,0xab9423a7);
260 &R3( 0,$B,$C,$D,$A,$X,51,21,0xfc93a039);
261 &R3( 0,$A,$B,$C,$D,$X,52, 6,0x655b59c3);
262 &R3( 0,$D,$A,$B,$C,$X,53,10,0x8f0ccc92);
263 &R3( 0,$C,$D,$A,$B,$X,54,15,0xffeff47d);
264 &R3( 0,$B,$C,$D,$A,$X,55,21,0x85845dd1);
265 &R3( 0,$A,$B,$C,$D,$X,56, 6,0x6fa87e4f);
266 &R3( 0,$D,$A,$B,$C,$X,57,10,0xfe2ce6e0);
267 &R3( 0,$C,$D,$A,$B,$X,58,15,0xa3014314);
268 &R3( 0,$B,$C,$D,$A,$X,59,21,0x4e0811a1);
269 &R3( 0,$A,$B,$C,$D,$X,60, 6,0xf7537e82);
270 &R3( 0,$D,$A,$B,$C,$X,61,10,0xbd3af235);
271 &R3( 0,$C,$D,$A,$B,$X,62,15,0x2ad7d2bb);
272 &R3( 2,$B,$C,$D,$A,$X,63,21,0xeb86d391);
273
274 # &mov($tmp2,&wparam(0)); # done in the last R3
275 # &mov($tmp1, &DWP( 0,$tmp2,"",0)); # done is the last R3
276
277 &add($A,$tmp1);
278 &mov($tmp1, &DWP( 4,$tmp2,"",0));
279
280 &add($B,$tmp1);
281 &mov($tmp1, &DWP( 8,$tmp2,"",0));
282
283 &add($C,$tmp1);
284 &mov($tmp1, &DWP(12,$tmp2,"",0));
285
286 &add($D,$tmp1);
287 &mov(&DWP( 0,$tmp2,"",0),$A);
288
289 &mov(&DWP( 4,$tmp2,"",0),$B);
290 &mov($tmp1,&swtmp(0)) unless $normal;
291
292 &mov(&DWP( 8,$tmp2,"",0),$C);
293 &mov(&DWP(12,$tmp2,"",0),$D);
294
295 &cmp($tmp1,$X) unless $normal; # check count
296 &jae(&label("start")) unless $normal;
297
298 &pop("eax"); # pop the temp variable off the stack
299 &pop("ebx");
300 &pop("ebp");
301 &pop("edi");
302 &pop("esi");
303 &ret();
304 &function_end_B($name);
305 }
306
diff --git a/src/lib/libcrypto/md5/asm/md5-x86_64.pl b/src/lib/libcrypto/md5/asm/md5-x86_64.pl
deleted file mode 100755
index 9a6fa67224..0000000000
--- a/src/lib/libcrypto/md5/asm/md5-x86_64.pl
+++ /dev/null
@@ -1,245 +0,0 @@
1#!/usr/bin/perl -w
2#
3# MD5 optimized for AMD64.
4#
5# Author: Marc Bevand <bevand_m (at) epita.fr>
6# Licence: I hereby disclaim the copyright on this code and place it
7# in the public domain.
8#
9
10use strict;
11
12my $code;
13
14# round1_step() does:
15# dst = x + ((dst + F(x,y,z) + X[k] + T_i) <<< s)
16# %r10d = X[k_next]
17# %r11d = z' (copy of z for the next step)
18# Each round1_step() takes about 5.71 clocks (9 instructions, 1.58 IPC)
19sub round1_step
20{
21 my ($pos, $dst, $x, $y, $z, $k_next, $T_i, $s) = @_;
22 $code .= " mov 0*4(%rsi), %r10d /* (NEXT STEP) X[0] */\n" if ($pos == -1);
23 $code .= " mov %edx, %r11d /* (NEXT STEP) z' = %edx */\n" if ($pos == -1);
24 $code .= <<EOF;
25 xor $y, %r11d /* y ^ ... */
26 lea $T_i($dst,%r10d),$dst /* Const + dst + ... */
27 and $x, %r11d /* x & ... */
28 xor $z, %r11d /* z ^ ... */
29 mov $k_next*4(%rsi),%r10d /* (NEXT STEP) X[$k_next] */
30 add %r11d, $dst /* dst += ... */
31 rol \$$s, $dst /* dst <<< s */
32 mov $y, %r11d /* (NEXT STEP) z' = $y */
33 add $x, $dst /* dst += x */
34EOF
35}
36
37# round2_step() does:
38# dst = x + ((dst + G(x,y,z) + X[k] + T_i) <<< s)
39# %r10d = X[k_next]
40# %r11d = y' (copy of y for the next step)
41# Each round2_step() takes about 6.22 clocks (9 instructions, 1.45 IPC)
42sub round2_step
43{
44 my ($pos, $dst, $x, $y, $z, $k_next, $T_i, $s) = @_;
45 $code .= " mov 1*4(%rsi), %r10d /* (NEXT STEP) X[1] */\n" if ($pos == -1);
46 $code .= " mov %ecx, %r11d /* (NEXT STEP) y' = %ecx */\n" if ($pos == -1);
47 $code .= <<EOF;
48 xor $x, %r11d /* x ^ ... */
49 lea $T_i($dst,%r10d),$dst /* Const + dst + ... */
50 and $z, %r11d /* z & ... */
51 xor $y, %r11d /* y ^ ... */
52 mov $k_next*4(%rsi),%r10d /* (NEXT STEP) X[$k_next] */
53 add %r11d, $dst /* dst += ... */
54 rol \$$s, $dst /* dst <<< s */
55 mov $x, %r11d /* (NEXT STEP) y' = $x */
56 add $x, $dst /* dst += x */
57EOF
58}
59
60# round3_step() does:
61# dst = x + ((dst + H(x,y,z) + X[k] + T_i) <<< s)
62# %r10d = X[k_next]
63# %r11d = y' (copy of y for the next step)
64# Each round3_step() takes about 4.26 clocks (8 instructions, 1.88 IPC)
65sub round3_step
66{
67 my ($pos, $dst, $x, $y, $z, $k_next, $T_i, $s) = @_;
68 $code .= " mov 5*4(%rsi), %r10d /* (NEXT STEP) X[5] */\n" if ($pos == -1);
69 $code .= " mov %ecx, %r11d /* (NEXT STEP) y' = %ecx */\n" if ($pos == -1);
70 $code .= <<EOF;
71 lea $T_i($dst,%r10d),$dst /* Const + dst + ... */
72 mov $k_next*4(%rsi),%r10d /* (NEXT STEP) X[$k_next] */
73 xor $z, %r11d /* z ^ ... */
74 xor $x, %r11d /* x ^ ... */
75 add %r11d, $dst /* dst += ... */
76 rol \$$s, $dst /* dst <<< s */
77 mov $x, %r11d /* (NEXT STEP) y' = $x */
78 add $x, $dst /* dst += x */
79EOF
80}
81
82# round4_step() does:
83# dst = x + ((dst + I(x,y,z) + X[k] + T_i) <<< s)
84# %r10d = X[k_next]
85# %r11d = not z' (copy of not z for the next step)
86# Each round4_step() takes about 5.27 clocks (9 instructions, 1.71 IPC)
87sub round4_step
88{
89 my ($pos, $dst, $x, $y, $z, $k_next, $T_i, $s) = @_;
90 $code .= " mov 0*4(%rsi), %r10d /* (NEXT STEP) X[0] */\n" if ($pos == -1);
91 $code .= " mov \$0xffffffff, %r11d\n" if ($pos == -1);
92 $code .= " xor %edx, %r11d /* (NEXT STEP) not z' = not %edx*/\n"
93 if ($pos == -1);
94 $code .= <<EOF;
95 lea $T_i($dst,%r10d),$dst /* Const + dst + ... */
96 or $x, %r11d /* x | ... */
97 xor $y, %r11d /* y ^ ... */
98 add %r11d, $dst /* dst += ... */
99 mov $k_next*4(%rsi),%r10d /* (NEXT STEP) X[$k_next] */
100 mov \$0xffffffff, %r11d
101 rol \$$s, $dst /* dst <<< s */
102 xor $y, %r11d /* (NEXT STEP) not z' = not $y */
103 add $x, $dst /* dst += x */
104EOF
105}
106
107my $output = shift;
108open STDOUT,"| $^X ../perlasm/x86_64-xlate.pl $output";
109
110$code .= <<EOF;
111.text
112.align 16
113
114.globl md5_block_asm_data_order
115.type md5_block_asm_data_order,\@function,3
116md5_block_asm_data_order:
117 push %rbp
118 push %rbx
119 push %r14
120 push %r15
121
122 # rdi = arg #1 (ctx, MD5_CTX pointer)
123 # rsi = arg #2 (ptr, data pointer)
124 # rdx = arg #3 (nbr, number of 16-word blocks to process)
125 mov %rdi, %rbp # rbp = ctx
126 shl \$6, %rdx # rdx = nbr in bytes
127 lea (%rsi,%rdx), %rdi # rdi = end
128 mov 0*4(%rbp), %eax # eax = ctx->A
129 mov 1*4(%rbp), %ebx # ebx = ctx->B
130 mov 2*4(%rbp), %ecx # ecx = ctx->C
131 mov 3*4(%rbp), %edx # edx = ctx->D
132 # end is 'rdi'
133 # ptr is 'rsi'
134 # A is 'eax'
135 # B is 'ebx'
136 # C is 'ecx'
137 # D is 'edx'
138
139 cmp %rdi, %rsi # cmp end with ptr
140 je .Lend # jmp if ptr == end
141
142 # BEGIN of loop over 16-word blocks
143.Lloop: # save old values of A, B, C, D
144 mov %eax, %r8d
145 mov %ebx, %r9d
146 mov %ecx, %r14d
147 mov %edx, %r15d
148EOF
149round1_step(-1,'%eax','%ebx','%ecx','%edx', '1','0xd76aa478', '7');
150round1_step( 0,'%edx','%eax','%ebx','%ecx', '2','0xe8c7b756','12');
151round1_step( 0,'%ecx','%edx','%eax','%ebx', '3','0x242070db','17');
152round1_step( 0,'%ebx','%ecx','%edx','%eax', '4','0xc1bdceee','22');
153round1_step( 0,'%eax','%ebx','%ecx','%edx', '5','0xf57c0faf', '7');
154round1_step( 0,'%edx','%eax','%ebx','%ecx', '6','0x4787c62a','12');
155round1_step( 0,'%ecx','%edx','%eax','%ebx', '7','0xa8304613','17');
156round1_step( 0,'%ebx','%ecx','%edx','%eax', '8','0xfd469501','22');
157round1_step( 0,'%eax','%ebx','%ecx','%edx', '9','0x698098d8', '7');
158round1_step( 0,'%edx','%eax','%ebx','%ecx','10','0x8b44f7af','12');
159round1_step( 0,'%ecx','%edx','%eax','%ebx','11','0xffff5bb1','17');
160round1_step( 0,'%ebx','%ecx','%edx','%eax','12','0x895cd7be','22');
161round1_step( 0,'%eax','%ebx','%ecx','%edx','13','0x6b901122', '7');
162round1_step( 0,'%edx','%eax','%ebx','%ecx','14','0xfd987193','12');
163round1_step( 0,'%ecx','%edx','%eax','%ebx','15','0xa679438e','17');
164round1_step( 1,'%ebx','%ecx','%edx','%eax', '0','0x49b40821','22');
165
166round2_step(-1,'%eax','%ebx','%ecx','%edx', '6','0xf61e2562', '5');
167round2_step( 0,'%edx','%eax','%ebx','%ecx','11','0xc040b340', '9');
168round2_step( 0,'%ecx','%edx','%eax','%ebx', '0','0x265e5a51','14');
169round2_step( 0,'%ebx','%ecx','%edx','%eax', '5','0xe9b6c7aa','20');
170round2_step( 0,'%eax','%ebx','%ecx','%edx','10','0xd62f105d', '5');
171round2_step( 0,'%edx','%eax','%ebx','%ecx','15', '0x2441453', '9');
172round2_step( 0,'%ecx','%edx','%eax','%ebx', '4','0xd8a1e681','14');
173round2_step( 0,'%ebx','%ecx','%edx','%eax', '9','0xe7d3fbc8','20');
174round2_step( 0,'%eax','%ebx','%ecx','%edx','14','0x21e1cde6', '5');
175round2_step( 0,'%edx','%eax','%ebx','%ecx', '3','0xc33707d6', '9');
176round2_step( 0,'%ecx','%edx','%eax','%ebx', '8','0xf4d50d87','14');
177round2_step( 0,'%ebx','%ecx','%edx','%eax','13','0x455a14ed','20');
178round2_step( 0,'%eax','%ebx','%ecx','%edx', '2','0xa9e3e905', '5');
179round2_step( 0,'%edx','%eax','%ebx','%ecx', '7','0xfcefa3f8', '9');
180round2_step( 0,'%ecx','%edx','%eax','%ebx','12','0x676f02d9','14');
181round2_step( 1,'%ebx','%ecx','%edx','%eax', '0','0x8d2a4c8a','20');
182
183round3_step(-1,'%eax','%ebx','%ecx','%edx', '8','0xfffa3942', '4');
184round3_step( 0,'%edx','%eax','%ebx','%ecx','11','0x8771f681','11');
185round3_step( 0,'%ecx','%edx','%eax','%ebx','14','0x6d9d6122','16');
186round3_step( 0,'%ebx','%ecx','%edx','%eax', '1','0xfde5380c','23');
187round3_step( 0,'%eax','%ebx','%ecx','%edx', '4','0xa4beea44', '4');
188round3_step( 0,'%edx','%eax','%ebx','%ecx', '7','0x4bdecfa9','11');
189round3_step( 0,'%ecx','%edx','%eax','%ebx','10','0xf6bb4b60','16');
190round3_step( 0,'%ebx','%ecx','%edx','%eax','13','0xbebfbc70','23');
191round3_step( 0,'%eax','%ebx','%ecx','%edx', '0','0x289b7ec6', '4');
192round3_step( 0,'%edx','%eax','%ebx','%ecx', '3','0xeaa127fa','11');
193round3_step( 0,'%ecx','%edx','%eax','%ebx', '6','0xd4ef3085','16');
194round3_step( 0,'%ebx','%ecx','%edx','%eax', '9', '0x4881d05','23');
195round3_step( 0,'%eax','%ebx','%ecx','%edx','12','0xd9d4d039', '4');
196round3_step( 0,'%edx','%eax','%ebx','%ecx','15','0xe6db99e5','11');
197round3_step( 0,'%ecx','%edx','%eax','%ebx', '2','0x1fa27cf8','16');
198round3_step( 1,'%ebx','%ecx','%edx','%eax', '0','0xc4ac5665','23');
199
200round4_step(-1,'%eax','%ebx','%ecx','%edx', '7','0xf4292244', '6');
201round4_step( 0,'%edx','%eax','%ebx','%ecx','14','0x432aff97','10');
202round4_step( 0,'%ecx','%edx','%eax','%ebx', '5','0xab9423a7','15');
203round4_step( 0,'%ebx','%ecx','%edx','%eax','12','0xfc93a039','21');
204round4_step( 0,'%eax','%ebx','%ecx','%edx', '3','0x655b59c3', '6');
205round4_step( 0,'%edx','%eax','%ebx','%ecx','10','0x8f0ccc92','10');
206round4_step( 0,'%ecx','%edx','%eax','%ebx', '1','0xffeff47d','15');
207round4_step( 0,'%ebx','%ecx','%edx','%eax', '8','0x85845dd1','21');
208round4_step( 0,'%eax','%ebx','%ecx','%edx','15','0x6fa87e4f', '6');
209round4_step( 0,'%edx','%eax','%ebx','%ecx', '6','0xfe2ce6e0','10');
210round4_step( 0,'%ecx','%edx','%eax','%ebx','13','0xa3014314','15');
211round4_step( 0,'%ebx','%ecx','%edx','%eax', '4','0x4e0811a1','21');
212round4_step( 0,'%eax','%ebx','%ecx','%edx','11','0xf7537e82', '6');
213round4_step( 0,'%edx','%eax','%ebx','%ecx', '2','0xbd3af235','10');
214round4_step( 0,'%ecx','%edx','%eax','%ebx', '9','0x2ad7d2bb','15');
215round4_step( 1,'%ebx','%ecx','%edx','%eax', '0','0xeb86d391','21');
216$code .= <<EOF;
217 # add old values of A, B, C, D
218 add %r8d, %eax
219 add %r9d, %ebx
220 add %r14d, %ecx
221 add %r15d, %edx
222
223 # loop control
224 add \$64, %rsi # ptr += 64
225 cmp %rdi, %rsi # cmp end with ptr
226 jb .Lloop # jmp if ptr < end
227 # END of loop over 16-word blocks
228
229.Lend:
230 mov %eax, 0*4(%rbp) # ctx->A = A
231 mov %ebx, 1*4(%rbp) # ctx->B = B
232 mov %ecx, 2*4(%rbp) # ctx->C = C
233 mov %edx, 3*4(%rbp) # ctx->D = D
234
235 pop %r15
236 pop %r14
237 pop %rbx
238 pop %rbp
239 ret
240.size md5_block_asm_data_order,.-md5_block_asm_data_order
241EOF
242
243print $code;
244
245close STDOUT;
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-04-07 00:24:37 +0000