Complete integration of SPARC assembler DES implementation. Tested on Solaris
[oweals/openssl.git] / crypto / des / asm / crypt586.pl
1 #!/usr/local/bin/perl
2 #
3 # The inner loop instruction sequence and the IP/FP modifications are from
4 # Svend Olaf Mikkelsen <svolaf@inet.uni-c.dk>
5 # I've added the stuff needed for crypt() but I've not worried about making
6 # things perfect.
7 #
8
9 push(@INC,"perlasm","../../perlasm");
10 require "x86asm.pl";
11
12 &asm_init($ARGV[0],"crypt586.pl");
13
14 $L="edi";
15 $R="esi";
16
17 &external_label("DES_SPtrans");
18 &fcrypt_body("fcrypt_body");
19 &asm_finish();
20
21 sub fcrypt_body
22         {
23         local($name,$do_ip)=@_;
24
25         &function_begin($name,"EXTRN   _DES_SPtrans:DWORD");
26
27         &comment("");
28         &comment("Load the 2 words");
29         $trans="ebp";
30
31         &xor(   $L,     $L);
32         &xor(   $R,     $R);
33
34         # PIC-ification:-)
35         if ($cpp)       { &picmeup("edx","DES_SPtrans");   }
36         else            { &lea("edx",&DWP("DES_SPtrans")); }
37         &push("edx");   # becomes &swtmp(1)
38         #
39         &mov($trans,&wparam(1)); # reloaded with DES_SPtrans in D_ENCRYPT
40
41         &push(&DWC(25)); # add a variable
42
43         &set_label("start");
44         for ($i=0; $i<16; $i+=2)
45                 {
46                 &comment("");
47                 &comment("Round $i");
48                 &D_ENCRYPT($i,$L,$R,$i*2,$trans,"eax","ebx","ecx","edx");
49
50                 &comment("");
51                 &comment("Round ".sprintf("%d",$i+1));
52                 &D_ENCRYPT($i+1,$R,$L,($i+1)*2,$trans,"eax","ebx","ecx","edx");
53                 }
54          &mov("ebx",    &swtmp(0));
55         &mov("eax",     $L);
56          &dec("ebx");
57         &mov($L,        $R);
58          &mov($R,       "eax");
59         &mov(&swtmp(0), "ebx");
60          &jnz(&label("start"));
61
62         &comment("");
63         &comment("FP");
64         &mov("edx",&wparam(0));
65
66         &FP_new($R,$L,"eax",3);
67         &mov(&DWP(0,"edx","",0),"eax");
68         &mov(&DWP(4,"edx","",0),$L);
69
70         &add("esp",8);  # remove variables
71
72         &function_end($name);
73         }
74
75 sub D_ENCRYPT
76         {
77         local($r,$L,$R,$S,$trans,$u,$tmp1,$tmp2,$t)=@_;
78
79         &mov(   $u,             &wparam(2));                    # 2
80         &mov(   $t,             $R);
81         &shr(   $t,             16);                            # 1
82         &mov(   $tmp2,          &wparam(3));                    # 2
83         &xor(   $t,             $R);                            # 1
84
85         &and(   $u,             $t);                            # 2
86         &and(   $t,             $tmp2);                         # 2
87
88         &mov(   $tmp1,          $u);
89         &shl(   $tmp1,          16);                            # 1
90         &mov(   $tmp2,          $t);
91         &shl(   $tmp2,          16);                            # 1
92         &xor(   $u,             $tmp1);                         # 2
93         &xor(   $t,             $tmp2);                         # 2
94         &mov(   $tmp1,          &DWP(&n2a($S*4),$trans,"",0));  # 2
95         &xor(   $u,             $tmp1);
96         &mov(   $tmp2,          &DWP(&n2a(($S+1)*4),$trans,"",0));      # 2
97         &xor(   $u,             $R);
98         &xor(   $t,             $R);
99         &xor(   $t,             $tmp2);
100
101         &and(   $u,             "0xfcfcfcfc"    );              # 2
102         &xor(   $tmp1,          $tmp1);                         # 1
103         &and(   $t,             "0xcfcfcfcf"    );              # 2
104         &xor(   $tmp2,          $tmp2); 
105         &movb(  &LB($tmp1),     &LB($u) );
106         &movb(  &LB($tmp2),     &HB($u) );
107         &rotr(  $t,             4               );
108         &mov(   $trans,         &swtmp(1));
109         &xor(   $L,             &DWP("     ",$trans,$tmp1,0));
110         &movb(  &LB($tmp1),     &LB($t) );
111         &xor(   $L,             &DWP("0x200",$trans,$tmp2,0));
112         &movb(  &LB($tmp2),     &HB($t) );
113         &shr(   $u,             16);
114         &xor(   $L,             &DWP("0x100",$trans,$tmp1,0));
115         &movb(  &LB($tmp1),     &HB($u) );
116         &shr(   $t,             16);
117         &xor(   $L,             &DWP("0x300",$trans,$tmp2,0));
118         &movb(  &LB($tmp2),     &HB($t) );
119         &and(   $u,             "0xff"  );
120         &and(   $t,             "0xff"  );
121         &mov(   $tmp1,          &DWP("0x600",$trans,$tmp1,0));
122         &xor(   $L,             $tmp1);
123         &mov(   $tmp1,          &DWP("0x700",$trans,$tmp2,0));
124         &xor(   $L,             $tmp1);
125         &mov(   $tmp1,          &DWP("0x400",$trans,$u,0));
126         &xor(   $L,             $tmp1);
127         &mov(   $tmp1,          &DWP("0x500",$trans,$t,0));
128         &xor(   $L,             $tmp1);
129         &mov(   $trans,         &wparam(1));
130         }
131
132 sub n2a
133         {
134         sprintf("%d",$_[0]);
135         }
136
137 # now has a side affect of rotating $a by $shift
138 sub R_PERM_OP
139         {
140         local($a,$b,$tt,$shift,$mask,$last)=@_;
141
142         &rotl(  $a,             $shift          ) if ($shift != 0);
143         &mov(   $tt,            $a              );
144         &xor(   $a,             $b              );
145         &and(   $a,             $mask           );
146         if ($notlast eq $b)
147                 {
148                 &xor(   $b,             $a              );
149                 &xor(   $tt,            $a              );
150                 }
151         else
152                 {
153                 &xor(   $tt,            $a              );
154                 &xor(   $b,             $a              );
155                 }
156         &comment("");
157         }
158
159 sub IP_new
160         {
161         local($l,$r,$tt,$lr)=@_;
162
163         &R_PERM_OP($l,$r,$tt, 4,"0xf0f0f0f0",$l);
164         &R_PERM_OP($r,$tt,$l,20,"0xfff0000f",$l);
165         &R_PERM_OP($l,$tt,$r,14,"0x33333333",$r);
166         &R_PERM_OP($tt,$r,$l,22,"0x03fc03fc",$r);
167         &R_PERM_OP($l,$r,$tt, 9,"0xaaaaaaaa",$r);
168         
169         if ($lr != 3)
170                 {
171                 if (($lr-3) < 0)
172                         { &rotr($tt,    3-$lr); }
173                 else    { &rotl($tt,    $lr-3); }
174                 }
175         if ($lr != 2)
176                 {
177                 if (($lr-2) < 0)
178                         { &rotr($r,     2-$lr); }
179                 else    { &rotl($r,     $lr-2); }
180                 }
181         }
182
183 sub FP_new
184         {
185         local($l,$r,$tt,$lr)=@_;
186
187         if ($lr != 2)
188                 {
189                 if (($lr-2) < 0)
190                         { &rotl($r,     2-$lr); }
191                 else    { &rotr($r,     $lr-2); }
192                 }
193         if ($lr != 3)
194                 {
195                 if (($lr-3) < 0)
196                         { &rotl($l,     3-$lr); }
197                 else    { &rotr($l,     $lr-3); }
198                 }
199
200         &R_PERM_OP($l,$r,$tt, 0,"0xaaaaaaaa",$r);
201         &R_PERM_OP($tt,$r,$l,23,"0x03fc03fc",$r);
202         &R_PERM_OP($l,$r,$tt,10,"0x33333333",$l);
203         &R_PERM_OP($r,$tt,$l,18,"0xfff0000f",$l);
204         &R_PERM_OP($l,$tt,$r,12,"0xf0f0f0f0",$r);
205         &rotr($tt       , 4);
206         }
207