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 #
6
7 $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
8 push(@INC,"${dir}","${dir}../../perlasm");
9 require "x86asm.pl";
10 require "cbc.pl";
11 require "desboth.pl";
12
13 # base code is in microsft
14 # op dest, source
15 # format.
16 #
17
18 &asm_init($ARGV[0],"des-586.pl");
19
20 $L="edi";
21 $R="esi";
22 $trans="ebp";
23 $small_footprint=1 if (grep(/\-DOPENSSL_SMALL_FOOTPRINT/,@ARGV));
24 # one can discuss setting this variable to 1 unconditionally, as
25 # the folded loop is only 3% slower than unrolled, but >7 times smaller
26
27 &public_label("DES_SPtrans");
28
29 &DES_encrypt_internal();
30 &DES_decrypt_internal();
31 &DES_encrypt("DES_encrypt1",1);
32 &DES_encrypt("DES_encrypt2",0);
33 &DES_encrypt3("DES_encrypt3",1);
34 &DES_encrypt3("DES_decrypt3",0);
35 &cbc("DES_ncbc_encrypt","DES_encrypt1","DES_encrypt1",0,4,5,3,5,-1);
36 &cbc("DES_ede3_cbc_encrypt","DES_encrypt3","DES_decrypt3",0,6,7,3,4,5);
37 &DES_SPtrans();
38
39 &asm_finish();
40
41 sub DES_encrypt_internal()
42 {
43 &function_begin_B("_x86_DES_encrypt");
44
45 if ($small_footprint)
46 {
47 &lea("edx",&DWP(128,"ecx"));
48 &push("edx");
49 &push("ecx");
50 &set_label("eloop");
51 &D_ENCRYPT(0,$L,$R,0,$trans,"eax","ebx","ecx","edx",&swtmp(0));
52 &comment("");
53 &D_ENCRYPT(1,$R,$L,2,$trans,"eax","ebx","ecx","edx",&swtmp(0));
54 &comment("");
55 &add("ecx",16);
56 &cmp("ecx",&swtmp(1));
57 &mov(&swtmp(0),"ecx");
58 &jb(&label("eloop"));
59 &add("esp",8);
60 }
61 else
62 {
63 &push("ecx");
64 for ($i=0; $i<16; $i+=2)
65 {
66 &comment("Round $i");
67 &D_ENCRYPT($i,$L,$R,$i*2,$trans,"eax","ebx","ecx","edx",&swtmp(0));
68 &comment("Round ".sprintf("%d",$i+1));
69 &D_ENCRYPT($i+1,$R,$L,($i+1)*2,$trans,"eax","ebx","ecx","edx",&swtmp(0));
70 }
71 &add("esp",4);
72 }
73 &ret();
74
75 &function_end_B("_x86_DES_encrypt");
76 }
77
78 sub DES_decrypt_internal()
79 {
80 &function_begin_B("_x86_DES_decrypt");
81
82 if ($small_footprint)
83 {
84 &push("ecx");
85 &lea("ecx",&DWP(128,"ecx"));
86 &push("ecx");
87 &set_label("dloop");
88 &D_ENCRYPT(0,$L,$R,-2,$trans,"eax","ebx","ecx","edx",&swtmp(0));
89 &comment("");
90 &D_ENCRYPT(1,$R,$L,-4,$trans,"eax","ebx","ecx","edx",&swtmp(0));
91 &comment("");
92 &sub("ecx",16);
93 &cmp("ecx",&swtmp(1));
94 &mov(&swtmp(0),"ecx");
95 &ja(&label("dloop"));
96 &add("esp",8);
97 }
98 else
99 {
100 &push("ecx");
101 for ($i=15; $i>0; $i-=2)
102 {
103 &comment("Round $i");
104 &D_ENCRYPT(15-$i,$L,$R,$i*2,$trans,"eax","ebx","ecx","edx",&swtmp(0));
105 &comment("Round ".sprintf("%d",$i-1));
106 &D_ENCRYPT(15-$i+1,$R,$L,($i-1)*2,$trans,"eax","ebx","ecx","edx",&swtmp(0));
107 }
108 &add("esp",4);
109 }
110 &ret();
111
112 &function_end_B("_x86_DES_decrypt");
113 }
114
115 sub DES_encrypt
116 {
117 local($name,$do_ip)=@_;
118
119 &function_begin_B($name);
120
121 &push("esi");
122 &push("edi");
123
124 &comment("");
125 &comment("Load the 2 words");
126
127 if ($do_ip)
128 {
129 &mov($R,&wparam(0));
130 &xor( "ecx", "ecx" );
131
132 &push("ebx");
133 &push("ebp");
134
135 &mov("eax",&DWP(0,$R,"",0));
136 &mov("ebx",&wparam(2)); # get encrypt flag
137 &mov($L,&DWP(4,$R,"",0));
138 &comment("");
139 &comment("IP");
140 &IP_new("eax",$L,$R,3);
141 }
142 else
143 {
144 &mov("eax",&wparam(0));
145 &xor( "ecx", "ecx" );
146
147 &push("ebx");
148 &push("ebp");
149
150 &mov($R,&DWP(0,"eax","",0));
151 &mov("ebx",&wparam(2)); # get encrypt flag
152 &rotl($R,3);
153 &mov($L,&DWP(4,"eax","",0));
154 &rotl($L,3);
155 }
156
157 # PIC-ification:-)
158 &call (&label("pic_point"));
159 &set_label("pic_point");
160 &blindpop($trans);
161 &lea ($trans,&DWP(&label("DES_SPtrans")."-".&label("pic_point"),$trans));
162
163 &mov( "ecx", &wparam(1) );
164
165 &cmp("ebx","0");
166 &je(&label("decrypt"));
167 &call("_x86_DES_encrypt");
168 &jmp(&label("done"));
169 &set_label("decrypt");
170 &call("_x86_DES_decrypt");
171 &set_label("done");
172
173 if ($do_ip)
174 {
175 &comment("");
176 &comment("FP");
177 &mov("edx",&wparam(0));
178 &FP_new($L,$R,"eax",3);
179
180 &mov(&DWP(0,"edx","",0),"eax");
181 &mov(&DWP(4,"edx","",0),$R);
182 }
183 else
184 {
185 &comment("");
186 &comment("Fixup");
187 &rotr($L,3); # r
188 &mov("eax",&wparam(0));
189 &rotr($R,3); # l
190 &mov(&DWP(0,"eax","",0),$L);
191 &mov(&DWP(4,"eax","",0),$R);
192 }
193
194 &pop("ebp");
195 &pop("ebx");
196 &pop("edi");
197 &pop("esi");
198 &ret();
199
200 &function_end_B($name);
201 }
202
203 sub D_ENCRYPT
204 {
205 local($r,$L,$R,$S,$trans,$u,$tmp1,$tmp2,$t,$wp1)=@_;
206
207 &mov( $u, &DWP(&n2a($S*4),$tmp2,"",0));
208 &xor( $tmp1, $tmp1);
209 &mov( $t, &DWP(&n2a(($S+1)*4),$tmp2,"",0));
210 &xor( $u, $R);
211 &xor( $tmp2, $tmp2);
212 &xor( $t, $R);
213 &and( $u, "0xfcfcfcfc" );
214 &and( $t, "0xcfcfcfcf" );
215 &movb( &LB($tmp1), &LB($u) );
216 &movb( &LB($tmp2), &HB($u) );
217 &rotr( $t, 4 );
218 &xor( $L, &DWP(" ",$trans,$tmp1,0));
219 &movb( &LB($tmp1), &LB($t) );
220 &xor( $L, &DWP("0x200",$trans,$tmp2,0));
221 &movb( &LB($tmp2), &HB($t) );
222 &shr( $u, 16);
223 &xor( $L, &DWP("0x100",$trans,$tmp1,0));
224 &movb( &LB($tmp1), &HB($u) );
225 &shr( $t, 16);
226 &xor( $L, &DWP("0x300",$trans,$tmp2,0));
227 &movb( &LB($tmp2), &HB($t) );
228 &and( $u, "0xff" );
229 &and( $t, "0xff" );
230 &xor( $L, &DWP("0x600",$trans,$tmp1,0));
231 &xor( $L, &DWP("0x700",$trans,$tmp2,0));
232 &mov( $tmp2, $wp1 );
233 &xor( $L, &DWP("0x400",$trans,$u,0));
234 &xor( $L, &DWP("0x500",$trans,$t,0));
235 }
236
237 sub n2a
238 {
239 sprintf("%d",$_[0]);
240 }
241
242 # now has a side affect of rotating $a by $shift
243 sub R_PERM_OP
244 {
245 local($a,$b,$tt,$shift,$mask,$last)=@_;
246
247 &rotl( $a, $shift ) if ($shift != 0);
248 &mov( $tt, $a );
249 &xor( $a, $b );
250 &and( $a, $mask );
251 # This can never succeed, and besides it is difficult to see what the
252 # idea was - Ben 13 Feb 99
253 if (!$last eq $b)
254 {
255 &xor( $b, $a );
256 &xor( $tt, $a );
257 }
258 else
259 {
260 &xor( $tt, $a );
261 &xor( $b, $a );
262 }
263 &comment("");
264 }
265
266 sub IP_new
267 {
268 local($l,$r,$tt,$lr)=@_;
269
270 &R_PERM_OP($l,$r,$tt, 4,"0xf0f0f0f0",$l);
271 &R_PERM_OP($r,$tt,$l,20,"0xfff0000f",$l);
272 &R_PERM_OP($l,$tt,$r,14,"0x33333333",$r);
273 &R_PERM_OP($tt,$r,$l,22,"0x03fc03fc",$r);
274 &R_PERM_OP($l,$r,$tt, 9,"0xaaaaaaaa",$r);
275
276 if ($lr != 3)
277 {
278 if (($lr-3) < 0)
279 { &rotr($tt, 3-$lr); }
280 else { &rotl($tt, $lr-3); }
281 }
282 if ($lr != 2)
283 {
284 if (($lr-2) < 0)
285 { &rotr($r, 2-$lr); }
286 else { &rotl($r, $lr-2); }
287 }
288 }
289
290 sub FP_new
291 {
292 local($l,$r,$tt,$lr)=@_;
293
294 if ($lr != 2)
295 {
296 if (($lr-2) < 0)
297 { &rotl($r, 2-$lr); }
298 else { &rotr($r, $lr-2); }
299 }
300 if ($lr != 3)
301 {
302 if (($lr-3) < 0)
303 { &rotl($l, 3-$lr); }
304 else { &rotr($l, $lr-3); }
305 }
306
307 &R_PERM_OP($l,$r,$tt, 0,"0xaaaaaaaa",$r);
308 &R_PERM_OP($tt,$r,$l,23,"0x03fc03fc",$r);
309 &R_PERM_OP($l,$r,$tt,10,"0x33333333",$l);
310 &R_PERM_OP($r,$tt,$l,18,"0xfff0000f",$l);
311 &R_PERM_OP($l,$tt,$r,12,"0xf0f0f0f0",$r);
312 &rotr($tt , 4);
313 }
314
315 sub DES_SPtrans
316 {
317 &set_label("DES_SPtrans",64);
318 &data_word(0x02080800, 0x00080000, 0x02000002, 0x02080802);
319 &data_word(0x02000000, 0x00080802, 0x00080002, 0x02000002);
320 &data_word(0x00080802, 0x02080800, 0x02080000, 0x00000802);
321 &data_word(0x02000802, 0x02000000, 0x00000000, 0x00080002);
322 &data_word(0x00080000, 0x00000002, 0x02000800, 0x00080800);
323 &data_word(0x02080802, 0x02080000, 0x00000802, 0x02000800);
324 &data_word(0x00000002, 0x00000800, 0x00080800, 0x02080002);
325 &data_word(0x00000800, 0x02000802, 0x02080002, 0x00000000);
326 &data_word(0x00000000, 0x02080802, 0x02000800, 0x00080002);
327 &data_word(0x02080800, 0x00080000, 0x00000802, 0x02000800);
328 &data_word(0x02080002, 0x00000800, 0x00080800, 0x02000002);
329 &data_word(0x00080802, 0x00000002, 0x02000002, 0x02080000);
330 &data_word(0x02080802, 0x00080800, 0x02080000, 0x02000802);
331 &data_word(0x02000000, 0x00000802, 0x00080002, 0x00000000);
332 &data_word(0x00080000, 0x02000000, 0x02000802, 0x02080800);
333 &data_word(0x00000002, 0x02080002, 0x00000800, 0x00080802);
334 # nibble 1
335 &data_word(0x40108010, 0x00000000, 0x00108000, 0x40100000);
336 &data_word(0x40000010, 0x00008010, 0x40008000, 0x00108000);
337 &data_word(0x00008000, 0x40100010, 0x00000010, 0x40008000);
338 &data_word(0x00100010, 0x40108000, 0x40100000, 0x00000010);
339 &data_word(0x00100000, 0x40008010, 0x40100010, 0x00008000);
340 &data_word(0x00108010, 0x40000000, 0x00000000, 0x00100010);
341 &data_word(0x40008010, 0x00108010, 0x40108000, 0x40000010);
342 &data_word(0x40000000, 0x00100000, 0x00008010, 0x40108010);
343 &data_word(0x00100010, 0x40108000, 0x40008000, 0x00108010);
344 &data_word(0x40108010, 0x00100010, 0x40000010, 0x00000000);
345 &data_word(0x40000000, 0x00008010, 0x00100000, 0x40100010);
346 &data_word(0x00008000, 0x40000000, 0x00108010, 0x40008010);
347 &data_word(0x40108000, 0x00008000, 0x00000000, 0x40000010);
348 &data_word(0x00000010, 0x40108010, 0x00108000, 0x40100000);
349 &data_word(0x40100010, 0x00100000, 0x00008010, 0x40008000);
350 &data_word(0x40008010, 0x00000010, 0x40100000, 0x00108000);
351 # nibble 2
352 &data_word(0x04000001, 0x04040100, 0x00000100, 0x04000101);
353 &data_word(0x00040001, 0x04000000, 0x04000101, 0x00040100);
354 &data_word(0x04000100, 0x00040000, 0x04040000, 0x00000001);
355 &data_word(0x04040101, 0x00000101, 0x00000001, 0x04040001);
356 &data_word(0x00000000, 0x00040001, 0x04040100, 0x00000100);
357 &data_word(0x00000101, 0x04040101, 0x00040000, 0x04000001);
358 &data_word(0x04040001, 0x04000100, 0x00040101, 0x04040000);
359 &data_word(0x00040100, 0x00000000, 0x04000000, 0x00040101);
360 &data_word(0x04040100, 0x00000100, 0x00000001, 0x00040000);
361 &data_word(0x00000101, 0x00040001, 0x04040000, 0x04000101);
362 &data_word(0x00000000, 0x04040100, 0x00040100, 0x04040001);
363 &data_word(0x00040001, 0x04000000, 0x04040101, 0x00000001);
364 &data_word(0x00040101, 0x04000001, 0x04000000, 0x04040101);
365 &data_word(0x00040000, 0x04000100, 0x04000101, 0x00040100);
366 &data_word(0x04000100, 0x00000000, 0x04040001, 0x00000101);
367 &data_word(0x04000001, 0x00040101, 0x00000100, 0x04040000);
368 # nibble 3
369 &data_word(0x00401008, 0x10001000, 0x00000008, 0x10401008);
370 &data_word(0x00000000, 0x10400000, 0x10001008, 0x00400008);
371 &data_word(0x10401000, 0x10000008, 0x10000000, 0x00001008);
372 &data_word(0x10000008, 0x00401008, 0x00400000, 0x10000000);
373 &data_word(0x10400008, 0x00401000, 0x00001000, 0x00000008);
374 &data_word(0x00401000, 0x10001008, 0x10400000, 0x00001000);
375 &data_word(0x00001008, 0x00000000, 0x00400008, 0x10401000);
376 &data_word(0x10001000, 0x10400008, 0x10401008, 0x00400000);
377 &data_word(0x10400008, 0x00001008, 0x00400000, 0x10000008);
378 &data_word(0x00401000, 0x10001000, 0x00000008, 0x10400000);
379 &data_word(0x10001008, 0x00000000, 0x00001000, 0x00400008);
380 &data_word(0x00000000, 0x10400008, 0x10401000, 0x00001000);
381 &data_word(0x10000000, 0x10401008, 0x00401008, 0x00400000);
382 &data_word(0x10401008, 0x00000008, 0x10001000, 0x00401008);
383 &data_word(0x00400008, 0x00401000, 0x10400000, 0x10001008);
384 &data_word(0x00001008, 0x10000000, 0x10000008, 0x10401000);
385 # nibble 4
386 &data_word(0x08000000, 0x00010000, 0x00000400, 0x08010420);
387 &data_word(0x08010020, 0x08000400, 0x00010420, 0x08010000);
388 &data_word(0x00010000, 0x00000020, 0x08000020, 0x00010400);
389 &data_word(0x08000420, 0x08010020, 0x08010400, 0x00000000);
390 &data_word(0x00010400, 0x08000000, 0x00010020, 0x00000420);
391 &data_word(0x08000400, 0x00010420, 0x00000000, 0x08000020);
392 &data_word(0x00000020, 0x08000420, 0x08010420, 0x00010020);
393 &data_word(0x08010000, 0x00000400, 0x00000420, 0x08010400);
394 &data_word(0x08010400, 0x08000420, 0x00010020, 0x08010000);
395 &data_word(0x00010000, 0x00000020, 0x08000020, 0x08000400);
396 &data_word(0x08000000, 0x00010400, 0x08010420, 0x00000000);
397 &data_word(0x00010420, 0x08000000, 0x00000400, 0x00010020);
398 &data_word(0x08000420, 0x00000400, 0x00000000, 0x08010420);
399 &data_word(0x08010020, 0x08010400, 0x00000420, 0x00010000);
400 &data_word(0x00010400, 0x08010020, 0x08000400, 0x00000420);
401 &data_word(0x00000020, 0x00010420, 0x08010000, 0x08000020);
402 # nibble 5
403 &data_word(0x80000040, 0x00200040, 0x00000000, 0x80202000);
404 &data_word(0x00200040, 0x00002000, 0x80002040, 0x00200000);
405 &data_word(0x00002040, 0x80202040, 0x00202000, 0x80000000);
406 &data_word(0x80002000, 0x80000040, 0x80200000, 0x00202040);
407 &data_word(0x00200000, 0x80002040, 0x80200040, 0x00000000);
408 &data_word(0x00002000, 0x00000040, 0x80202000, 0x80200040);
409 &data_word(0x80202040, 0x80200000, 0x80000000, 0x00002040);
410 &data_word(0x00000040, 0x00202000, 0x00202040, 0x80002000);
411 &data_word(0x00002040, 0x80000000, 0x80002000, 0x00202040);
412 &data_word(0x80202000, 0x00200040, 0x00000000, 0x80002000);
413 &data_word(0x80000000, 0x00002000, 0x80200040, 0x00200000);
414 &data_word(0x00200040, 0x80202040, 0x00202000, 0x00000040);
415 &data_word(0x80202040, 0x00202000, 0x00200000, 0x80002040);
416 &data_word(0x80000040, 0x80200000, 0x00202040, 0x00000000);
417 &data_word(0x00002000, 0x80000040, 0x80002040, 0x80202000);
418 &data_word(0x80200000, 0x00002040, 0x00000040, 0x80200040);
419 # nibble 6
420 &data_word(0x00004000, 0x00000200, 0x01000200, 0x01000004);
421 &data_word(0x01004204, 0x00004004, 0x00004200, 0x00000000);
422 &data_word(0x01000000, 0x01000204, 0x00000204, 0x01004000);
423 &data_word(0x00000004, 0x01004200, 0x01004000, 0x00000204);
424 &data_word(0x01000204, 0x00004000, 0x00004004, 0x01004204);
425 &data_word(0x00000000, 0x01000200, 0x01000004, 0x00004200);
426 &data_word(0x01004004, 0x00004204, 0x01004200, 0x00000004);
427 &data_word(0x00004204, 0x01004004, 0x00000200, 0x01000000);
428 &data_word(0x00004204, 0x01004000, 0x01004004, 0x00000204);
429 &data_word(0x00004000, 0x00000200, 0x01000000, 0x01004004);
430 &data_word(0x01000204, 0x00004204, 0x00004200, 0x00000000);
431 &data_word(0x00000200, 0x01000004, 0x00000004, 0x01000200);
432 &data_word(0x00000000, 0x01000204, 0x01000200, 0x00004200);
433 &data_word(0x00000204, 0x00004000, 0x01004204, 0x01000000);
434 &data_word(0x01004200, 0x00000004, 0x00004004, 0x01004204);
435 &data_word(0x01000004, 0x01004200, 0x01004000, 0x00004004);
436 # nibble 7
437 &data_word(0x20800080, 0x20820000, 0x00020080, 0x00000000);
438 &data_word(0x20020000, 0x00800080, 0x20800000, 0x20820080);
439 &data_word(0x00000080, 0x20000000, 0x00820000, 0x00020080);
440 &data_word(0x00820080, 0x20020080, 0x20000080, 0x20800000);
441 &data_word(0x00020000, 0x00820080, 0x00800080, 0x20020000);
442 &data_word(0x20820080, 0x20000080, 0x00000000, 0x00820000);
443 &data_word(0x20000000, 0x00800000, 0x20020080, 0x20800080);
444 &data_word(0x00800000, 0x00020000, 0x20820000, 0x00000080);
445 &data_word(0x00800000, 0x00020000, 0x20000080, 0x20820080);
446 &data_word(0x00020080, 0x20000000, 0x00000000, 0x00820000);
447 &data_word(0x20800080, 0x20020080, 0x20020000, 0x00800080);
448 &data_word(0x20820000, 0x00000080, 0x00800080, 0x20020000);
449 &data_word(0x20820080, 0x00800000, 0x20800000, 0x20000080);
450 &data_word(0x00820000, 0x00020080, 0x20020080, 0x20800000);
451 &data_word(0x00000080, 0x20820000, 0x00820080, 0x00000000);
452 &data_word(0x20000000, 0x20800080, 0x00020000, 0x00820080);
453 }