summaryrefslogtreecommitdiff
path: root/src/lib/libcrypto/bn
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--src/lib/libcrypto/bn/asm/bn-586.pl384
-rw-r--r--src/lib/libcrypto/bn/asm/co-586.pl286
-rw-r--r--src/lib/libcrypto/bn/asm/pa-risc2.s1618
-rw-r--r--src/lib/libcrypto/bn/asm/pa-risc2W.s1605
-rw-r--r--src/lib/libcrypto/bn/asm/sparcv8.S1458
-rw-r--r--src/lib/libcrypto/bn/asm/sparcv8plus.S1535
-rw-r--r--src/lib/libcrypto/bn/asm/x86.pl28
-rw-r--r--src/lib/libcrypto/bn/asm/x86/add.pl76
-rw-r--r--src/lib/libcrypto/bn/asm/x86/comba.pl277
-rw-r--r--src/lib/libcrypto/bn/asm/x86/div.pl15
-rw-r--r--src/lib/libcrypto/bn/asm/x86/mul.pl77
-rw-r--r--src/lib/libcrypto/bn/asm/x86/mul_add.pl87
-rw-r--r--src/lib/libcrypto/bn/asm/x86/sqr.pl60
-rw-r--r--src/lib/libcrypto/bn/asm/x86/sub.pl76
-rw-r--r--src/lib/libcrypto/bn/bn.h520
-rw-r--r--src/lib/libcrypto/bn/bn_add.c307
-rw-r--r--src/lib/libcrypto/bn/bn_asm.c832
-rw-r--r--src/lib/libcrypto/bn/bn_blind.c144
-rw-r--r--src/lib/libcrypto/bn/bn_ctx.c144
-rw-r--r--src/lib/libcrypto/bn/bn_div.c381
-rw-r--r--src/lib/libcrypto/bn/bn_err.c124
-rw-r--r--src/lib/libcrypto/bn/bn_exp.c696
-rw-r--r--src/lib/libcrypto/bn/bn_exp2.c300
-rw-r--r--src/lib/libcrypto/bn/bn_gcd.c210
-rw-r--r--src/lib/libcrypto/bn/bn_lcl.h419
-rw-r--r--src/lib/libcrypto/bn/bn_lib.c762
-rw-r--r--src/lib/libcrypto/bn/bn_mont.c346
-rw-r--r--src/lib/libcrypto/bn/bn_mpi.c129
-rw-r--r--src/lib/libcrypto/bn/bn_mul.c794
-rw-r--r--src/lib/libcrypto/bn/bn_prime.c465
-rw-r--r--src/lib/libcrypto/bn/bn_prime.h325
-rw-r--r--src/lib/libcrypto/bn/bn_prime.pl117
-rw-r--r--src/lib/libcrypto/bn/bn_print.c332
-rw-r--r--src/lib/libcrypto/bn/bn_rand.c223
-rw-r--r--src/lib/libcrypto/bn/bn_recp.c220
-rw-r--r--src/lib/libcrypto/bn/bn_shift.c205
-rw-r--r--src/lib/libcrypto/bn/bn_sqr.c288
-rw-r--r--src/lib/libcrypto/bn/bn_word.c199
38 files changed, 0 insertions, 16064 deletions
diff --git a/src/lib/libcrypto/bn/asm/bn-586.pl b/src/lib/libcrypto/bn/asm/bn-586.pl
deleted file mode 100644
index 5191bed273..0000000000
--- a/src/lib/libcrypto/bn/asm/bn-586.pl
+++ /dev/null
@@ -1,384 +0,0 @@
1#!/usr/local/bin/perl
2
3push(@INC,"perlasm","../../perlasm");
4require "x86asm.pl";
5
6&asm_init($ARGV[0],$0);
7
8&bn_mul_add_words("bn_mul_add_words");
9&bn_mul_words("bn_mul_words");
10&bn_sqr_words("bn_sqr_words");
11&bn_div_words("bn_div_words");
12&bn_add_words("bn_add_words");
13&bn_sub_words("bn_sub_words");
14
15&asm_finish();
16
17sub bn_mul_add_words
18 {
19 local($name)=@_;
20
21 &function_begin($name,"");
22
23 &comment("");
24 $Low="eax";
25 $High="edx";
26 $a="ebx";
27 $w="ebp";
28 $r="edi";
29 $c="esi";
30
31 &xor($c,$c); # clear carry
32 &mov($r,&wparam(0)); #
33
34 &mov("ecx",&wparam(2)); #
35 &mov($a,&wparam(1)); #
36
37 &and("ecx",0xfffffff8); # num / 8
38 &mov($w,&wparam(3)); #
39
40 &push("ecx"); # Up the stack for a tmp variable
41
42 &jz(&label("maw_finish"));
43
44 &set_label("maw_loop",0);
45
46 &mov(&swtmp(0),"ecx"); #
47
48 for ($i=0; $i<32; $i+=4)
49 {
50 &comment("Round $i");
51
52 &mov("eax",&DWP($i,$a,"",0)); # *a
53 &mul($w); # *a * w
54 &add("eax",$c); # L(t)+= *r
55 &mov($c,&DWP($i,$r,"",0)); # L(t)+= *r
56 &adc("edx",0); # H(t)+=carry
57 &add("eax",$c); # L(t)+=c
58 &adc("edx",0); # H(t)+=carry
59 &mov(&DWP($i,$r,"",0),"eax"); # *r= L(t);
60 &mov($c,"edx"); # c= H(t);
61 }
62
63 &comment("");
64 &mov("ecx",&swtmp(0)); #
65 &add($a,32);
66 &add($r,32);
67 &sub("ecx",8);
68 &jnz(&label("maw_loop"));
69
70 &set_label("maw_finish",0);
71 &mov("ecx",&wparam(2)); # get num
72 &and("ecx",7);
73 &jnz(&label("maw_finish2")); # helps branch prediction
74 &jmp(&label("maw_end"));
75
76 &set_label("maw_finish2",1);
77 for ($i=0; $i<7; $i++)
78 {
79 &comment("Tail Round $i");
80 &mov("eax",&DWP($i*4,$a,"",0));# *a
81 &mul($w); # *a * w
82 &add("eax",$c); # L(t)+=c
83 &mov($c,&DWP($i*4,$r,"",0)); # L(t)+= *r
84 &adc("edx",0); # H(t)+=carry
85 &add("eax",$c);
86 &adc("edx",0); # H(t)+=carry
87 &dec("ecx") if ($i != 7-1);
88 &mov(&DWP($i*4,$r,"",0),"eax"); # *r= L(t);
89 &mov($c,"edx"); # c= H(t);
90 &jz(&label("maw_end")) if ($i != 7-1);
91 }
92 &set_label("maw_end",0);
93 &mov("eax",$c);
94
95 &pop("ecx"); # clear variable from
96
97 &function_end($name);
98 }
99
100sub bn_mul_words
101 {
102 local($name)=@_;
103
104 &function_begin($name,"");
105
106 &comment("");
107 $Low="eax";
108 $High="edx";
109 $a="ebx";
110 $w="ecx";
111 $r="edi";
112 $c="esi";
113 $num="ebp";
114
115 &xor($c,$c); # clear carry
116 &mov($r,&wparam(0)); #
117 &mov($a,&wparam(1)); #
118 &mov($num,&wparam(2)); #
119 &mov($w,&wparam(3)); #
120
121 &and($num,0xfffffff8); # num / 8
122 &jz(&label("mw_finish"));
123
124 &set_label("mw_loop",0);
125 for ($i=0; $i<32; $i+=4)
126 {
127 &comment("Round $i");
128
129 &mov("eax",&DWP($i,$a,"",0)); # *a
130 &mul($w); # *a * w
131 &add("eax",$c); # L(t)+=c
132 # XXX
133
134 &adc("edx",0); # H(t)+=carry
135 &mov(&DWP($i,$r,"",0),"eax"); # *r= L(t);
136
137 &mov($c,"edx"); # c= H(t);
138 }
139
140 &comment("");
141 &add($a,32);
142 &add($r,32);
143 &sub($num,8);
144 &jz(&label("mw_finish"));
145 &jmp(&label("mw_loop"));
146
147 &set_label("mw_finish",0);
148 &mov($num,&wparam(2)); # get num
149 &and($num,7);
150 &jnz(&label("mw_finish2"));
151 &jmp(&label("mw_end"));
152
153 &set_label("mw_finish2",1);
154 for ($i=0; $i<7; $i++)
155 {
156 &comment("Tail Round $i");
157 &mov("eax",&DWP($i*4,$a,"",0));# *a
158 &mul($w); # *a * w
159 &add("eax",$c); # L(t)+=c
160 # XXX
161 &adc("edx",0); # H(t)+=carry
162 &mov(&DWP($i*4,$r,"",0),"eax");# *r= L(t);
163 &mov($c,"edx"); # c= H(t);
164 &dec($num) if ($i != 7-1);
165 &jz(&label("mw_end")) if ($i != 7-1);
166 }
167 &set_label("mw_end",0);
168 &mov("eax",$c);
169
170 &function_end($name);
171 }
172
173sub bn_sqr_words
174 {
175 local($name)=@_;
176
177 &function_begin($name,"");
178
179 &comment("");
180 $r="esi";
181 $a="edi";
182 $num="ebx";
183
184 &mov($r,&wparam(0)); #
185 &mov($a,&wparam(1)); #
186 &mov($num,&wparam(2)); #
187
188 &and($num,0xfffffff8); # num / 8
189 &jz(&label("sw_finish"));
190
191 &set_label("sw_loop",0);
192 for ($i=0; $i<32; $i+=4)
193 {
194 &comment("Round $i");
195 &mov("eax",&DWP($i,$a,"",0)); # *a
196 # XXX
197 &mul("eax"); # *a * *a
198 &mov(&DWP($i*2,$r,"",0),"eax"); #
199 &mov(&DWP($i*2+4,$r,"",0),"edx");#
200 }
201
202 &comment("");
203 &add($a,32);
204 &add($r,64);
205 &sub($num,8);
206 &jnz(&label("sw_loop"));
207
208 &set_label("sw_finish",0);
209 &mov($num,&wparam(2)); # get num
210 &and($num,7);
211 &jz(&label("sw_end"));
212
213 for ($i=0; $i<7; $i++)
214 {
215 &comment("Tail Round $i");
216 &mov("eax",&DWP($i*4,$a,"",0)); # *a
217 # XXX
218 &mul("eax"); # *a * *a
219 &mov(&DWP($i*8,$r,"",0),"eax"); #
220 &dec($num) if ($i != 7-1);
221 &mov(&DWP($i*8+4,$r,"",0),"edx");
222 &jz(&label("sw_end")) if ($i != 7-1);
223 }
224 &set_label("sw_end",0);
225
226 &function_end($name);
227 }
228
229sub bn_div_words
230 {
231 local($name)=@_;
232
233 &function_begin($name,"");
234 &mov("edx",&wparam(0)); #
235 &mov("eax",&wparam(1)); #
236 &mov("ebx",&wparam(2)); #
237 &div("ebx");
238 &function_end($name);
239 }
240
241sub bn_add_words
242 {
243 local($name)=@_;
244
245 &function_begin($name,"");
246
247 &comment("");
248 $a="esi";
249 $b="edi";
250 $c="eax";
251 $r="ebx";
252 $tmp1="ecx";
253 $tmp2="edx";
254 $num="ebp";
255
256 &mov($r,&wparam(0)); # get r
257 &mov($a,&wparam(1)); # get a
258 &mov($b,&wparam(2)); # get b
259 &mov($num,&wparam(3)); # get num
260 &xor($c,$c); # clear carry
261 &and($num,0xfffffff8); # num / 8
262
263 &jz(&label("aw_finish"));
264
265 &set_label("aw_loop",0);
266 for ($i=0; $i<8; $i++)
267 {
268 &comment("Round $i");
269
270 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
271 &mov($tmp2,&DWP($i*4,$b,"",0)); # *b
272 &add($tmp1,$c);
273 &mov($c,0);
274 &adc($c,$c);
275 &add($tmp1,$tmp2);
276 &adc($c,0);
277 &mov(&DWP($i*4,$r,"",0),$tmp1); # *r
278 }
279
280 &comment("");
281 &add($a,32);
282 &add($b,32);
283 &add($r,32);
284 &sub($num,8);
285 &jnz(&label("aw_loop"));
286
287 &set_label("aw_finish",0);
288 &mov($num,&wparam(3)); # get num
289 &and($num,7);
290 &jz(&label("aw_end"));
291
292 for ($i=0; $i<7; $i++)
293 {
294 &comment("Tail Round $i");
295 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
296 &mov($tmp2,&DWP($i*4,$b,"",0));# *b
297 &add($tmp1,$c);
298 &mov($c,0);
299 &adc($c,$c);
300 &add($tmp1,$tmp2);
301 &adc($c,0);
302 &dec($num) if ($i != 6);
303 &mov(&DWP($i*4,$r,"",0),$tmp1); # *a
304 &jz(&label("aw_end")) if ($i != 6);
305 }
306 &set_label("aw_end",0);
307
308# &mov("eax",$c); # $c is "eax"
309
310 &function_end($name);
311 }
312
313sub bn_sub_words
314 {
315 local($name)=@_;
316
317 &function_begin($name,"");
318
319 &comment("");
320 $a="esi";
321 $b="edi";
322 $c="eax";
323 $r="ebx";
324 $tmp1="ecx";
325 $tmp2="edx";
326 $num="ebp";
327
328 &mov($r,&wparam(0)); # get r
329 &mov($a,&wparam(1)); # get a
330 &mov($b,&wparam(2)); # get b
331 &mov($num,&wparam(3)); # get num
332 &xor($c,$c); # clear carry
333 &and($num,0xfffffff8); # num / 8
334
335 &jz(&label("aw_finish"));
336
337 &set_label("aw_loop",0);
338 for ($i=0; $i<8; $i++)
339 {
340 &comment("Round $i");
341
342 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
343 &mov($tmp2,&DWP($i*4,$b,"",0)); # *b
344 &sub($tmp1,$c);
345 &mov($c,0);
346 &adc($c,$c);
347 &sub($tmp1,$tmp2);
348 &adc($c,0);
349 &mov(&DWP($i*4,$r,"",0),$tmp1); # *r
350 }
351
352 &comment("");
353 &add($a,32);
354 &add($b,32);
355 &add($r,32);
356 &sub($num,8);
357 &jnz(&label("aw_loop"));
358
359 &set_label("aw_finish",0);
360 &mov($num,&wparam(3)); # get num
361 &and($num,7);
362 &jz(&label("aw_end"));
363
364 for ($i=0; $i<7; $i++)
365 {
366 &comment("Tail Round $i");
367 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
368 &mov($tmp2,&DWP($i*4,$b,"",0));# *b
369 &sub($tmp1,$c);
370 &mov($c,0);
371 &adc($c,$c);
372 &sub($tmp1,$tmp2);
373 &adc($c,0);
374 &dec($num) if ($i != 6);
375 &mov(&DWP($i*4,$r,"",0),$tmp1); # *a
376 &jz(&label("aw_end")) if ($i != 6);
377 }
378 &set_label("aw_end",0);
379
380# &mov("eax",$c); # $c is "eax"
381
382 &function_end($name);
383 }
384
diff --git a/src/lib/libcrypto/bn/asm/co-586.pl b/src/lib/libcrypto/bn/asm/co-586.pl
deleted file mode 100644
index 5d962cb957..0000000000
--- a/src/lib/libcrypto/bn/asm/co-586.pl
+++ /dev/null
@@ -1,286 +0,0 @@
1#!/usr/local/bin/perl
2
3push(@INC,"perlasm","../../perlasm");
4require "x86asm.pl";
5
6&asm_init($ARGV[0],$0);
7
8&bn_mul_comba("bn_mul_comba8",8);
9&bn_mul_comba("bn_mul_comba4",4);
10&bn_sqr_comba("bn_sqr_comba8",8);
11&bn_sqr_comba("bn_sqr_comba4",4);
12
13&asm_finish();
14
15sub mul_add_c
16 {
17 local($a,$ai,$b,$bi,$c0,$c1,$c2,$pos,$i,$na,$nb)=@_;
18
19 # pos == -1 if eax and edx are pre-loaded, 0 to load from next
20 # words, and 1 if load return value
21
22 &comment("mul a[$ai]*b[$bi]");
23
24 # "eax" and "edx" will always be pre-loaded.
25 # &mov("eax",&DWP($ai*4,$a,"",0)) ;
26 # &mov("edx",&DWP($bi*4,$b,"",0));
27
28 &mul("edx");
29 &add($c0,"eax");
30 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 0; # laod next a
31 &mov("eax",&wparam(0)) if $pos > 0; # load r[]
32 ###
33 &adc($c1,"edx");
34 &mov("edx",&DWP(($nb)*4,$b,"",0)) if $pos == 0; # laod next b
35 &mov("edx",&DWP(($nb)*4,$b,"",0)) if $pos == 1; # laod next b
36 ###
37 &adc($c2,0);
38 # is pos > 1, it means it is the last loop
39 &mov(&DWP($i*4,"eax","",0),$c0) if $pos > 0; # save r[];
40 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 1; # laod next a
41 }
42
43sub sqr_add_c
44 {
45 local($r,$a,$ai,$bi,$c0,$c1,$c2,$pos,$i,$na,$nb)=@_;
46
47 # pos == -1 if eax and edx are pre-loaded, 0 to load from next
48 # words, and 1 if load return value
49
50 &comment("sqr a[$ai]*a[$bi]");
51
52 # "eax" and "edx" will always be pre-loaded.
53 # &mov("eax",&DWP($ai*4,$a,"",0)) ;
54 # &mov("edx",&DWP($bi*4,$b,"",0));
55
56 if ($ai == $bi)
57 { &mul("eax");}
58 else
59 { &mul("edx");}
60 &add($c0,"eax");
61 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 0; # load next a
62 ###
63 &adc($c1,"edx");
64 &mov("edx",&DWP(($nb)*4,$a,"",0)) if ($pos == 1) && ($na != $nb);
65 ###
66 &adc($c2,0);
67 # is pos > 1, it means it is the last loop
68 &mov(&DWP($i*4,$r,"",0),$c0) if $pos > 0; # save r[];
69 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 1; # load next b
70 }
71
72sub sqr_add_c2
73 {
74 local($r,$a,$ai,$bi,$c0,$c1,$c2,$pos,$i,$na,$nb)=@_;
75
76 # pos == -1 if eax and edx are pre-loaded, 0 to load from next
77 # words, and 1 if load return value
78
79 &comment("sqr a[$ai]*a[$bi]");
80
81 # "eax" and "edx" will always be pre-loaded.
82 # &mov("eax",&DWP($ai*4,$a,"",0)) ;
83 # &mov("edx",&DWP($bi*4,$a,"",0));
84
85 if ($ai == $bi)
86 { &mul("eax");}
87 else
88 { &mul("edx");}
89 &add("eax","eax");
90 ###
91 &adc("edx","edx");
92 ###
93 &adc($c2,0);
94 &add($c0,"eax");
95 &adc($c1,"edx");
96 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 0; # load next a
97 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 1; # load next b
98 &adc($c2,0);
99 &mov(&DWP($i*4,$r,"",0),$c0) if $pos > 0; # save r[];
100 &mov("edx",&DWP(($nb)*4,$a,"",0)) if ($pos <= 1) && ($na != $nb);
101 ###
102 }
103
104sub bn_mul_comba
105 {
106 local($name,$num)=@_;
107 local($a,$b,$c0,$c1,$c2);
108 local($i,$as,$ae,$bs,$be,$ai,$bi);
109 local($tot,$end);
110
111 &function_begin_B($name,"");
112
113 $c0="ebx";
114 $c1="ecx";
115 $c2="ebp";
116 $a="esi";
117 $b="edi";
118
119 $as=0;
120 $ae=0;
121 $bs=0;
122 $be=0;
123 $tot=$num+$num-1;
124
125 &push("esi");
126 &mov($a,&wparam(1));
127 &push("edi");
128 &mov($b,&wparam(2));
129 &push("ebp");
130 &push("ebx");
131
132 &xor($c0,$c0);
133 &mov("eax",&DWP(0,$a,"",0)); # load the first word
134 &xor($c1,$c1);
135 &mov("edx",&DWP(0,$b,"",0)); # load the first second
136
137 for ($i=0; $i<$tot; $i++)
138 {
139 $ai=$as;
140 $bi=$bs;
141 $end=$be+1;
142
143 &comment("################## Calculate word $i");
144
145 for ($j=$bs; $j<$end; $j++)
146 {
147 &xor($c2,$c2) if ($j == $bs);
148 if (($j+1) == $end)
149 {
150 $v=1;
151 $v=2 if (($i+1) == $tot);
152 }
153 else
154 { $v=0; }
155 if (($j+1) != $end)
156 {
157 $na=($ai-1);
158 $nb=($bi+1);
159 }
160 else
161 {
162 $na=$as+($i < ($num-1));
163 $nb=$bs+($i >= ($num-1));
164 }
165#printf STDERR "[$ai,$bi] -> [$na,$nb]\n";
166 &mul_add_c($a,$ai,$b,$bi,$c0,$c1,$c2,$v,$i,$na,$nb);
167 if ($v)
168 {
169 &comment("saved r[$i]");
170 # &mov("eax",&wparam(0));
171 # &mov(&DWP($i*4,"eax","",0),$c0);
172 ($c0,$c1,$c2)=($c1,$c2,$c0);
173 }
174 $ai--;
175 $bi++;
176 }
177 $as++ if ($i < ($num-1));
178 $ae++ if ($i >= ($num-1));
179
180 $bs++ if ($i >= ($num-1));
181 $be++ if ($i < ($num-1));
182 }
183 &comment("save r[$i]");
184 # &mov("eax",&wparam(0));
185 &mov(&DWP($i*4,"eax","",0),$c0);
186
187 &pop("ebx");
188 &pop("ebp");
189 &pop("edi");
190 &pop("esi");
191 &ret();
192 &function_end_B($name);
193 }
194
195sub bn_sqr_comba
196 {
197 local($name,$num)=@_;
198 local($r,$a,$c0,$c1,$c2)=@_;
199 local($i,$as,$ae,$bs,$be,$ai,$bi);
200 local($b,$tot,$end,$half);
201
202 &function_begin_B($name,"");
203
204 $c0="ebx";
205 $c1="ecx";
206 $c2="ebp";
207 $a="esi";
208 $r="edi";
209
210 &push("esi");
211 &push("edi");
212 &push("ebp");
213 &push("ebx");
214 &mov($r,&wparam(0));
215 &mov($a,&wparam(1));
216 &xor($c0,$c0);
217 &xor($c1,$c1);
218 &mov("eax",&DWP(0,$a,"",0)); # load the first word
219
220 $as=0;
221 $ae=0;
222 $bs=0;
223 $be=0;
224 $tot=$num+$num-1;
225
226 for ($i=0; $i<$tot; $i++)
227 {
228 $ai=$as;
229 $bi=$bs;
230 $end=$be+1;
231
232 &comment("############### Calculate word $i");
233 for ($j=$bs; $j<$end; $j++)
234 {
235 &xor($c2,$c2) if ($j == $bs);
236 if (($ai-1) < ($bi+1))
237 {
238 $v=1;
239 $v=2 if ($i+1) == $tot;
240 }
241 else
242 { $v=0; }
243 if (!$v)
244 {
245 $na=$ai-1;
246 $nb=$bi+1;
247 }
248 else
249 {
250 $na=$as+($i < ($num-1));
251 $nb=$bs+($i >= ($num-1));
252 }
253 if ($ai == $bi)
254 {
255 &sqr_add_c($r,$a,$ai,$bi,
256 $c0,$c1,$c2,$v,$i,$na,$nb);
257 }
258 else
259 {
260 &sqr_add_c2($r,$a,$ai,$bi,
261 $c0,$c1,$c2,$v,$i,$na,$nb);
262 }
263 if ($v)
264 {
265 &comment("saved r[$i]");
266 #&mov(&DWP($i*4,$r,"",0),$c0);
267 ($c0,$c1,$c2)=($c1,$c2,$c0);
268 last;
269 }
270 $ai--;
271 $bi++;
272 }
273 $as++ if ($i < ($num-1));
274 $ae++ if ($i >= ($num-1));
275
276 $bs++ if ($i >= ($num-1));
277 $be++ if ($i < ($num-1));
278 }
279 &mov(&DWP($i*4,$r,"",0),$c0);
280 &pop("ebx");
281 &pop("ebp");
282 &pop("edi");
283 &pop("esi");
284 &ret();
285 &function_end_B($name);
286 }
diff --git a/src/lib/libcrypto/bn/asm/pa-risc2.s b/src/lib/libcrypto/bn/asm/pa-risc2.s
deleted file mode 100644
index af9730d062..0000000000
--- a/src/lib/libcrypto/bn/asm/pa-risc2.s
+++ /dev/null
@@ -1,1618 +0,0 @@
1;
2; PA-RISC 2.0 implementation of bn_asm code, based on the
3; 64-bit version of the code. This code is effectively the
4; same as the 64-bit version except the register model is
5; slightly different given all values must be 32-bit between
6; function calls. Thus the 64-bit return values are returned
7; in %ret0 and %ret1 vs just %ret0 as is done in 64-bit
8;
9;
10; This code is approximately 2x faster than the C version
11; for RSA/DSA.
12;
13; See http://devresource.hp.com/ for more details on the PA-RISC
14; architecture. Also see the book "PA-RISC 2.0 Architecture"
15; by Gerry Kane for information on the instruction set architecture.
16;
17; Code written by Chris Ruemmler (with some help from the HP C
18; compiler).
19;
20; The code compiles with HP's assembler
21;
22
23 .level 2.0N
24 .space $TEXT$
25 .subspa $CODE$,QUAD=0,ALIGN=8,ACCESS=0x2c,CODE_ONLY
26
27;
28; Global Register definitions used for the routines.
29;
30; Some information about HP's runtime architecture for 32-bits.
31;
32; "Caller save" means the calling function must save the register
33; if it wants the register to be preserved.
34; "Callee save" means if a function uses the register, it must save
35; the value before using it.
36;
37; For the floating point registers
38;
39; "caller save" registers: fr4-fr11, fr22-fr31
40; "callee save" registers: fr12-fr21
41; "special" registers: fr0-fr3 (status and exception registers)
42;
43; For the integer registers
44; value zero : r0
45; "caller save" registers: r1,r19-r26
46; "callee save" registers: r3-r18
47; return register : r2 (rp)
48; return values ; r28,r29 (ret0,ret1)
49; Stack pointer ; r30 (sp)
50; millicode return ptr ; r31 (also a caller save register)
51
52
53;
54; Arguments to the routines
55;
56r_ptr .reg %r26
57a_ptr .reg %r25
58b_ptr .reg %r24
59num .reg %r24
60n .reg %r23
61
62;
63; Note that the "w" argument for bn_mul_add_words and bn_mul_words
64; is passed on the stack at a delta of -56 from the top of stack
65; as the routine is entered.
66;
67
68;
69; Globals used in some routines
70;
71
72top_overflow .reg %r23
73high_mask .reg %r22 ; value 0xffffffff80000000L
74
75
76;------------------------------------------------------------------------------
77;
78; bn_mul_add_words
79;
80;BN_ULONG bn_mul_add_words(BN_ULONG *r_ptr, BN_ULONG *a_ptr,
81; int num, BN_ULONG w)
82;
83; arg0 = r_ptr
84; arg1 = a_ptr
85; arg3 = num
86; -56(sp) = w
87;
88; Local register definitions
89;
90
91fm1 .reg %fr22
92fm .reg %fr23
93ht_temp .reg %fr24
94ht_temp_1 .reg %fr25
95lt_temp .reg %fr26
96lt_temp_1 .reg %fr27
97fm1_1 .reg %fr28
98fm_1 .reg %fr29
99
100fw_h .reg %fr7L
101fw_l .reg %fr7R
102fw .reg %fr7
103
104fht_0 .reg %fr8L
105flt_0 .reg %fr8R
106t_float_0 .reg %fr8
107
108fht_1 .reg %fr9L
109flt_1 .reg %fr9R
110t_float_1 .reg %fr9
111
112tmp_0 .reg %r31
113tmp_1 .reg %r21
114m_0 .reg %r20
115m_1 .reg %r19
116ht_0 .reg %r1
117ht_1 .reg %r3
118lt_0 .reg %r4
119lt_1 .reg %r5
120m1_0 .reg %r6
121m1_1 .reg %r7
122rp_val .reg %r8
123rp_val_1 .reg %r9
124
125bn_mul_add_words
126 .export bn_mul_add_words,entry,NO_RELOCATION,LONG_RETURN
127 .proc
128 .callinfo frame=128
129 .entry
130 .align 64
131
132 STD %r3,0(%sp) ; save r3
133 STD %r4,8(%sp) ; save r4
134 NOP ; Needed to make the loop 16-byte aligned
135 NOP ; needed to make the loop 16-byte aligned
136
137 STD %r5,16(%sp) ; save r5
138 NOP
139 STD %r6,24(%sp) ; save r6
140 STD %r7,32(%sp) ; save r7
141
142 STD %r8,40(%sp) ; save r8
143 STD %r9,48(%sp) ; save r9
144 COPY %r0,%ret1 ; return 0 by default
145 DEPDI,Z 1,31,1,top_overflow ; top_overflow = 1 << 32
146
147 CMPIB,>= 0,num,bn_mul_add_words_exit ; if (num <= 0) then exit
148 LDO 128(%sp),%sp ; bump stack
149
150 ;
151 ; The loop is unrolled twice, so if there is only 1 number
152 ; then go straight to the cleanup code.
153 ;
154 CMPIB,= 1,num,bn_mul_add_words_single_top
155 FLDD -184(%sp),fw ; (-56-128) load up w into fw (fw_h/fw_l)
156
157 ;
158 ; This loop is unrolled 2 times (64-byte aligned as well)
159 ;
160 ; PA-RISC 2.0 chips have two fully pipelined multipliers, thus
161 ; two 32-bit mutiplies can be issued per cycle.
162 ;
163bn_mul_add_words_unroll2
164
165 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
166 FLDD 8(a_ptr),t_float_1 ; load up 64-bit value (fr8L) ht(L)/lt(R)
167 LDD 0(r_ptr),rp_val ; rp[0]
168 LDD 8(r_ptr),rp_val_1 ; rp[1]
169
170 XMPYU fht_0,fw_l,fm1 ; m1[0] = fht_0*fw_l
171 XMPYU fht_1,fw_l,fm1_1 ; m1[1] = fht_1*fw_l
172 FSTD fm1,-16(%sp) ; -16(sp) = m1[0]
173 FSTD fm1_1,-48(%sp) ; -48(sp) = m1[1]
174
175 XMPYU flt_0,fw_h,fm ; m[0] = flt_0*fw_h
176 XMPYU flt_1,fw_h,fm_1 ; m[1] = flt_1*fw_h
177 FSTD fm,-8(%sp) ; -8(sp) = m[0]
178 FSTD fm_1,-40(%sp) ; -40(sp) = m[1]
179
180 XMPYU fht_0,fw_h,ht_temp ; ht_temp = fht_0*fw_h
181 XMPYU fht_1,fw_h,ht_temp_1 ; ht_temp_1 = fht_1*fw_h
182 FSTD ht_temp,-24(%sp) ; -24(sp) = ht_temp
183 FSTD ht_temp_1,-56(%sp) ; -56(sp) = ht_temp_1
184
185 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
186 XMPYU flt_1,fw_l,lt_temp_1 ; lt_temp = lt*fw_l
187 FSTD lt_temp,-32(%sp) ; -32(sp) = lt_temp
188 FSTD lt_temp_1,-64(%sp) ; -64(sp) = lt_temp_1
189
190 LDD -8(%sp),m_0 ; m[0]
191 LDD -40(%sp),m_1 ; m[1]
192 LDD -16(%sp),m1_0 ; m1[0]
193 LDD -48(%sp),m1_1 ; m1[1]
194
195 LDD -24(%sp),ht_0 ; ht[0]
196 LDD -56(%sp),ht_1 ; ht[1]
197 ADD,L m1_0,m_0,tmp_0 ; tmp_0 = m[0] + m1[0];
198 ADD,L m1_1,m_1,tmp_1 ; tmp_1 = m[1] + m1[1];
199
200 LDD -32(%sp),lt_0
201 LDD -64(%sp),lt_1
202 CMPCLR,*>>= tmp_0,m1_0, %r0 ; if (m[0] < m1[0])
203 ADD,L ht_0,top_overflow,ht_0 ; ht[0] += (1<<32)
204
205 CMPCLR,*>>= tmp_1,m1_1,%r0 ; if (m[1] < m1[1])
206 ADD,L ht_1,top_overflow,ht_1 ; ht[1] += (1<<32)
207 EXTRD,U tmp_0,31,32,m_0 ; m[0]>>32
208 DEPD,Z tmp_0,31,32,m1_0 ; m1[0] = m[0]<<32
209
210 EXTRD,U tmp_1,31,32,m_1 ; m[1]>>32
211 DEPD,Z tmp_1,31,32,m1_1 ; m1[1] = m[1]<<32
212 ADD,L ht_0,m_0,ht_0 ; ht[0]+= (m[0]>>32)
213 ADD,L ht_1,m_1,ht_1 ; ht[1]+= (m[1]>>32)
214
215 ADD lt_0,m1_0,lt_0 ; lt[0] = lt[0]+m1[0];
216 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
217 ADD lt_1,m1_1,lt_1 ; lt[1] = lt[1]+m1[1];
218 ADD,DC ht_1,%r0,ht_1 ; ht[1]++
219
220 ADD %ret1,lt_0,lt_0 ; lt[0] = lt[0] + c;
221 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
222 ADD lt_0,rp_val,lt_0 ; lt[0] = lt[0]+rp[0]
223 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
224
225 LDO -2(num),num ; num = num - 2;
226 ADD ht_0,lt_1,lt_1 ; lt[1] = lt[1] + ht_0 (c);
227 ADD,DC ht_1,%r0,ht_1 ; ht[1]++
228 STD lt_0,0(r_ptr) ; rp[0] = lt[0]
229
230 ADD lt_1,rp_val_1,lt_1 ; lt[1] = lt[1]+rp[1]
231 ADD,DC ht_1,%r0,%ret1 ; ht[1]++
232 LDO 16(a_ptr),a_ptr ; a_ptr += 2
233
234 STD lt_1,8(r_ptr) ; rp[1] = lt[1]
235 CMPIB,<= 2,num,bn_mul_add_words_unroll2 ; go again if more to do
236 LDO 16(r_ptr),r_ptr ; r_ptr += 2
237
238 CMPIB,=,N 0,num,bn_mul_add_words_exit ; are we done, or cleanup last one
239
240 ;
241 ; Top of loop aligned on 64-byte boundary
242 ;
243bn_mul_add_words_single_top
244 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
245 LDD 0(r_ptr),rp_val ; rp[0]
246 LDO 8(a_ptr),a_ptr ; a_ptr++
247 XMPYU fht_0,fw_l,fm1 ; m1 = ht*fw_l
248 FSTD fm1,-16(%sp) ; -16(sp) = m1
249 XMPYU flt_0,fw_h,fm ; m = lt*fw_h
250 FSTD fm,-8(%sp) ; -8(sp) = m
251 XMPYU fht_0,fw_h,ht_temp ; ht_temp = ht*fw_h
252 FSTD ht_temp,-24(%sp) ; -24(sp) = ht
253 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
254 FSTD lt_temp,-32(%sp) ; -32(sp) = lt
255
256 LDD -8(%sp),m_0
257 LDD -16(%sp),m1_0 ; m1 = temp1
258 ADD,L m_0,m1_0,tmp_0 ; tmp_0 = m + m1;
259 LDD -24(%sp),ht_0
260 LDD -32(%sp),lt_0
261
262 CMPCLR,*>>= tmp_0,m1_0,%r0 ; if (m < m1)
263 ADD,L ht_0,top_overflow,ht_0 ; ht += (1<<32)
264
265 EXTRD,U tmp_0,31,32,m_0 ; m>>32
266 DEPD,Z tmp_0,31,32,m1_0 ; m1 = m<<32
267
268 ADD,L ht_0,m_0,ht_0 ; ht+= (m>>32)
269 ADD lt_0,m1_0,tmp_0 ; tmp_0 = lt+m1;
270 ADD,DC ht_0,%r0,ht_0 ; ht++
271 ADD %ret1,tmp_0,lt_0 ; lt = lt + c;
272 ADD,DC ht_0,%r0,ht_0 ; ht++
273 ADD lt_0,rp_val,lt_0 ; lt = lt+rp[0]
274 ADD,DC ht_0,%r0,%ret1 ; ht++
275 STD lt_0,0(r_ptr) ; rp[0] = lt
276
277bn_mul_add_words_exit
278 .EXIT
279
280 EXTRD,U %ret1,31,32,%ret0 ; for 32-bit, return in ret0/ret1
281 LDD -80(%sp),%r9 ; restore r9
282 LDD -88(%sp),%r8 ; restore r8
283 LDD -96(%sp),%r7 ; restore r7
284 LDD -104(%sp),%r6 ; restore r6
285 LDD -112(%sp),%r5 ; restore r5
286 LDD -120(%sp),%r4 ; restore r4
287 BVE (%rp)
288 LDD,MB -128(%sp),%r3 ; restore r3
289 .PROCEND ;in=23,24,25,26,29;out=28;
290
291;----------------------------------------------------------------------------
292;
293;BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
294;
295; arg0 = rp
296; arg1 = ap
297; arg3 = num
298; w on stack at -56(sp)
299
300bn_mul_words
301 .proc
302 .callinfo frame=128
303 .entry
304 .EXPORT bn_mul_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
305 .align 64
306
307 STD %r3,0(%sp) ; save r3
308 STD %r4,8(%sp) ; save r4
309 NOP
310 STD %r5,16(%sp) ; save r5
311
312 STD %r6,24(%sp) ; save r6
313 STD %r7,32(%sp) ; save r7
314 COPY %r0,%ret1 ; return 0 by default
315 DEPDI,Z 1,31,1,top_overflow ; top_overflow = 1 << 32
316
317 CMPIB,>= 0,num,bn_mul_words_exit
318 LDO 128(%sp),%sp ; bump stack
319
320 ;
321 ; See if only 1 word to do, thus just do cleanup
322 ;
323 CMPIB,= 1,num,bn_mul_words_single_top
324 FLDD -184(%sp),fw ; (-56-128) load up w into fw (fw_h/fw_l)
325
326 ;
327 ; This loop is unrolled 2 times (64-byte aligned as well)
328 ;
329 ; PA-RISC 2.0 chips have two fully pipelined multipliers, thus
330 ; two 32-bit mutiplies can be issued per cycle.
331 ;
332bn_mul_words_unroll2
333
334 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
335 FLDD 8(a_ptr),t_float_1 ; load up 64-bit value (fr8L) ht(L)/lt(R)
336 XMPYU fht_0,fw_l,fm1 ; m1[0] = fht_0*fw_l
337 XMPYU fht_1,fw_l,fm1_1 ; m1[1] = ht*fw_l
338
339 FSTD fm1,-16(%sp) ; -16(sp) = m1
340 FSTD fm1_1,-48(%sp) ; -48(sp) = m1
341 XMPYU flt_0,fw_h,fm ; m = lt*fw_h
342 XMPYU flt_1,fw_h,fm_1 ; m = lt*fw_h
343
344 FSTD fm,-8(%sp) ; -8(sp) = m
345 FSTD fm_1,-40(%sp) ; -40(sp) = m
346 XMPYU fht_0,fw_h,ht_temp ; ht_temp = fht_0*fw_h
347 XMPYU fht_1,fw_h,ht_temp_1 ; ht_temp = ht*fw_h
348
349 FSTD ht_temp,-24(%sp) ; -24(sp) = ht
350 FSTD ht_temp_1,-56(%sp) ; -56(sp) = ht
351 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
352 XMPYU flt_1,fw_l,lt_temp_1 ; lt_temp = lt*fw_l
353
354 FSTD lt_temp,-32(%sp) ; -32(sp) = lt
355 FSTD lt_temp_1,-64(%sp) ; -64(sp) = lt
356 LDD -8(%sp),m_0
357 LDD -40(%sp),m_1
358
359 LDD -16(%sp),m1_0
360 LDD -48(%sp),m1_1
361 LDD -24(%sp),ht_0
362 LDD -56(%sp),ht_1
363
364 ADD,L m1_0,m_0,tmp_0 ; tmp_0 = m + m1;
365 ADD,L m1_1,m_1,tmp_1 ; tmp_1 = m + m1;
366 LDD -32(%sp),lt_0
367 LDD -64(%sp),lt_1
368
369 CMPCLR,*>>= tmp_0,m1_0, %r0 ; if (m < m1)
370 ADD,L ht_0,top_overflow,ht_0 ; ht += (1<<32)
371 CMPCLR,*>>= tmp_1,m1_1,%r0 ; if (m < m1)
372 ADD,L ht_1,top_overflow,ht_1 ; ht += (1<<32)
373
374 EXTRD,U tmp_0,31,32,m_0 ; m>>32
375 DEPD,Z tmp_0,31,32,m1_0 ; m1 = m<<32
376 EXTRD,U tmp_1,31,32,m_1 ; m>>32
377 DEPD,Z tmp_1,31,32,m1_1 ; m1 = m<<32
378
379 ADD,L ht_0,m_0,ht_0 ; ht+= (m>>32)
380 ADD,L ht_1,m_1,ht_1 ; ht+= (m>>32)
381 ADD lt_0,m1_0,lt_0 ; lt = lt+m1;
382 ADD,DC ht_0,%r0,ht_0 ; ht++
383
384 ADD lt_1,m1_1,lt_1 ; lt = lt+m1;
385 ADD,DC ht_1,%r0,ht_1 ; ht++
386 ADD %ret1,lt_0,lt_0 ; lt = lt + c (ret1);
387 ADD,DC ht_0,%r0,ht_0 ; ht++
388
389 ADD ht_0,lt_1,lt_1 ; lt = lt + c (ht_0)
390 ADD,DC ht_1,%r0,ht_1 ; ht++
391 STD lt_0,0(r_ptr) ; rp[0] = lt
392 STD lt_1,8(r_ptr) ; rp[1] = lt
393
394 COPY ht_1,%ret1 ; carry = ht
395 LDO -2(num),num ; num = num - 2;
396 LDO 16(a_ptr),a_ptr ; ap += 2
397 CMPIB,<= 2,num,bn_mul_words_unroll2
398 LDO 16(r_ptr),r_ptr ; rp++
399
400 CMPIB,=,N 0,num,bn_mul_words_exit ; are we done?
401
402 ;
403 ; Top of loop aligned on 64-byte boundary
404 ;
405bn_mul_words_single_top
406 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
407
408 XMPYU fht_0,fw_l,fm1 ; m1 = ht*fw_l
409 FSTD fm1,-16(%sp) ; -16(sp) = m1
410 XMPYU flt_0,fw_h,fm ; m = lt*fw_h
411 FSTD fm,-8(%sp) ; -8(sp) = m
412 XMPYU fht_0,fw_h,ht_temp ; ht_temp = ht*fw_h
413 FSTD ht_temp,-24(%sp) ; -24(sp) = ht
414 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
415 FSTD lt_temp,-32(%sp) ; -32(sp) = lt
416
417 LDD -8(%sp),m_0
418 LDD -16(%sp),m1_0
419 ADD,L m_0,m1_0,tmp_0 ; tmp_0 = m + m1;
420 LDD -24(%sp),ht_0
421 LDD -32(%sp),lt_0
422
423 CMPCLR,*>>= tmp_0,m1_0,%r0 ; if (m < m1)
424 ADD,L ht_0,top_overflow,ht_0 ; ht += (1<<32)
425
426 EXTRD,U tmp_0,31,32,m_0 ; m>>32
427 DEPD,Z tmp_0,31,32,m1_0 ; m1 = m<<32
428
429 ADD,L ht_0,m_0,ht_0 ; ht+= (m>>32)
430 ADD lt_0,m1_0,lt_0 ; lt= lt+m1;
431 ADD,DC ht_0,%r0,ht_0 ; ht++
432
433 ADD %ret1,lt_0,lt_0 ; lt = lt + c;
434 ADD,DC ht_0,%r0,ht_0 ; ht++
435
436 COPY ht_0,%ret1 ; copy carry
437 STD lt_0,0(r_ptr) ; rp[0] = lt
438
439bn_mul_words_exit
440 .EXIT
441 EXTRD,U %ret1,31,32,%ret0 ; for 32-bit, return in ret0/ret1
442 LDD -96(%sp),%r7 ; restore r7
443 LDD -104(%sp),%r6 ; restore r6
444 LDD -112(%sp),%r5 ; restore r5
445 LDD -120(%sp),%r4 ; restore r4
446 BVE (%rp)
447 LDD,MB -128(%sp),%r3 ; restore r3
448 .PROCEND
449
450;----------------------------------------------------------------------------
451;
452;void bn_sqr_words(BN_ULONG *rp, BN_ULONG *ap, int num)
453;
454; arg0 = rp
455; arg1 = ap
456; arg2 = num
457;
458
459bn_sqr_words
460 .proc
461 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
462 .EXPORT bn_sqr_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
463 .entry
464 .align 64
465
466 STD %r3,0(%sp) ; save r3
467 STD %r4,8(%sp) ; save r4
468 NOP
469 STD %r5,16(%sp) ; save r5
470
471 CMPIB,>= 0,num,bn_sqr_words_exit
472 LDO 128(%sp),%sp ; bump stack
473
474 ;
475 ; If only 1, the goto straight to cleanup
476 ;
477 CMPIB,= 1,num,bn_sqr_words_single_top
478 DEPDI,Z -1,32,33,high_mask ; Create Mask 0xffffffff80000000L
479
480 ;
481 ; This loop is unrolled 2 times (64-byte aligned as well)
482 ;
483
484bn_sqr_words_unroll2
485 FLDD 0(a_ptr),t_float_0 ; a[0]
486 FLDD 8(a_ptr),t_float_1 ; a[1]
487 XMPYU fht_0,flt_0,fm ; m[0]
488 XMPYU fht_1,flt_1,fm_1 ; m[1]
489
490 FSTD fm,-24(%sp) ; store m[0]
491 FSTD fm_1,-56(%sp) ; store m[1]
492 XMPYU flt_0,flt_0,lt_temp ; lt[0]
493 XMPYU flt_1,flt_1,lt_temp_1 ; lt[1]
494
495 FSTD lt_temp,-16(%sp) ; store lt[0]
496 FSTD lt_temp_1,-48(%sp) ; store lt[1]
497 XMPYU fht_0,fht_0,ht_temp ; ht[0]
498 XMPYU fht_1,fht_1,ht_temp_1 ; ht[1]
499
500 FSTD ht_temp,-8(%sp) ; store ht[0]
501 FSTD ht_temp_1,-40(%sp) ; store ht[1]
502 LDD -24(%sp),m_0
503 LDD -56(%sp),m_1
504
505 AND m_0,high_mask,tmp_0 ; m[0] & Mask
506 AND m_1,high_mask,tmp_1 ; m[1] & Mask
507 DEPD,Z m_0,30,31,m_0 ; m[0] << 32+1
508 DEPD,Z m_1,30,31,m_1 ; m[1] << 32+1
509
510 LDD -16(%sp),lt_0
511 LDD -48(%sp),lt_1
512 EXTRD,U tmp_0,32,33,tmp_0 ; tmp_0 = m[0]&Mask >> 32-1
513 EXTRD,U tmp_1,32,33,tmp_1 ; tmp_1 = m[1]&Mask >> 32-1
514
515 LDD -8(%sp),ht_0
516 LDD -40(%sp),ht_1
517 ADD,L ht_0,tmp_0,ht_0 ; ht[0] += tmp_0
518 ADD,L ht_1,tmp_1,ht_1 ; ht[1] += tmp_1
519
520 ADD lt_0,m_0,lt_0 ; lt = lt+m
521 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
522 STD lt_0,0(r_ptr) ; rp[0] = lt[0]
523 STD ht_0,8(r_ptr) ; rp[1] = ht[1]
524
525 ADD lt_1,m_1,lt_1 ; lt = lt+m
526 ADD,DC ht_1,%r0,ht_1 ; ht[1]++
527 STD lt_1,16(r_ptr) ; rp[2] = lt[1]
528 STD ht_1,24(r_ptr) ; rp[3] = ht[1]
529
530 LDO -2(num),num ; num = num - 2;
531 LDO 16(a_ptr),a_ptr ; ap += 2
532 CMPIB,<= 2,num,bn_sqr_words_unroll2
533 LDO 32(r_ptr),r_ptr ; rp += 4
534
535 CMPIB,=,N 0,num,bn_sqr_words_exit ; are we done?
536
537 ;
538 ; Top of loop aligned on 64-byte boundary
539 ;
540bn_sqr_words_single_top
541 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
542
543 XMPYU fht_0,flt_0,fm ; m
544 FSTD fm,-24(%sp) ; store m
545
546 XMPYU flt_0,flt_0,lt_temp ; lt
547 FSTD lt_temp,-16(%sp) ; store lt
548
549 XMPYU fht_0,fht_0,ht_temp ; ht
550 FSTD ht_temp,-8(%sp) ; store ht
551
552 LDD -24(%sp),m_0 ; load m
553 AND m_0,high_mask,tmp_0 ; m & Mask
554 DEPD,Z m_0,30,31,m_0 ; m << 32+1
555 LDD -16(%sp),lt_0 ; lt
556
557 LDD -8(%sp),ht_0 ; ht
558 EXTRD,U tmp_0,32,33,tmp_0 ; tmp_0 = m&Mask >> 32-1
559 ADD m_0,lt_0,lt_0 ; lt = lt+m
560 ADD,L ht_0,tmp_0,ht_0 ; ht += tmp_0
561 ADD,DC ht_0,%r0,ht_0 ; ht++
562
563 STD lt_0,0(r_ptr) ; rp[0] = lt
564 STD ht_0,8(r_ptr) ; rp[1] = ht
565
566bn_sqr_words_exit
567 .EXIT
568 LDD -112(%sp),%r5 ; restore r5
569 LDD -120(%sp),%r4 ; restore r4
570 BVE (%rp)
571 LDD,MB -128(%sp),%r3
572 .PROCEND ;in=23,24,25,26,29;out=28;
573
574
575;----------------------------------------------------------------------------
576;
577;BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
578;
579; arg0 = rp
580; arg1 = ap
581; arg2 = bp
582; arg3 = n
583
584t .reg %r22
585b .reg %r21
586l .reg %r20
587
588bn_add_words
589 .proc
590 .entry
591 .callinfo
592 .EXPORT bn_add_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
593 .align 64
594
595 CMPIB,>= 0,n,bn_add_words_exit
596 COPY %r0,%ret1 ; return 0 by default
597
598 ;
599 ; If 2 or more numbers do the loop
600 ;
601 CMPIB,= 1,n,bn_add_words_single_top
602 NOP
603
604 ;
605 ; This loop is unrolled 2 times (64-byte aligned as well)
606 ;
607bn_add_words_unroll2
608 LDD 0(a_ptr),t
609 LDD 0(b_ptr),b
610 ADD t,%ret1,t ; t = t+c;
611 ADD,DC %r0,%r0,%ret1 ; set c to carry
612 ADD t,b,l ; l = t + b[0]
613 ADD,DC %ret1,%r0,%ret1 ; c+= carry
614 STD l,0(r_ptr)
615
616 LDD 8(a_ptr),t
617 LDD 8(b_ptr),b
618 ADD t,%ret1,t ; t = t+c;
619 ADD,DC %r0,%r0,%ret1 ; set c to carry
620 ADD t,b,l ; l = t + b[0]
621 ADD,DC %ret1,%r0,%ret1 ; c+= carry
622 STD l,8(r_ptr)
623
624 LDO -2(n),n
625 LDO 16(a_ptr),a_ptr
626 LDO 16(b_ptr),b_ptr
627
628 CMPIB,<= 2,n,bn_add_words_unroll2
629 LDO 16(r_ptr),r_ptr
630
631 CMPIB,=,N 0,n,bn_add_words_exit ; are we done?
632
633bn_add_words_single_top
634 LDD 0(a_ptr),t
635 LDD 0(b_ptr),b
636
637 ADD t,%ret1,t ; t = t+c;
638 ADD,DC %r0,%r0,%ret1 ; set c to carry (could use CMPCLR??)
639 ADD t,b,l ; l = t + b[0]
640 ADD,DC %ret1,%r0,%ret1 ; c+= carry
641 STD l,0(r_ptr)
642
643bn_add_words_exit
644 .EXIT
645 BVE (%rp)
646 EXTRD,U %ret1,31,32,%ret0 ; for 32-bit, return in ret0/ret1
647 .PROCEND ;in=23,24,25,26,29;out=28;
648
649;----------------------------------------------------------------------------
650;
651;BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
652;
653; arg0 = rp
654; arg1 = ap
655; arg2 = bp
656; arg3 = n
657
658t1 .reg %r22
659t2 .reg %r21
660sub_tmp1 .reg %r20
661sub_tmp2 .reg %r19
662
663
664bn_sub_words
665 .proc
666 .callinfo
667 .EXPORT bn_sub_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
668 .entry
669 .align 64
670
671 CMPIB,>= 0,n,bn_sub_words_exit
672 COPY %r0,%ret1 ; return 0 by default
673
674 ;
675 ; If 2 or more numbers do the loop
676 ;
677 CMPIB,= 1,n,bn_sub_words_single_top
678 NOP
679
680 ;
681 ; This loop is unrolled 2 times (64-byte aligned as well)
682 ;
683bn_sub_words_unroll2
684 LDD 0(a_ptr),t1
685 LDD 0(b_ptr),t2
686 SUB t1,t2,sub_tmp1 ; t3 = t1-t2;
687 SUB sub_tmp1,%ret1,sub_tmp1 ; t3 = t3- c;
688
689 CMPCLR,*>> t1,t2,sub_tmp2 ; clear if t1 > t2
690 LDO 1(%r0),sub_tmp2
691
692 CMPCLR,*= t1,t2,%r0
693 COPY sub_tmp2,%ret1
694 STD sub_tmp1,0(r_ptr)
695
696 LDD 8(a_ptr),t1
697 LDD 8(b_ptr),t2
698 SUB t1,t2,sub_tmp1 ; t3 = t1-t2;
699 SUB sub_tmp1,%ret1,sub_tmp1 ; t3 = t3- c;
700 CMPCLR,*>> t1,t2,sub_tmp2 ; clear if t1 > t2
701 LDO 1(%r0),sub_tmp2
702
703 CMPCLR,*= t1,t2,%r0
704 COPY sub_tmp2,%ret1
705 STD sub_tmp1,8(r_ptr)
706
707 LDO -2(n),n
708 LDO 16(a_ptr),a_ptr
709 LDO 16(b_ptr),b_ptr
710
711 CMPIB,<= 2,n,bn_sub_words_unroll2
712 LDO 16(r_ptr),r_ptr
713
714 CMPIB,=,N 0,n,bn_sub_words_exit ; are we done?
715
716bn_sub_words_single_top
717 LDD 0(a_ptr),t1
718 LDD 0(b_ptr),t2
719 SUB t1,t2,sub_tmp1 ; t3 = t1-t2;
720 SUB sub_tmp1,%ret1,sub_tmp1 ; t3 = t3- c;
721 CMPCLR,*>> t1,t2,sub_tmp2 ; clear if t1 > t2
722 LDO 1(%r0),sub_tmp2
723
724 CMPCLR,*= t1,t2,%r0
725 COPY sub_tmp2,%ret1
726
727 STD sub_tmp1,0(r_ptr)
728
729bn_sub_words_exit
730 .EXIT
731 BVE (%rp)
732 EXTRD,U %ret1,31,32,%ret0 ; for 32-bit, return in ret0/ret1
733 .PROCEND ;in=23,24,25,26,29;out=28;
734
735;------------------------------------------------------------------------------
736;
737; unsigned long bn_div_words(unsigned long h, unsigned long l, unsigned long d)
738;
739; arg0 = h
740; arg1 = l
741; arg2 = d
742;
743; This is mainly just output from the HP C compiler.
744;
745;------------------------------------------------------------------------------
746bn_div_words
747 .PROC
748 .EXPORT bn_div_words,ENTRY,PRIV_LEV=3,ARGW0=GR,ARGW1=GR,ARGW2=GR,ARGW3=GR,RTNVAL=GR,LONG_RETURN
749 .IMPORT BN_num_bits_word,CODE
750 .IMPORT __iob,DATA
751 .IMPORT fprintf,CODE
752 .IMPORT abort,CODE
753 .IMPORT $$div2U,MILLICODE
754 .CALLINFO CALLER,FRAME=144,ENTRY_GR=%r9,SAVE_RP,ARGS_SAVED,ORDERING_AWARE
755 .ENTRY
756 STW %r2,-20(%r30) ;offset 0x8ec
757 STW,MA %r3,192(%r30) ;offset 0x8f0
758 STW %r4,-188(%r30) ;offset 0x8f4
759 DEPD %r5,31,32,%r6 ;offset 0x8f8
760 STD %r6,-184(%r30) ;offset 0x8fc
761 DEPD %r7,31,32,%r8 ;offset 0x900
762 STD %r8,-176(%r30) ;offset 0x904
763 STW %r9,-168(%r30) ;offset 0x908
764 LDD -248(%r30),%r3 ;offset 0x90c
765 COPY %r26,%r4 ;offset 0x910
766 COPY %r24,%r5 ;offset 0x914
767 DEPD %r25,31,32,%r4 ;offset 0x918
768 CMPB,*<> %r3,%r0,$0006000C ;offset 0x91c
769 DEPD %r23,31,32,%r5 ;offset 0x920
770 MOVIB,TR -1,%r29,$00060002 ;offset 0x924
771 EXTRD,U %r29,31,32,%r28 ;offset 0x928
772$0006002A
773 LDO -1(%r29),%r29 ;offset 0x92c
774 SUB %r23,%r7,%r23 ;offset 0x930
775$00060024
776 SUB %r4,%r31,%r25 ;offset 0x934
777 AND %r25,%r19,%r26 ;offset 0x938
778 CMPB,*<>,N %r0,%r26,$00060046 ;offset 0x93c
779 DEPD,Z %r25,31,32,%r20 ;offset 0x940
780 OR %r20,%r24,%r21 ;offset 0x944
781 CMPB,*<<,N %r21,%r23,$0006002A ;offset 0x948
782 SUB %r31,%r2,%r31 ;offset 0x94c
783$00060046
784$0006002E
785 DEPD,Z %r23,31,32,%r25 ;offset 0x950
786 EXTRD,U %r23,31,32,%r26 ;offset 0x954
787 AND %r25,%r19,%r24 ;offset 0x958
788 ADD,L %r31,%r26,%r31 ;offset 0x95c
789 CMPCLR,*>>= %r5,%r24,%r0 ;offset 0x960
790 LDO 1(%r31),%r31 ;offset 0x964
791$00060032
792 CMPB,*<<=,N %r31,%r4,$00060036 ;offset 0x968
793 LDO -1(%r29),%r29 ;offset 0x96c
794 ADD,L %r4,%r3,%r4 ;offset 0x970
795$00060036
796 ADDIB,=,N -1,%r8,$D0 ;offset 0x974
797 SUB %r5,%r24,%r28 ;offset 0x978
798$0006003A
799 SUB %r4,%r31,%r24 ;offset 0x97c
800 SHRPD %r24,%r28,32,%r4 ;offset 0x980
801 DEPD,Z %r29,31,32,%r9 ;offset 0x984
802 DEPD,Z %r28,31,32,%r5 ;offset 0x988
803$0006001C
804 EXTRD,U %r4,31,32,%r31 ;offset 0x98c
805 CMPB,*<>,N %r31,%r2,$00060020 ;offset 0x990
806 MOVB,TR %r6,%r29,$D1 ;offset 0x994
807 STD %r29,-152(%r30) ;offset 0x998
808$0006000C
809 EXTRD,U %r3,31,32,%r25 ;offset 0x99c
810 COPY %r3,%r26 ;offset 0x9a0
811 EXTRD,U %r3,31,32,%r9 ;offset 0x9a4
812 EXTRD,U %r4,31,32,%r8 ;offset 0x9a8
813 .CALL ARGW0=GR,ARGW1=GR,RTNVAL=GR ;in=25,26;out=28;
814 B,L BN_num_bits_word,%r2 ;offset 0x9ac
815 EXTRD,U %r5,31,32,%r7 ;offset 0x9b0
816 LDI 64,%r20 ;offset 0x9b4
817 DEPD %r7,31,32,%r5 ;offset 0x9b8
818 DEPD %r8,31,32,%r4 ;offset 0x9bc
819 DEPD %r9,31,32,%r3 ;offset 0x9c0
820 CMPB,= %r28,%r20,$00060012 ;offset 0x9c4
821 COPY %r28,%r24 ;offset 0x9c8
822 MTSARCM %r24 ;offset 0x9cc
823 DEPDI,Z -1,%sar,1,%r19 ;offset 0x9d0
824 CMPB,*>>,N %r4,%r19,$D2 ;offset 0x9d4
825$00060012
826 SUBI 64,%r24,%r31 ;offset 0x9d8
827 CMPCLR,*<< %r4,%r3,%r0 ;offset 0x9dc
828 SUB %r4,%r3,%r4 ;offset 0x9e0
829$00060016
830 CMPB,= %r31,%r0,$0006001A ;offset 0x9e4
831 COPY %r0,%r9 ;offset 0x9e8
832 MTSARCM %r31 ;offset 0x9ec
833 DEPD,Z %r3,%sar,64,%r3 ;offset 0x9f0
834 SUBI 64,%r31,%r26 ;offset 0x9f4
835 MTSAR %r26 ;offset 0x9f8
836 SHRPD %r4,%r5,%sar,%r4 ;offset 0x9fc
837 MTSARCM %r31 ;offset 0xa00
838 DEPD,Z %r5,%sar,64,%r5 ;offset 0xa04
839$0006001A
840 DEPDI,Z -1,31,32,%r19 ;offset 0xa08
841 AND %r3,%r19,%r29 ;offset 0xa0c
842 EXTRD,U %r29,31,32,%r2 ;offset 0xa10
843 DEPDI,Z -1,63,32,%r6 ;offset 0xa14
844 MOVIB,TR 2,%r8,$0006001C ;offset 0xa18
845 EXTRD,U %r3,63,32,%r7 ;offset 0xa1c
846$D2
847 ADDIL LR'__iob-$global$,%r27,%r1 ;offset 0xa20
848 LDIL LR'C$7,%r21 ;offset 0xa24
849 LDO RR'__iob-$global$+32(%r1),%r26 ;offset 0xa28
850 .CALL ARGW0=GR,ARGW1=GR,ARGW2=GR,RTNVAL=GR ;in=24,25,26;out=28;
851 B,L fprintf,%r2 ;offset 0xa2c
852 LDO RR'C$7(%r21),%r25 ;offset 0xa30
853 .CALL ;
854 B,L abort,%r2 ;offset 0xa34
855 NOP ;offset 0xa38
856 B $D3 ;offset 0xa3c
857 LDW -212(%r30),%r2 ;offset 0xa40
858$00060020
859 COPY %r4,%r26 ;offset 0xa44
860 EXTRD,U %r4,31,32,%r25 ;offset 0xa48
861 COPY %r2,%r24 ;offset 0xa4c
862 .CALL ;in=23,24,25,26;out=20,21,22,28,29; (MILLICALL)
863 B,L $$div2U,%r31 ;offset 0xa50
864 EXTRD,U %r2,31,32,%r23 ;offset 0xa54
865 DEPD %r28,31,32,%r29 ;offset 0xa58
866$00060022
867 STD %r29,-152(%r30) ;offset 0xa5c
868$D1
869 AND %r5,%r19,%r24 ;offset 0xa60
870 EXTRD,U %r24,31,32,%r24 ;offset 0xa64
871 STW %r2,-160(%r30) ;offset 0xa68
872 STW %r7,-128(%r30) ;offset 0xa6c
873 FLDD -152(%r30),%fr4 ;offset 0xa70
874 FLDD -152(%r30),%fr7 ;offset 0xa74
875 FLDW -160(%r30),%fr8L ;offset 0xa78
876 FLDW -128(%r30),%fr5L ;offset 0xa7c
877 XMPYU %fr8L,%fr7L,%fr10 ;offset 0xa80
878 FSTD %fr10,-136(%r30) ;offset 0xa84
879 XMPYU %fr8L,%fr7R,%fr22 ;offset 0xa88
880 FSTD %fr22,-144(%r30) ;offset 0xa8c
881 XMPYU %fr5L,%fr4L,%fr11 ;offset 0xa90
882 XMPYU %fr5L,%fr4R,%fr23 ;offset 0xa94
883 FSTD %fr11,-112(%r30) ;offset 0xa98
884 FSTD %fr23,-120(%r30) ;offset 0xa9c
885 LDD -136(%r30),%r28 ;offset 0xaa0
886 DEPD,Z %r28,31,32,%r31 ;offset 0xaa4
887 LDD -144(%r30),%r20 ;offset 0xaa8
888 ADD,L %r20,%r31,%r31 ;offset 0xaac
889 LDD -112(%r30),%r22 ;offset 0xab0
890 DEPD,Z %r22,31,32,%r22 ;offset 0xab4
891 LDD -120(%r30),%r21 ;offset 0xab8
892 B $00060024 ;offset 0xabc
893 ADD,L %r21,%r22,%r23 ;offset 0xac0
894$D0
895 OR %r9,%r29,%r29 ;offset 0xac4
896$00060040
897 EXTRD,U %r29,31,32,%r28 ;offset 0xac8
898$00060002
899$L2
900 LDW -212(%r30),%r2 ;offset 0xacc
901$D3
902 LDW -168(%r30),%r9 ;offset 0xad0
903 LDD -176(%r30),%r8 ;offset 0xad4
904 EXTRD,U %r8,31,32,%r7 ;offset 0xad8
905 LDD -184(%r30),%r6 ;offset 0xadc
906 EXTRD,U %r6,31,32,%r5 ;offset 0xae0
907 LDW -188(%r30),%r4 ;offset 0xae4
908 BVE (%r2) ;offset 0xae8
909 .EXIT
910 LDW,MB -192(%r30),%r3 ;offset 0xaec
911 .PROCEND ;in=23,25;out=28,29;fpin=105,107;
912
913
914
915
916;----------------------------------------------------------------------------
917;
918; Registers to hold 64-bit values to manipulate. The "L" part
919; of the register corresponds to the upper 32-bits, while the "R"
920; part corresponds to the lower 32-bits
921;
922; Note, that when using b6 and b7, the code must save these before
923; using them because they are callee save registers
924;
925;
926; Floating point registers to use to save values that
927; are manipulated. These don't collide with ftemp1-6 and
928; are all caller save registers
929;
930a0 .reg %fr22
931a0L .reg %fr22L
932a0R .reg %fr22R
933
934a1 .reg %fr23
935a1L .reg %fr23L
936a1R .reg %fr23R
937
938a2 .reg %fr24
939a2L .reg %fr24L
940a2R .reg %fr24R
941
942a3 .reg %fr25
943a3L .reg %fr25L
944a3R .reg %fr25R
945
946a4 .reg %fr26
947a4L .reg %fr26L
948a4R .reg %fr26R
949
950a5 .reg %fr27
951a5L .reg %fr27L
952a5R .reg %fr27R
953
954a6 .reg %fr28
955a6L .reg %fr28L
956a6R .reg %fr28R
957
958a7 .reg %fr29
959a7L .reg %fr29L
960a7R .reg %fr29R
961
962b0 .reg %fr30
963b0L .reg %fr30L
964b0R .reg %fr30R
965
966b1 .reg %fr31
967b1L .reg %fr31L
968b1R .reg %fr31R
969
970;
971; Temporary floating point variables, these are all caller save
972; registers
973;
974ftemp1 .reg %fr4
975ftemp2 .reg %fr5
976ftemp3 .reg %fr6
977ftemp4 .reg %fr7
978
979;
980; The B set of registers when used.
981;
982
983b2 .reg %fr8
984b2L .reg %fr8L
985b2R .reg %fr8R
986
987b3 .reg %fr9
988b3L .reg %fr9L
989b3R .reg %fr9R
990
991b4 .reg %fr10
992b4L .reg %fr10L
993b4R .reg %fr10R
994
995b5 .reg %fr11
996b5L .reg %fr11L
997b5R .reg %fr11R
998
999b6 .reg %fr12
1000b6L .reg %fr12L
1001b6R .reg %fr12R
1002
1003b7 .reg %fr13
1004b7L .reg %fr13L
1005b7R .reg %fr13R
1006
1007c1 .reg %r21 ; only reg
1008temp1 .reg %r20 ; only reg
1009temp2 .reg %r19 ; only reg
1010temp3 .reg %r31 ; only reg
1011
1012m1 .reg %r28
1013c2 .reg %r23
1014high_one .reg %r1
1015ht .reg %r6
1016lt .reg %r5
1017m .reg %r4
1018c3 .reg %r3
1019
1020SQR_ADD_C .macro A0L,A0R,C1,C2,C3
1021 XMPYU A0L,A0R,ftemp1 ; m
1022 FSTD ftemp1,-24(%sp) ; store m
1023
1024 XMPYU A0R,A0R,ftemp2 ; lt
1025 FSTD ftemp2,-16(%sp) ; store lt
1026
1027 XMPYU A0L,A0L,ftemp3 ; ht
1028 FSTD ftemp3,-8(%sp) ; store ht
1029
1030 LDD -24(%sp),m ; load m
1031 AND m,high_mask,temp2 ; m & Mask
1032 DEPD,Z m,30,31,temp3 ; m << 32+1
1033 LDD -16(%sp),lt ; lt
1034
1035 LDD -8(%sp),ht ; ht
1036 EXTRD,U temp2,32,33,temp1 ; temp1 = m&Mask >> 32-1
1037 ADD temp3,lt,lt ; lt = lt+m
1038 ADD,L ht,temp1,ht ; ht += temp1
1039 ADD,DC ht,%r0,ht ; ht++
1040
1041 ADD C1,lt,C1 ; c1=c1+lt
1042 ADD,DC ht,%r0,ht ; ht++
1043
1044 ADD C2,ht,C2 ; c2=c2+ht
1045 ADD,DC C3,%r0,C3 ; c3++
1046.endm
1047
1048SQR_ADD_C2 .macro A0L,A0R,A1L,A1R,C1,C2,C3
1049 XMPYU A0L,A1R,ftemp1 ; m1 = bl*ht
1050 FSTD ftemp1,-16(%sp) ;
1051 XMPYU A0R,A1L,ftemp2 ; m = bh*lt
1052 FSTD ftemp2,-8(%sp) ;
1053 XMPYU A0R,A1R,ftemp3 ; lt = bl*lt
1054 FSTD ftemp3,-32(%sp)
1055 XMPYU A0L,A1L,ftemp4 ; ht = bh*ht
1056 FSTD ftemp4,-24(%sp) ;
1057
1058 LDD -8(%sp),m ; r21 = m
1059 LDD -16(%sp),m1 ; r19 = m1
1060 ADD,L m,m1,m ; m+m1
1061
1062 DEPD,Z m,31,32,temp3 ; (m+m1<<32)
1063 LDD -24(%sp),ht ; r24 = ht
1064
1065 CMPCLR,*>>= m,m1,%r0 ; if (m < m1)
1066 ADD,L ht,high_one,ht ; ht+=high_one
1067
1068 EXTRD,U m,31,32,temp1 ; m >> 32
1069 LDD -32(%sp),lt ; lt
1070 ADD,L ht,temp1,ht ; ht+= m>>32
1071 ADD lt,temp3,lt ; lt = lt+m1
1072 ADD,DC ht,%r0,ht ; ht++
1073
1074 ADD ht,ht,ht ; ht=ht+ht;
1075 ADD,DC C3,%r0,C3 ; add in carry (c3++)
1076
1077 ADD lt,lt,lt ; lt=lt+lt;
1078 ADD,DC ht,%r0,ht ; add in carry (ht++)
1079
1080 ADD C1,lt,C1 ; c1=c1+lt
1081 ADD,DC,*NUV ht,%r0,ht ; add in carry (ht++)
1082 LDO 1(C3),C3 ; bump c3 if overflow,nullify otherwise
1083
1084 ADD C2,ht,C2 ; c2 = c2 + ht
1085 ADD,DC C3,%r0,C3 ; add in carry (c3++)
1086.endm
1087
1088;
1089;void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
1090; arg0 = r_ptr
1091; arg1 = a_ptr
1092;
1093
1094bn_sqr_comba8
1095 .PROC
1096 .CALLINFO FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1097 .EXPORT bn_sqr_comba8,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1098 .ENTRY
1099 .align 64
1100
1101 STD %r3,0(%sp) ; save r3
1102 STD %r4,8(%sp) ; save r4
1103 STD %r5,16(%sp) ; save r5
1104 STD %r6,24(%sp) ; save r6
1105
1106 ;
1107 ; Zero out carries
1108 ;
1109 COPY %r0,c1
1110 COPY %r0,c2
1111 COPY %r0,c3
1112
1113 LDO 128(%sp),%sp ; bump stack
1114 DEPDI,Z -1,32,33,high_mask ; Create Mask 0xffffffff80000000L
1115 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1116
1117 ;
1118 ; Load up all of the values we are going to use
1119 ;
1120 FLDD 0(a_ptr),a0
1121 FLDD 8(a_ptr),a1
1122 FLDD 16(a_ptr),a2
1123 FLDD 24(a_ptr),a3
1124 FLDD 32(a_ptr),a4
1125 FLDD 40(a_ptr),a5
1126 FLDD 48(a_ptr),a6
1127 FLDD 56(a_ptr),a7
1128
1129 SQR_ADD_C a0L,a0R,c1,c2,c3
1130 STD c1,0(r_ptr) ; r[0] = c1;
1131 COPY %r0,c1
1132
1133 SQR_ADD_C2 a1L,a1R,a0L,a0R,c2,c3,c1
1134 STD c2,8(r_ptr) ; r[1] = c2;
1135 COPY %r0,c2
1136
1137 SQR_ADD_C a1L,a1R,c3,c1,c2
1138 SQR_ADD_C2 a2L,a2R,a0L,a0R,c3,c1,c2
1139 STD c3,16(r_ptr) ; r[2] = c3;
1140 COPY %r0,c3
1141
1142 SQR_ADD_C2 a3L,a3R,a0L,a0R,c1,c2,c3
1143 SQR_ADD_C2 a2L,a2R,a1L,a1R,c1,c2,c3
1144 STD c1,24(r_ptr) ; r[3] = c1;
1145 COPY %r0,c1
1146
1147 SQR_ADD_C a2L,a2R,c2,c3,c1
1148 SQR_ADD_C2 a3L,a3R,a1L,a1R,c2,c3,c1
1149 SQR_ADD_C2 a4L,a4R,a0L,a0R,c2,c3,c1
1150 STD c2,32(r_ptr) ; r[4] = c2;
1151 COPY %r0,c2
1152
1153 SQR_ADD_C2 a5L,a5R,a0L,a0R,c3,c1,c2
1154 SQR_ADD_C2 a4L,a4R,a1L,a1R,c3,c1,c2
1155 SQR_ADD_C2 a3L,a3R,a2L,a2R,c3,c1,c2
1156 STD c3,40(r_ptr) ; r[5] = c3;
1157 COPY %r0,c3
1158
1159 SQR_ADD_C a3L,a3R,c1,c2,c3
1160 SQR_ADD_C2 a4L,a4R,a2L,a2R,c1,c2,c3
1161 SQR_ADD_C2 a5L,a5R,a1L,a1R,c1,c2,c3
1162 SQR_ADD_C2 a6L,a6R,a0L,a0R,c1,c2,c3
1163 STD c1,48(r_ptr) ; r[6] = c1;
1164 COPY %r0,c1
1165
1166 SQR_ADD_C2 a7L,a7R,a0L,a0R,c2,c3,c1
1167 SQR_ADD_C2 a6L,a6R,a1L,a1R,c2,c3,c1
1168 SQR_ADD_C2 a5L,a5R,a2L,a2R,c2,c3,c1
1169 SQR_ADD_C2 a4L,a4R,a3L,a3R,c2,c3,c1
1170 STD c2,56(r_ptr) ; r[7] = c2;
1171 COPY %r0,c2
1172
1173 SQR_ADD_C a4L,a4R,c3,c1,c2
1174 SQR_ADD_C2 a5L,a5R,a3L,a3R,c3,c1,c2
1175 SQR_ADD_C2 a6L,a6R,a2L,a2R,c3,c1,c2
1176 SQR_ADD_C2 a7L,a7R,a1L,a1R,c3,c1,c2
1177 STD c3,64(r_ptr) ; r[8] = c3;
1178 COPY %r0,c3
1179
1180 SQR_ADD_C2 a7L,a7R,a2L,a2R,c1,c2,c3
1181 SQR_ADD_C2 a6L,a6R,a3L,a3R,c1,c2,c3
1182 SQR_ADD_C2 a5L,a5R,a4L,a4R,c1,c2,c3
1183 STD c1,72(r_ptr) ; r[9] = c1;
1184 COPY %r0,c1
1185
1186 SQR_ADD_C a5L,a5R,c2,c3,c1
1187 SQR_ADD_C2 a6L,a6R,a4L,a4R,c2,c3,c1
1188 SQR_ADD_C2 a7L,a7R,a3L,a3R,c2,c3,c1
1189 STD c2,80(r_ptr) ; r[10] = c2;
1190 COPY %r0,c2
1191
1192 SQR_ADD_C2 a7L,a7R,a4L,a4R,c3,c1,c2
1193 SQR_ADD_C2 a6L,a6R,a5L,a5R,c3,c1,c2
1194 STD c3,88(r_ptr) ; r[11] = c3;
1195 COPY %r0,c3
1196
1197 SQR_ADD_C a6L,a6R,c1,c2,c3
1198 SQR_ADD_C2 a7L,a7R,a5L,a5R,c1,c2,c3
1199 STD c1,96(r_ptr) ; r[12] = c1;
1200 COPY %r0,c1
1201
1202 SQR_ADD_C2 a7L,a7R,a6L,a6R,c2,c3,c1
1203 STD c2,104(r_ptr) ; r[13] = c2;
1204 COPY %r0,c2
1205
1206 SQR_ADD_C a7L,a7R,c3,c1,c2
1207 STD c3, 112(r_ptr) ; r[14] = c3
1208 STD c1, 120(r_ptr) ; r[15] = c1
1209
1210 .EXIT
1211 LDD -104(%sp),%r6 ; restore r6
1212 LDD -112(%sp),%r5 ; restore r5
1213 LDD -120(%sp),%r4 ; restore r4
1214 BVE (%rp)
1215 LDD,MB -128(%sp),%r3
1216
1217 .PROCEND
1218
1219;-----------------------------------------------------------------------------
1220;
1221;void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
1222; arg0 = r_ptr
1223; arg1 = a_ptr
1224;
1225
1226bn_sqr_comba4
1227 .proc
1228 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1229 .EXPORT bn_sqr_comba4,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1230 .entry
1231 .align 64
1232 STD %r3,0(%sp) ; save r3
1233 STD %r4,8(%sp) ; save r4
1234 STD %r5,16(%sp) ; save r5
1235 STD %r6,24(%sp) ; save r6
1236
1237 ;
1238 ; Zero out carries
1239 ;
1240 COPY %r0,c1
1241 COPY %r0,c2
1242 COPY %r0,c3
1243
1244 LDO 128(%sp),%sp ; bump stack
1245 DEPDI,Z -1,32,33,high_mask ; Create Mask 0xffffffff80000000L
1246 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1247
1248 ;
1249 ; Load up all of the values we are going to use
1250 ;
1251 FLDD 0(a_ptr),a0
1252 FLDD 8(a_ptr),a1
1253 FLDD 16(a_ptr),a2
1254 FLDD 24(a_ptr),a3
1255 FLDD 32(a_ptr),a4
1256 FLDD 40(a_ptr),a5
1257 FLDD 48(a_ptr),a6
1258 FLDD 56(a_ptr),a7
1259
1260 SQR_ADD_C a0L,a0R,c1,c2,c3
1261
1262 STD c1,0(r_ptr) ; r[0] = c1;
1263 COPY %r0,c1
1264
1265 SQR_ADD_C2 a1L,a1R,a0L,a0R,c2,c3,c1
1266
1267 STD c2,8(r_ptr) ; r[1] = c2;
1268 COPY %r0,c2
1269
1270 SQR_ADD_C a1L,a1R,c3,c1,c2
1271 SQR_ADD_C2 a2L,a2R,a0L,a0R,c3,c1,c2
1272
1273 STD c3,16(r_ptr) ; r[2] = c3;
1274 COPY %r0,c3
1275
1276 SQR_ADD_C2 a3L,a3R,a0L,a0R,c1,c2,c3
1277 SQR_ADD_C2 a2L,a2R,a1L,a1R,c1,c2,c3
1278
1279 STD c1,24(r_ptr) ; r[3] = c1;
1280 COPY %r0,c1
1281
1282 SQR_ADD_C a2L,a2R,c2,c3,c1
1283 SQR_ADD_C2 a3L,a3R,a1L,a1R,c2,c3,c1
1284
1285 STD c2,32(r_ptr) ; r[4] = c2;
1286 COPY %r0,c2
1287
1288 SQR_ADD_C2 a3L,a3R,a2L,a2R,c3,c1,c2
1289 STD c3,40(r_ptr) ; r[5] = c3;
1290 COPY %r0,c3
1291
1292 SQR_ADD_C a3L,a3R,c1,c2,c3
1293 STD c1,48(r_ptr) ; r[6] = c1;
1294 STD c2,56(r_ptr) ; r[7] = c2;
1295
1296 .EXIT
1297 LDD -104(%sp),%r6 ; restore r6
1298 LDD -112(%sp),%r5 ; restore r5
1299 LDD -120(%sp),%r4 ; restore r4
1300 BVE (%rp)
1301 LDD,MB -128(%sp),%r3
1302
1303 .PROCEND
1304
1305
1306;---------------------------------------------------------------------------
1307
1308MUL_ADD_C .macro A0L,A0R,B0L,B0R,C1,C2,C3
1309 XMPYU A0L,B0R,ftemp1 ; m1 = bl*ht
1310 FSTD ftemp1,-16(%sp) ;
1311 XMPYU A0R,B0L,ftemp2 ; m = bh*lt
1312 FSTD ftemp2,-8(%sp) ;
1313 XMPYU A0R,B0R,ftemp3 ; lt = bl*lt
1314 FSTD ftemp3,-32(%sp)
1315 XMPYU A0L,B0L,ftemp4 ; ht = bh*ht
1316 FSTD ftemp4,-24(%sp) ;
1317
1318 LDD -8(%sp),m ; r21 = m
1319 LDD -16(%sp),m1 ; r19 = m1
1320 ADD,L m,m1,m ; m+m1
1321
1322 DEPD,Z m,31,32,temp3 ; (m+m1<<32)
1323 LDD -24(%sp),ht ; r24 = ht
1324
1325 CMPCLR,*>>= m,m1,%r0 ; if (m < m1)
1326 ADD,L ht,high_one,ht ; ht+=high_one
1327
1328 EXTRD,U m,31,32,temp1 ; m >> 32
1329 LDD -32(%sp),lt ; lt
1330 ADD,L ht,temp1,ht ; ht+= m>>32
1331 ADD lt,temp3,lt ; lt = lt+m1
1332 ADD,DC ht,%r0,ht ; ht++
1333
1334 ADD C1,lt,C1 ; c1=c1+lt
1335 ADD,DC ht,%r0,ht ; bump c3 if overflow,nullify otherwise
1336
1337 ADD C2,ht,C2 ; c2 = c2 + ht
1338 ADD,DC C3,%r0,C3 ; add in carry (c3++)
1339.endm
1340
1341
1342;
1343;void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
1344; arg0 = r_ptr
1345; arg1 = a_ptr
1346; arg2 = b_ptr
1347;
1348
1349bn_mul_comba8
1350 .proc
1351 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1352 .EXPORT bn_mul_comba8,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1353 .entry
1354 .align 64
1355
1356 STD %r3,0(%sp) ; save r3
1357 STD %r4,8(%sp) ; save r4
1358 STD %r5,16(%sp) ; save r5
1359 STD %r6,24(%sp) ; save r6
1360 FSTD %fr12,32(%sp) ; save r6
1361 FSTD %fr13,40(%sp) ; save r7
1362
1363 ;
1364 ; Zero out carries
1365 ;
1366 COPY %r0,c1
1367 COPY %r0,c2
1368 COPY %r0,c3
1369
1370 LDO 128(%sp),%sp ; bump stack
1371 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1372
1373 ;
1374 ; Load up all of the values we are going to use
1375 ;
1376 FLDD 0(a_ptr),a0
1377 FLDD 8(a_ptr),a1
1378 FLDD 16(a_ptr),a2
1379 FLDD 24(a_ptr),a3
1380 FLDD 32(a_ptr),a4
1381 FLDD 40(a_ptr),a5
1382 FLDD 48(a_ptr),a6
1383 FLDD 56(a_ptr),a7
1384
1385 FLDD 0(b_ptr),b0
1386 FLDD 8(b_ptr),b1
1387 FLDD 16(b_ptr),b2
1388 FLDD 24(b_ptr),b3
1389 FLDD 32(b_ptr),b4
1390 FLDD 40(b_ptr),b5
1391 FLDD 48(b_ptr),b6
1392 FLDD 56(b_ptr),b7
1393
1394 MUL_ADD_C a0L,a0R,b0L,b0R,c1,c2,c3
1395 STD c1,0(r_ptr)
1396 COPY %r0,c1
1397
1398 MUL_ADD_C a0L,a0R,b1L,b1R,c2,c3,c1
1399 MUL_ADD_C a1L,a1R,b0L,b0R,c2,c3,c1
1400 STD c2,8(r_ptr)
1401 COPY %r0,c2
1402
1403 MUL_ADD_C a2L,a2R,b0L,b0R,c3,c1,c2
1404 MUL_ADD_C a1L,a1R,b1L,b1R,c3,c1,c2
1405 MUL_ADD_C a0L,a0R,b2L,b2R,c3,c1,c2
1406 STD c3,16(r_ptr)
1407 COPY %r0,c3
1408
1409 MUL_ADD_C a0L,a0R,b3L,b3R,c1,c2,c3
1410 MUL_ADD_C a1L,a1R,b2L,b2R,c1,c2,c3
1411 MUL_ADD_C a2L,a2R,b1L,b1R,c1,c2,c3
1412 MUL_ADD_C a3L,a3R,b0L,b0R,c1,c2,c3
1413 STD c1,24(r_ptr)
1414 COPY %r0,c1
1415
1416 MUL_ADD_C a4L,a4R,b0L,b0R,c2,c3,c1
1417 MUL_ADD_C a3L,a3R,b1L,b1R,c2,c3,c1
1418 MUL_ADD_C a2L,a2R,b2L,b2R,c2,c3,c1
1419 MUL_ADD_C a1L,a1R,b3L,b3R,c2,c3,c1
1420 MUL_ADD_C a0L,a0R,b4L,b4R,c2,c3,c1
1421 STD c2,32(r_ptr)
1422 COPY %r0,c2
1423
1424 MUL_ADD_C a0L,a0R,b5L,b5R,c3,c1,c2
1425 MUL_ADD_C a1L,a1R,b4L,b4R,c3,c1,c2
1426 MUL_ADD_C a2L,a2R,b3L,b3R,c3,c1,c2
1427 MUL_ADD_C a3L,a3R,b2L,b2R,c3,c1,c2
1428 MUL_ADD_C a4L,a4R,b1L,b1R,c3,c1,c2
1429 MUL_ADD_C a5L,a5R,b0L,b0R,c3,c1,c2
1430 STD c3,40(r_ptr)
1431 COPY %r0,c3
1432
1433 MUL_ADD_C a6L,a6R,b0L,b0R,c1,c2,c3
1434 MUL_ADD_C a5L,a5R,b1L,b1R,c1,c2,c3
1435 MUL_ADD_C a4L,a4R,b2L,b2R,c1,c2,c3
1436 MUL_ADD_C a3L,a3R,b3L,b3R,c1,c2,c3
1437 MUL_ADD_C a2L,a2R,b4L,b4R,c1,c2,c3
1438 MUL_ADD_C a1L,a1R,b5L,b5R,c1,c2,c3
1439 MUL_ADD_C a0L,a0R,b6L,b6R,c1,c2,c3
1440 STD c1,48(r_ptr)
1441 COPY %r0,c1
1442
1443 MUL_ADD_C a0L,a0R,b7L,b7R,c2,c3,c1
1444 MUL_ADD_C a1L,a1R,b6L,b6R,c2,c3,c1
1445 MUL_ADD_C a2L,a2R,b5L,b5R,c2,c3,c1
1446 MUL_ADD_C a3L,a3R,b4L,b4R,c2,c3,c1
1447 MUL_ADD_C a4L,a4R,b3L,b3R,c2,c3,c1
1448 MUL_ADD_C a5L,a5R,b2L,b2R,c2,c3,c1
1449 MUL_ADD_C a6L,a6R,b1L,b1R,c2,c3,c1
1450 MUL_ADD_C a7L,a7R,b0L,b0R,c2,c3,c1
1451 STD c2,56(r_ptr)
1452 COPY %r0,c2
1453
1454 MUL_ADD_C a7L,a7R,b1L,b1R,c3,c1,c2
1455 MUL_ADD_C a6L,a6R,b2L,b2R,c3,c1,c2
1456 MUL_ADD_C a5L,a5R,b3L,b3R,c3,c1,c2
1457 MUL_ADD_C a4L,a4R,b4L,b4R,c3,c1,c2
1458 MUL_ADD_C a3L,a3R,b5L,b5R,c3,c1,c2
1459 MUL_ADD_C a2L,a2R,b6L,b6R,c3,c1,c2
1460 MUL_ADD_C a1L,a1R,b7L,b7R,c3,c1,c2
1461 STD c3,64(r_ptr)
1462 COPY %r0,c3
1463
1464 MUL_ADD_C a2L,a2R,b7L,b7R,c1,c2,c3
1465 MUL_ADD_C a3L,a3R,b6L,b6R,c1,c2,c3
1466 MUL_ADD_C a4L,a4R,b5L,b5R,c1,c2,c3
1467 MUL_ADD_C a5L,a5R,b4L,b4R,c1,c2,c3
1468 MUL_ADD_C a6L,a6R,b3L,b3R,c1,c2,c3
1469 MUL_ADD_C a7L,a7R,b2L,b2R,c1,c2,c3
1470 STD c1,72(r_ptr)
1471 COPY %r0,c1
1472
1473 MUL_ADD_C a7L,a7R,b3L,b3R,c2,c3,c1
1474 MUL_ADD_C a6L,a6R,b4L,b4R,c2,c3,c1
1475 MUL_ADD_C a5L,a5R,b5L,b5R,c2,c3,c1
1476 MUL_ADD_C a4L,a4R,b6L,b6R,c2,c3,c1
1477 MUL_ADD_C a3L,a3R,b7L,b7R,c2,c3,c1
1478 STD c2,80(r_ptr)
1479 COPY %r0,c2
1480
1481 MUL_ADD_C a4L,a4R,b7L,b7R,c3,c1,c2
1482 MUL_ADD_C a5L,a5R,b6L,b6R,c3,c1,c2
1483 MUL_ADD_C a6L,a6R,b5L,b5R,c3,c1,c2
1484 MUL_ADD_C a7L,a7R,b4L,b4R,c3,c1,c2
1485 STD c3,88(r_ptr)
1486 COPY %r0,c3
1487
1488 MUL_ADD_C a7L,a7R,b5L,b5R,c1,c2,c3
1489 MUL_ADD_C a6L,a6R,b6L,b6R,c1,c2,c3
1490 MUL_ADD_C a5L,a5R,b7L,b7R,c1,c2,c3
1491 STD c1,96(r_ptr)
1492 COPY %r0,c1
1493
1494 MUL_ADD_C a6L,a6R,b7L,b7R,c2,c3,c1
1495 MUL_ADD_C a7L,a7R,b6L,b6R,c2,c3,c1
1496 STD c2,104(r_ptr)
1497 COPY %r0,c2
1498
1499 MUL_ADD_C a7L,a7R,b7L,b7R,c3,c1,c2
1500 STD c3,112(r_ptr)
1501 STD c1,120(r_ptr)
1502
1503 .EXIT
1504 FLDD -88(%sp),%fr13
1505 FLDD -96(%sp),%fr12
1506 LDD -104(%sp),%r6 ; restore r6
1507 LDD -112(%sp),%r5 ; restore r5
1508 LDD -120(%sp),%r4 ; restore r4
1509 BVE (%rp)
1510 LDD,MB -128(%sp),%r3
1511
1512 .PROCEND
1513
1514;-----------------------------------------------------------------------------
1515;
1516;void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
1517; arg0 = r_ptr
1518; arg1 = a_ptr
1519; arg2 = b_ptr
1520;
1521
1522bn_mul_comba4
1523 .proc
1524 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1525 .EXPORT bn_mul_comba4,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1526 .entry
1527 .align 64
1528
1529 STD %r3,0(%sp) ; save r3
1530 STD %r4,8(%sp) ; save r4
1531 STD %r5,16(%sp) ; save r5
1532 STD %r6,24(%sp) ; save r6
1533 FSTD %fr12,32(%sp) ; save r6
1534 FSTD %fr13,40(%sp) ; save r7
1535
1536 ;
1537 ; Zero out carries
1538 ;
1539 COPY %r0,c1
1540 COPY %r0,c2
1541 COPY %r0,c3
1542
1543 LDO 128(%sp),%sp ; bump stack
1544 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1545
1546 ;
1547 ; Load up all of the values we are going to use
1548 ;
1549 FLDD 0(a_ptr),a0
1550 FLDD 8(a_ptr),a1
1551 FLDD 16(a_ptr),a2
1552 FLDD 24(a_ptr),a3
1553
1554 FLDD 0(b_ptr),b0
1555 FLDD 8(b_ptr),b1
1556 FLDD 16(b_ptr),b2
1557 FLDD 24(b_ptr),b3
1558
1559 MUL_ADD_C a0L,a0R,b0L,b0R,c1,c2,c3
1560 STD c1,0(r_ptr)
1561 COPY %r0,c1
1562
1563 MUL_ADD_C a0L,a0R,b1L,b1R,c2,c3,c1
1564 MUL_ADD_C a1L,a1R,b0L,b0R,c2,c3,c1
1565 STD c2,8(r_ptr)
1566 COPY %r0,c2
1567
1568 MUL_ADD_C a2L,a2R,b0L,b0R,c3,c1,c2
1569 MUL_ADD_C a1L,a1R,b1L,b1R,c3,c1,c2
1570 MUL_ADD_C a0L,a0R,b2L,b2R,c3,c1,c2
1571 STD c3,16(r_ptr)
1572 COPY %r0,c3
1573
1574 MUL_ADD_C a0L,a0R,b3L,b3R,c1,c2,c3
1575 MUL_ADD_C a1L,a1R,b2L,b2R,c1,c2,c3
1576 MUL_ADD_C a2L,a2R,b1L,b1R,c1,c2,c3
1577 MUL_ADD_C a3L,a3R,b0L,b0R,c1,c2,c3
1578 STD c1,24(r_ptr)
1579 COPY %r0,c1
1580
1581 MUL_ADD_C a3L,a3R,b1L,b1R,c2,c3,c1
1582 MUL_ADD_C a2L,a2R,b2L,b2R,c2,c3,c1
1583 MUL_ADD_C a1L,a1R,b3L,b3R,c2,c3,c1
1584 STD c2,32(r_ptr)
1585 COPY %r0,c2
1586
1587 MUL_ADD_C a2L,a2R,b3L,b3R,c3,c1,c2
1588 MUL_ADD_C a3L,a3R,b2L,b2R,c3,c1,c2
1589 STD c3,40(r_ptr)
1590 COPY %r0,c3
1591
1592 MUL_ADD_C a3L,a3R,b3L,b3R,c1,c2,c3
1593 STD c1,48(r_ptr)
1594 STD c2,56(r_ptr)
1595
1596 .EXIT
1597 FLDD -88(%sp),%fr13
1598 FLDD -96(%sp),%fr12
1599 LDD -104(%sp),%r6 ; restore r6
1600 LDD -112(%sp),%r5 ; restore r5
1601 LDD -120(%sp),%r4 ; restore r4
1602 BVE (%rp)
1603 LDD,MB -128(%sp),%r3
1604
1605 .PROCEND
1606
1607
1608 .SPACE $TEXT$
1609 .SUBSPA $CODE$
1610 .SPACE $PRIVATE$,SORT=16
1611 .IMPORT $global$,DATA
1612 .SPACE $TEXT$
1613 .SUBSPA $CODE$
1614 .SUBSPA $LIT$,ACCESS=0x2c
1615C$7
1616 .ALIGN 8
1617 .STRINGZ "Division would overflow (%d)\n"
1618 .END
diff --git a/src/lib/libcrypto/bn/asm/pa-risc2W.s b/src/lib/libcrypto/bn/asm/pa-risc2W.s
deleted file mode 100644
index a99545754d..0000000000
--- a/src/lib/libcrypto/bn/asm/pa-risc2W.s
+++ /dev/null
@@ -1,1605 +0,0 @@
1;
2; PA-RISC 64-bit implementation of bn_asm code
3;
4; This code is approximately 2x faster than the C version
5; for RSA/DSA.
6;
7; See http://devresource.hp.com/ for more details on the PA-RISC
8; architecture. Also see the book "PA-RISC 2.0 Architecture"
9; by Gerry Kane for information on the instruction set architecture.
10;
11; Code written by Chris Ruemmler (with some help from the HP C
12; compiler).
13;
14; The code compiles with HP's assembler
15;
16
17 .level 2.0W
18 .space $TEXT$
19 .subspa $CODE$,QUAD=0,ALIGN=8,ACCESS=0x2c,CODE_ONLY
20
21;
22; Global Register definitions used for the routines.
23;
24; Some information about HP's runtime architecture for 64-bits.
25;
26; "Caller save" means the calling function must save the register
27; if it wants the register to be preserved.
28; "Callee save" means if a function uses the register, it must save
29; the value before using it.
30;
31; For the floating point registers
32;
33; "caller save" registers: fr4-fr11, fr22-fr31
34; "callee save" registers: fr12-fr21
35; "special" registers: fr0-fr3 (status and exception registers)
36;
37; For the integer registers
38; value zero : r0
39; "caller save" registers: r1,r19-r26
40; "callee save" registers: r3-r18
41; return register : r2 (rp)
42; return values ; r28 (ret0,ret1)
43; Stack pointer ; r30 (sp)
44; global data pointer ; r27 (dp)
45; argument pointer ; r29 (ap)
46; millicode return ptr ; r31 (also a caller save register)
47
48
49;
50; Arguments to the routines
51;
52r_ptr .reg %r26
53a_ptr .reg %r25
54b_ptr .reg %r24
55num .reg %r24
56w .reg %r23
57n .reg %r23
58
59
60;
61; Globals used in some routines
62;
63
64top_overflow .reg %r29
65high_mask .reg %r22 ; value 0xffffffff80000000L
66
67
68;------------------------------------------------------------------------------
69;
70; bn_mul_add_words
71;
72;BN_ULONG bn_mul_add_words(BN_ULONG *r_ptr, BN_ULONG *a_ptr,
73; int num, BN_ULONG w)
74;
75; arg0 = r_ptr
76; arg1 = a_ptr
77; arg2 = num
78; arg3 = w
79;
80; Local register definitions
81;
82
83fm1 .reg %fr22
84fm .reg %fr23
85ht_temp .reg %fr24
86ht_temp_1 .reg %fr25
87lt_temp .reg %fr26
88lt_temp_1 .reg %fr27
89fm1_1 .reg %fr28
90fm_1 .reg %fr29
91
92fw_h .reg %fr7L
93fw_l .reg %fr7R
94fw .reg %fr7
95
96fht_0 .reg %fr8L
97flt_0 .reg %fr8R
98t_float_0 .reg %fr8
99
100fht_1 .reg %fr9L
101flt_1 .reg %fr9R
102t_float_1 .reg %fr9
103
104tmp_0 .reg %r31
105tmp_1 .reg %r21
106m_0 .reg %r20
107m_1 .reg %r19
108ht_0 .reg %r1
109ht_1 .reg %r3
110lt_0 .reg %r4
111lt_1 .reg %r5
112m1_0 .reg %r6
113m1_1 .reg %r7
114rp_val .reg %r8
115rp_val_1 .reg %r9
116
117bn_mul_add_words
118 .export bn_mul_add_words,entry,NO_RELOCATION,LONG_RETURN
119 .proc
120 .callinfo frame=128
121 .entry
122 .align 64
123
124 STD %r3,0(%sp) ; save r3
125 STD %r4,8(%sp) ; save r4
126 NOP ; Needed to make the loop 16-byte aligned
127 NOP ; Needed to make the loop 16-byte aligned
128
129 STD %r5,16(%sp) ; save r5
130 STD %r6,24(%sp) ; save r6
131 STD %r7,32(%sp) ; save r7
132 STD %r8,40(%sp) ; save r8
133
134 STD %r9,48(%sp) ; save r9
135 COPY %r0,%ret0 ; return 0 by default
136 DEPDI,Z 1,31,1,top_overflow ; top_overflow = 1 << 32
137 STD w,56(%sp) ; store w on stack
138
139 CMPIB,>= 0,num,bn_mul_add_words_exit ; if (num <= 0) then exit
140 LDO 128(%sp),%sp ; bump stack
141
142 ;
143 ; The loop is unrolled twice, so if there is only 1 number
144 ; then go straight to the cleanup code.
145 ;
146 CMPIB,= 1,num,bn_mul_add_words_single_top
147 FLDD -72(%sp),fw ; load up w into fp register fw (fw_h/fw_l)
148
149 ;
150 ; This loop is unrolled 2 times (64-byte aligned as well)
151 ;
152 ; PA-RISC 2.0 chips have two fully pipelined multipliers, thus
153 ; two 32-bit mutiplies can be issued per cycle.
154 ;
155bn_mul_add_words_unroll2
156
157 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
158 FLDD 8(a_ptr),t_float_1 ; load up 64-bit value (fr8L) ht(L)/lt(R)
159 LDD 0(r_ptr),rp_val ; rp[0]
160 LDD 8(r_ptr),rp_val_1 ; rp[1]
161
162 XMPYU fht_0,fw_l,fm1 ; m1[0] = fht_0*fw_l
163 XMPYU fht_1,fw_l,fm1_1 ; m1[1] = fht_1*fw_l
164 FSTD fm1,-16(%sp) ; -16(sp) = m1[0]
165 FSTD fm1_1,-48(%sp) ; -48(sp) = m1[1]
166
167 XMPYU flt_0,fw_h,fm ; m[0] = flt_0*fw_h
168 XMPYU flt_1,fw_h,fm_1 ; m[1] = flt_1*fw_h
169 FSTD fm,-8(%sp) ; -8(sp) = m[0]
170 FSTD fm_1,-40(%sp) ; -40(sp) = m[1]
171
172 XMPYU fht_0,fw_h,ht_temp ; ht_temp = fht_0*fw_h
173 XMPYU fht_1,fw_h,ht_temp_1 ; ht_temp_1 = fht_1*fw_h
174 FSTD ht_temp,-24(%sp) ; -24(sp) = ht_temp
175 FSTD ht_temp_1,-56(%sp) ; -56(sp) = ht_temp_1
176
177 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
178 XMPYU flt_1,fw_l,lt_temp_1 ; lt_temp = lt*fw_l
179 FSTD lt_temp,-32(%sp) ; -32(sp) = lt_temp
180 FSTD lt_temp_1,-64(%sp) ; -64(sp) = lt_temp_1
181
182 LDD -8(%sp),m_0 ; m[0]
183 LDD -40(%sp),m_1 ; m[1]
184 LDD -16(%sp),m1_0 ; m1[0]
185 LDD -48(%sp),m1_1 ; m1[1]
186
187 LDD -24(%sp),ht_0 ; ht[0]
188 LDD -56(%sp),ht_1 ; ht[1]
189 ADD,L m1_0,m_0,tmp_0 ; tmp_0 = m[0] + m1[0];
190 ADD,L m1_1,m_1,tmp_1 ; tmp_1 = m[1] + m1[1];
191
192 LDD -32(%sp),lt_0
193 LDD -64(%sp),lt_1
194 CMPCLR,*>>= tmp_0,m1_0, %r0 ; if (m[0] < m1[0])
195 ADD,L ht_0,top_overflow,ht_0 ; ht[0] += (1<<32)
196
197 CMPCLR,*>>= tmp_1,m1_1,%r0 ; if (m[1] < m1[1])
198 ADD,L ht_1,top_overflow,ht_1 ; ht[1] += (1<<32)
199 EXTRD,U tmp_0,31,32,m_0 ; m[0]>>32
200 DEPD,Z tmp_0,31,32,m1_0 ; m1[0] = m[0]<<32
201
202 EXTRD,U tmp_1,31,32,m_1 ; m[1]>>32
203 DEPD,Z tmp_1,31,32,m1_1 ; m1[1] = m[1]<<32
204 ADD,L ht_0,m_0,ht_0 ; ht[0]+= (m[0]>>32)
205 ADD,L ht_1,m_1,ht_1 ; ht[1]+= (m[1]>>32)
206
207 ADD lt_0,m1_0,lt_0 ; lt[0] = lt[0]+m1[0];
208 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
209 ADD lt_1,m1_1,lt_1 ; lt[1] = lt[1]+m1[1];
210 ADD,DC ht_1,%r0,ht_1 ; ht[1]++
211
212 ADD %ret0,lt_0,lt_0 ; lt[0] = lt[0] + c;
213 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
214 ADD lt_0,rp_val,lt_0 ; lt[0] = lt[0]+rp[0]
215 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
216
217 LDO -2(num),num ; num = num - 2;
218 ADD ht_0,lt_1,lt_1 ; lt[1] = lt[1] + ht_0 (c);
219 ADD,DC ht_1,%r0,ht_1 ; ht[1]++
220 STD lt_0,0(r_ptr) ; rp[0] = lt[0]
221
222 ADD lt_1,rp_val_1,lt_1 ; lt[1] = lt[1]+rp[1]
223 ADD,DC ht_1,%r0,%ret0 ; ht[1]++
224 LDO 16(a_ptr),a_ptr ; a_ptr += 2
225
226 STD lt_1,8(r_ptr) ; rp[1] = lt[1]
227 CMPIB,<= 2,num,bn_mul_add_words_unroll2 ; go again if more to do
228 LDO 16(r_ptr),r_ptr ; r_ptr += 2
229
230 CMPIB,=,N 0,num,bn_mul_add_words_exit ; are we done, or cleanup last one
231
232 ;
233 ; Top of loop aligned on 64-byte boundary
234 ;
235bn_mul_add_words_single_top
236 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
237 LDD 0(r_ptr),rp_val ; rp[0]
238 LDO 8(a_ptr),a_ptr ; a_ptr++
239 XMPYU fht_0,fw_l,fm1 ; m1 = ht*fw_l
240 FSTD fm1,-16(%sp) ; -16(sp) = m1
241 XMPYU flt_0,fw_h,fm ; m = lt*fw_h
242 FSTD fm,-8(%sp) ; -8(sp) = m
243 XMPYU fht_0,fw_h,ht_temp ; ht_temp = ht*fw_h
244 FSTD ht_temp,-24(%sp) ; -24(sp) = ht
245 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
246 FSTD lt_temp,-32(%sp) ; -32(sp) = lt
247
248 LDD -8(%sp),m_0
249 LDD -16(%sp),m1_0 ; m1 = temp1
250 ADD,L m_0,m1_0,tmp_0 ; tmp_0 = m + m1;
251 LDD -24(%sp),ht_0
252 LDD -32(%sp),lt_0
253
254 CMPCLR,*>>= tmp_0,m1_0,%r0 ; if (m < m1)
255 ADD,L ht_0,top_overflow,ht_0 ; ht += (1<<32)
256
257 EXTRD,U tmp_0,31,32,m_0 ; m>>32
258 DEPD,Z tmp_0,31,32,m1_0 ; m1 = m<<32
259
260 ADD,L ht_0,m_0,ht_0 ; ht+= (m>>32)
261 ADD lt_0,m1_0,tmp_0 ; tmp_0 = lt+m1;
262 ADD,DC ht_0,%r0,ht_0 ; ht++
263 ADD %ret0,tmp_0,lt_0 ; lt = lt + c;
264 ADD,DC ht_0,%r0,ht_0 ; ht++
265 ADD lt_0,rp_val,lt_0 ; lt = lt+rp[0]
266 ADD,DC ht_0,%r0,%ret0 ; ht++
267 STD lt_0,0(r_ptr) ; rp[0] = lt
268
269bn_mul_add_words_exit
270 .EXIT
271 LDD -80(%sp),%r9 ; restore r9
272 LDD -88(%sp),%r8 ; restore r8
273 LDD -96(%sp),%r7 ; restore r7
274 LDD -104(%sp),%r6 ; restore r6
275 LDD -112(%sp),%r5 ; restore r5
276 LDD -120(%sp),%r4 ; restore r4
277 BVE (%rp)
278 LDD,MB -128(%sp),%r3 ; restore r3
279 .PROCEND ;in=23,24,25,26,29;out=28;
280
281;----------------------------------------------------------------------------
282;
283;BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
284;
285; arg0 = rp
286; arg1 = ap
287; arg2 = num
288; arg3 = w
289
290bn_mul_words
291 .proc
292 .callinfo frame=128
293 .entry
294 .EXPORT bn_mul_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
295 .align 64
296
297 STD %r3,0(%sp) ; save r3
298 STD %r4,8(%sp) ; save r4
299 STD %r5,16(%sp) ; save r5
300 STD %r6,24(%sp) ; save r6
301
302 STD %r7,32(%sp) ; save r7
303 COPY %r0,%ret0 ; return 0 by default
304 DEPDI,Z 1,31,1,top_overflow ; top_overflow = 1 << 32
305 STD w,56(%sp) ; w on stack
306
307 CMPIB,>= 0,num,bn_mul_words_exit
308 LDO 128(%sp),%sp ; bump stack
309
310 ;
311 ; See if only 1 word to do, thus just do cleanup
312 ;
313 CMPIB,= 1,num,bn_mul_words_single_top
314 FLDD -72(%sp),fw ; load up w into fp register fw (fw_h/fw_l)
315
316 ;
317 ; This loop is unrolled 2 times (64-byte aligned as well)
318 ;
319 ; PA-RISC 2.0 chips have two fully pipelined multipliers, thus
320 ; two 32-bit mutiplies can be issued per cycle.
321 ;
322bn_mul_words_unroll2
323
324 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
325 FLDD 8(a_ptr),t_float_1 ; load up 64-bit value (fr8L) ht(L)/lt(R)
326 XMPYU fht_0,fw_l,fm1 ; m1[0] = fht_0*fw_l
327 XMPYU fht_1,fw_l,fm1_1 ; m1[1] = ht*fw_l
328
329 FSTD fm1,-16(%sp) ; -16(sp) = m1
330 FSTD fm1_1,-48(%sp) ; -48(sp) = m1
331 XMPYU flt_0,fw_h,fm ; m = lt*fw_h
332 XMPYU flt_1,fw_h,fm_1 ; m = lt*fw_h
333
334 FSTD fm,-8(%sp) ; -8(sp) = m
335 FSTD fm_1,-40(%sp) ; -40(sp) = m
336 XMPYU fht_0,fw_h,ht_temp ; ht_temp = fht_0*fw_h
337 XMPYU fht_1,fw_h,ht_temp_1 ; ht_temp = ht*fw_h
338
339 FSTD ht_temp,-24(%sp) ; -24(sp) = ht
340 FSTD ht_temp_1,-56(%sp) ; -56(sp) = ht
341 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
342 XMPYU flt_1,fw_l,lt_temp_1 ; lt_temp = lt*fw_l
343
344 FSTD lt_temp,-32(%sp) ; -32(sp) = lt
345 FSTD lt_temp_1,-64(%sp) ; -64(sp) = lt
346 LDD -8(%sp),m_0
347 LDD -40(%sp),m_1
348
349 LDD -16(%sp),m1_0
350 LDD -48(%sp),m1_1
351 LDD -24(%sp),ht_0
352 LDD -56(%sp),ht_1
353
354 ADD,L m1_0,m_0,tmp_0 ; tmp_0 = m + m1;
355 ADD,L m1_1,m_1,tmp_1 ; tmp_1 = m + m1;
356 LDD -32(%sp),lt_0
357 LDD -64(%sp),lt_1
358
359 CMPCLR,*>>= tmp_0,m1_0, %r0 ; if (m < m1)
360 ADD,L ht_0,top_overflow,ht_0 ; ht += (1<<32)
361 CMPCLR,*>>= tmp_1,m1_1,%r0 ; if (m < m1)
362 ADD,L ht_1,top_overflow,ht_1 ; ht += (1<<32)
363
364 EXTRD,U tmp_0,31,32,m_0 ; m>>32
365 DEPD,Z tmp_0,31,32,m1_0 ; m1 = m<<32
366 EXTRD,U tmp_1,31,32,m_1 ; m>>32
367 DEPD,Z tmp_1,31,32,m1_1 ; m1 = m<<32
368
369 ADD,L ht_0,m_0,ht_0 ; ht+= (m>>32)
370 ADD,L ht_1,m_1,ht_1 ; ht+= (m>>32)
371 ADD lt_0,m1_0,lt_0 ; lt = lt+m1;
372 ADD,DC ht_0,%r0,ht_0 ; ht++
373
374 ADD lt_1,m1_1,lt_1 ; lt = lt+m1;
375 ADD,DC ht_1,%r0,ht_1 ; ht++
376 ADD %ret0,lt_0,lt_0 ; lt = lt + c (ret0);
377 ADD,DC ht_0,%r0,ht_0 ; ht++
378
379 ADD ht_0,lt_1,lt_1 ; lt = lt + c (ht_0)
380 ADD,DC ht_1,%r0,ht_1 ; ht++
381 STD lt_0,0(r_ptr) ; rp[0] = lt
382 STD lt_1,8(r_ptr) ; rp[1] = lt
383
384 COPY ht_1,%ret0 ; carry = ht
385 LDO -2(num),num ; num = num - 2;
386 LDO 16(a_ptr),a_ptr ; ap += 2
387 CMPIB,<= 2,num,bn_mul_words_unroll2
388 LDO 16(r_ptr),r_ptr ; rp++
389
390 CMPIB,=,N 0,num,bn_mul_words_exit ; are we done?
391
392 ;
393 ; Top of loop aligned on 64-byte boundary
394 ;
395bn_mul_words_single_top
396 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
397
398 XMPYU fht_0,fw_l,fm1 ; m1 = ht*fw_l
399 FSTD fm1,-16(%sp) ; -16(sp) = m1
400 XMPYU flt_0,fw_h,fm ; m = lt*fw_h
401 FSTD fm,-8(%sp) ; -8(sp) = m
402 XMPYU fht_0,fw_h,ht_temp ; ht_temp = ht*fw_h
403 FSTD ht_temp,-24(%sp) ; -24(sp) = ht
404 XMPYU flt_0,fw_l,lt_temp ; lt_temp = lt*fw_l
405 FSTD lt_temp,-32(%sp) ; -32(sp) = lt
406
407 LDD -8(%sp),m_0
408 LDD -16(%sp),m1_0
409 ADD,L m_0,m1_0,tmp_0 ; tmp_0 = m + m1;
410 LDD -24(%sp),ht_0
411 LDD -32(%sp),lt_0
412
413 CMPCLR,*>>= tmp_0,m1_0,%r0 ; if (m < m1)
414 ADD,L ht_0,top_overflow,ht_0 ; ht += (1<<32)
415
416 EXTRD,U tmp_0,31,32,m_0 ; m>>32
417 DEPD,Z tmp_0,31,32,m1_0 ; m1 = m<<32
418
419 ADD,L ht_0,m_0,ht_0 ; ht+= (m>>32)
420 ADD lt_0,m1_0,lt_0 ; lt= lt+m1;
421 ADD,DC ht_0,%r0,ht_0 ; ht++
422
423 ADD %ret0,lt_0,lt_0 ; lt = lt + c;
424 ADD,DC ht_0,%r0,ht_0 ; ht++
425
426 COPY ht_0,%ret0 ; copy carry
427 STD lt_0,0(r_ptr) ; rp[0] = lt
428
429bn_mul_words_exit
430 .EXIT
431 LDD -96(%sp),%r7 ; restore r7
432 LDD -104(%sp),%r6 ; restore r6
433 LDD -112(%sp),%r5 ; restore r5
434 LDD -120(%sp),%r4 ; restore r4
435 BVE (%rp)
436 LDD,MB -128(%sp),%r3 ; restore r3
437 .PROCEND ;in=23,24,25,26,29;out=28;
438
439;----------------------------------------------------------------------------
440;
441;void bn_sqr_words(BN_ULONG *rp, BN_ULONG *ap, int num)
442;
443; arg0 = rp
444; arg1 = ap
445; arg2 = num
446;
447
448bn_sqr_words
449 .proc
450 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
451 .EXPORT bn_sqr_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
452 .entry
453 .align 64
454
455 STD %r3,0(%sp) ; save r3
456 STD %r4,8(%sp) ; save r4
457 NOP
458 STD %r5,16(%sp) ; save r5
459
460 CMPIB,>= 0,num,bn_sqr_words_exit
461 LDO 128(%sp),%sp ; bump stack
462
463 ;
464 ; If only 1, the goto straight to cleanup
465 ;
466 CMPIB,= 1,num,bn_sqr_words_single_top
467 DEPDI,Z -1,32,33,high_mask ; Create Mask 0xffffffff80000000L
468
469 ;
470 ; This loop is unrolled 2 times (64-byte aligned as well)
471 ;
472
473bn_sqr_words_unroll2
474 FLDD 0(a_ptr),t_float_0 ; a[0]
475 FLDD 8(a_ptr),t_float_1 ; a[1]
476 XMPYU fht_0,flt_0,fm ; m[0]
477 XMPYU fht_1,flt_1,fm_1 ; m[1]
478
479 FSTD fm,-24(%sp) ; store m[0]
480 FSTD fm_1,-56(%sp) ; store m[1]
481 XMPYU flt_0,flt_0,lt_temp ; lt[0]
482 XMPYU flt_1,flt_1,lt_temp_1 ; lt[1]
483
484 FSTD lt_temp,-16(%sp) ; store lt[0]
485 FSTD lt_temp_1,-48(%sp) ; store lt[1]
486 XMPYU fht_0,fht_0,ht_temp ; ht[0]
487 XMPYU fht_1,fht_1,ht_temp_1 ; ht[1]
488
489 FSTD ht_temp,-8(%sp) ; store ht[0]
490 FSTD ht_temp_1,-40(%sp) ; store ht[1]
491 LDD -24(%sp),m_0
492 LDD -56(%sp),m_1
493
494 AND m_0,high_mask,tmp_0 ; m[0] & Mask
495 AND m_1,high_mask,tmp_1 ; m[1] & Mask
496 DEPD,Z m_0,30,31,m_0 ; m[0] << 32+1
497 DEPD,Z m_1,30,31,m_1 ; m[1] << 32+1
498
499 LDD -16(%sp),lt_0
500 LDD -48(%sp),lt_1
501 EXTRD,U tmp_0,32,33,tmp_0 ; tmp_0 = m[0]&Mask >> 32-1
502 EXTRD,U tmp_1,32,33,tmp_1 ; tmp_1 = m[1]&Mask >> 32-1
503
504 LDD -8(%sp),ht_0
505 LDD -40(%sp),ht_1
506 ADD,L ht_0,tmp_0,ht_0 ; ht[0] += tmp_0
507 ADD,L ht_1,tmp_1,ht_1 ; ht[1] += tmp_1
508
509 ADD lt_0,m_0,lt_0 ; lt = lt+m
510 ADD,DC ht_0,%r0,ht_0 ; ht[0]++
511 STD lt_0,0(r_ptr) ; rp[0] = lt[0]
512 STD ht_0,8(r_ptr) ; rp[1] = ht[1]
513
514 ADD lt_1,m_1,lt_1 ; lt = lt+m
515 ADD,DC ht_1,%r0,ht_1 ; ht[1]++
516 STD lt_1,16(r_ptr) ; rp[2] = lt[1]
517 STD ht_1,24(r_ptr) ; rp[3] = ht[1]
518
519 LDO -2(num),num ; num = num - 2;
520 LDO 16(a_ptr),a_ptr ; ap += 2
521 CMPIB,<= 2,num,bn_sqr_words_unroll2
522 LDO 32(r_ptr),r_ptr ; rp += 4
523
524 CMPIB,=,N 0,num,bn_sqr_words_exit ; are we done?
525
526 ;
527 ; Top of loop aligned on 64-byte boundary
528 ;
529bn_sqr_words_single_top
530 FLDD 0(a_ptr),t_float_0 ; load up 64-bit value (fr8L) ht(L)/lt(R)
531
532 XMPYU fht_0,flt_0,fm ; m
533 FSTD fm,-24(%sp) ; store m
534
535 XMPYU flt_0,flt_0,lt_temp ; lt
536 FSTD lt_temp,-16(%sp) ; store lt
537
538 XMPYU fht_0,fht_0,ht_temp ; ht
539 FSTD ht_temp,-8(%sp) ; store ht
540
541 LDD -24(%sp),m_0 ; load m
542 AND m_0,high_mask,tmp_0 ; m & Mask
543 DEPD,Z m_0,30,31,m_0 ; m << 32+1
544 LDD -16(%sp),lt_0 ; lt
545
546 LDD -8(%sp),ht_0 ; ht
547 EXTRD,U tmp_0,32,33,tmp_0 ; tmp_0 = m&Mask >> 32-1
548 ADD m_0,lt_0,lt_0 ; lt = lt+m
549 ADD,L ht_0,tmp_0,ht_0 ; ht += tmp_0
550 ADD,DC ht_0,%r0,ht_0 ; ht++
551
552 STD lt_0,0(r_ptr) ; rp[0] = lt
553 STD ht_0,8(r_ptr) ; rp[1] = ht
554
555bn_sqr_words_exit
556 .EXIT
557 LDD -112(%sp),%r5 ; restore r5
558 LDD -120(%sp),%r4 ; restore r4
559 BVE (%rp)
560 LDD,MB -128(%sp),%r3
561 .PROCEND ;in=23,24,25,26,29;out=28;
562
563
564;----------------------------------------------------------------------------
565;
566;BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
567;
568; arg0 = rp
569; arg1 = ap
570; arg2 = bp
571; arg3 = n
572
573t .reg %r22
574b .reg %r21
575l .reg %r20
576
577bn_add_words
578 .proc
579 .entry
580 .callinfo
581 .EXPORT bn_add_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
582 .align 64
583
584 CMPIB,>= 0,n,bn_add_words_exit
585 COPY %r0,%ret0 ; return 0 by default
586
587 ;
588 ; If 2 or more numbers do the loop
589 ;
590 CMPIB,= 1,n,bn_add_words_single_top
591 NOP
592
593 ;
594 ; This loop is unrolled 2 times (64-byte aligned as well)
595 ;
596bn_add_words_unroll2
597 LDD 0(a_ptr),t
598 LDD 0(b_ptr),b
599 ADD t,%ret0,t ; t = t+c;
600 ADD,DC %r0,%r0,%ret0 ; set c to carry
601 ADD t,b,l ; l = t + b[0]
602 ADD,DC %ret0,%r0,%ret0 ; c+= carry
603 STD l,0(r_ptr)
604
605 LDD 8(a_ptr),t
606 LDD 8(b_ptr),b
607 ADD t,%ret0,t ; t = t+c;
608 ADD,DC %r0,%r0,%ret0 ; set c to carry
609 ADD t,b,l ; l = t + b[0]
610 ADD,DC %ret0,%r0,%ret0 ; c+= carry
611 STD l,8(r_ptr)
612
613 LDO -2(n),n
614 LDO 16(a_ptr),a_ptr
615 LDO 16(b_ptr),b_ptr
616
617 CMPIB,<= 2,n,bn_add_words_unroll2
618 LDO 16(r_ptr),r_ptr
619
620 CMPIB,=,N 0,n,bn_add_words_exit ; are we done?
621
622bn_add_words_single_top
623 LDD 0(a_ptr),t
624 LDD 0(b_ptr),b
625
626 ADD t,%ret0,t ; t = t+c;
627 ADD,DC %r0,%r0,%ret0 ; set c to carry (could use CMPCLR??)
628 ADD t,b,l ; l = t + b[0]
629 ADD,DC %ret0,%r0,%ret0 ; c+= carry
630 STD l,0(r_ptr)
631
632bn_add_words_exit
633 .EXIT
634 BVE (%rp)
635 NOP
636 .PROCEND ;in=23,24,25,26,29;out=28;
637
638;----------------------------------------------------------------------------
639;
640;BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
641;
642; arg0 = rp
643; arg1 = ap
644; arg2 = bp
645; arg3 = n
646
647t1 .reg %r22
648t2 .reg %r21
649sub_tmp1 .reg %r20
650sub_tmp2 .reg %r19
651
652
653bn_sub_words
654 .proc
655 .callinfo
656 .EXPORT bn_sub_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
657 .entry
658 .align 64
659
660 CMPIB,>= 0,n,bn_sub_words_exit
661 COPY %r0,%ret0 ; return 0 by default
662
663 ;
664 ; If 2 or more numbers do the loop
665 ;
666 CMPIB,= 1,n,bn_sub_words_single_top
667 NOP
668
669 ;
670 ; This loop is unrolled 2 times (64-byte aligned as well)
671 ;
672bn_sub_words_unroll2
673 LDD 0(a_ptr),t1
674 LDD 0(b_ptr),t2
675 SUB t1,t2,sub_tmp1 ; t3 = t1-t2;
676 SUB sub_tmp1,%ret0,sub_tmp1 ; t3 = t3- c;
677
678 CMPCLR,*>> t1,t2,sub_tmp2 ; clear if t1 > t2
679 LDO 1(%r0),sub_tmp2
680
681 CMPCLR,*= t1,t2,%r0
682 COPY sub_tmp2,%ret0
683 STD sub_tmp1,0(r_ptr)
684
685 LDD 8(a_ptr),t1
686 LDD 8(b_ptr),t2
687 SUB t1,t2,sub_tmp1 ; t3 = t1-t2;
688 SUB sub_tmp1,%ret0,sub_tmp1 ; t3 = t3- c;
689 CMPCLR,*>> t1,t2,sub_tmp2 ; clear if t1 > t2
690 LDO 1(%r0),sub_tmp2
691
692 CMPCLR,*= t1,t2,%r0
693 COPY sub_tmp2,%ret0
694 STD sub_tmp1,8(r_ptr)
695
696 LDO -2(n),n
697 LDO 16(a_ptr),a_ptr
698 LDO 16(b_ptr),b_ptr
699
700 CMPIB,<= 2,n,bn_sub_words_unroll2
701 LDO 16(r_ptr),r_ptr
702
703 CMPIB,=,N 0,n,bn_sub_words_exit ; are we done?
704
705bn_sub_words_single_top
706 LDD 0(a_ptr),t1
707 LDD 0(b_ptr),t2
708 SUB t1,t2,sub_tmp1 ; t3 = t1-t2;
709 SUB sub_tmp1,%ret0,sub_tmp1 ; t3 = t3- c;
710 CMPCLR,*>> t1,t2,sub_tmp2 ; clear if t1 > t2
711 LDO 1(%r0),sub_tmp2
712
713 CMPCLR,*= t1,t2,%r0
714 COPY sub_tmp2,%ret0
715
716 STD sub_tmp1,0(r_ptr)
717
718bn_sub_words_exit
719 .EXIT
720 BVE (%rp)
721 NOP
722 .PROCEND ;in=23,24,25,26,29;out=28;
723
724;------------------------------------------------------------------------------
725;
726; unsigned long bn_div_words(unsigned long h, unsigned long l, unsigned long d)
727;
728; arg0 = h
729; arg1 = l
730; arg2 = d
731;
732; This is mainly just modified assembly from the compiler, thus the
733; lack of variable names.
734;
735;------------------------------------------------------------------------------
736bn_div_words
737 .proc
738 .callinfo CALLER,FRAME=272,ENTRY_GR=%r10,SAVE_RP,ARGS_SAVED,ORDERING_AWARE
739 .EXPORT bn_div_words,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
740 .IMPORT BN_num_bits_word,CODE,NO_RELOCATION
741 .IMPORT __iob,DATA
742 .IMPORT fprintf,CODE,NO_RELOCATION
743 .IMPORT abort,CODE,NO_RELOCATION
744 .IMPORT $$div2U,MILLICODE
745 .entry
746 STD %r2,-16(%r30)
747 STD,MA %r3,352(%r30)
748 STD %r4,-344(%r30)
749 STD %r5,-336(%r30)
750 STD %r6,-328(%r30)
751 STD %r7,-320(%r30)
752 STD %r8,-312(%r30)
753 STD %r9,-304(%r30)
754 STD %r10,-296(%r30)
755
756 STD %r27,-288(%r30) ; save gp
757
758 COPY %r24,%r3 ; save d
759 COPY %r26,%r4 ; save h (high 64-bits)
760 LDO -1(%r0),%ret0 ; return -1 by default
761
762 CMPB,*= %r0,%arg2,$D3 ; if (d == 0)
763 COPY %r25,%r5 ; save l (low 64-bits)
764
765 LDO -48(%r30),%r29 ; create ap
766 .CALL ;in=26,29;out=28;
767 B,L BN_num_bits_word,%r2
768 COPY %r3,%r26
769 LDD -288(%r30),%r27 ; restore gp
770 LDI 64,%r21
771
772 CMPB,= %r21,%ret0,$00000012 ;if (i == 64) (forward)
773 COPY %ret0,%r24 ; i
774 MTSARCM %r24
775 DEPDI,Z -1,%sar,1,%r29
776 CMPB,*<<,N %r29,%r4,bn_div_err_case ; if (h > 1<<i) (forward)
777
778$00000012
779 SUBI 64,%r24,%r31 ; i = 64 - i;
780 CMPCLR,*<< %r4,%r3,%r0 ; if (h >= d)
781 SUB %r4,%r3,%r4 ; h -= d
782 CMPB,= %r31,%r0,$0000001A ; if (i)
783 COPY %r0,%r10 ; ret = 0
784 MTSARCM %r31 ; i to shift
785 DEPD,Z %r3,%sar,64,%r3 ; d <<= i;
786 SUBI 64,%r31,%r19 ; 64 - i; redundent
787 MTSAR %r19 ; (64 -i) to shift
788 SHRPD %r4,%r5,%sar,%r4 ; l>> (64-i)
789 MTSARCM %r31 ; i to shift
790 DEPD,Z %r5,%sar,64,%r5 ; l <<= i;
791
792$0000001A
793 DEPDI,Z -1,31,32,%r19
794 EXTRD,U %r3,31,32,%r6 ; dh=(d&0xfff)>>32
795 EXTRD,U %r3,63,32,%r8 ; dl = d&0xffffff
796 LDO 2(%r0),%r9
797 STD %r3,-280(%r30) ; "d" to stack
798
799$0000001C
800 DEPDI,Z -1,63,32,%r29 ;
801 EXTRD,U %r4,31,32,%r31 ; h >> 32
802 CMPB,*=,N %r31,%r6,$D2 ; if ((h>>32) != dh)(forward) div
803 COPY %r4,%r26
804 EXTRD,U %r4,31,32,%r25
805 COPY %r6,%r24
806 .CALL ;in=23,24,25,26;out=20,21,22,28,29; (MILLICALL)
807 B,L $$div2U,%r2
808 EXTRD,U %r6,31,32,%r23
809 DEPD %r28,31,32,%r29
810$D2
811 STD %r29,-272(%r30) ; q
812 AND %r5,%r19,%r24 ; t & 0xffffffff00000000;
813 EXTRD,U %r24,31,32,%r24 ; ???
814 FLDD -272(%r30),%fr7 ; q
815 FLDD -280(%r30),%fr8 ; d
816 XMPYU %fr8L,%fr7L,%fr10
817 FSTD %fr10,-256(%r30)
818 XMPYU %fr8L,%fr7R,%fr22
819 FSTD %fr22,-264(%r30)
820 XMPYU %fr8R,%fr7L,%fr11
821 XMPYU %fr8R,%fr7R,%fr23
822 FSTD %fr11,-232(%r30)
823 FSTD %fr23,-240(%r30)
824 LDD -256(%r30),%r28
825 DEPD,Z %r28,31,32,%r2
826 LDD -264(%r30),%r20
827 ADD,L %r20,%r2,%r31
828 LDD -232(%r30),%r22
829 DEPD,Z %r22,31,32,%r22
830 LDD -240(%r30),%r21
831 B $00000024 ; enter loop
832 ADD,L %r21,%r22,%r23
833
834$0000002A
835 LDO -1(%r29),%r29
836 SUB %r23,%r8,%r23
837$00000024
838 SUB %r4,%r31,%r25
839 AND %r25,%r19,%r26
840 CMPB,*<>,N %r0,%r26,$00000046 ; (forward)
841 DEPD,Z %r25,31,32,%r20
842 OR %r20,%r24,%r21
843 CMPB,*<<,N %r21,%r23,$0000002A ;(backward)
844 SUB %r31,%r6,%r31
845;-------------Break path---------------------
846
847$00000046
848 DEPD,Z %r23,31,32,%r25 ;tl
849 EXTRD,U %r23,31,32,%r26 ;t
850 AND %r25,%r19,%r24 ;tl = (tl<<32)&0xfffffff0000000L
851 ADD,L %r31,%r26,%r31 ;th += t;
852 CMPCLR,*>>= %r5,%r24,%r0 ;if (l<tl)
853 LDO 1(%r31),%r31 ; th++;
854 CMPB,*<<=,N %r31,%r4,$00000036 ;if (n < th) (forward)
855 LDO -1(%r29),%r29 ;q--;
856 ADD,L %r4,%r3,%r4 ;h += d;
857$00000036
858 ADDIB,=,N -1,%r9,$D1 ;if (--count == 0) break (forward)
859 SUB %r5,%r24,%r28 ; l -= tl;
860 SUB %r4,%r31,%r24 ; h -= th;
861 SHRPD %r24,%r28,32,%r4 ; h = ((h<<32)|(l>>32));
862 DEPD,Z %r29,31,32,%r10 ; ret = q<<32
863 b $0000001C
864 DEPD,Z %r28,31,32,%r5 ; l = l << 32
865
866$D1
867 OR %r10,%r29,%r28 ; ret |= q
868$D3
869 LDD -368(%r30),%r2
870$D0
871 LDD -296(%r30),%r10
872 LDD -304(%r30),%r9
873 LDD -312(%r30),%r8
874 LDD -320(%r30),%r7
875 LDD -328(%r30),%r6
876 LDD -336(%r30),%r5
877 LDD -344(%r30),%r4
878 BVE (%r2)
879 .EXIT
880 LDD,MB -352(%r30),%r3
881
882bn_div_err_case
883 MFIA %r6
884 ADDIL L'bn_div_words-bn_div_err_case,%r6,%r1
885 LDO R'bn_div_words-bn_div_err_case(%r1),%r6
886 ADDIL LT'__iob,%r27,%r1
887 LDD RT'__iob(%r1),%r26
888 ADDIL L'C$4-bn_div_words,%r6,%r1
889 LDO R'C$4-bn_div_words(%r1),%r25
890 LDO 64(%r26),%r26
891 .CALL ;in=24,25,26,29;out=28;
892 B,L fprintf,%r2
893 LDO -48(%r30),%r29
894 LDD -288(%r30),%r27
895 .CALL ;in=29;
896 B,L abort,%r2
897 LDO -48(%r30),%r29
898 LDD -288(%r30),%r27
899 B $D0
900 LDD -368(%r30),%r2
901 .PROCEND ;in=24,25,26,29;out=28;
902
903;----------------------------------------------------------------------------
904;
905; Registers to hold 64-bit values to manipulate. The "L" part
906; of the register corresponds to the upper 32-bits, while the "R"
907; part corresponds to the lower 32-bits
908;
909; Note, that when using b6 and b7, the code must save these before
910; using them because they are callee save registers
911;
912;
913; Floating point registers to use to save values that
914; are manipulated. These don't collide with ftemp1-6 and
915; are all caller save registers
916;
917a0 .reg %fr22
918a0L .reg %fr22L
919a0R .reg %fr22R
920
921a1 .reg %fr23
922a1L .reg %fr23L
923a1R .reg %fr23R
924
925a2 .reg %fr24
926a2L .reg %fr24L
927a2R .reg %fr24R
928
929a3 .reg %fr25
930a3L .reg %fr25L
931a3R .reg %fr25R
932
933a4 .reg %fr26
934a4L .reg %fr26L
935a4R .reg %fr26R
936
937a5 .reg %fr27
938a5L .reg %fr27L
939a5R .reg %fr27R
940
941a6 .reg %fr28
942a6L .reg %fr28L
943a6R .reg %fr28R
944
945a7 .reg %fr29
946a7L .reg %fr29L
947a7R .reg %fr29R
948
949b0 .reg %fr30
950b0L .reg %fr30L
951b0R .reg %fr30R
952
953b1 .reg %fr31
954b1L .reg %fr31L
955b1R .reg %fr31R
956
957;
958; Temporary floating point variables, these are all caller save
959; registers
960;
961ftemp1 .reg %fr4
962ftemp2 .reg %fr5
963ftemp3 .reg %fr6
964ftemp4 .reg %fr7
965
966;
967; The B set of registers when used.
968;
969
970b2 .reg %fr8
971b2L .reg %fr8L
972b2R .reg %fr8R
973
974b3 .reg %fr9
975b3L .reg %fr9L
976b3R .reg %fr9R
977
978b4 .reg %fr10
979b4L .reg %fr10L
980b4R .reg %fr10R
981
982b5 .reg %fr11
983b5L .reg %fr11L
984b5R .reg %fr11R
985
986b6 .reg %fr12
987b6L .reg %fr12L
988b6R .reg %fr12R
989
990b7 .reg %fr13
991b7L .reg %fr13L
992b7R .reg %fr13R
993
994c1 .reg %r21 ; only reg
995temp1 .reg %r20 ; only reg
996temp2 .reg %r19 ; only reg
997temp3 .reg %r31 ; only reg
998
999m1 .reg %r28
1000c2 .reg %r23
1001high_one .reg %r1
1002ht .reg %r6
1003lt .reg %r5
1004m .reg %r4
1005c3 .reg %r3
1006
1007SQR_ADD_C .macro A0L,A0R,C1,C2,C3
1008 XMPYU A0L,A0R,ftemp1 ; m
1009 FSTD ftemp1,-24(%sp) ; store m
1010
1011 XMPYU A0R,A0R,ftemp2 ; lt
1012 FSTD ftemp2,-16(%sp) ; store lt
1013
1014 XMPYU A0L,A0L,ftemp3 ; ht
1015 FSTD ftemp3,-8(%sp) ; store ht
1016
1017 LDD -24(%sp),m ; load m
1018 AND m,high_mask,temp2 ; m & Mask
1019 DEPD,Z m,30,31,temp3 ; m << 32+1
1020 LDD -16(%sp),lt ; lt
1021
1022 LDD -8(%sp),ht ; ht
1023 EXTRD,U temp2,32,33,temp1 ; temp1 = m&Mask >> 32-1
1024 ADD temp3,lt,lt ; lt = lt+m
1025 ADD,L ht,temp1,ht ; ht += temp1
1026 ADD,DC ht,%r0,ht ; ht++
1027
1028 ADD C1,lt,C1 ; c1=c1+lt
1029 ADD,DC ht,%r0,ht ; ht++
1030
1031 ADD C2,ht,C2 ; c2=c2+ht
1032 ADD,DC C3,%r0,C3 ; c3++
1033.endm
1034
1035SQR_ADD_C2 .macro A0L,A0R,A1L,A1R,C1,C2,C3
1036 XMPYU A0L,A1R,ftemp1 ; m1 = bl*ht
1037 FSTD ftemp1,-16(%sp) ;
1038 XMPYU A0R,A1L,ftemp2 ; m = bh*lt
1039 FSTD ftemp2,-8(%sp) ;
1040 XMPYU A0R,A1R,ftemp3 ; lt = bl*lt
1041 FSTD ftemp3,-32(%sp)
1042 XMPYU A0L,A1L,ftemp4 ; ht = bh*ht
1043 FSTD ftemp4,-24(%sp) ;
1044
1045 LDD -8(%sp),m ; r21 = m
1046 LDD -16(%sp),m1 ; r19 = m1
1047 ADD,L m,m1,m ; m+m1
1048
1049 DEPD,Z m,31,32,temp3 ; (m+m1<<32)
1050 LDD -24(%sp),ht ; r24 = ht
1051
1052 CMPCLR,*>>= m,m1,%r0 ; if (m < m1)
1053 ADD,L ht,high_one,ht ; ht+=high_one
1054
1055 EXTRD,U m,31,32,temp1 ; m >> 32
1056 LDD -32(%sp),lt ; lt
1057 ADD,L ht,temp1,ht ; ht+= m>>32
1058 ADD lt,temp3,lt ; lt = lt+m1
1059 ADD,DC ht,%r0,ht ; ht++
1060
1061 ADD ht,ht,ht ; ht=ht+ht;
1062 ADD,DC C3,%r0,C3 ; add in carry (c3++)
1063
1064 ADD lt,lt,lt ; lt=lt+lt;
1065 ADD,DC ht,%r0,ht ; add in carry (ht++)
1066
1067 ADD C1,lt,C1 ; c1=c1+lt
1068 ADD,DC,*NUV ht,%r0,ht ; add in carry (ht++)
1069 LDO 1(C3),C3 ; bump c3 if overflow,nullify otherwise
1070
1071 ADD C2,ht,C2 ; c2 = c2 + ht
1072 ADD,DC C3,%r0,C3 ; add in carry (c3++)
1073.endm
1074
1075;
1076;void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
1077; arg0 = r_ptr
1078; arg1 = a_ptr
1079;
1080
1081bn_sqr_comba8
1082 .PROC
1083 .CALLINFO FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1084 .EXPORT bn_sqr_comba8,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1085 .ENTRY
1086 .align 64
1087
1088 STD %r3,0(%sp) ; save r3
1089 STD %r4,8(%sp) ; save r4
1090 STD %r5,16(%sp) ; save r5
1091 STD %r6,24(%sp) ; save r6
1092
1093 ;
1094 ; Zero out carries
1095 ;
1096 COPY %r0,c1
1097 COPY %r0,c2
1098 COPY %r0,c3
1099
1100 LDO 128(%sp),%sp ; bump stack
1101 DEPDI,Z -1,32,33,high_mask ; Create Mask 0xffffffff80000000L
1102 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1103
1104 ;
1105 ; Load up all of the values we are going to use
1106 ;
1107 FLDD 0(a_ptr),a0
1108 FLDD 8(a_ptr),a1
1109 FLDD 16(a_ptr),a2
1110 FLDD 24(a_ptr),a3
1111 FLDD 32(a_ptr),a4
1112 FLDD 40(a_ptr),a5
1113 FLDD 48(a_ptr),a6
1114 FLDD 56(a_ptr),a7
1115
1116 SQR_ADD_C a0L,a0R,c1,c2,c3
1117 STD c1,0(r_ptr) ; r[0] = c1;
1118 COPY %r0,c1
1119
1120 SQR_ADD_C2 a1L,a1R,a0L,a0R,c2,c3,c1
1121 STD c2,8(r_ptr) ; r[1] = c2;
1122 COPY %r0,c2
1123
1124 SQR_ADD_C a1L,a1R,c3,c1,c2
1125 SQR_ADD_C2 a2L,a2R,a0L,a0R,c3,c1,c2
1126 STD c3,16(r_ptr) ; r[2] = c3;
1127 COPY %r0,c3
1128
1129 SQR_ADD_C2 a3L,a3R,a0L,a0R,c1,c2,c3
1130 SQR_ADD_C2 a2L,a2R,a1L,a1R,c1,c2,c3
1131 STD c1,24(r_ptr) ; r[3] = c1;
1132 COPY %r0,c1
1133
1134 SQR_ADD_C a2L,a2R,c2,c3,c1
1135 SQR_ADD_C2 a3L,a3R,a1L,a1R,c2,c3,c1
1136 SQR_ADD_C2 a4L,a4R,a0L,a0R,c2,c3,c1
1137 STD c2,32(r_ptr) ; r[4] = c2;
1138 COPY %r0,c2
1139
1140 SQR_ADD_C2 a5L,a5R,a0L,a0R,c3,c1,c2
1141 SQR_ADD_C2 a4L,a4R,a1L,a1R,c3,c1,c2
1142 SQR_ADD_C2 a3L,a3R,a2L,a2R,c3,c1,c2
1143 STD c3,40(r_ptr) ; r[5] = c3;
1144 COPY %r0,c3
1145
1146 SQR_ADD_C a3L,a3R,c1,c2,c3
1147 SQR_ADD_C2 a4L,a4R,a2L,a2R,c1,c2,c3
1148 SQR_ADD_C2 a5L,a5R,a1L,a1R,c1,c2,c3
1149 SQR_ADD_C2 a6L,a6R,a0L,a0R,c1,c2,c3
1150 STD c1,48(r_ptr) ; r[6] = c1;
1151 COPY %r0,c1
1152
1153 SQR_ADD_C2 a7L,a7R,a0L,a0R,c2,c3,c1
1154 SQR_ADD_C2 a6L,a6R,a1L,a1R,c2,c3,c1
1155 SQR_ADD_C2 a5L,a5R,a2L,a2R,c2,c3,c1
1156 SQR_ADD_C2 a4L,a4R,a3L,a3R,c2,c3,c1
1157 STD c2,56(r_ptr) ; r[7] = c2;
1158 COPY %r0,c2
1159
1160 SQR_ADD_C a4L,a4R,c3,c1,c2
1161 SQR_ADD_C2 a5L,a5R,a3L,a3R,c3,c1,c2
1162 SQR_ADD_C2 a6L,a6R,a2L,a2R,c3,c1,c2
1163 SQR_ADD_C2 a7L,a7R,a1L,a1R,c3,c1,c2
1164 STD c3,64(r_ptr) ; r[8] = c3;
1165 COPY %r0,c3
1166
1167 SQR_ADD_C2 a7L,a7R,a2L,a2R,c1,c2,c3
1168 SQR_ADD_C2 a6L,a6R,a3L,a3R,c1,c2,c3
1169 SQR_ADD_C2 a5L,a5R,a4L,a4R,c1,c2,c3
1170 STD c1,72(r_ptr) ; r[9] = c1;
1171 COPY %r0,c1
1172
1173 SQR_ADD_C a5L,a5R,c2,c3,c1
1174 SQR_ADD_C2 a6L,a6R,a4L,a4R,c2,c3,c1
1175 SQR_ADD_C2 a7L,a7R,a3L,a3R,c2,c3,c1
1176 STD c2,80(r_ptr) ; r[10] = c2;
1177 COPY %r0,c2
1178
1179 SQR_ADD_C2 a7L,a7R,a4L,a4R,c3,c1,c2
1180 SQR_ADD_C2 a6L,a6R,a5L,a5R,c3,c1,c2
1181 STD c3,88(r_ptr) ; r[11] = c3;
1182 COPY %r0,c3
1183
1184 SQR_ADD_C a6L,a6R,c1,c2,c3
1185 SQR_ADD_C2 a7L,a7R,a5L,a5R,c1,c2,c3
1186 STD c1,96(r_ptr) ; r[12] = c1;
1187 COPY %r0,c1
1188
1189 SQR_ADD_C2 a7L,a7R,a6L,a6R,c2,c3,c1
1190 STD c2,104(r_ptr) ; r[13] = c2;
1191 COPY %r0,c2
1192
1193 SQR_ADD_C a7L,a7R,c3,c1,c2
1194 STD c3, 112(r_ptr) ; r[14] = c3
1195 STD c1, 120(r_ptr) ; r[15] = c1
1196
1197 .EXIT
1198 LDD -104(%sp),%r6 ; restore r6
1199 LDD -112(%sp),%r5 ; restore r5
1200 LDD -120(%sp),%r4 ; restore r4
1201 BVE (%rp)
1202 LDD,MB -128(%sp),%r3
1203
1204 .PROCEND
1205
1206;-----------------------------------------------------------------------------
1207;
1208;void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
1209; arg0 = r_ptr
1210; arg1 = a_ptr
1211;
1212
1213bn_sqr_comba4
1214 .proc
1215 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1216 .EXPORT bn_sqr_comba4,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1217 .entry
1218 .align 64
1219 STD %r3,0(%sp) ; save r3
1220 STD %r4,8(%sp) ; save r4
1221 STD %r5,16(%sp) ; save r5
1222 STD %r6,24(%sp) ; save r6
1223
1224 ;
1225 ; Zero out carries
1226 ;
1227 COPY %r0,c1
1228 COPY %r0,c2
1229 COPY %r0,c3
1230
1231 LDO 128(%sp),%sp ; bump stack
1232 DEPDI,Z -1,32,33,high_mask ; Create Mask 0xffffffff80000000L
1233 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1234
1235 ;
1236 ; Load up all of the values we are going to use
1237 ;
1238 FLDD 0(a_ptr),a0
1239 FLDD 8(a_ptr),a1
1240 FLDD 16(a_ptr),a2
1241 FLDD 24(a_ptr),a3
1242 FLDD 32(a_ptr),a4
1243 FLDD 40(a_ptr),a5
1244 FLDD 48(a_ptr),a6
1245 FLDD 56(a_ptr),a7
1246
1247 SQR_ADD_C a0L,a0R,c1,c2,c3
1248
1249 STD c1,0(r_ptr) ; r[0] = c1;
1250 COPY %r0,c1
1251
1252 SQR_ADD_C2 a1L,a1R,a0L,a0R,c2,c3,c1
1253
1254 STD c2,8(r_ptr) ; r[1] = c2;
1255 COPY %r0,c2
1256
1257 SQR_ADD_C a1L,a1R,c3,c1,c2
1258 SQR_ADD_C2 a2L,a2R,a0L,a0R,c3,c1,c2
1259
1260 STD c3,16(r_ptr) ; r[2] = c3;
1261 COPY %r0,c3
1262
1263 SQR_ADD_C2 a3L,a3R,a0L,a0R,c1,c2,c3
1264 SQR_ADD_C2 a2L,a2R,a1L,a1R,c1,c2,c3
1265
1266 STD c1,24(r_ptr) ; r[3] = c1;
1267 COPY %r0,c1
1268
1269 SQR_ADD_C a2L,a2R,c2,c3,c1
1270 SQR_ADD_C2 a3L,a3R,a1L,a1R,c2,c3,c1
1271
1272 STD c2,32(r_ptr) ; r[4] = c2;
1273 COPY %r0,c2
1274
1275 SQR_ADD_C2 a3L,a3R,a2L,a2R,c3,c1,c2
1276 STD c3,40(r_ptr) ; r[5] = c3;
1277 COPY %r0,c3
1278
1279 SQR_ADD_C a3L,a3R,c1,c2,c3
1280 STD c1,48(r_ptr) ; r[6] = c1;
1281 STD c2,56(r_ptr) ; r[7] = c2;
1282
1283 .EXIT
1284 LDD -104(%sp),%r6 ; restore r6
1285 LDD -112(%sp),%r5 ; restore r5
1286 LDD -120(%sp),%r4 ; restore r4
1287 BVE (%rp)
1288 LDD,MB -128(%sp),%r3
1289
1290 .PROCEND
1291
1292
1293;---------------------------------------------------------------------------
1294
1295MUL_ADD_C .macro A0L,A0R,B0L,B0R,C1,C2,C3
1296 XMPYU A0L,B0R,ftemp1 ; m1 = bl*ht
1297 FSTD ftemp1,-16(%sp) ;
1298 XMPYU A0R,B0L,ftemp2 ; m = bh*lt
1299 FSTD ftemp2,-8(%sp) ;
1300 XMPYU A0R,B0R,ftemp3 ; lt = bl*lt
1301 FSTD ftemp3,-32(%sp)
1302 XMPYU A0L,B0L,ftemp4 ; ht = bh*ht
1303 FSTD ftemp4,-24(%sp) ;
1304
1305 LDD -8(%sp),m ; r21 = m
1306 LDD -16(%sp),m1 ; r19 = m1
1307 ADD,L m,m1,m ; m+m1
1308
1309 DEPD,Z m,31,32,temp3 ; (m+m1<<32)
1310 LDD -24(%sp),ht ; r24 = ht
1311
1312 CMPCLR,*>>= m,m1,%r0 ; if (m < m1)
1313 ADD,L ht,high_one,ht ; ht+=high_one
1314
1315 EXTRD,U m,31,32,temp1 ; m >> 32
1316 LDD -32(%sp),lt ; lt
1317 ADD,L ht,temp1,ht ; ht+= m>>32
1318 ADD lt,temp3,lt ; lt = lt+m1
1319 ADD,DC ht,%r0,ht ; ht++
1320
1321 ADD C1,lt,C1 ; c1=c1+lt
1322 ADD,DC ht,%r0,ht ; bump c3 if overflow,nullify otherwise
1323
1324 ADD C2,ht,C2 ; c2 = c2 + ht
1325 ADD,DC C3,%r0,C3 ; add in carry (c3++)
1326.endm
1327
1328
1329;
1330;void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
1331; arg0 = r_ptr
1332; arg1 = a_ptr
1333; arg2 = b_ptr
1334;
1335
1336bn_mul_comba8
1337 .proc
1338 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1339 .EXPORT bn_mul_comba8,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1340 .entry
1341 .align 64
1342
1343 STD %r3,0(%sp) ; save r3
1344 STD %r4,8(%sp) ; save r4
1345 STD %r5,16(%sp) ; save r5
1346 STD %r6,24(%sp) ; save r6
1347 FSTD %fr12,32(%sp) ; save r6
1348 FSTD %fr13,40(%sp) ; save r7
1349
1350 ;
1351 ; Zero out carries
1352 ;
1353 COPY %r0,c1
1354 COPY %r0,c2
1355 COPY %r0,c3
1356
1357 LDO 128(%sp),%sp ; bump stack
1358 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1359
1360 ;
1361 ; Load up all of the values we are going to use
1362 ;
1363 FLDD 0(a_ptr),a0
1364 FLDD 8(a_ptr),a1
1365 FLDD 16(a_ptr),a2
1366 FLDD 24(a_ptr),a3
1367 FLDD 32(a_ptr),a4
1368 FLDD 40(a_ptr),a5
1369 FLDD 48(a_ptr),a6
1370 FLDD 56(a_ptr),a7
1371
1372 FLDD 0(b_ptr),b0
1373 FLDD 8(b_ptr),b1
1374 FLDD 16(b_ptr),b2
1375 FLDD 24(b_ptr),b3
1376 FLDD 32(b_ptr),b4
1377 FLDD 40(b_ptr),b5
1378 FLDD 48(b_ptr),b6
1379 FLDD 56(b_ptr),b7
1380
1381 MUL_ADD_C a0L,a0R,b0L,b0R,c1,c2,c3
1382 STD c1,0(r_ptr)
1383 COPY %r0,c1
1384
1385 MUL_ADD_C a0L,a0R,b1L,b1R,c2,c3,c1
1386 MUL_ADD_C a1L,a1R,b0L,b0R,c2,c3,c1
1387 STD c2,8(r_ptr)
1388 COPY %r0,c2
1389
1390 MUL_ADD_C a2L,a2R,b0L,b0R,c3,c1,c2
1391 MUL_ADD_C a1L,a1R,b1L,b1R,c3,c1,c2
1392 MUL_ADD_C a0L,a0R,b2L,b2R,c3,c1,c2
1393 STD c3,16(r_ptr)
1394 COPY %r0,c3
1395
1396 MUL_ADD_C a0L,a0R,b3L,b3R,c1,c2,c3
1397 MUL_ADD_C a1L,a1R,b2L,b2R,c1,c2,c3
1398 MUL_ADD_C a2L,a2R,b1L,b1R,c1,c2,c3
1399 MUL_ADD_C a3L,a3R,b0L,b0R,c1,c2,c3
1400 STD c1,24(r_ptr)
1401 COPY %r0,c1
1402
1403 MUL_ADD_C a4L,a4R,b0L,b0R,c2,c3,c1
1404 MUL_ADD_C a3L,a3R,b1L,b1R,c2,c3,c1
1405 MUL_ADD_C a2L,a2R,b2L,b2R,c2,c3,c1
1406 MUL_ADD_C a1L,a1R,b3L,b3R,c2,c3,c1
1407 MUL_ADD_C a0L,a0R,b4L,b4R,c2,c3,c1
1408 STD c2,32(r_ptr)
1409 COPY %r0,c2
1410
1411 MUL_ADD_C a0L,a0R,b5L,b5R,c3,c1,c2
1412 MUL_ADD_C a1L,a1R,b4L,b4R,c3,c1,c2
1413 MUL_ADD_C a2L,a2R,b3L,b3R,c3,c1,c2
1414 MUL_ADD_C a3L,a3R,b2L,b2R,c3,c1,c2
1415 MUL_ADD_C a4L,a4R,b1L,b1R,c3,c1,c2
1416 MUL_ADD_C a5L,a5R,b0L,b0R,c3,c1,c2
1417 STD c3,40(r_ptr)
1418 COPY %r0,c3
1419
1420 MUL_ADD_C a6L,a6R,b0L,b0R,c1,c2,c3
1421 MUL_ADD_C a5L,a5R,b1L,b1R,c1,c2,c3
1422 MUL_ADD_C a4L,a4R,b2L,b2R,c1,c2,c3
1423 MUL_ADD_C a3L,a3R,b3L,b3R,c1,c2,c3
1424 MUL_ADD_C a2L,a2R,b4L,b4R,c1,c2,c3
1425 MUL_ADD_C a1L,a1R,b5L,b5R,c1,c2,c3
1426 MUL_ADD_C a0L,a0R,b6L,b6R,c1,c2,c3
1427 STD c1,48(r_ptr)
1428 COPY %r0,c1
1429
1430 MUL_ADD_C a0L,a0R,b7L,b7R,c2,c3,c1
1431 MUL_ADD_C a1L,a1R,b6L,b6R,c2,c3,c1
1432 MUL_ADD_C a2L,a2R,b5L,b5R,c2,c3,c1
1433 MUL_ADD_C a3L,a3R,b4L,b4R,c2,c3,c1
1434 MUL_ADD_C a4L,a4R,b3L,b3R,c2,c3,c1
1435 MUL_ADD_C a5L,a5R,b2L,b2R,c2,c3,c1
1436 MUL_ADD_C a6L,a6R,b1L,b1R,c2,c3,c1
1437 MUL_ADD_C a7L,a7R,b0L,b0R,c2,c3,c1
1438 STD c2,56(r_ptr)
1439 COPY %r0,c2
1440
1441 MUL_ADD_C a7L,a7R,b1L,b1R,c3,c1,c2
1442 MUL_ADD_C a6L,a6R,b2L,b2R,c3,c1,c2
1443 MUL_ADD_C a5L,a5R,b3L,b3R,c3,c1,c2
1444 MUL_ADD_C a4L,a4R,b4L,b4R,c3,c1,c2
1445 MUL_ADD_C a3L,a3R,b5L,b5R,c3,c1,c2
1446 MUL_ADD_C a2L,a2R,b6L,b6R,c3,c1,c2
1447 MUL_ADD_C a1L,a1R,b7L,b7R,c3,c1,c2
1448 STD c3,64(r_ptr)
1449 COPY %r0,c3
1450
1451 MUL_ADD_C a2L,a2R,b7L,b7R,c1,c2,c3
1452 MUL_ADD_C a3L,a3R,b6L,b6R,c1,c2,c3
1453 MUL_ADD_C a4L,a4R,b5L,b5R,c1,c2,c3
1454 MUL_ADD_C a5L,a5R,b4L,b4R,c1,c2,c3
1455 MUL_ADD_C a6L,a6R,b3L,b3R,c1,c2,c3
1456 MUL_ADD_C a7L,a7R,b2L,b2R,c1,c2,c3
1457 STD c1,72(r_ptr)
1458 COPY %r0,c1
1459
1460 MUL_ADD_C a7L,a7R,b3L,b3R,c2,c3,c1
1461 MUL_ADD_C a6L,a6R,b4L,b4R,c2,c3,c1
1462 MUL_ADD_C a5L,a5R,b5L,b5R,c2,c3,c1
1463 MUL_ADD_C a4L,a4R,b6L,b6R,c2,c3,c1
1464 MUL_ADD_C a3L,a3R,b7L,b7R,c2,c3,c1
1465 STD c2,80(r_ptr)
1466 COPY %r0,c2
1467
1468 MUL_ADD_C a4L,a4R,b7L,b7R,c3,c1,c2
1469 MUL_ADD_C a5L,a5R,b6L,b6R,c3,c1,c2
1470 MUL_ADD_C a6L,a6R,b5L,b5R,c3,c1,c2
1471 MUL_ADD_C a7L,a7R,b4L,b4R,c3,c1,c2
1472 STD c3,88(r_ptr)
1473 COPY %r0,c3
1474
1475 MUL_ADD_C a7L,a7R,b5L,b5R,c1,c2,c3
1476 MUL_ADD_C a6L,a6R,b6L,b6R,c1,c2,c3
1477 MUL_ADD_C a5L,a5R,b7L,b7R,c1,c2,c3
1478 STD c1,96(r_ptr)
1479 COPY %r0,c1
1480
1481 MUL_ADD_C a6L,a6R,b7L,b7R,c2,c3,c1
1482 MUL_ADD_C a7L,a7R,b6L,b6R,c2,c3,c1
1483 STD c2,104(r_ptr)
1484 COPY %r0,c2
1485
1486 MUL_ADD_C a7L,a7R,b7L,b7R,c3,c1,c2
1487 STD c3,112(r_ptr)
1488 STD c1,120(r_ptr)
1489
1490 .EXIT
1491 FLDD -88(%sp),%fr13
1492 FLDD -96(%sp),%fr12
1493 LDD -104(%sp),%r6 ; restore r6
1494 LDD -112(%sp),%r5 ; restore r5
1495 LDD -120(%sp),%r4 ; restore r4
1496 BVE (%rp)
1497 LDD,MB -128(%sp),%r3
1498
1499 .PROCEND
1500
1501;-----------------------------------------------------------------------------
1502;
1503;void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
1504; arg0 = r_ptr
1505; arg1 = a_ptr
1506; arg2 = b_ptr
1507;
1508
1509bn_mul_comba4
1510 .proc
1511 .callinfo FRAME=128,ENTRY_GR=%r3,ARGS_SAVED,ORDERING_AWARE
1512 .EXPORT bn_mul_comba4,ENTRY,PRIV_LEV=3,NO_RELOCATION,LONG_RETURN
1513 .entry
1514 .align 64
1515
1516 STD %r3,0(%sp) ; save r3
1517 STD %r4,8(%sp) ; save r4
1518 STD %r5,16(%sp) ; save r5
1519 STD %r6,24(%sp) ; save r6
1520 FSTD %fr12,32(%sp) ; save r6
1521 FSTD %fr13,40(%sp) ; save r7
1522
1523 ;
1524 ; Zero out carries
1525 ;
1526 COPY %r0,c1
1527 COPY %r0,c2
1528 COPY %r0,c3
1529
1530 LDO 128(%sp),%sp ; bump stack
1531 DEPDI,Z 1,31,1,high_one ; Create Value 1 << 32
1532
1533 ;
1534 ; Load up all of the values we are going to use
1535 ;
1536 FLDD 0(a_ptr),a0
1537 FLDD 8(a_ptr),a1
1538 FLDD 16(a_ptr),a2
1539 FLDD 24(a_ptr),a3
1540
1541 FLDD 0(b_ptr),b0
1542 FLDD 8(b_ptr),b1
1543 FLDD 16(b_ptr),b2
1544 FLDD 24(b_ptr),b3
1545
1546 MUL_ADD_C a0L,a0R,b0L,b0R,c1,c2,c3
1547 STD c1,0(r_ptr)
1548 COPY %r0,c1
1549
1550 MUL_ADD_C a0L,a0R,b1L,b1R,c2,c3,c1
1551 MUL_ADD_C a1L,a1R,b0L,b0R,c2,c3,c1
1552 STD c2,8(r_ptr)
1553 COPY %r0,c2
1554
1555 MUL_ADD_C a2L,a2R,b0L,b0R,c3,c1,c2
1556 MUL_ADD_C a1L,a1R,b1L,b1R,c3,c1,c2
1557 MUL_ADD_C a0L,a0R,b2L,b2R,c3,c1,c2
1558 STD c3,16(r_ptr)
1559 COPY %r0,c3
1560
1561 MUL_ADD_C a0L,a0R,b3L,b3R,c1,c2,c3
1562 MUL_ADD_C a1L,a1R,b2L,b2R,c1,c2,c3
1563 MUL_ADD_C a2L,a2R,b1L,b1R,c1,c2,c3
1564 MUL_ADD_C a3L,a3R,b0L,b0R,c1,c2,c3
1565 STD c1,24(r_ptr)
1566 COPY %r0,c1
1567
1568 MUL_ADD_C a3L,a3R,b1L,b1R,c2,c3,c1
1569 MUL_ADD_C a2L,a2R,b2L,b2R,c2,c3,c1
1570 MUL_ADD_C a1L,a1R,b3L,b3R,c2,c3,c1
1571 STD c2,32(r_ptr)
1572 COPY %r0,c2
1573
1574 MUL_ADD_C a2L,a2R,b3L,b3R,c3,c1,c2
1575 MUL_ADD_C a3L,a3R,b2L,b2R,c3,c1,c2
1576 STD c3,40(r_ptr)
1577 COPY %r0,c3
1578
1579 MUL_ADD_C a3L,a3R,b3L,b3R,c1,c2,c3
1580 STD c1,48(r_ptr)
1581 STD c2,56(r_ptr)
1582
1583 .EXIT
1584 FLDD -88(%sp),%fr13
1585 FLDD -96(%sp),%fr12
1586 LDD -104(%sp),%r6 ; restore r6
1587 LDD -112(%sp),%r5 ; restore r5
1588 LDD -120(%sp),%r4 ; restore r4
1589 BVE (%rp)
1590 LDD,MB -128(%sp),%r3
1591
1592 .PROCEND
1593
1594
1595 .SPACE $TEXT$
1596 .SUBSPA $CODE$
1597 .SPACE $PRIVATE$,SORT=16
1598 .IMPORT $global$,DATA
1599 .SPACE $TEXT$
1600 .SUBSPA $CODE$
1601 .SUBSPA $LIT$,ACCESS=0x2c
1602C$4
1603 .ALIGN 8
1604 .STRINGZ "Division would overflow (%d)\n"
1605 .END
diff --git a/src/lib/libcrypto/bn/asm/sparcv8.S b/src/lib/libcrypto/bn/asm/sparcv8.S
deleted file mode 100644
index 88c5dc480a..0000000000
--- a/src/lib/libcrypto/bn/asm/sparcv8.S
+++ /dev/null
@@ -1,1458 +0,0 @@
1.ident "sparcv8.s, Version 1.4"
2.ident "SPARC v8 ISA artwork by Andy Polyakov <appro@fy.chalmers.se>"
3
4/*
5 * ====================================================================
6 * Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
7 * project.
8 *
9 * Rights for redistribution and usage in source and binary forms are
10 * granted according to the OpenSSL license. Warranty of any kind is
11 * disclaimed.
12 * ====================================================================
13 */
14
15/*
16 * This is my modest contributon to OpenSSL project (see
17 * http://www.openssl.org/ for more information about it) and is
18 * a drop-in SuperSPARC ISA replacement for crypto/bn/bn_asm.c
19 * module. For updates see http://fy.chalmers.se/~appro/hpe/.
20 *
21 * See bn_asm.sparc.v8plus.S for more details.
22 */
23
24/*
25 * Revision history.
26 *
27 * 1.1 - new loop unrolling model(*);
28 * 1.2 - made gas friendly;
29 * 1.3 - fixed problem with /usr/ccs/lib/cpp;
30 * 1.4 - some retunes;
31 *
32 * (*) see bn_asm.sparc.v8plus.S for details
33 */
34
35.section ".text",#alloc,#execinstr
36.file "bn_asm.sparc.v8.S"
37
38.align 32
39
40.global bn_mul_add_words
41/*
42 * BN_ULONG bn_mul_add_words(rp,ap,num,w)
43 * BN_ULONG *rp,*ap;
44 * int num;
45 * BN_ULONG w;
46 */
47bn_mul_add_words:
48 cmp %o2,0
49 bg,a .L_bn_mul_add_words_proceed
50 ld [%o1],%g2
51 retl
52 clr %o0
53
54.L_bn_mul_add_words_proceed:
55 andcc %o2,-4,%g0
56 bz .L_bn_mul_add_words_tail
57 clr %o5
58
59.L_bn_mul_add_words_loop:
60 ld [%o0],%o4
61 ld [%o1+4],%g3
62 umul %o3,%g2,%g2
63 rd %y,%g1
64 addcc %o4,%o5,%o4
65 addx %g1,0,%g1
66 addcc %o4,%g2,%o4
67 st %o4,[%o0]
68 addx %g1,0,%o5
69
70 ld [%o0+4],%o4
71 ld [%o1+8],%g2
72 umul %o3,%g3,%g3
73 dec 4,%o2
74 rd %y,%g1
75 addcc %o4,%o5,%o4
76 addx %g1,0,%g1
77 addcc %o4,%g3,%o4
78 st %o4,[%o0+4]
79 addx %g1,0,%o5
80
81 ld [%o0+8],%o4
82 ld [%o1+12],%g3
83 umul %o3,%g2,%g2
84 inc 16,%o1
85 rd %y,%g1
86 addcc %o4,%o5,%o4
87 addx %g1,0,%g1
88 addcc %o4,%g2,%o4
89 st %o4,[%o0+8]
90 addx %g1,0,%o5
91
92 ld [%o0+12],%o4
93 umul %o3,%g3,%g3
94 inc 16,%o0
95 rd %y,%g1
96 addcc %o4,%o5,%o4
97 addx %g1,0,%g1
98 addcc %o4,%g3,%o4
99 st %o4,[%o0-4]
100 addx %g1,0,%o5
101 andcc %o2,-4,%g0
102 bnz,a .L_bn_mul_add_words_loop
103 ld [%o1],%g2
104
105 tst %o2
106 bnz,a .L_bn_mul_add_words_tail
107 ld [%o1],%g2
108.L_bn_mul_add_words_return:
109 retl
110 mov %o5,%o0
111 nop
112
113.L_bn_mul_add_words_tail:
114 ld [%o0],%o4
115 umul %o3,%g2,%g2
116 addcc %o4,%o5,%o4
117 rd %y,%g1
118 addx %g1,0,%g1
119 addcc %o4,%g2,%o4
120 addx %g1,0,%o5
121 deccc %o2
122 bz .L_bn_mul_add_words_return
123 st %o4,[%o0]
124
125 ld [%o1+4],%g2
126 ld [%o0+4],%o4
127 umul %o3,%g2,%g2
128 rd %y,%g1
129 addcc %o4,%o5,%o4
130 addx %g1,0,%g1
131 addcc %o4,%g2,%o4
132 addx %g1,0,%o5
133 deccc %o2
134 bz .L_bn_mul_add_words_return
135 st %o4,[%o0+4]
136
137 ld [%o1+8],%g2
138 ld [%o0+8],%o4
139 umul %o3,%g2,%g2
140 rd %y,%g1
141 addcc %o4,%o5,%o4
142 addx %g1,0,%g1
143 addcc %o4,%g2,%o4
144 st %o4,[%o0+8]
145 retl
146 addx %g1,0,%o0
147
148.type bn_mul_add_words,#function
149.size bn_mul_add_words,(.-bn_mul_add_words)
150
151.align 32
152
153.global bn_mul_words
154/*
155 * BN_ULONG bn_mul_words(rp,ap,num,w)
156 * BN_ULONG *rp,*ap;
157 * int num;
158 * BN_ULONG w;
159 */
160bn_mul_words:
161 cmp %o2,0
162 bg,a .L_bn_mul_words_proceeed
163 ld [%o1],%g2
164 retl
165 clr %o0
166
167.L_bn_mul_words_proceeed:
168 andcc %o2,-4,%g0
169 bz .L_bn_mul_words_tail
170 clr %o5
171
172.L_bn_mul_words_loop:
173 ld [%o1+4],%g3
174 umul %o3,%g2,%g2
175 addcc %g2,%o5,%g2
176 rd %y,%g1
177 addx %g1,0,%o5
178 st %g2,[%o0]
179
180 ld [%o1+8],%g2
181 umul %o3,%g3,%g3
182 addcc %g3,%o5,%g3
183 rd %y,%g1
184 dec 4,%o2
185 addx %g1,0,%o5
186 st %g3,[%o0+4]
187
188 ld [%o1+12],%g3
189 umul %o3,%g2,%g2
190 addcc %g2,%o5,%g2
191 rd %y,%g1
192 inc 16,%o1
193 st %g2,[%o0+8]
194 addx %g1,0,%o5
195
196 umul %o3,%g3,%g3
197 addcc %g3,%o5,%g3
198 rd %y,%g1
199 inc 16,%o0
200 addx %g1,0,%o5
201 st %g3,[%o0-4]
202 andcc %o2,-4,%g0
203 nop
204 bnz,a .L_bn_mul_words_loop
205 ld [%o1],%g2
206
207 tst %o2
208 bnz,a .L_bn_mul_words_tail
209 ld [%o1],%g2
210.L_bn_mul_words_return:
211 retl
212 mov %o5,%o0
213 nop
214
215.L_bn_mul_words_tail:
216 umul %o3,%g2,%g2
217 addcc %g2,%o5,%g2
218 rd %y,%g1
219 addx %g1,0,%o5
220 deccc %o2
221 bz .L_bn_mul_words_return
222 st %g2,[%o0]
223 nop
224
225 ld [%o1+4],%g2
226 umul %o3,%g2,%g2
227 addcc %g2,%o5,%g2
228 rd %y,%g1
229 addx %g1,0,%o5
230 deccc %o2
231 bz .L_bn_mul_words_return
232 st %g2,[%o0+4]
233
234 ld [%o1+8],%g2
235 umul %o3,%g2,%g2
236 addcc %g2,%o5,%g2
237 rd %y,%g1
238 st %g2,[%o0+8]
239 retl
240 addx %g1,0,%o0
241
242.type bn_mul_words,#function
243.size bn_mul_words,(.-bn_mul_words)
244
245.align 32
246.global bn_sqr_words
247/*
248 * void bn_sqr_words(r,a,n)
249 * BN_ULONG *r,*a;
250 * int n;
251 */
252bn_sqr_words:
253 cmp %o2,0
254 bg,a .L_bn_sqr_words_proceeed
255 ld [%o1],%g2
256 retl
257 clr %o0
258
259.L_bn_sqr_words_proceeed:
260 andcc %o2,-4,%g0
261 bz .L_bn_sqr_words_tail
262 clr %o5
263
264.L_bn_sqr_words_loop:
265 ld [%o1+4],%g3
266 umul %g2,%g2,%o4
267 st %o4,[%o0]
268 rd %y,%o5
269 st %o5,[%o0+4]
270
271 ld [%o1+8],%g2
272 umul %g3,%g3,%o4
273 dec 4,%o2
274 st %o4,[%o0+8]
275 rd %y,%o5
276 st %o5,[%o0+12]
277 nop
278
279 ld [%o1+12],%g3
280 umul %g2,%g2,%o4
281 st %o4,[%o0+16]
282 rd %y,%o5
283 inc 16,%o1
284 st %o5,[%o0+20]
285
286 umul %g3,%g3,%o4
287 inc 32,%o0
288 st %o4,[%o0-8]
289 rd %y,%o5
290 st %o5,[%o0-4]
291 andcc %o2,-4,%g2
292 bnz,a .L_bn_sqr_words_loop
293 ld [%o1],%g2
294
295 tst %o2
296 nop
297 bnz,a .L_bn_sqr_words_tail
298 ld [%o1],%g2
299.L_bn_sqr_words_return:
300 retl
301 clr %o0
302
303.L_bn_sqr_words_tail:
304 umul %g2,%g2,%o4
305 st %o4,[%o0]
306 deccc %o2
307 rd %y,%o5
308 bz .L_bn_sqr_words_return
309 st %o5,[%o0+4]
310
311 ld [%o1+4],%g2
312 umul %g2,%g2,%o4
313 st %o4,[%o0+8]
314 deccc %o2
315 rd %y,%o5
316 nop
317 bz .L_bn_sqr_words_return
318 st %o5,[%o0+12]
319
320 ld [%o1+8],%g2
321 umul %g2,%g2,%o4
322 st %o4,[%o0+16]
323 rd %y,%o5
324 st %o5,[%o0+20]
325 retl
326 clr %o0
327
328.type bn_sqr_words,#function
329.size bn_sqr_words,(.-bn_sqr_words)
330
331.align 32
332
333.global bn_div_words
334/*
335 * BN_ULONG bn_div_words(h,l,d)
336 * BN_ULONG h,l,d;
337 */
338bn_div_words:
339 wr %o0,%y
340 udiv %o1,%o2,%o0
341 retl
342 nop
343
344.type bn_div_words,#function
345.size bn_div_words,(.-bn_div_words)
346
347.align 32
348
349.global bn_add_words
350/*
351 * BN_ULONG bn_add_words(rp,ap,bp,n)
352 * BN_ULONG *rp,*ap,*bp;
353 * int n;
354 */
355bn_add_words:
356 cmp %o3,0
357 bg,a .L_bn_add_words_proceed
358 ld [%o1],%o4
359 retl
360 clr %o0
361
362.L_bn_add_words_proceed:
363 andcc %o3,-4,%g0
364 bz .L_bn_add_words_tail
365 clr %g1
366 ba .L_bn_add_words_warn_loop
367 addcc %g0,0,%g0 ! clear carry flag
368
369.L_bn_add_words_loop:
370 ld [%o1],%o4
371.L_bn_add_words_warn_loop:
372 ld [%o2],%o5
373 ld [%o1+4],%g3
374 ld [%o2+4],%g4
375 dec 4,%o3
376 addxcc %o5,%o4,%o5
377 st %o5,[%o0]
378
379 ld [%o1+8],%o4
380 ld [%o2+8],%o5
381 inc 16,%o1
382 addxcc %g3,%g4,%g3
383 st %g3,[%o0+4]
384
385 ld [%o1-4],%g3
386 ld [%o2+12],%g4
387 inc 16,%o2
388 addxcc %o5,%o4,%o5
389 st %o5,[%o0+8]
390
391 inc 16,%o0
392 addxcc %g3,%g4,%g3
393 st %g3,[%o0-4]
394 addx %g0,0,%g1
395 andcc %o3,-4,%g0
396 bnz,a .L_bn_add_words_loop
397 addcc %g1,-1,%g0
398
399 tst %o3
400 bnz,a .L_bn_add_words_tail
401 ld [%o1],%o4
402.L_bn_add_words_return:
403 retl
404 mov %g1,%o0
405
406.L_bn_add_words_tail:
407 addcc %g1,-1,%g0
408 ld [%o2],%o5
409 addxcc %o5,%o4,%o5
410 addx %g0,0,%g1
411 deccc %o3
412 bz .L_bn_add_words_return
413 st %o5,[%o0]
414
415 ld [%o1+4],%o4
416 addcc %g1,-1,%g0
417 ld [%o2+4],%o5
418 addxcc %o5,%o4,%o5
419 addx %g0,0,%g1
420 deccc %o3
421 bz .L_bn_add_words_return
422 st %o5,[%o0+4]
423
424 ld [%o1+8],%o4
425 addcc %g1,-1,%g0
426 ld [%o2+8],%o5
427 addxcc %o5,%o4,%o5
428 st %o5,[%o0+8]
429 retl
430 addx %g0,0,%o0
431
432.type bn_add_words,#function
433.size bn_add_words,(.-bn_add_words)
434
435.align 32
436
437.global bn_sub_words
438/*
439 * BN_ULONG bn_sub_words(rp,ap,bp,n)
440 * BN_ULONG *rp,*ap,*bp;
441 * int n;
442 */
443bn_sub_words:
444 cmp %o3,0
445 bg,a .L_bn_sub_words_proceed
446 ld [%o1],%o4
447 retl
448 clr %o0
449
450.L_bn_sub_words_proceed:
451 andcc %o3,-4,%g0
452 bz .L_bn_sub_words_tail
453 clr %g1
454 ba .L_bn_sub_words_warm_loop
455 addcc %g0,0,%g0 ! clear carry flag
456
457.L_bn_sub_words_loop:
458 ld [%o1],%o4
459.L_bn_sub_words_warm_loop:
460 ld [%o2],%o5
461 ld [%o1+4],%g3
462 ld [%o2+4],%g4
463 dec 4,%o3
464 subxcc %o4,%o5,%o5
465 st %o5,[%o0]
466
467 ld [%o1+8],%o4
468 ld [%o2+8],%o5
469 inc 16,%o1
470 subxcc %g3,%g4,%g4
471 st %g4,[%o0+4]
472
473 ld [%o1-4],%g3
474 ld [%o2+12],%g4
475 inc 16,%o2
476 subxcc %o4,%o5,%o5
477 st %o5,[%o0+8]
478
479 inc 16,%o0
480 subxcc %g3,%g4,%g4
481 st %g4,[%o0-4]
482 addx %g0,0,%g1
483 andcc %o3,-4,%g0
484 bnz,a .L_bn_sub_words_loop
485 addcc %g1,-1,%g0
486
487 tst %o3
488 nop
489 bnz,a .L_bn_sub_words_tail
490 ld [%o1],%o4
491.L_bn_sub_words_return:
492 retl
493 mov %g1,%o0
494
495.L_bn_sub_words_tail:
496 addcc %g1,-1,%g0
497 ld [%o2],%o5
498 subxcc %o4,%o5,%o5
499 addx %g0,0,%g1
500 deccc %o3
501 bz .L_bn_sub_words_return
502 st %o5,[%o0]
503 nop
504
505 ld [%o1+4],%o4
506 addcc %g1,-1,%g0
507 ld [%o2+4],%o5
508 subxcc %o4,%o5,%o5
509 addx %g0,0,%g1
510 deccc %o3
511 bz .L_bn_sub_words_return
512 st %o5,[%o0+4]
513
514 ld [%o1+8],%o4
515 addcc %g1,-1,%g0
516 ld [%o2+8],%o5
517 subxcc %o4,%o5,%o5
518 st %o5,[%o0+8]
519 retl
520 addx %g0,0,%o0
521
522.type bn_sub_words,#function
523.size bn_sub_words,(.-bn_sub_words)
524
525#define FRAME_SIZE -96
526
527/*
528 * Here is register usage map for *all* routines below.
529 */
530#define t_1 %o0
531#define t_2 %o1
532#define c_1 %o2
533#define c_2 %o3
534#define c_3 %o4
535
536#define ap(I) [%i1+4*I]
537#define bp(I) [%i2+4*I]
538#define rp(I) [%i0+4*I]
539
540#define a_0 %l0
541#define a_1 %l1
542#define a_2 %l2
543#define a_3 %l3
544#define a_4 %l4
545#define a_5 %l5
546#define a_6 %l6
547#define a_7 %l7
548
549#define b_0 %i3
550#define b_1 %i4
551#define b_2 %i5
552#define b_3 %o5
553#define b_4 %g1
554#define b_5 %g2
555#define b_6 %g3
556#define b_7 %g4
557
558.align 32
559.global bn_mul_comba8
560/*
561 * void bn_mul_comba8(r,a,b)
562 * BN_ULONG *r,*a,*b;
563 */
564bn_mul_comba8:
565 save %sp,FRAME_SIZE,%sp
566 ld ap(0),a_0
567 ld bp(0),b_0
568 umul a_0,b_0,c_1 !=!mul_add_c(a[0],b[0],c1,c2,c3);
569 ld bp(1),b_1
570 rd %y,c_2
571 st c_1,rp(0) !r[0]=c1;
572
573 umul a_0,b_1,t_1 !=!mul_add_c(a[0],b[1],c2,c3,c1);
574 ld ap(1),a_1
575 addcc c_2,t_1,c_2
576 rd %y,t_2
577 addxcc %g0,t_2,c_3 !=
578 addx %g0,%g0,c_1
579 ld ap(2),a_2
580 umul a_1,b_0,t_1 !mul_add_c(a[1],b[0],c2,c3,c1);
581 addcc c_2,t_1,c_2 !=
582 rd %y,t_2
583 addxcc c_3,t_2,c_3
584 st c_2,rp(1) !r[1]=c2;
585 addx c_1,%g0,c_1 !=
586
587 umul a_2,b_0,t_1 !mul_add_c(a[2],b[0],c3,c1,c2);
588 addcc c_3,t_1,c_3
589 rd %y,t_2
590 addxcc c_1,t_2,c_1 !=
591 addx %g0,%g0,c_2
592 ld bp(2),b_2
593 umul a_1,b_1,t_1 !mul_add_c(a[1],b[1],c3,c1,c2);
594 addcc c_3,t_1,c_3 !=
595 rd %y,t_2
596 addxcc c_1,t_2,c_1
597 ld bp(3),b_3
598 addx c_2,%g0,c_2 !=
599 umul a_0,b_2,t_1 !mul_add_c(a[0],b[2],c3,c1,c2);
600 addcc c_3,t_1,c_3
601 rd %y,t_2
602 addxcc c_1,t_2,c_1 !=
603 addx c_2,%g0,c_2
604 st c_3,rp(2) !r[2]=c3;
605
606 umul a_0,b_3,t_1 !mul_add_c(a[0],b[3],c1,c2,c3);
607 addcc c_1,t_1,c_1 !=
608 rd %y,t_2
609 addxcc c_2,t_2,c_2
610 addx %g0,%g0,c_3
611 umul a_1,b_2,t_1 !=!mul_add_c(a[1],b[2],c1,c2,c3);
612 addcc c_1,t_1,c_1
613 rd %y,t_2
614 addxcc c_2,t_2,c_2
615 addx c_3,%g0,c_3 !=
616 ld ap(3),a_3
617 umul a_2,b_1,t_1 !mul_add_c(a[2],b[1],c1,c2,c3);
618 addcc c_1,t_1,c_1
619 rd %y,t_2 !=
620 addxcc c_2,t_2,c_2
621 addx c_3,%g0,c_3
622 ld ap(4),a_4
623 umul a_3,b_0,t_1 !mul_add_c(a[3],b[0],c1,c2,c3);!=
624 addcc c_1,t_1,c_1
625 rd %y,t_2
626 addxcc c_2,t_2,c_2
627 addx c_3,%g0,c_3 !=
628 st c_1,rp(3) !r[3]=c1;
629
630 umul a_4,b_0,t_1 !mul_add_c(a[4],b[0],c2,c3,c1);
631 addcc c_2,t_1,c_2
632 rd %y,t_2 !=
633 addxcc c_3,t_2,c_3
634 addx %g0,%g0,c_1
635 umul a_3,b_1,t_1 !mul_add_c(a[3],b[1],c2,c3,c1);
636 addcc c_2,t_1,c_2 !=
637 rd %y,t_2
638 addxcc c_3,t_2,c_3
639 addx c_1,%g0,c_1
640 umul a_2,b_2,t_1 !=!mul_add_c(a[2],b[2],c2,c3,c1);
641 addcc c_2,t_1,c_2
642 rd %y,t_2
643 addxcc c_3,t_2,c_3
644 addx c_1,%g0,c_1 !=
645 ld bp(4),b_4
646 umul a_1,b_3,t_1 !mul_add_c(a[1],b[3],c2,c3,c1);
647 addcc c_2,t_1,c_2
648 rd %y,t_2 !=
649 addxcc c_3,t_2,c_3
650 addx c_1,%g0,c_1
651 ld bp(5),b_5
652 umul a_0,b_4,t_1 !=!mul_add_c(a[0],b[4],c2,c3,c1);
653 addcc c_2,t_1,c_2
654 rd %y,t_2
655 addxcc c_3,t_2,c_3
656 addx c_1,%g0,c_1 !=
657 st c_2,rp(4) !r[4]=c2;
658
659 umul a_0,b_5,t_1 !mul_add_c(a[0],b[5],c3,c1,c2);
660 addcc c_3,t_1,c_3
661 rd %y,t_2 !=
662 addxcc c_1,t_2,c_1
663 addx %g0,%g0,c_2
664 umul a_1,b_4,t_1 !mul_add_c(a[1],b[4],c3,c1,c2);
665 addcc c_3,t_1,c_3 !=
666 rd %y,t_2
667 addxcc c_1,t_2,c_1
668 addx c_2,%g0,c_2
669 umul a_2,b_3,t_1 !=!mul_add_c(a[2],b[3],c3,c1,c2);
670 addcc c_3,t_1,c_3
671 rd %y,t_2
672 addxcc c_1,t_2,c_1
673 addx c_2,%g0,c_2 !=
674 umul a_3,b_2,t_1 !mul_add_c(a[3],b[2],c3,c1,c2);
675 addcc c_3,t_1,c_3
676 rd %y,t_2
677 addxcc c_1,t_2,c_1 !=
678 addx c_2,%g0,c_2
679 ld ap(5),a_5
680 umul a_4,b_1,t_1 !mul_add_c(a[4],b[1],c3,c1,c2);
681 addcc c_3,t_1,c_3 !=
682 rd %y,t_2
683 addxcc c_1,t_2,c_1
684 ld ap(6),a_6
685 addx c_2,%g0,c_2 !=
686 umul a_5,b_0,t_1 !mul_add_c(a[5],b[0],c3,c1,c2);
687 addcc c_3,t_1,c_3
688 rd %y,t_2
689 addxcc c_1,t_2,c_1 !=
690 addx c_2,%g0,c_2
691 st c_3,rp(5) !r[5]=c3;
692
693 umul a_6,b_0,t_1 !mul_add_c(a[6],b[0],c1,c2,c3);
694 addcc c_1,t_1,c_1 !=
695 rd %y,t_2
696 addxcc c_2,t_2,c_2
697 addx %g0,%g0,c_3
698 umul a_5,b_1,t_1 !=!mul_add_c(a[5],b[1],c1,c2,c3);
699 addcc c_1,t_1,c_1
700 rd %y,t_2
701 addxcc c_2,t_2,c_2
702 addx c_3,%g0,c_3 !=
703 umul a_4,b_2,t_1 !mul_add_c(a[4],b[2],c1,c2,c3);
704 addcc c_1,t_1,c_1
705 rd %y,t_2
706 addxcc c_2,t_2,c_2 !=
707 addx c_3,%g0,c_3
708 umul a_3,b_3,t_1 !mul_add_c(a[3],b[3],c1,c2,c3);
709 addcc c_1,t_1,c_1
710 rd %y,t_2 !=
711 addxcc c_2,t_2,c_2
712 addx c_3,%g0,c_3
713 umul a_2,b_4,t_1 !mul_add_c(a[2],b[4],c1,c2,c3);
714 addcc c_1,t_1,c_1 !=
715 rd %y,t_2
716 addxcc c_2,t_2,c_2
717 ld bp(6),b_6
718 addx c_3,%g0,c_3 !=
719 umul a_1,b_5,t_1 !mul_add_c(a[1],b[5],c1,c2,c3);
720 addcc c_1,t_1,c_1
721 rd %y,t_2
722 addxcc c_2,t_2,c_2 !=
723 addx c_3,%g0,c_3
724 ld bp(7),b_7
725 umul a_0,b_6,t_1 !mul_add_c(a[0],b[6],c1,c2,c3);
726 addcc c_1,t_1,c_1 !=
727 rd %y,t_2
728 addxcc c_2,t_2,c_2
729 st c_1,rp(6) !r[6]=c1;
730 addx c_3,%g0,c_3 !=
731
732 umul a_0,b_7,t_1 !mul_add_c(a[0],b[7],c2,c3,c1);
733 addcc c_2,t_1,c_2
734 rd %y,t_2
735 addxcc c_3,t_2,c_3 !=
736 addx %g0,%g0,c_1
737 umul a_1,b_6,t_1 !mul_add_c(a[1],b[6],c2,c3,c1);
738 addcc c_2,t_1,c_2
739 rd %y,t_2 !=
740 addxcc c_3,t_2,c_3
741 addx c_1,%g0,c_1
742 umul a_2,b_5,t_1 !mul_add_c(a[2],b[5],c2,c3,c1);
743 addcc c_2,t_1,c_2 !=
744 rd %y,t_2
745 addxcc c_3,t_2,c_3
746 addx c_1,%g0,c_1
747 umul a_3,b_4,t_1 !=!mul_add_c(a[3],b[4],c2,c3,c1);
748 addcc c_2,t_1,c_2
749 rd %y,t_2
750 addxcc c_3,t_2,c_3
751 addx c_1,%g0,c_1 !=
752 umul a_4,b_3,t_1 !mul_add_c(a[4],b[3],c2,c3,c1);
753 addcc c_2,t_1,c_2
754 rd %y,t_2
755 addxcc c_3,t_2,c_3 !=
756 addx c_1,%g0,c_1
757 umul a_5,b_2,t_1 !mul_add_c(a[5],b[2],c2,c3,c1);
758 addcc c_2,t_1,c_2
759 rd %y,t_2 !=
760 addxcc c_3,t_2,c_3
761 addx c_1,%g0,c_1
762 ld ap(7),a_7
763 umul a_6,b_1,t_1 !=!mul_add_c(a[6],b[1],c2,c3,c1);
764 addcc c_2,t_1,c_2
765 rd %y,t_2
766 addxcc c_3,t_2,c_3
767 addx c_1,%g0,c_1 !=
768 umul a_7,b_0,t_1 !mul_add_c(a[7],b[0],c2,c3,c1);
769 addcc c_2,t_1,c_2
770 rd %y,t_2
771 addxcc c_3,t_2,c_3 !=
772 addx c_1,%g0,c_1
773 st c_2,rp(7) !r[7]=c2;
774
775 umul a_7,b_1,t_1 !mul_add_c(a[7],b[1],c3,c1,c2);
776 addcc c_3,t_1,c_3 !=
777 rd %y,t_2
778 addxcc c_1,t_2,c_1
779 addx %g0,%g0,c_2
780 umul a_6,b_2,t_1 !=!mul_add_c(a[6],b[2],c3,c1,c2);
781 addcc c_3,t_1,c_3
782 rd %y,t_2
783 addxcc c_1,t_2,c_1
784 addx c_2,%g0,c_2 !=
785 umul a_5,b_3,t_1 !mul_add_c(a[5],b[3],c3,c1,c2);
786 addcc c_3,t_1,c_3
787 rd %y,t_2
788 addxcc c_1,t_2,c_1 !=
789 addx c_2,%g0,c_2
790 umul a_4,b_4,t_1 !mul_add_c(a[4],b[4],c3,c1,c2);
791 addcc c_3,t_1,c_3
792 rd %y,t_2 !=
793 addxcc c_1,t_2,c_1
794 addx c_2,%g0,c_2
795 umul a_3,b_5,t_1 !mul_add_c(a[3],b[5],c3,c1,c2);
796 addcc c_3,t_1,c_3 !=
797 rd %y,t_2
798 addxcc c_1,t_2,c_1
799 addx c_2,%g0,c_2
800 umul a_2,b_6,t_1 !=!mul_add_c(a[2],b[6],c3,c1,c2);
801 addcc c_3,t_1,c_3
802 rd %y,t_2
803 addxcc c_1,t_2,c_1
804 addx c_2,%g0,c_2 !=
805 umul a_1,b_7,t_1 !mul_add_c(a[1],b[7],c3,c1,c2);
806 addcc c_3,t_1,c_3
807 rd %y,t_2
808 addxcc c_1,t_2,c_1 !
809 addx c_2,%g0,c_2
810 st c_3,rp(8) !r[8]=c3;
811
812 umul a_2,b_7,t_1 !mul_add_c(a[2],b[7],c1,c2,c3);
813 addcc c_1,t_1,c_1 !=
814 rd %y,t_2
815 addxcc c_2,t_2,c_2
816 addx %g0,%g0,c_3
817 umul a_3,b_6,t_1 !=!mul_add_c(a[3],b[6],c1,c2,c3);
818 addcc c_1,t_1,c_1
819 rd %y,t_2
820 addxcc c_2,t_2,c_2
821 addx c_3,%g0,c_3 !=
822 umul a_4,b_5,t_1 !mul_add_c(a[4],b[5],c1,c2,c3);
823 addcc c_1,t_1,c_1
824 rd %y,t_2
825 addxcc c_2,t_2,c_2 !=
826 addx c_3,%g0,c_3
827 umul a_5,b_4,t_1 !mul_add_c(a[5],b[4],c1,c2,c3);
828 addcc c_1,t_1,c_1
829 rd %y,t_2 !=
830 addxcc c_2,t_2,c_2
831 addx c_3,%g0,c_3
832 umul a_6,b_3,t_1 !mul_add_c(a[6],b[3],c1,c2,c3);
833 addcc c_1,t_1,c_1 !=
834 rd %y,t_2
835 addxcc c_2,t_2,c_2
836 addx c_3,%g0,c_3
837 umul a_7,b_2,t_1 !=!mul_add_c(a[7],b[2],c1,c2,c3);
838 addcc c_1,t_1,c_1
839 rd %y,t_2
840 addxcc c_2,t_2,c_2
841 addx c_3,%g0,c_3 !=
842 st c_1,rp(9) !r[9]=c1;
843
844 umul a_7,b_3,t_1 !mul_add_c(a[7],b[3],c2,c3,c1);
845 addcc c_2,t_1,c_2
846 rd %y,t_2 !=
847 addxcc c_3,t_2,c_3
848 addx %g0,%g0,c_1
849 umul a_6,b_4,t_1 !mul_add_c(a[6],b[4],c2,c3,c1);
850 addcc c_2,t_1,c_2 !=
851 rd %y,t_2
852 addxcc c_3,t_2,c_3
853 addx c_1,%g0,c_1
854 umul a_5,b_5,t_1 !=!mul_add_c(a[5],b[5],c2,c3,c1);
855 addcc c_2,t_1,c_2
856 rd %y,t_2
857 addxcc c_3,t_2,c_3
858 addx c_1,%g0,c_1 !=
859 umul a_4,b_6,t_1 !mul_add_c(a[4],b[6],c2,c3,c1);
860 addcc c_2,t_1,c_2
861 rd %y,t_2
862 addxcc c_3,t_2,c_3 !=
863 addx c_1,%g0,c_1
864 umul a_3,b_7,t_1 !mul_add_c(a[3],b[7],c2,c3,c1);
865 addcc c_2,t_1,c_2
866 rd %y,t_2 !=
867 addxcc c_3,t_2,c_3
868 addx c_1,%g0,c_1
869 st c_2,rp(10) !r[10]=c2;
870
871 umul a_4,b_7,t_1 !=!mul_add_c(a[4],b[7],c3,c1,c2);
872 addcc c_3,t_1,c_3
873 rd %y,t_2
874 addxcc c_1,t_2,c_1
875 addx %g0,%g0,c_2 !=
876 umul a_5,b_6,t_1 !mul_add_c(a[5],b[6],c3,c1,c2);
877 addcc c_3,t_1,c_3
878 rd %y,t_2
879 addxcc c_1,t_2,c_1 !=
880 addx c_2,%g0,c_2
881 umul a_6,b_5,t_1 !mul_add_c(a[6],b[5],c3,c1,c2);
882 addcc c_3,t_1,c_3
883 rd %y,t_2 !=
884 addxcc c_1,t_2,c_1
885 addx c_2,%g0,c_2
886 umul a_7,b_4,t_1 !mul_add_c(a[7],b[4],c3,c1,c2);
887 addcc c_3,t_1,c_3 !=
888 rd %y,t_2
889 addxcc c_1,t_2,c_1
890 st c_3,rp(11) !r[11]=c3;
891 addx c_2,%g0,c_2 !=
892
893 umul a_7,b_5,t_1 !mul_add_c(a[7],b[5],c1,c2,c3);
894 addcc c_1,t_1,c_1
895 rd %y,t_2
896 addxcc c_2,t_2,c_2 !=
897 addx %g0,%g0,c_3
898 umul a_6,b_6,t_1 !mul_add_c(a[6],b[6],c1,c2,c3);
899 addcc c_1,t_1,c_1
900 rd %y,t_2 !=
901 addxcc c_2,t_2,c_2
902 addx c_3,%g0,c_3
903 umul a_5,b_7,t_1 !mul_add_c(a[5],b[7],c1,c2,c3);
904 addcc c_1,t_1,c_1 !=
905 rd %y,t_2
906 addxcc c_2,t_2,c_2
907 st c_1,rp(12) !r[12]=c1;
908 addx c_3,%g0,c_3 !=
909
910 umul a_6,b_7,t_1 !mul_add_c(a[6],b[7],c2,c3,c1);
911 addcc c_2,t_1,c_2
912 rd %y,t_2
913 addxcc c_3,t_2,c_3 !=
914 addx %g0,%g0,c_1
915 umul a_7,b_6,t_1 !mul_add_c(a[7],b[6],c2,c3,c1);
916 addcc c_2,t_1,c_2
917 rd %y,t_2 !=
918 addxcc c_3,t_2,c_3
919 addx c_1,%g0,c_1
920 st c_2,rp(13) !r[13]=c2;
921
922 umul a_7,b_7,t_1 !=!mul_add_c(a[7],b[7],c3,c1,c2);
923 addcc c_3,t_1,c_3
924 rd %y,t_2
925 addxcc c_1,t_2,c_1
926 nop !=
927 st c_3,rp(14) !r[14]=c3;
928 st c_1,rp(15) !r[15]=c1;
929
930 ret
931 restore %g0,%g0,%o0
932
933.type bn_mul_comba8,#function
934.size bn_mul_comba8,(.-bn_mul_comba8)
935
936.align 32
937
938.global bn_mul_comba4
939/*
940 * void bn_mul_comba4(r,a,b)
941 * BN_ULONG *r,*a,*b;
942 */
943bn_mul_comba4:
944 save %sp,FRAME_SIZE,%sp
945 ld ap(0),a_0
946 ld bp(0),b_0
947 umul a_0,b_0,c_1 !=!mul_add_c(a[0],b[0],c1,c2,c3);
948 ld bp(1),b_1
949 rd %y,c_2
950 st c_1,rp(0) !r[0]=c1;
951
952 umul a_0,b_1,t_1 !=!mul_add_c(a[0],b[1],c2,c3,c1);
953 ld ap(1),a_1
954 addcc c_2,t_1,c_2
955 rd %y,t_2 !=
956 addxcc %g0,t_2,c_3
957 addx %g0,%g0,c_1
958 ld ap(2),a_2
959 umul a_1,b_0,t_1 !=!mul_add_c(a[1],b[0],c2,c3,c1);
960 addcc c_2,t_1,c_2
961 rd %y,t_2
962 addxcc c_3,t_2,c_3
963 addx c_1,%g0,c_1 !=
964 st c_2,rp(1) !r[1]=c2;
965
966 umul a_2,b_0,t_1 !mul_add_c(a[2],b[0],c3,c1,c2);
967 addcc c_3,t_1,c_3
968 rd %y,t_2 !=
969 addxcc c_1,t_2,c_1
970 addx %g0,%g0,c_2
971 ld bp(2),b_2
972 umul a_1,b_1,t_1 !=!mul_add_c(a[1],b[1],c3,c1,c2);
973 addcc c_3,t_1,c_3
974 rd %y,t_2
975 addxcc c_1,t_2,c_1
976 addx c_2,%g0,c_2 !=
977 ld bp(3),b_3
978 umul a_0,b_2,t_1 !mul_add_c(a[0],b[2],c3,c1,c2);
979 addcc c_3,t_1,c_3
980 rd %y,t_2 !=
981 addxcc c_1,t_2,c_1
982 addx c_2,%g0,c_2
983 st c_3,rp(2) !r[2]=c3;
984
985 umul a_0,b_3,t_1 !=!mul_add_c(a[0],b[3],c1,c2,c3);
986 addcc c_1,t_1,c_1
987 rd %y,t_2
988 addxcc c_2,t_2,c_2
989 addx %g0,%g0,c_3 !=
990 umul a_1,b_2,t_1 !mul_add_c(a[1],b[2],c1,c2,c3);
991 addcc c_1,t_1,c_1
992 rd %y,t_2
993 addxcc c_2,t_2,c_2 !=
994 addx c_3,%g0,c_3
995 ld ap(3),a_3
996 umul a_2,b_1,t_1 !mul_add_c(a[2],b[1],c1,c2,c3);
997 addcc c_1,t_1,c_1 !=
998 rd %y,t_2
999 addxcc c_2,t_2,c_2
1000 addx c_3,%g0,c_3
1001 umul a_3,b_0,t_1 !=!mul_add_c(a[3],b[0],c1,c2,c3);
1002 addcc c_1,t_1,c_1
1003 rd %y,t_2
1004 addxcc c_2,t_2,c_2
1005 addx c_3,%g0,c_3 !=
1006 st c_1,rp(3) !r[3]=c1;
1007
1008 umul a_3,b_1,t_1 !mul_add_c(a[3],b[1],c2,c3,c1);
1009 addcc c_2,t_1,c_2
1010 rd %y,t_2 !=
1011 addxcc c_3,t_2,c_3
1012 addx %g0,%g0,c_1
1013 umul a_2,b_2,t_1 !mul_add_c(a[2],b[2],c2,c3,c1);
1014 addcc c_2,t_1,c_2 !=
1015 rd %y,t_2
1016 addxcc c_3,t_2,c_3
1017 addx c_1,%g0,c_1
1018 umul a_1,b_3,t_1 !=!mul_add_c(a[1],b[3],c2,c3,c1);
1019 addcc c_2,t_1,c_2
1020 rd %y,t_2
1021 addxcc c_3,t_2,c_3
1022 addx c_1,%g0,c_1 !=
1023 st c_2,rp(4) !r[4]=c2;
1024
1025 umul a_2,b_3,t_1 !mul_add_c(a[2],b[3],c3,c1,c2);
1026 addcc c_3,t_1,c_3
1027 rd %y,t_2 !=
1028 addxcc c_1,t_2,c_1
1029 addx %g0,%g0,c_2
1030 umul a_3,b_2,t_1 !mul_add_c(a[3],b[2],c3,c1,c2);
1031 addcc c_3,t_1,c_3 !=
1032 rd %y,t_2
1033 addxcc c_1,t_2,c_1
1034 st c_3,rp(5) !r[5]=c3;
1035 addx c_2,%g0,c_2 !=
1036
1037 umul a_3,b_3,t_1 !mul_add_c(a[3],b[3],c1,c2,c3);
1038 addcc c_1,t_1,c_1
1039 rd %y,t_2
1040 addxcc c_2,t_2,c_2 !=
1041 st c_1,rp(6) !r[6]=c1;
1042 st c_2,rp(7) !r[7]=c2;
1043
1044 ret
1045 restore %g0,%g0,%o0
1046
1047.type bn_mul_comba4,#function
1048.size bn_mul_comba4,(.-bn_mul_comba4)
1049
1050.align 32
1051
1052.global bn_sqr_comba8
1053bn_sqr_comba8:
1054 save %sp,FRAME_SIZE,%sp
1055 ld ap(0),a_0
1056 ld ap(1),a_1
1057 umul a_0,a_0,c_1 !=!sqr_add_c(a,0,c1,c2,c3);
1058 rd %y,c_2
1059 st c_1,rp(0) !r[0]=c1;
1060
1061 ld ap(2),a_2
1062 umul a_0,a_1,t_1 !=!sqr_add_c2(a,1,0,c2,c3,c1);
1063 addcc c_2,t_1,c_2
1064 rd %y,t_2
1065 addxcc %g0,t_2,c_3
1066 addx %g0,%g0,c_1 !=
1067 addcc c_2,t_1,c_2
1068 addxcc c_3,t_2,c_3
1069 st c_2,rp(1) !r[1]=c2;
1070 addx c_1,%g0,c_1 !=
1071
1072 umul a_2,a_0,t_1 !sqr_add_c2(a,2,0,c3,c1,c2);
1073 addcc c_3,t_1,c_3
1074 rd %y,t_2
1075 addxcc c_1,t_2,c_1 !=
1076 addx %g0,%g0,c_2
1077 addcc c_3,t_1,c_3
1078 addxcc c_1,t_2,c_1
1079 addx c_2,%g0,c_2 !=
1080 ld ap(3),a_3
1081 umul a_1,a_1,t_1 !sqr_add_c(a,1,c3,c1,c2);
1082 addcc c_3,t_1,c_3
1083 rd %y,t_2 !=
1084 addxcc c_1,t_2,c_1
1085 addx c_2,%g0,c_2
1086 st c_3,rp(2) !r[2]=c3;
1087
1088 umul a_0,a_3,t_1 !=!sqr_add_c2(a,3,0,c1,c2,c3);
1089 addcc c_1,t_1,c_1
1090 rd %y,t_2
1091 addxcc c_2,t_2,c_2
1092 addx %g0,%g0,c_3 !=
1093 addcc c_1,t_1,c_1
1094 addxcc c_2,t_2,c_2
1095 ld ap(4),a_4
1096 addx c_3,%g0,c_3 !=
1097 umul a_1,a_2,t_1 !sqr_add_c2(a,2,1,c1,c2,c3);
1098 addcc c_1,t_1,c_1
1099 rd %y,t_2
1100 addxcc c_2,t_2,c_2 !=
1101 addx c_3,%g0,c_3
1102 addcc c_1,t_1,c_1
1103 addxcc c_2,t_2,c_2
1104 addx c_3,%g0,c_3 !=
1105 st c_1,rp(3) !r[3]=c1;
1106
1107 umul a_4,a_0,t_1 !sqr_add_c2(a,4,0,c2,c3,c1);
1108 addcc c_2,t_1,c_2
1109 rd %y,t_2 !=
1110 addxcc c_3,t_2,c_3
1111 addx %g0,%g0,c_1
1112 addcc c_2,t_1,c_2
1113 addxcc c_3,t_2,c_3 !=
1114 addx c_1,%g0,c_1
1115 umul a_3,a_1,t_1 !sqr_add_c2(a,3,1,c2,c3,c1);
1116 addcc c_2,t_1,c_2
1117 rd %y,t_2 !=
1118 addxcc c_3,t_2,c_3
1119 addx c_1,%g0,c_1
1120 addcc c_2,t_1,c_2
1121 addxcc c_3,t_2,c_3 !=
1122 addx c_1,%g0,c_1
1123 ld ap(5),a_5
1124 umul a_2,a_2,t_1 !sqr_add_c(a,2,c2,c3,c1);
1125 addcc c_2,t_1,c_2 !=
1126 rd %y,t_2
1127 addxcc c_3,t_2,c_3
1128 st c_2,rp(4) !r[4]=c2;
1129 addx c_1,%g0,c_1 !=
1130
1131 umul a_0,a_5,t_1 !sqr_add_c2(a,5,0,c3,c1,c2);
1132 addcc c_3,t_1,c_3
1133 rd %y,t_2
1134 addxcc c_1,t_2,c_1 !=
1135 addx %g0,%g0,c_2
1136 addcc c_3,t_1,c_3
1137 addxcc c_1,t_2,c_1
1138 addx c_2,%g0,c_2 !=
1139 umul a_1,a_4,t_1 !sqr_add_c2(a,4,1,c3,c1,c2);
1140 addcc c_3,t_1,c_3
1141 rd %y,t_2
1142 addxcc c_1,t_2,c_1 !=
1143 addx c_2,%g0,c_2
1144 addcc c_3,t_1,c_3
1145 addxcc c_1,t_2,c_1
1146 addx c_2,%g0,c_2 !=
1147 ld ap(6),a_6
1148 umul a_2,a_3,t_1 !sqr_add_c2(a,3,2,c3,c1,c2);
1149 addcc c_3,t_1,c_3
1150 rd %y,t_2 !=
1151 addxcc c_1,t_2,c_1
1152 addx c_2,%g0,c_2
1153 addcc c_3,t_1,c_3
1154 addxcc c_1,t_2,c_1 !=
1155 addx c_2,%g0,c_2
1156 st c_3,rp(5) !r[5]=c3;
1157
1158 umul a_6,a_0,t_1 !sqr_add_c2(a,6,0,c1,c2,c3);
1159 addcc c_1,t_1,c_1 !=
1160 rd %y,t_2
1161 addxcc c_2,t_2,c_2
1162 addx %g0,%g0,c_3
1163 addcc c_1,t_1,c_1 !=
1164 addxcc c_2,t_2,c_2
1165 addx c_3,%g0,c_3
1166 umul a_5,a_1,t_1 !sqr_add_c2(a,5,1,c1,c2,c3);
1167 addcc c_1,t_1,c_1 !=
1168 rd %y,t_2
1169 addxcc c_2,t_2,c_2
1170 addx c_3,%g0,c_3
1171 addcc c_1,t_1,c_1 !=
1172 addxcc c_2,t_2,c_2
1173 addx c_3,%g0,c_3
1174 umul a_4,a_2,t_1 !sqr_add_c2(a,4,2,c1,c2,c3);
1175 addcc c_1,t_1,c_1 !=
1176 rd %y,t_2
1177 addxcc c_2,t_2,c_2
1178 addx c_3,%g0,c_3
1179 addcc c_1,t_1,c_1 !=
1180 addxcc c_2,t_2,c_2
1181 addx c_3,%g0,c_3
1182 ld ap(7),a_7
1183 umul a_3,a_3,t_1 !=!sqr_add_c(a,3,c1,c2,c3);
1184 addcc c_1,t_1,c_1
1185 rd %y,t_2
1186 addxcc c_2,t_2,c_2
1187 addx c_3,%g0,c_3 !=
1188 st c_1,rp(6) !r[6]=c1;
1189
1190 umul a_0,a_7,t_1 !sqr_add_c2(a,7,0,c2,c3,c1);
1191 addcc c_2,t_1,c_2
1192 rd %y,t_2 !=
1193 addxcc c_3,t_2,c_3
1194 addx %g0,%g0,c_1
1195 addcc c_2,t_1,c_2
1196 addxcc c_3,t_2,c_3 !=
1197 addx c_1,%g0,c_1
1198 umul a_1,a_6,t_1 !sqr_add_c2(a,6,1,c2,c3,c1);
1199 addcc c_2,t_1,c_2
1200 rd %y,t_2 !=
1201 addxcc c_3,t_2,c_3
1202 addx c_1,%g0,c_1
1203 addcc c_2,t_1,c_2
1204 addxcc c_3,t_2,c_3 !=
1205 addx c_1,%g0,c_1
1206 umul a_2,a_5,t_1 !sqr_add_c2(a,5,2,c2,c3,c1);
1207 addcc c_2,t_1,c_2
1208 rd %y,t_2 !=
1209 addxcc c_3,t_2,c_3
1210 addx c_1,%g0,c_1
1211 addcc c_2,t_1,c_2
1212 addxcc c_3,t_2,c_3 !=
1213 addx c_1,%g0,c_1
1214 umul a_3,a_4,t_1 !sqr_add_c2(a,4,3,c2,c3,c1);
1215 addcc c_2,t_1,c_2
1216 rd %y,t_2 !=
1217 addxcc c_3,t_2,c_3
1218 addx c_1,%g0,c_1
1219 addcc c_2,t_1,c_2
1220 addxcc c_3,t_2,c_3 !=
1221 addx c_1,%g0,c_1
1222 st c_2,rp(7) !r[7]=c2;
1223
1224 umul a_7,a_1,t_1 !sqr_add_c2(a,7,1,c3,c1,c2);
1225 addcc c_3,t_1,c_3 !=
1226 rd %y,t_2
1227 addxcc c_1,t_2,c_1
1228 addx %g0,%g0,c_2
1229 addcc c_3,t_1,c_3 !=
1230 addxcc c_1,t_2,c_1
1231 addx c_2,%g0,c_2
1232 umul a_6,a_2,t_1 !sqr_add_c2(a,6,2,c3,c1,c2);
1233 addcc c_3,t_1,c_3 !=
1234 rd %y,t_2
1235 addxcc c_1,t_2,c_1
1236 addx c_2,%g0,c_2
1237 addcc c_3,t_1,c_3 !=
1238 addxcc c_1,t_2,c_1
1239 addx c_2,%g0,c_2
1240 umul a_5,a_3,t_1 !sqr_add_c2(a,5,3,c3,c1,c2);
1241 addcc c_3,t_1,c_3 !=
1242 rd %y,t_2
1243 addxcc c_1,t_2,c_1
1244 addx c_2,%g0,c_2
1245 addcc c_3,t_1,c_3 !=
1246 addxcc c_1,t_2,c_1
1247 addx c_2,%g0,c_2
1248 umul a_4,a_4,t_1 !sqr_add_c(a,4,c3,c1,c2);
1249 addcc c_3,t_1,c_3 !=
1250 rd %y,t_2
1251 addxcc c_1,t_2,c_1
1252 st c_3,rp(8) !r[8]=c3;
1253 addx c_2,%g0,c_2 !=
1254
1255 umul a_2,a_7,t_1 !sqr_add_c2(a,7,2,c1,c2,c3);
1256 addcc c_1,t_1,c_1
1257 rd %y,t_2
1258 addxcc c_2,t_2,c_2 !=
1259 addx %g0,%g0,c_3
1260 addcc c_1,t_1,c_1
1261 addxcc c_2,t_2,c_2
1262 addx c_3,%g0,c_3 !=
1263 umul a_3,a_6,t_1 !sqr_add_c2(a,6,3,c1,c2,c3);
1264 addcc c_1,t_1,c_1
1265 rd %y,t_2
1266 addxcc c_2,t_2,c_2 !=
1267 addx c_3,%g0,c_3
1268 addcc c_1,t_1,c_1
1269 addxcc c_2,t_2,c_2
1270 addx c_3,%g0,c_3 !=
1271 umul a_4,a_5,t_1 !sqr_add_c2(a,5,4,c1,c2,c3);
1272 addcc c_1,t_1,c_1
1273 rd %y,t_2
1274 addxcc c_2,t_2,c_2 !=
1275 addx c_3,%g0,c_3
1276 addcc c_1,t_1,c_1
1277 addxcc c_2,t_2,c_2
1278 addx c_3,%g0,c_3 !=
1279 st c_1,rp(9) !r[9]=c1;
1280
1281 umul a_7,a_3,t_1 !sqr_add_c2(a,7,3,c2,c3,c1);
1282 addcc c_2,t_1,c_2
1283 rd %y,t_2 !=
1284 addxcc c_3,t_2,c_3
1285 addx %g0,%g0,c_1
1286 addcc c_2,t_1,c_2
1287 addxcc c_3,t_2,c_3 !=
1288 addx c_1,%g0,c_1
1289 umul a_6,a_4,t_1 !sqr_add_c2(a,6,4,c2,c3,c1);
1290 addcc c_2,t_1,c_2
1291 rd %y,t_2 !=
1292 addxcc c_3,t_2,c_3
1293 addx c_1,%g0,c_1
1294 addcc c_2,t_1,c_2
1295 addxcc c_3,t_2,c_3 !=
1296 addx c_1,%g0,c_1
1297 umul a_5,a_5,t_1 !sqr_add_c(a,5,c2,c3,c1);
1298 addcc c_2,t_1,c_2
1299 rd %y,t_2 !=
1300 addxcc c_3,t_2,c_3
1301 addx c_1,%g0,c_1
1302 st c_2,rp(10) !r[10]=c2;
1303
1304 umul a_4,a_7,t_1 !=!sqr_add_c2(a,7,4,c3,c1,c2);
1305 addcc c_3,t_1,c_3
1306 rd %y,t_2
1307 addxcc c_1,t_2,c_1
1308 addx %g0,%g0,c_2 !=
1309 addcc c_3,t_1,c_3
1310 addxcc c_1,t_2,c_1
1311 addx c_2,%g0,c_2
1312 umul a_5,a_6,t_1 !=!sqr_add_c2(a,6,5,c3,c1,c2);
1313 addcc c_3,t_1,c_3
1314 rd %y,t_2
1315 addxcc c_1,t_2,c_1
1316 addx c_2,%g0,c_2 !=
1317 addcc c_3,t_1,c_3
1318 addxcc c_1,t_2,c_1
1319 st c_3,rp(11) !r[11]=c3;
1320 addx c_2,%g0,c_2 !=
1321
1322 umul a_7,a_5,t_1 !sqr_add_c2(a,7,5,c1,c2,c3);
1323 addcc c_1,t_1,c_1
1324 rd %y,t_2
1325 addxcc c_2,t_2,c_2 !=
1326 addx %g0,%g0,c_3
1327 addcc c_1,t_1,c_1
1328 addxcc c_2,t_2,c_2
1329 addx c_3,%g0,c_3 !=
1330 umul a_6,a_6,t_1 !sqr_add_c(a,6,c1,c2,c3);
1331 addcc c_1,t_1,c_1
1332 rd %y,t_2
1333 addxcc c_2,t_2,c_2 !=
1334 addx c_3,%g0,c_3
1335 st c_1,rp(12) !r[12]=c1;
1336
1337 umul a_6,a_7,t_1 !sqr_add_c2(a,7,6,c2,c3,c1);
1338 addcc c_2,t_1,c_2 !=
1339 rd %y,t_2
1340 addxcc c_3,t_2,c_3
1341 addx %g0,%g0,c_1
1342 addcc c_2,t_1,c_2 !=
1343 addxcc c_3,t_2,c_3
1344 st c_2,rp(13) !r[13]=c2;
1345 addx c_1,%g0,c_1 !=
1346
1347 umul a_7,a_7,t_1 !sqr_add_c(a,7,c3,c1,c2);
1348 addcc c_3,t_1,c_3
1349 rd %y,t_2
1350 addxcc c_1,t_2,c_1 !=
1351 st c_3,rp(14) !r[14]=c3;
1352 st c_1,rp(15) !r[15]=c1;
1353
1354 ret
1355 restore %g0,%g0,%o0
1356
1357.type bn_sqr_comba8,#function
1358.size bn_sqr_comba8,(.-bn_sqr_comba8)
1359
1360.align 32
1361
1362.global bn_sqr_comba4
1363/*
1364 * void bn_sqr_comba4(r,a)
1365 * BN_ULONG *r,*a;
1366 */
1367bn_sqr_comba4:
1368 save %sp,FRAME_SIZE,%sp
1369 ld ap(0),a_0
1370 umul a_0,a_0,c_1 !sqr_add_c(a,0,c1,c2,c3);
1371 ld ap(1),a_1 !=
1372 rd %y,c_2
1373 st c_1,rp(0) !r[0]=c1;
1374
1375 ld ap(2),a_2
1376 umul a_0,a_1,t_1 !=!sqr_add_c2(a,1,0,c2,c3,c1);
1377 addcc c_2,t_1,c_2
1378 rd %y,t_2
1379 addxcc %g0,t_2,c_3
1380 addx %g0,%g0,c_1 !=
1381 addcc c_2,t_1,c_2
1382 addxcc c_3,t_2,c_3
1383 addx c_1,%g0,c_1 !=
1384 st c_2,rp(1) !r[1]=c2;
1385
1386 umul a_2,a_0,t_1 !sqr_add_c2(a,2,0,c3,c1,c2);
1387 addcc c_3,t_1,c_3
1388 rd %y,t_2 !=
1389 addxcc c_1,t_2,c_1
1390 addx %g0,%g0,c_2
1391 addcc c_3,t_1,c_3
1392 addxcc c_1,t_2,c_1 !=
1393 addx c_2,%g0,c_2
1394 ld ap(3),a_3
1395 umul a_1,a_1,t_1 !sqr_add_c(a,1,c3,c1,c2);
1396 addcc c_3,t_1,c_3 !=
1397 rd %y,t_2
1398 addxcc c_1,t_2,c_1
1399 st c_3,rp(2) !r[2]=c3;
1400 addx c_2,%g0,c_2 !=
1401
1402 umul a_0,a_3,t_1 !sqr_add_c2(a,3,0,c1,c2,c3);
1403 addcc c_1,t_1,c_1
1404 rd %y,t_2
1405 addxcc c_2,t_2,c_2 !=
1406 addx %g0,%g0,c_3
1407 addcc c_1,t_1,c_1
1408 addxcc c_2,t_2,c_2
1409 addx c_3,%g0,c_3 !=
1410 umul a_1,a_2,t_1 !sqr_add_c2(a,2,1,c1,c2,c3);
1411 addcc c_1,t_1,c_1
1412 rd %y,t_2
1413 addxcc c_2,t_2,c_2 !=
1414 addx c_3,%g0,c_3
1415 addcc c_1,t_1,c_1
1416 addxcc c_2,t_2,c_2
1417 addx c_3,%g0,c_3 !=
1418 st c_1,rp(3) !r[3]=c1;
1419
1420 umul a_3,a_1,t_1 !sqr_add_c2(a,3,1,c2,c3,c1);
1421 addcc c_2,t_1,c_2
1422 rd %y,t_2 !=
1423 addxcc c_3,t_2,c_3
1424 addx %g0,%g0,c_1
1425 addcc c_2,t_1,c_2
1426 addxcc c_3,t_2,c_3 !=
1427 addx c_1,%g0,c_1
1428 umul a_2,a_2,t_1 !sqr_add_c(a,2,c2,c3,c1);
1429 addcc c_2,t_1,c_2
1430 rd %y,t_2 !=
1431 addxcc c_3,t_2,c_3
1432 addx c_1,%g0,c_1
1433 st c_2,rp(4) !r[4]=c2;
1434
1435 umul a_2,a_3,t_1 !=!sqr_add_c2(a,3,2,c3,c1,c2);
1436 addcc c_3,t_1,c_3
1437 rd %y,t_2
1438 addxcc c_1,t_2,c_1
1439 addx %g0,%g0,c_2 !=
1440 addcc c_3,t_1,c_3
1441 addxcc c_1,t_2,c_1
1442 st c_3,rp(5) !r[5]=c3;
1443 addx c_2,%g0,c_2 !=
1444
1445 umul a_3,a_3,t_1 !sqr_add_c(a,3,c1,c2,c3);
1446 addcc c_1,t_1,c_1
1447 rd %y,t_2
1448 addxcc c_2,t_2,c_2 !=
1449 st c_1,rp(6) !r[6]=c1;
1450 st c_2,rp(7) !r[7]=c2;
1451
1452 ret
1453 restore %g0,%g0,%o0
1454
1455.type bn_sqr_comba4,#function
1456.size bn_sqr_comba4,(.-bn_sqr_comba4)
1457
1458.align 32
diff --git a/src/lib/libcrypto/bn/asm/sparcv8plus.S b/src/lib/libcrypto/bn/asm/sparcv8plus.S
deleted file mode 100644
index 0074dfdb75..0000000000
--- a/src/lib/libcrypto/bn/asm/sparcv8plus.S
+++ /dev/null
@@ -1,1535 +0,0 @@
1.ident "sparcv8plus.s, Version 1.4"
2.ident "SPARC v9 ISA artwork by Andy Polyakov <appro@fy.chalmers.se>"
3
4/*
5 * ====================================================================
6 * Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
7 * project.
8 *
9 * Rights for redistribution and usage in source and binary forms are
10 * granted according to the OpenSSL license. Warranty of any kind is
11 * disclaimed.
12 * ====================================================================
13 */
14
15/*
16 * This is my modest contributon to OpenSSL project (see
17 * http://www.openssl.org/ for more information about it) and is
18 * a drop-in UltraSPARC ISA replacement for crypto/bn/bn_asm.c
19 * module. For updates see http://fy.chalmers.se/~appro/hpe/.
20 *
21 * Questions-n-answers.
22 *
23 * Q. How to compile?
24 * A. With SC4.x/SC5.x:
25 *
26 * cc -xarch=v8plus -c bn_asm.sparc.v8plus.S -o bn_asm.o
27 *
28 * and with gcc:
29 *
30 * gcc -mcpu=ultrasparc -c bn_asm.sparc.v8plus.S -o bn_asm.o
31 *
32 * or if above fails (it does if you have gas installed):
33 *
34 * gcc -E bn_asm.sparc.v8plus.S | as -xarch=v8plus /dev/fd/0 -o bn_asm.o
35 *
36 * Quick-n-dirty way to fuse the module into the library.
37 * Provided that the library is already configured and built
38 * (in 0.9.2 case with no-asm option):
39 *
40 * # cd crypto/bn
41 * # cp /some/place/bn_asm.sparc.v8plus.S .
42 * # cc -xarch=v8plus -c bn_asm.sparc.v8plus.S -o bn_asm.o
43 * # make
44 * # cd ../..
45 * # make; make test
46 *
47 * Quick-n-dirty way to get rid of it:
48 *
49 * # cd crypto/bn
50 * # touch bn_asm.c
51 * # make
52 * # cd ../..
53 * # make; make test
54 *
55 * Q. V8plus achitecture? What kind of beast is that?
56 * A. Well, it's rather a programming model than an architecture...
57 * It's actually v9-compliant, i.e. *any* UltraSPARC, CPU under
58 * special conditions, namely when kernel doesn't preserve upper
59 * 32 bits of otherwise 64-bit registers during a context switch.
60 *
61 * Q. Why just UltraSPARC? What about SuperSPARC?
62 * A. Original release did target UltraSPARC only. Now SuperSPARC
63 * version is provided along. Both version share bn_*comba[48]
64 * implementations (see comment later in code for explanation).
65 * But what's so special about this UltraSPARC implementation?
66 * Why didn't I let compiler do the job? Trouble is that most of
67 * available compilers (well, SC5.0 is the only exception) don't
68 * attempt to take advantage of UltraSPARC's 64-bitness under
69 * 32-bit kernels even though it's perfectly possible (see next
70 * question).
71 *
72 * Q. 64-bit registers under 32-bit kernels? Didn't you just say it
73 * doesn't work?
74 * A. You can't adress *all* registers as 64-bit wide:-( The catch is
75 * that you actually may rely upon %o0-%o5 and %g1-%g4 being fully
76 * preserved if you're in a leaf function, i.e. such never calling
77 * any other functions. All functions in this module are leaf and
78 * 10 registers is a handful. And as a matter of fact none-"comba"
79 * routines don't require even that much and I could even afford to
80 * not allocate own stack frame for 'em:-)
81 *
82 * Q. What about 64-bit kernels?
83 * A. What about 'em? Just kidding:-) Pure 64-bit version is currently
84 * under evaluation and development...
85 *
86 * Q. What about shared libraries?
87 * A. What about 'em? Kidding again:-) Code does *not* contain any
88 * code position dependencies and it's safe to include it into
89 * shared library as is.
90 *
91 * Q. How much faster does it go?
92 * A. Do you have a good benchmark? In either case below is what I
93 * experience with crypto/bn/expspeed.c test program:
94 *
95 * v8plus module on U10/300MHz against bn_asm.c compiled with:
96 *
97 * cc-5.0 -xarch=v8plus -xO5 -xdepend +7-12%
98 * cc-4.2 -xarch=v8plus -xO5 -xdepend +25-35%
99 * egcs-1.1.2 -mcpu=ultrasparc -O3 +35-45%
100 *
101 * v8 module on SS10/60MHz against bn_asm.c compiled with:
102 *
103 * cc-5.0 -xarch=v8 -xO5 -xdepend +7-10%
104 * cc-4.2 -xarch=v8 -xO5 -xdepend +10%
105 * egcs-1.1.2 -mv8 -O3 +35-45%
106 *
107 * As you can see it's damn hard to beat the new Sun C compiler
108 * and it's in first place GNU C users who will appreciate this
109 * assembler implementation:-)
110 */
111
112/*
113 * Revision history.
114 *
115 * 1.0 - initial release;
116 * 1.1 - new loop unrolling model(*);
117 * - some more fine tuning;
118 * 1.2 - made gas friendly;
119 * - updates to documentation concerning v9;
120 * - new performance comparison matrix;
121 * 1.3 - fixed problem with /usr/ccs/lib/cpp;
122 * 1.4 - native V9 bn_*_comba[48] implementation (15% more efficient)
123 * resulting in slight overall performance kick;
124 * - some retunes;
125 * - support for GNU as added;
126 *
127 * (*) Originally unrolled loop looked like this:
128 * for (;;) {
129 * op(p+0); if (--n==0) break;
130 * op(p+1); if (--n==0) break;
131 * op(p+2); if (--n==0) break;
132 * op(p+3); if (--n==0) break;
133 * p+=4;
134 * }
135 * I unroll according to following:
136 * while (n&~3) {
137 * op(p+0); op(p+1); op(p+2); op(p+3);
138 * p+=4; n=-4;
139 * }
140 * if (n) {
141 * op(p+0); if (--n==0) return;
142 * op(p+2); if (--n==0) return;
143 * op(p+3); return;
144 * }
145 */
146
147/*
148 * GNU assembler can't stand stuw:-(
149 */
150#define stuw st
151
152.section ".text",#alloc,#execinstr
153.file "bn_asm.sparc.v8plus.S"
154
155.align 32
156
157.global bn_mul_add_words
158/*
159 * BN_ULONG bn_mul_add_words(rp,ap,num,w)
160 * BN_ULONG *rp,*ap;
161 * int num;
162 * BN_ULONG w;
163 */
164bn_mul_add_words:
165 brgz,a %o2,.L_bn_mul_add_words_proceed
166 lduw [%o1],%g2
167 retl
168 clr %o0
169
170.L_bn_mul_add_words_proceed:
171 srl %o3,%g0,%o3 ! clruw %o3
172 andcc %o2,-4,%g0
173 bz,pn %icc,.L_bn_mul_add_words_tail
174 clr %o5
175
176.L_bn_mul_add_words_loop: ! wow! 32 aligned!
177 lduw [%o0],%g1
178 lduw [%o1+4],%g3
179 mulx %o3,%g2,%g2
180 add %g1,%o5,%o4
181 nop
182 add %o4,%g2,%o4
183 stuw %o4,[%o0]
184 srlx %o4,32,%o5
185
186 lduw [%o0+4],%g1
187 lduw [%o1+8],%g2
188 mulx %o3,%g3,%g3
189 add %g1,%o5,%o4
190 dec 4,%o2
191 add %o4,%g3,%o4
192 stuw %o4,[%o0+4]
193 srlx %o4,32,%o5
194
195 lduw [%o0+8],%g1
196 lduw [%o1+12],%g3
197 mulx %o3,%g2,%g2
198 add %g1,%o5,%o4
199 inc 16,%o1
200 add %o4,%g2,%o4
201 stuw %o4,[%o0+8]
202 srlx %o4,32,%o5
203
204 lduw [%o0+12],%g1
205 mulx %o3,%g3,%g3
206 add %g1,%o5,%o4
207 inc 16,%o0
208 add %o4,%g3,%o4
209 andcc %o2,-4,%g0
210 stuw %o4,[%o0-4]
211 srlx %o4,32,%o5
212 bnz,a,pt %icc,.L_bn_mul_add_words_loop
213 lduw [%o1],%g2
214
215 brnz,a,pn %o2,.L_bn_mul_add_words_tail
216 lduw [%o1],%g2
217.L_bn_mul_add_words_return:
218 retl
219 mov %o5,%o0
220
221.L_bn_mul_add_words_tail:
222 lduw [%o0],%g1
223 mulx %o3,%g2,%g2
224 add %g1,%o5,%o4
225 dec %o2
226 add %o4,%g2,%o4
227 srlx %o4,32,%o5
228 brz,pt %o2,.L_bn_mul_add_words_return
229 stuw %o4,[%o0]
230
231 lduw [%o1+4],%g2
232 lduw [%o0+4],%g1
233 mulx %o3,%g2,%g2
234 add %g1,%o5,%o4
235 dec %o2
236 add %o4,%g2,%o4
237 srlx %o4,32,%o5
238 brz,pt %o2,.L_bn_mul_add_words_return
239 stuw %o4,[%o0+4]
240
241 lduw [%o1+8],%g2
242 lduw [%o0+8],%g1
243 mulx %o3,%g2,%g2
244 add %g1,%o5,%o4
245 add %o4,%g2,%o4
246 stuw %o4,[%o0+8]
247 retl
248 srlx %o4,32,%o0
249
250.type bn_mul_add_words,#function
251.size bn_mul_add_words,(.-bn_mul_add_words)
252
253.align 32
254
255.global bn_mul_words
256/*
257 * BN_ULONG bn_mul_words(rp,ap,num,w)
258 * BN_ULONG *rp,*ap;
259 * int num;
260 * BN_ULONG w;
261 */
262bn_mul_words:
263 brgz,a %o2,.L_bn_mul_words_proceeed
264 lduw [%o1],%g2
265 retl
266 clr %o0
267
268.L_bn_mul_words_proceeed:
269 srl %o3,%g0,%o3 ! clruw %o3
270 andcc %o2,-4,%g0
271 bz,pn %icc,.L_bn_mul_words_tail
272 clr %o5
273
274.L_bn_mul_words_loop: ! wow! 32 aligned!
275 lduw [%o1+4],%g3
276 mulx %o3,%g2,%g2
277 add %g2,%o5,%o4
278 nop
279 stuw %o4,[%o0]
280 srlx %o4,32,%o5
281
282 lduw [%o1+8],%g2
283 mulx %o3,%g3,%g3
284 add %g3,%o5,%o4
285 dec 4,%o2
286 stuw %o4,[%o0+4]
287 srlx %o4,32,%o5
288
289 lduw [%o1+12],%g3
290 mulx %o3,%g2,%g2
291 add %g2,%o5,%o4
292 inc 16,%o1
293 stuw %o4,[%o0+8]
294 srlx %o4,32,%o5
295
296 mulx %o3,%g3,%g3
297 add %g3,%o5,%o4
298 inc 16,%o0
299 stuw %o4,[%o0-4]
300 srlx %o4,32,%o5
301 andcc %o2,-4,%g0
302 bnz,a,pt %icc,.L_bn_mul_words_loop
303 lduw [%o1],%g2
304 nop
305 nop
306
307 brnz,a,pn %o2,.L_bn_mul_words_tail
308 lduw [%o1],%g2
309.L_bn_mul_words_return:
310 retl
311 mov %o5,%o0
312
313.L_bn_mul_words_tail:
314 mulx %o3,%g2,%g2
315 add %g2,%o5,%o4
316 dec %o2
317 srlx %o4,32,%o5
318 brz,pt %o2,.L_bn_mul_words_return
319 stuw %o4,[%o0]
320
321 lduw [%o1+4],%g2
322 mulx %o3,%g2,%g2
323 add %g2,%o5,%o4
324 dec %o2
325 srlx %o4,32,%o5
326 brz,pt %o2,.L_bn_mul_words_return
327 stuw %o4,[%o0+4]
328
329 lduw [%o1+8],%g2
330 mulx %o3,%g2,%g2
331 add %g2,%o5,%o4
332 stuw %o4,[%o0+8]
333 retl
334 srlx %o4,32,%o0
335
336.type bn_mul_words,#function
337.size bn_mul_words,(.-bn_mul_words)
338
339.align 32
340.global bn_sqr_words
341/*
342 * void bn_sqr_words(r,a,n)
343 * BN_ULONG *r,*a;
344 * int n;
345 */
346bn_sqr_words:
347 brgz,a %o2,.L_bn_sqr_words_proceeed
348 lduw [%o1],%g2
349 retl
350 clr %o0
351
352.L_bn_sqr_words_proceeed:
353 andcc %o2,-4,%g0
354 nop
355 bz,pn %icc,.L_bn_sqr_words_tail
356 nop
357
358.L_bn_sqr_words_loop: ! wow! 32 aligned!
359 lduw [%o1+4],%g3
360 mulx %g2,%g2,%o4
361 stuw %o4,[%o0]
362 srlx %o4,32,%o5
363 stuw %o5,[%o0+4]
364 nop
365
366 lduw [%o1+8],%g2
367 mulx %g3,%g3,%o4
368 dec 4,%o2
369 stuw %o4,[%o0+8]
370 srlx %o4,32,%o5
371 stuw %o5,[%o0+12]
372
373 lduw [%o1+12],%g3
374 mulx %g2,%g2,%o4
375 srlx %o4,32,%o5
376 stuw %o4,[%o0+16]
377 inc 16,%o1
378 stuw %o5,[%o0+20]
379
380 mulx %g3,%g3,%o4
381 inc 32,%o0
382 stuw %o4,[%o0-8]
383 srlx %o4,32,%o5
384 andcc %o2,-4,%g2
385 stuw %o5,[%o0-4]
386 bnz,a,pt %icc,.L_bn_sqr_words_loop
387 lduw [%o1],%g2
388 nop
389
390 brnz,a,pn %o2,.L_bn_sqr_words_tail
391 lduw [%o1],%g2
392.L_bn_sqr_words_return:
393 retl
394 clr %o0
395
396.L_bn_sqr_words_tail:
397 mulx %g2,%g2,%o4
398 dec %o2
399 stuw %o4,[%o0]
400 srlx %o4,32,%o5
401 brz,pt %o2,.L_bn_sqr_words_return
402 stuw %o5,[%o0+4]
403
404 lduw [%o1+4],%g2
405 mulx %g2,%g2,%o4
406 dec %o2
407 stuw %o4,[%o0+8]
408 srlx %o4,32,%o5
409 brz,pt %o2,.L_bn_sqr_words_return
410 stuw %o5,[%o0+12]
411
412 lduw [%o1+8],%g2
413 mulx %g2,%g2,%o4
414 srlx %o4,32,%o5
415 stuw %o4,[%o0+16]
416 stuw %o5,[%o0+20]
417 retl
418 clr %o0
419
420.type bn_sqr_words,#function
421.size bn_sqr_words,(.-bn_sqr_words)
422
423.align 32
424.global bn_div_words
425/*
426 * BN_ULONG bn_div_words(h,l,d)
427 * BN_ULONG h,l,d;
428 */
429bn_div_words:
430 sllx %o0,32,%o0
431 or %o0,%o1,%o0
432 udivx %o0,%o2,%o0
433 retl
434 srl %o0,%g0,%o0 ! clruw %o0
435
436.type bn_div_words,#function
437.size bn_div_words,(.-bn_div_words)
438
439.align 32
440
441.global bn_add_words
442/*
443 * BN_ULONG bn_add_words(rp,ap,bp,n)
444 * BN_ULONG *rp,*ap,*bp;
445 * int n;
446 */
447bn_add_words:
448 brgz,a %o3,.L_bn_add_words_proceed
449 lduw [%o1],%o4
450 retl
451 clr %o0
452
453.L_bn_add_words_proceed:
454 andcc %o3,-4,%g0
455 bz,pn %icc,.L_bn_add_words_tail
456 addcc %g0,0,%g0 ! clear carry flag
457 nop
458
459.L_bn_add_words_loop: ! wow! 32 aligned!
460 dec 4,%o3
461 lduw [%o2],%o5
462 lduw [%o1+4],%g1
463 lduw [%o2+4],%g2
464 lduw [%o1+8],%g3
465 lduw [%o2+8],%g4
466 addccc %o5,%o4,%o5
467 stuw %o5,[%o0]
468
469 lduw [%o1+12],%o4
470 lduw [%o2+12],%o5
471 inc 16,%o1
472 addccc %g1,%g2,%g1
473 stuw %g1,[%o0+4]
474
475 inc 16,%o2
476 addccc %g3,%g4,%g3
477 stuw %g3,[%o0+8]
478
479 inc 16,%o0
480 addccc %o5,%o4,%o5
481 stuw %o5,[%o0-4]
482 and %o3,-4,%g1
483 brnz,a,pt %g1,.L_bn_add_words_loop
484 lduw [%o1],%o4
485
486 brnz,a,pn %o3,.L_bn_add_words_tail
487 lduw [%o1],%o4
488.L_bn_add_words_return:
489 clr %o0
490 retl
491 movcs %icc,1,%o0
492 nop
493
494.L_bn_add_words_tail:
495 lduw [%o2],%o5
496 dec %o3
497 addccc %o5,%o4,%o5
498 brz,pt %o3,.L_bn_add_words_return
499 stuw %o5,[%o0]
500
501 lduw [%o1+4],%o4
502 lduw [%o2+4],%o5
503 dec %o3
504 addccc %o5,%o4,%o5
505 brz,pt %o3,.L_bn_add_words_return
506 stuw %o5,[%o0+4]
507
508 lduw [%o1+8],%o4
509 lduw [%o2+8],%o5
510 addccc %o5,%o4,%o5
511 stuw %o5,[%o0+8]
512 clr %o0
513 retl
514 movcs %icc,1,%o0
515
516.type bn_add_words,#function
517.size bn_add_words,(.-bn_add_words)
518
519.global bn_sub_words
520/*
521 * BN_ULONG bn_sub_words(rp,ap,bp,n)
522 * BN_ULONG *rp,*ap,*bp;
523 * int n;
524 */
525bn_sub_words:
526 brgz,a %o3,.L_bn_sub_words_proceed
527 lduw [%o1],%o4
528 retl
529 clr %o0
530
531.L_bn_sub_words_proceed:
532 andcc %o3,-4,%g0
533 bz,pn %icc,.L_bn_sub_words_tail
534 addcc %g0,0,%g0 ! clear carry flag
535 nop
536
537.L_bn_sub_words_loop: ! wow! 32 aligned!
538 dec 4,%o3
539 lduw [%o2],%o5
540 lduw [%o1+4],%g1
541 lduw [%o2+4],%g2
542 lduw [%o1+8],%g3
543 lduw [%o2+8],%g4
544 subccc %o4,%o5,%o5
545 stuw %o5,[%o0]
546
547 lduw [%o1+12],%o4
548 lduw [%o2+12],%o5
549 inc 16,%o1
550 subccc %g1,%g2,%g2
551 stuw %g2,[%o0+4]
552
553 inc 16,%o2
554 subccc %g3,%g4,%g4
555 stuw %g4,[%o0+8]
556
557 inc 16,%o0
558 subccc %o4,%o5,%o5
559 stuw %o5,[%o0-4]
560 and %o3,-4,%g1
561 brnz,a,pt %g1,.L_bn_sub_words_loop
562 lduw [%o1],%o4
563
564 brnz,a,pn %o3,.L_bn_sub_words_tail
565 lduw [%o1],%o4
566.L_bn_sub_words_return:
567 clr %o0
568 retl
569 movcs %icc,1,%o0
570 nop
571
572.L_bn_sub_words_tail: ! wow! 32 aligned!
573 lduw [%o2],%o5
574 dec %o3
575 subccc %o4,%o5,%o5
576 brz,pt %o3,.L_bn_sub_words_return
577 stuw %o5,[%o0]
578
579 lduw [%o1+4],%o4
580 lduw [%o2+4],%o5
581 dec %o3
582 subccc %o4,%o5,%o5
583 brz,pt %o3,.L_bn_sub_words_return
584 stuw %o5,[%o0+4]
585
586 lduw [%o1+8],%o4
587 lduw [%o2+8],%o5
588 subccc %o4,%o5,%o5
589 stuw %o5,[%o0+8]
590 clr %o0
591 retl
592 movcs %icc,1,%o0
593
594.type bn_sub_words,#function
595.size bn_sub_words,(.-bn_sub_words)
596
597/*
598 * Code below depends on the fact that upper parts of the %l0-%l7
599 * and %i0-%i7 are zeroed by kernel after context switch. In
600 * previous versions this comment stated that "the trouble is that
601 * it's not feasible to implement the mumbo-jumbo in less V9
602 * instructions:-(" which apparently isn't true thanks to
603 * 'bcs,a %xcc,.+8; inc %rd' pair. But the performance improvement
604 * results not from the shorter code, but from elimination of
605 * multicycle none-pairable 'rd %y,%rd' instructions.
606 *
607 * Andy.
608 */
609
610#define FRAME_SIZE -96
611
612/*
613 * Here is register usage map for *all* routines below.
614 */
615#define t_1 %o0
616#define t_2 %o1
617#define c_12 %o2
618#define c_3 %o3
619
620#define ap(I) [%i1+4*I]
621#define bp(I) [%i2+4*I]
622#define rp(I) [%i0+4*I]
623
624#define a_0 %l0
625#define a_1 %l1
626#define a_2 %l2
627#define a_3 %l3
628#define a_4 %l4
629#define a_5 %l5
630#define a_6 %l6
631#define a_7 %l7
632
633#define b_0 %i3
634#define b_1 %i4
635#define b_2 %i5
636#define b_3 %o4
637#define b_4 %o5
638#define b_5 %o7
639#define b_6 %g1
640#define b_7 %g4
641
642.align 32
643.global bn_mul_comba8
644/*
645 * void bn_mul_comba8(r,a,b)
646 * BN_ULONG *r,*a,*b;
647 */
648bn_mul_comba8:
649 save %sp,FRAME_SIZE,%sp
650 mov 1,t_2
651 lduw ap(0),a_0
652 sllx t_2,32,t_2
653 lduw bp(0),b_0 !=
654 lduw bp(1),b_1
655 mulx a_0,b_0,t_1 !mul_add_c(a[0],b[0],c1,c2,c3);
656 srlx t_1,32,c_12
657 stuw t_1,rp(0) !=!r[0]=c1;
658
659 lduw ap(1),a_1
660 mulx a_0,b_1,t_1 !mul_add_c(a[0],b[1],c2,c3,c1);
661 addcc c_12,t_1,c_12
662 clr c_3 !=
663 bcs,a %xcc,.+8
664 add c_3,t_2,c_3
665 lduw ap(2),a_2
666 mulx a_1,b_0,t_1 !=!mul_add_c(a[1],b[0],c2,c3,c1);
667 addcc c_12,t_1,t_1
668 bcs,a %xcc,.+8
669 add c_3,t_2,c_3
670 srlx t_1,32,c_12 !=
671 stuw t_1,rp(1) !r[1]=c2;
672 or c_12,c_3,c_12
673
674 mulx a_2,b_0,t_1 !mul_add_c(a[2],b[0],c3,c1,c2);
675 addcc c_12,t_1,c_12 !=
676 clr c_3
677 bcs,a %xcc,.+8
678 add c_3,t_2,c_3
679 lduw bp(2),b_2 !=
680 mulx a_1,b_1,t_1 !mul_add_c(a[1],b[1],c3,c1,c2);
681 addcc c_12,t_1,c_12
682 bcs,a %xcc,.+8
683 add c_3,t_2,c_3 !=
684 lduw bp(3),b_3
685 mulx a_0,b_2,t_1 !mul_add_c(a[0],b[2],c3,c1,c2);
686 addcc c_12,t_1,t_1
687 bcs,a %xcc,.+8 !=
688 add c_3,t_2,c_3
689 srlx t_1,32,c_12
690 stuw t_1,rp(2) !r[2]=c3;
691 or c_12,c_3,c_12 !=
692
693 mulx a_0,b_3,t_1 !mul_add_c(a[0],b[3],c1,c2,c3);
694 addcc c_12,t_1,c_12
695 clr c_3
696 bcs,a %xcc,.+8 !=
697 add c_3,t_2,c_3
698 mulx a_1,b_2,t_1 !=!mul_add_c(a[1],b[2],c1,c2,c3);
699 addcc c_12,t_1,c_12
700 bcs,a %xcc,.+8 !=
701 add c_3,t_2,c_3
702 lduw ap(3),a_3
703 mulx a_2,b_1,t_1 !mul_add_c(a[2],b[1],c1,c2,c3);
704 addcc c_12,t_1,c_12 !=
705 bcs,a %xcc,.+8
706 add c_3,t_2,c_3
707 lduw ap(4),a_4
708 mulx a_3,b_0,t_1 !=!mul_add_c(a[3],b[0],c1,c2,c3);!=
709 addcc c_12,t_1,t_1
710 bcs,a %xcc,.+8
711 add c_3,t_2,c_3
712 srlx t_1,32,c_12 !=
713 stuw t_1,rp(3) !r[3]=c1;
714 or c_12,c_3,c_12
715
716 mulx a_4,b_0,t_1 !mul_add_c(a[4],b[0],c2,c3,c1);
717 addcc c_12,t_1,c_12 !=
718 clr c_3
719 bcs,a %xcc,.+8
720 add c_3,t_2,c_3
721 mulx a_3,b_1,t_1 !=!mul_add_c(a[3],b[1],c2,c3,c1);
722 addcc c_12,t_1,c_12
723 bcs,a %xcc,.+8
724 add c_3,t_2,c_3
725 mulx a_2,b_2,t_1 !=!mul_add_c(a[2],b[2],c2,c3,c1);
726 addcc c_12,t_1,c_12
727 bcs,a %xcc,.+8
728 add c_3,t_2,c_3
729 lduw bp(4),b_4 !=
730 mulx a_1,b_3,t_1 !mul_add_c(a[1],b[3],c2,c3,c1);
731 addcc c_12,t_1,c_12
732 bcs,a %xcc,.+8
733 add c_3,t_2,c_3 !=
734 lduw bp(5),b_5
735 mulx a_0,b_4,t_1 !mul_add_c(a[0],b[4],c2,c3,c1);
736 addcc c_12,t_1,t_1
737 bcs,a %xcc,.+8 !=
738 add c_3,t_2,c_3
739 srlx t_1,32,c_12
740 stuw t_1,rp(4) !r[4]=c2;
741 or c_12,c_3,c_12 !=
742
743 mulx a_0,b_5,t_1 !mul_add_c(a[0],b[5],c3,c1,c2);
744 addcc c_12,t_1,c_12
745 clr c_3
746 bcs,a %xcc,.+8 !=
747 add c_3,t_2,c_3
748 mulx a_1,b_4,t_1 !mul_add_c(a[1],b[4],c3,c1,c2);
749 addcc c_12,t_1,c_12
750 bcs,a %xcc,.+8 !=
751 add c_3,t_2,c_3
752 mulx a_2,b_3,t_1 !mul_add_c(a[2],b[3],c3,c1,c2);
753 addcc c_12,t_1,c_12
754 bcs,a %xcc,.+8 !=
755 add c_3,t_2,c_3
756 mulx a_3,b_2,t_1 !mul_add_c(a[3],b[2],c3,c1,c2);
757 addcc c_12,t_1,c_12
758 bcs,a %xcc,.+8 !=
759 add c_3,t_2,c_3
760 lduw ap(5),a_5
761 mulx a_4,b_1,t_1 !mul_add_c(a[4],b[1],c3,c1,c2);
762 addcc c_12,t_1,c_12 !=
763 bcs,a %xcc,.+8
764 add c_3,t_2,c_3
765 lduw ap(6),a_6
766 mulx a_5,b_0,t_1 !=!mul_add_c(a[5],b[0],c3,c1,c2);
767 addcc c_12,t_1,t_1
768 bcs,a %xcc,.+8
769 add c_3,t_2,c_3
770 srlx t_1,32,c_12 !=
771 stuw t_1,rp(5) !r[5]=c3;
772 or c_12,c_3,c_12
773
774 mulx a_6,b_0,t_1 !mul_add_c(a[6],b[0],c1,c2,c3);
775 addcc c_12,t_1,c_12 !=
776 clr c_3
777 bcs,a %xcc,.+8
778 add c_3,t_2,c_3
779 mulx a_5,b_1,t_1 !=!mul_add_c(a[5],b[1],c1,c2,c3);
780 addcc c_12,t_1,c_12
781 bcs,a %xcc,.+8
782 add c_3,t_2,c_3
783 mulx a_4,b_2,t_1 !=!mul_add_c(a[4],b[2],c1,c2,c3);
784 addcc c_12,t_1,c_12
785 bcs,a %xcc,.+8
786 add c_3,t_2,c_3
787 mulx a_3,b_3,t_1 !=!mul_add_c(a[3],b[3],c1,c2,c3);
788 addcc c_12,t_1,c_12
789 bcs,a %xcc,.+8
790 add c_3,t_2,c_3
791 mulx a_2,b_4,t_1 !=!mul_add_c(a[2],b[4],c1,c2,c3);
792 addcc c_12,t_1,c_12
793 bcs,a %xcc,.+8
794 add c_3,t_2,c_3
795 lduw bp(6),b_6 !=
796 mulx a_1,b_5,t_1 !mul_add_c(a[1],b[5],c1,c2,c3);
797 addcc c_12,t_1,c_12
798 bcs,a %xcc,.+8
799 add c_3,t_2,c_3 !=
800 lduw bp(7),b_7
801 mulx a_0,b_6,t_1 !mul_add_c(a[0],b[6],c1,c2,c3);
802 addcc c_12,t_1,t_1
803 bcs,a %xcc,.+8 !=
804 add c_3,t_2,c_3
805 srlx t_1,32,c_12
806 stuw t_1,rp(6) !r[6]=c1;
807 or c_12,c_3,c_12 !=
808
809 mulx a_0,b_7,t_1 !mul_add_c(a[0],b[7],c2,c3,c1);
810 addcc c_12,t_1,c_12
811 clr c_3
812 bcs,a %xcc,.+8 !=
813 add c_3,t_2,c_3
814 mulx a_1,b_6,t_1 !mul_add_c(a[1],b[6],c2,c3,c1);
815 addcc c_12,t_1,c_12
816 bcs,a %xcc,.+8 !=
817 add c_3,t_2,c_3
818 mulx a_2,b_5,t_1 !mul_add_c(a[2],b[5],c2,c3,c1);
819 addcc c_12,t_1,c_12
820 bcs,a %xcc,.+8 !=
821 add c_3,t_2,c_3
822 mulx a_3,b_4,t_1 !mul_add_c(a[3],b[4],c2,c3,c1);
823 addcc c_12,t_1,c_12
824 bcs,a %xcc,.+8 !=
825 add c_3,t_2,c_3
826 mulx a_4,b_3,t_1 !mul_add_c(a[4],b[3],c2,c3,c1);
827 addcc c_12,t_1,c_12
828 bcs,a %xcc,.+8 !=
829 add c_3,t_2,c_3
830 mulx a_5,b_2,t_1 !mul_add_c(a[5],b[2],c2,c3,c1);
831 addcc c_12,t_1,c_12
832 bcs,a %xcc,.+8 !=
833 add c_3,t_2,c_3
834 lduw ap(7),a_7
835 mulx a_6,b_1,t_1 !=!mul_add_c(a[6],b[1],c2,c3,c1);
836 addcc c_12,t_1,c_12
837 bcs,a %xcc,.+8
838 add c_3,t_2,c_3
839 mulx a_7,b_0,t_1 !=!mul_add_c(a[7],b[0],c2,c3,c1);
840 addcc c_12,t_1,t_1
841 bcs,a %xcc,.+8
842 add c_3,t_2,c_3
843 srlx t_1,32,c_12 !=
844 stuw t_1,rp(7) !r[7]=c2;
845 or c_12,c_3,c_12
846
847 mulx a_7,b_1,t_1 !=!mul_add_c(a[7],b[1],c3,c1,c2);
848 addcc c_12,t_1,c_12
849 clr c_3
850 bcs,a %xcc,.+8
851 add c_3,t_2,c_3 !=
852 mulx a_6,b_2,t_1 !mul_add_c(a[6],b[2],c3,c1,c2);
853 addcc c_12,t_1,c_12
854 bcs,a %xcc,.+8
855 add c_3,t_2,c_3 !=
856 mulx a_5,b_3,t_1 !mul_add_c(a[5],b[3],c3,c1,c2);
857 addcc c_12,t_1,c_12
858 bcs,a %xcc,.+8
859 add c_3,t_2,c_3 !=
860 mulx a_4,b_4,t_1 !mul_add_c(a[4],b[4],c3,c1,c2);
861 addcc c_12,t_1,c_12
862 bcs,a %xcc,.+8
863 add c_3,t_2,c_3 !=
864 mulx a_3,b_5,t_1 !mul_add_c(a[3],b[5],c3,c1,c2);
865 addcc c_12,t_1,c_12
866 bcs,a %xcc,.+8
867 add c_3,t_2,c_3 !=
868 mulx a_2,b_6,t_1 !mul_add_c(a[2],b[6],c3,c1,c2);
869 addcc c_12,t_1,c_12
870 bcs,a %xcc,.+8
871 add c_3,t_2,c_3 !=
872 mulx a_1,b_7,t_1 !mul_add_c(a[1],b[7],c3,c1,c2);
873 addcc c_12,t_1,t_1
874 bcs,a %xcc,.+8
875 add c_3,t_2,c_3 !=
876 srlx t_1,32,c_12
877 stuw t_1,rp(8) !r[8]=c3;
878 or c_12,c_3,c_12
879
880 mulx a_2,b_7,t_1 !=!mul_add_c(a[2],b[7],c1,c2,c3);
881 addcc c_12,t_1,c_12
882 clr c_3
883 bcs,a %xcc,.+8
884 add c_3,t_2,c_3 !=
885 mulx a_3,b_6,t_1 !mul_add_c(a[3],b[6],c1,c2,c3);
886 addcc c_12,t_1,c_12
887 bcs,a %xcc,.+8 !=
888 add c_3,t_2,c_3
889 mulx a_4,b_5,t_1 !mul_add_c(a[4],b[5],c1,c2,c3);
890 addcc c_12,t_1,c_12
891 bcs,a %xcc,.+8 !=
892 add c_3,t_2,c_3
893 mulx a_5,b_4,t_1 !mul_add_c(a[5],b[4],c1,c2,c3);
894 addcc c_12,t_1,c_12
895 bcs,a %xcc,.+8 !=
896 add c_3,t_2,c_3
897 mulx a_6,b_3,t_1 !mul_add_c(a[6],b[3],c1,c2,c3);
898 addcc c_12,t_1,c_12
899 bcs,a %xcc,.+8 !=
900 add c_3,t_2,c_3
901 mulx a_7,b_2,t_1 !mul_add_c(a[7],b[2],c1,c2,c3);
902 addcc c_12,t_1,t_1
903 bcs,a %xcc,.+8 !=
904 add c_3,t_2,c_3
905 srlx t_1,32,c_12
906 stuw t_1,rp(9) !r[9]=c1;
907 or c_12,c_3,c_12 !=
908
909 mulx a_7,b_3,t_1 !mul_add_c(a[7],b[3],c2,c3,c1);
910 addcc c_12,t_1,c_12
911 clr c_3
912 bcs,a %xcc,.+8 !=
913 add c_3,t_2,c_3
914 mulx a_6,b_4,t_1 !mul_add_c(a[6],b[4],c2,c3,c1);
915 addcc c_12,t_1,c_12
916 bcs,a %xcc,.+8 !=
917 add c_3,t_2,c_3
918 mulx a_5,b_5,t_1 !mul_add_c(a[5],b[5],c2,c3,c1);
919 addcc c_12,t_1,c_12
920 bcs,a %xcc,.+8 !=
921 add c_3,t_2,c_3
922 mulx a_4,b_6,t_1 !mul_add_c(a[4],b[6],c2,c3,c1);
923 addcc c_12,t_1,c_12
924 bcs,a %xcc,.+8 !=
925 add c_3,t_2,c_3
926 mulx a_3,b_7,t_1 !mul_add_c(a[3],b[7],c2,c3,c1);
927 addcc c_12,t_1,t_1
928 bcs,a %xcc,.+8 !=
929 add c_3,t_2,c_3
930 srlx t_1,32,c_12
931 stuw t_1,rp(10) !r[10]=c2;
932 or c_12,c_3,c_12 !=
933
934 mulx a_4,b_7,t_1 !mul_add_c(a[4],b[7],c3,c1,c2);
935 addcc c_12,t_1,c_12
936 clr c_3
937 bcs,a %xcc,.+8 !=
938 add c_3,t_2,c_3
939 mulx a_5,b_6,t_1 !mul_add_c(a[5],b[6],c3,c1,c2);
940 addcc c_12,t_1,c_12
941 bcs,a %xcc,.+8 !=
942 add c_3,t_2,c_3
943 mulx a_6,b_5,t_1 !mul_add_c(a[6],b[5],c3,c1,c2);
944 addcc c_12,t_1,c_12
945 bcs,a %xcc,.+8 !=
946 add c_3,t_2,c_3
947 mulx a_7,b_4,t_1 !mul_add_c(a[7],b[4],c3,c1,c2);
948 addcc c_12,t_1,t_1
949 bcs,a %xcc,.+8 !=
950 add c_3,t_2,c_3
951 srlx t_1,32,c_12
952 stuw t_1,rp(11) !r[11]=c3;
953 or c_12,c_3,c_12 !=
954
955 mulx a_7,b_5,t_1 !mul_add_c(a[7],b[5],c1,c2,c3);
956 addcc c_12,t_1,c_12
957 clr c_3
958 bcs,a %xcc,.+8 !=
959 add c_3,t_2,c_3
960 mulx a_6,b_6,t_1 !mul_add_c(a[6],b[6],c1,c2,c3);
961 addcc c_12,t_1,c_12
962 bcs,a %xcc,.+8 !=
963 add c_3,t_2,c_3
964 mulx a_5,b_7,t_1 !mul_add_c(a[5],b[7],c1,c2,c3);
965 addcc c_12,t_1,t_1
966 bcs,a %xcc,.+8 !=
967 add c_3,t_2,c_3
968 srlx t_1,32,c_12
969 stuw t_1,rp(12) !r[12]=c1;
970 or c_12,c_3,c_12 !=
971
972 mulx a_6,b_7,t_1 !mul_add_c(a[6],b[7],c2,c3,c1);
973 addcc c_12,t_1,c_12
974 clr c_3
975 bcs,a %xcc,.+8 !=
976 add c_3,t_2,c_3
977 mulx a_7,b_6,t_1 !mul_add_c(a[7],b[6],c2,c3,c1);
978 addcc c_12,t_1,t_1
979 bcs,a %xcc,.+8 !=
980 add c_3,t_2,c_3
981 srlx t_1,32,c_12
982 st t_1,rp(13) !r[13]=c2;
983 or c_12,c_3,c_12 !=
984
985 mulx a_7,b_7,t_1 !mul_add_c(a[7],b[7],c3,c1,c2);
986 addcc c_12,t_1,t_1
987 srlx t_1,32,c_12 !=
988 stuw t_1,rp(14) !r[14]=c3;
989 stuw c_12,rp(15) !r[15]=c1;
990
991 ret
992 restore %g0,%g0,%o0 !=
993
994.type bn_mul_comba8,#function
995.size bn_mul_comba8,(.-bn_mul_comba8)
996
997.align 32
998
999.global bn_mul_comba4
1000/*
1001 * void bn_mul_comba4(r,a,b)
1002 * BN_ULONG *r,*a,*b;
1003 */
1004bn_mul_comba4:
1005 save %sp,FRAME_SIZE,%sp
1006 lduw ap(0),a_0
1007 mov 1,t_2
1008 lduw bp(0),b_0
1009 sllx t_2,32,t_2 !=
1010 lduw bp(1),b_1
1011 mulx a_0,b_0,t_1 !mul_add_c(a[0],b[0],c1,c2,c3);
1012 srlx t_1,32,c_12
1013 stuw t_1,rp(0) !=!r[0]=c1;
1014
1015 lduw ap(1),a_1
1016 mulx a_0,b_1,t_1 !mul_add_c(a[0],b[1],c2,c3,c1);
1017 addcc c_12,t_1,c_12
1018 clr c_3 !=
1019 bcs,a %xcc,.+8
1020 add c_3,t_2,c_3
1021 lduw ap(2),a_2
1022 mulx a_1,b_0,t_1 !=!mul_add_c(a[1],b[0],c2,c3,c1);
1023 addcc c_12,t_1,t_1
1024 bcs,a %xcc,.+8
1025 add c_3,t_2,c_3
1026 srlx t_1,32,c_12 !=
1027 stuw t_1,rp(1) !r[1]=c2;
1028 or c_12,c_3,c_12
1029
1030 mulx a_2,b_0,t_1 !mul_add_c(a[2],b[0],c3,c1,c2);
1031 addcc c_12,t_1,c_12 !=
1032 clr c_3
1033 bcs,a %xcc,.+8
1034 add c_3,t_2,c_3
1035 lduw bp(2),b_2 !=
1036 mulx a_1,b_1,t_1 !mul_add_c(a[1],b[1],c3,c1,c2);
1037 addcc c_12,t_1,c_12
1038 bcs,a %xcc,.+8
1039 add c_3,t_2,c_3 !=
1040 lduw bp(3),b_3
1041 mulx a_0,b_2,t_1 !mul_add_c(a[0],b[2],c3,c1,c2);
1042 addcc c_12,t_1,t_1
1043 bcs,a %xcc,.+8 !=
1044 add c_3,t_2,c_3
1045 srlx t_1,32,c_12
1046 stuw t_1,rp(2) !r[2]=c3;
1047 or c_12,c_3,c_12 !=
1048
1049 mulx a_0,b_3,t_1 !mul_add_c(a[0],b[3],c1,c2,c3);
1050 addcc c_12,t_1,c_12
1051 clr c_3
1052 bcs,a %xcc,.+8 !=
1053 add c_3,t_2,c_3
1054 mulx a_1,b_2,t_1 !mul_add_c(a[1],b[2],c1,c2,c3);
1055 addcc c_12,t_1,c_12
1056 bcs,a %xcc,.+8 !=
1057 add c_3,t_2,c_3
1058 lduw ap(3),a_3
1059 mulx a_2,b_1,t_1 !mul_add_c(a[2],b[1],c1,c2,c3);
1060 addcc c_12,t_1,c_12 !=
1061 bcs,a %xcc,.+8
1062 add c_3,t_2,c_3
1063 mulx a_3,b_0,t_1 !mul_add_c(a[3],b[0],c1,c2,c3);!=
1064 addcc c_12,t_1,t_1 !=
1065 bcs,a %xcc,.+8
1066 add c_3,t_2,c_3
1067 srlx t_1,32,c_12
1068 stuw t_1,rp(3) !=!r[3]=c1;
1069 or c_12,c_3,c_12
1070
1071 mulx a_3,b_1,t_1 !mul_add_c(a[3],b[1],c2,c3,c1);
1072 addcc c_12,t_1,c_12
1073 clr c_3 !=
1074 bcs,a %xcc,.+8
1075 add c_3,t_2,c_3
1076 mulx a_2,b_2,t_1 !mul_add_c(a[2],b[2],c2,c3,c1);
1077 addcc c_12,t_1,c_12 !=
1078 bcs,a %xcc,.+8
1079 add c_3,t_2,c_3
1080 mulx a_1,b_3,t_1 !mul_add_c(a[1],b[3],c2,c3,c1);
1081 addcc c_12,t_1,t_1 !=
1082 bcs,a %xcc,.+8
1083 add c_3,t_2,c_3
1084 srlx t_1,32,c_12
1085 stuw t_1,rp(4) !=!r[4]=c2;
1086 or c_12,c_3,c_12
1087
1088 mulx a_2,b_3,t_1 !mul_add_c(a[2],b[3],c3,c1,c2);
1089 addcc c_12,t_1,c_12
1090 clr c_3 !=
1091 bcs,a %xcc,.+8
1092 add c_3,t_2,c_3
1093 mulx a_3,b_2,t_1 !mul_add_c(a[3],b[2],c3,c1,c2);
1094 addcc c_12,t_1,t_1 !=
1095 bcs,a %xcc,.+8
1096 add c_3,t_2,c_3
1097 srlx t_1,32,c_12
1098 stuw t_1,rp(5) !=!r[5]=c3;
1099 or c_12,c_3,c_12
1100
1101 mulx a_3,b_3,t_1 !mul_add_c(a[3],b[3],c1,c2,c3);
1102 addcc c_12,t_1,t_1
1103 srlx t_1,32,c_12 !=
1104 stuw t_1,rp(6) !r[6]=c1;
1105 stuw c_12,rp(7) !r[7]=c2;
1106
1107 ret
1108 restore %g0,%g0,%o0
1109
1110.type bn_mul_comba4,#function
1111.size bn_mul_comba4,(.-bn_mul_comba4)
1112
1113.align 32
1114
1115.global bn_sqr_comba8
1116bn_sqr_comba8:
1117 save %sp,FRAME_SIZE,%sp
1118 mov 1,t_2
1119 lduw ap(0),a_0
1120 sllx t_2,32,t_2
1121 lduw ap(1),a_1
1122 mulx a_0,a_0,t_1 !sqr_add_c(a,0,c1,c2,c3);
1123 srlx t_1,32,c_12
1124 stuw t_1,rp(0) !r[0]=c1;
1125
1126 lduw ap(2),a_2
1127 mulx a_0,a_1,t_1 !=!sqr_add_c2(a,1,0,c2,c3,c1);
1128 addcc c_12,t_1,c_12
1129 clr c_3
1130 bcs,a %xcc,.+8
1131 add c_3,t_2,c_3
1132 addcc c_12,t_1,t_1
1133 bcs,a %xcc,.+8
1134 add c_3,t_2,c_3
1135 srlx t_1,32,c_12
1136 stuw t_1,rp(1) !r[1]=c2;
1137 or c_12,c_3,c_12
1138
1139 mulx a_2,a_0,t_1 !sqr_add_c2(a,2,0,c3,c1,c2);
1140 addcc c_12,t_1,c_12
1141 clr c_3
1142 bcs,a %xcc,.+8
1143 add c_3,t_2,c_3
1144 addcc c_12,t_1,c_12
1145 bcs,a %xcc,.+8
1146 add c_3,t_2,c_3
1147 lduw ap(3),a_3
1148 mulx a_1,a_1,t_1 !sqr_add_c(a,1,c3,c1,c2);
1149 addcc c_12,t_1,t_1
1150 bcs,a %xcc,.+8
1151 add c_3,t_2,c_3
1152 srlx t_1,32,c_12
1153 stuw t_1,rp(2) !r[2]=c3;
1154 or c_12,c_3,c_12
1155
1156 mulx a_0,a_3,t_1 !sqr_add_c2(a,3,0,c1,c2,c3);
1157 addcc c_12,t_1,c_12
1158 clr c_3
1159 bcs,a %xcc,.+8
1160 add c_3,t_2,c_3
1161 addcc c_12,t_1,c_12
1162 bcs,a %xcc,.+8
1163 add c_3,t_2,c_3
1164 lduw ap(4),a_4
1165 mulx a_1,a_2,t_1 !sqr_add_c2(a,2,1,c1,c2,c3);
1166 addcc c_12,t_1,c_12
1167 bcs,a %xcc,.+8
1168 add c_3,t_2,c_3
1169 addcc c_12,t_1,t_1
1170 bcs,a %xcc,.+8
1171 add c_3,t_2,c_3
1172 srlx t_1,32,c_12
1173 st t_1,rp(3) !r[3]=c1;
1174 or c_12,c_3,c_12
1175
1176 mulx a_4,a_0,t_1 !sqr_add_c2(a,4,0,c2,c3,c1);
1177 addcc c_12,t_1,c_12
1178 clr c_3
1179 bcs,a %xcc,.+8
1180 add c_3,t_2,c_3
1181 addcc c_12,t_1,c_12
1182 bcs,a %xcc,.+8
1183 add c_3,t_2,c_3
1184 mulx a_3,a_1,t_1 !sqr_add_c2(a,3,1,c2,c3,c1);
1185 addcc c_12,t_1,c_12
1186 bcs,a %xcc,.+8
1187 add c_3,t_2,c_3
1188 addcc c_12,t_1,c_12
1189 bcs,a %xcc,.+8
1190 add c_3,t_2,c_3
1191 lduw ap(5),a_5
1192 mulx a_2,a_2,t_1 !sqr_add_c(a,2,c2,c3,c1);
1193 addcc c_12,t_1,t_1
1194 bcs,a %xcc,.+8
1195 add c_3,t_2,c_3
1196 srlx t_1,32,c_12
1197 stuw t_1,rp(4) !r[4]=c2;
1198 or c_12,c_3,c_12
1199
1200 mulx a_0,a_5,t_1 !sqr_add_c2(a,5,0,c3,c1,c2);
1201 addcc c_12,t_1,c_12
1202 clr c_3
1203 bcs,a %xcc,.+8
1204 add c_3,t_2,c_3
1205 addcc c_12,t_1,c_12
1206 bcs,a %xcc,.+8
1207 add c_3,t_2,c_3
1208 mulx a_1,a_4,t_1 !sqr_add_c2(a,4,1,c3,c1,c2);
1209 addcc c_12,t_1,c_12
1210 bcs,a %xcc,.+8
1211 add c_3,t_2,c_3
1212 addcc c_12,t_1,c_12
1213 bcs,a %xcc,.+8
1214 add c_3,t_2,c_3
1215 lduw ap(6),a_6
1216 mulx a_2,a_3,t_1 !sqr_add_c2(a,3,2,c3,c1,c2);
1217 addcc c_12,t_1,c_12
1218 bcs,a %xcc,.+8
1219 add c_3,t_2,c_3
1220 addcc c_12,t_1,t_1
1221 bcs,a %xcc,.+8
1222 add c_3,t_2,c_3
1223 srlx t_1,32,c_12
1224 stuw t_1,rp(5) !r[5]=c3;
1225 or c_12,c_3,c_12
1226
1227 mulx a_6,a_0,t_1 !sqr_add_c2(a,6,0,c1,c2,c3);
1228 addcc c_12,t_1,c_12
1229 clr c_3
1230 bcs,a %xcc,.+8
1231 add c_3,t_2,c_3
1232 addcc c_12,t_1,c_12
1233 bcs,a %xcc,.+8
1234 add c_3,t_2,c_3
1235 mulx a_5,a_1,t_1 !sqr_add_c2(a,5,1,c1,c2,c3);
1236 addcc c_12,t_1,c_12
1237 bcs,a %xcc,.+8
1238 add c_3,t_2,c_3
1239 addcc c_12,t_1,c_12
1240 bcs,a %xcc,.+8
1241 add c_3,t_2,c_3
1242 mulx a_4,a_2,t_1 !sqr_add_c2(a,4,2,c1,c2,c3);
1243 addcc c_12,t_1,c_12
1244 bcs,a %xcc,.+8
1245 add c_3,t_2,c_3
1246 addcc c_12,t_1,c_12
1247 bcs,a %xcc,.+8
1248 add c_3,t_2,c_3
1249 lduw ap(7),a_7
1250 mulx a_3,a_3,t_1 !=!sqr_add_c(a,3,c1,c2,c3);
1251 addcc c_12,t_1,t_1
1252 bcs,a %xcc,.+8
1253 add c_3,t_2,c_3
1254 srlx t_1,32,c_12
1255 stuw t_1,rp(6) !r[6]=c1;
1256 or c_12,c_3,c_12
1257
1258 mulx a_0,a_7,t_1 !sqr_add_c2(a,7,0,c2,c3,c1);
1259 addcc c_12,t_1,c_12
1260 clr c_3
1261 bcs,a %xcc,.+8
1262 add c_3,t_2,c_3
1263 addcc c_12,t_1,c_12
1264 bcs,a %xcc,.+8
1265 add c_3,t_2,c_3
1266 mulx a_1,a_6,t_1 !sqr_add_c2(a,6,1,c2,c3,c1);
1267 addcc c_12,t_1,c_12
1268 bcs,a %xcc,.+8
1269 add c_3,t_2,c_3
1270 addcc c_12,t_1,c_12
1271 bcs,a %xcc,.+8
1272 add c_3,t_2,c_3
1273 mulx a_2,a_5,t_1 !sqr_add_c2(a,5,2,c2,c3,c1);
1274 addcc c_12,t_1,c_12
1275 bcs,a %xcc,.+8
1276 add c_3,t_2,c_3
1277 addcc c_12,t_1,c_12
1278 bcs,a %xcc,.+8
1279 add c_3,t_2,c_3
1280 mulx a_3,a_4,t_1 !sqr_add_c2(a,4,3,c2,c3,c1);
1281 addcc c_12,t_1,c_12
1282 bcs,a %xcc,.+8
1283 add c_3,t_2,c_3
1284 addcc c_12,t_1,t_1
1285 bcs,a %xcc,.+8
1286 add c_3,t_2,c_3
1287 srlx t_1,32,c_12
1288 stuw t_1,rp(7) !r[7]=c2;
1289 or c_12,c_3,c_12
1290
1291 mulx a_7,a_1,t_1 !sqr_add_c2(a,7,1,c3,c1,c2);
1292 addcc c_12,t_1,c_12
1293 clr c_3
1294 bcs,a %xcc,.+8
1295 add c_3,t_2,c_3
1296 addcc c_12,t_1,c_12
1297 bcs,a %xcc,.+8
1298 add c_3,t_2,c_3
1299 mulx a_6,a_2,t_1 !sqr_add_c2(a,6,2,c3,c1,c2);
1300 addcc c_12,t_1,c_12
1301 bcs,a %xcc,.+8
1302 add c_3,t_2,c_3
1303 addcc c_12,t_1,c_12
1304 bcs,a %xcc,.+8
1305 add c_3,t_2,c_3
1306 mulx a_5,a_3,t_1 !sqr_add_c2(a,5,3,c3,c1,c2);
1307 addcc c_12,t_1,c_12
1308 bcs,a %xcc,.+8
1309 add c_3,t_2,c_3
1310 addcc c_12,t_1,c_12
1311 bcs,a %xcc,.+8
1312 add c_3,t_2,c_3
1313 mulx a_4,a_4,t_1 !sqr_add_c(a,4,c3,c1,c2);
1314 addcc c_12,t_1,t_1
1315 bcs,a %xcc,.+8
1316 add c_3,t_2,c_3
1317 srlx t_1,32,c_12
1318 stuw t_1,rp(8) !r[8]=c3;
1319 or c_12,c_3,c_12
1320
1321 mulx a_2,a_7,t_1 !sqr_add_c2(a,7,2,c1,c2,c3);
1322 addcc c_12,t_1,c_12
1323 clr c_3
1324 bcs,a %xcc,.+8
1325 add c_3,t_2,c_3
1326 addcc c_12,t_1,c_12
1327 bcs,a %xcc,.+8
1328 add c_3,t_2,c_3
1329 mulx a_3,a_6,t_1 !sqr_add_c2(a,6,3,c1,c2,c3);
1330 addcc c_12,t_1,c_12
1331 bcs,a %xcc,.+8
1332 add c_3,t_2,c_3
1333 addcc c_12,t_1,c_12
1334 bcs,a %xcc,.+8
1335 add c_3,t_2,c_3
1336 mulx a_4,a_5,t_1 !sqr_add_c2(a,5,4,c1,c2,c3);
1337 addcc c_12,t_1,c_12
1338 bcs,a %xcc,.+8
1339 add c_3,t_2,c_3
1340 addcc c_12,t_1,t_1
1341 bcs,a %xcc,.+8
1342 add c_3,t_2,c_3
1343 srlx t_1,32,c_12
1344 stuw t_1,rp(9) !r[9]=c1;
1345 or c_12,c_3,c_12
1346
1347 mulx a_7,a_3,t_1 !sqr_add_c2(a,7,3,c2,c3,c1);
1348 addcc c_12,t_1,c_12
1349 clr c_3
1350 bcs,a %xcc,.+8
1351 add c_3,t_2,c_3
1352 addcc c_12,t_1,c_12
1353 bcs,a %xcc,.+8
1354 add c_3,t_2,c_3
1355 mulx a_6,a_4,t_1 !sqr_add_c2(a,6,4,c2,c3,c1);
1356 addcc c_12,t_1,c_12
1357 bcs,a %xcc,.+8
1358 add c_3,t_2,c_3
1359 addcc c_12,t_1,c_12
1360 bcs,a %xcc,.+8
1361 add c_3,t_2,c_3
1362 mulx a_5,a_5,t_1 !sqr_add_c(a,5,c2,c3,c1);
1363 addcc c_12,t_1,t_1
1364 bcs,a %xcc,.+8
1365 add c_3,t_2,c_3
1366 srlx t_1,32,c_12
1367 stuw t_1,rp(10) !r[10]=c2;
1368 or c_12,c_3,c_12
1369
1370 mulx a_4,a_7,t_1 !sqr_add_c2(a,7,4,c3,c1,c2);
1371 addcc c_12,t_1,c_12
1372 clr c_3
1373 bcs,a %xcc,.+8
1374 add c_3,t_2,c_3
1375 addcc c_12,t_1,c_12
1376 bcs,a %xcc,.+8
1377 add c_3,t_2,c_3
1378 mulx a_5,a_6,t_1 !sqr_add_c2(a,6,5,c3,c1,c2);
1379 addcc c_12,t_1,c_12
1380 bcs,a %xcc,.+8
1381 add c_3,t_2,c_3
1382 addcc c_12,t_1,t_1
1383 bcs,a %xcc,.+8
1384 add c_3,t_2,c_3
1385 srlx t_1,32,c_12
1386 stuw t_1,rp(11) !r[11]=c3;
1387 or c_12,c_3,c_12
1388
1389 mulx a_7,a_5,t_1 !sqr_add_c2(a,7,5,c1,c2,c3);
1390 addcc c_12,t_1,c_12
1391 clr c_3
1392 bcs,a %xcc,.+8
1393 add c_3,t_2,c_3
1394 addcc c_12,t_1,c_12
1395 bcs,a %xcc,.+8
1396 add c_3,t_2,c_3
1397 mulx a_6,a_6,t_1 !sqr_add_c(a,6,c1,c2,c3);
1398 addcc c_12,t_1,t_1
1399 bcs,a %xcc,.+8
1400 add c_3,t_2,c_3
1401 srlx t_1,32,c_12
1402 stuw t_1,rp(12) !r[12]=c1;
1403 or c_12,c_3,c_12
1404
1405 mulx a_6,a_7,t_1 !sqr_add_c2(a,7,6,c2,c3,c1);
1406 addcc c_12,t_1,c_12
1407 clr c_3
1408 bcs,a %xcc,.+8
1409 add c_3,t_2,c_3
1410 addcc c_12,t_1,t_1
1411 bcs,a %xcc,.+8
1412 add c_3,t_2,c_3
1413 srlx t_1,32,c_12
1414 stuw t_1,rp(13) !r[13]=c2;
1415 or c_12,c_3,c_12
1416
1417 mulx a_7,a_7,t_1 !sqr_add_c(a,7,c3,c1,c2);
1418 addcc c_12,t_1,t_1
1419 srlx t_1,32,c_12
1420 stuw t_1,rp(14) !r[14]=c3;
1421 stuw c_12,rp(15) !r[15]=c1;
1422
1423 ret
1424 restore %g0,%g0,%o0
1425
1426.type bn_sqr_comba8,#function
1427.size bn_sqr_comba8,(.-bn_sqr_comba8)
1428
1429.align 32
1430
1431.global bn_sqr_comba4
1432/*
1433 * void bn_sqr_comba4(r,a)
1434 * BN_ULONG *r,*a;
1435 */
1436bn_sqr_comba4:
1437 save %sp,FRAME_SIZE,%sp
1438 mov 1,t_2
1439 lduw ap(0),a_0
1440 sllx t_2,32,t_2
1441 lduw ap(1),a_1
1442 mulx a_0,a_0,t_1 !sqr_add_c(a,0,c1,c2,c3);
1443 srlx t_1,32,c_12
1444 stuw t_1,rp(0) !r[0]=c1;
1445
1446 lduw ap(2),a_2
1447 mulx a_0,a_1,t_1 !sqr_add_c2(a,1,0,c2,c3,c1);
1448 addcc c_12,t_1,c_12
1449 clr c_3
1450 bcs,a %xcc,.+8
1451 add c_3,t_2,c_3
1452 addcc c_12,t_1,t_1
1453 bcs,a %xcc,.+8
1454 add c_3,t_2,c_3
1455 srlx t_1,32,c_12
1456 stuw t_1,rp(1) !r[1]=c2;
1457 or c_12,c_3,c_12
1458
1459 mulx a_2,a_0,t_1 !sqr_add_c2(a,2,0,c3,c1,c2);
1460 addcc c_12,t_1,c_12
1461 clr c_3
1462 bcs,a %xcc,.+8
1463 add c_3,t_2,c_3
1464 addcc c_12,t_1,c_12
1465 bcs,a %xcc,.+8
1466 add c_3,t_2,c_3
1467 lduw ap(3),a_3
1468 mulx a_1,a_1,t_1 !sqr_add_c(a,1,c3,c1,c2);
1469 addcc c_12,t_1,t_1
1470 bcs,a %xcc,.+8
1471 add c_3,t_2,c_3
1472 srlx t_1,32,c_12
1473 stuw t_1,rp(2) !r[2]=c3;
1474 or c_12,c_3,c_12
1475
1476 mulx a_0,a_3,t_1 !sqr_add_c2(a,3,0,c1,c2,c3);
1477 addcc c_12,t_1,c_12
1478 clr c_3
1479 bcs,a %xcc,.+8
1480 add c_3,t_2,c_3
1481 addcc c_12,t_1,c_12
1482 bcs,a %xcc,.+8
1483 add c_3,t_2,c_3
1484 mulx a_1,a_2,t_1 !sqr_add_c2(a,2,1,c1,c2,c3);
1485 addcc c_12,t_1,c_12
1486 bcs,a %xcc,.+8
1487 add c_3,t_2,c_3
1488 addcc c_12,t_1,t_1
1489 bcs,a %xcc,.+8
1490 add c_3,t_2,c_3
1491 srlx t_1,32,c_12
1492 stuw t_1,rp(3) !r[3]=c1;
1493 or c_12,c_3,c_12
1494
1495 mulx a_3,a_1,t_1 !sqr_add_c2(a,3,1,c2,c3,c1);
1496 addcc c_12,t_1,c_12
1497 clr c_3
1498 bcs,a %xcc,.+8
1499 add c_3,t_2,c_3
1500 addcc c_12,t_1,c_12
1501 bcs,a %xcc,.+8
1502 add c_3,t_2,c_3
1503 mulx a_2,a_2,t_1 !sqr_add_c(a,2,c2,c3,c1);
1504 addcc c_12,t_1,t_1
1505 bcs,a %xcc,.+8
1506 add c_3,t_2,c_3
1507 srlx t_1,32,c_12
1508 stuw t_1,rp(4) !r[4]=c2;
1509 or c_12,c_3,c_12
1510
1511 mulx a_2,a_3,t_1 !sqr_add_c2(a,3,2,c3,c1,c2);
1512 addcc c_12,t_1,c_12
1513 clr c_3
1514 bcs,a %xcc,.+8
1515 add c_3,t_2,c_3
1516 addcc c_12,t_1,t_1
1517 bcs,a %xcc,.+8
1518 add c_3,t_2,c_3
1519 srlx t_1,32,c_12
1520 stuw t_1,rp(5) !r[5]=c3;
1521 or c_12,c_3,c_12
1522
1523 mulx a_3,a_3,t_1 !sqr_add_c(a,3,c1,c2,c3);
1524 addcc c_12,t_1,t_1
1525 srlx t_1,32,c_12
1526 stuw t_1,rp(6) !r[6]=c1;
1527 stuw c_12,rp(7) !r[7]=c2;
1528
1529 ret
1530 restore %g0,%g0,%o0
1531
1532.type bn_sqr_comba4,#function
1533.size bn_sqr_comba4,(.-bn_sqr_comba4)
1534
1535.align 32
diff --git a/src/lib/libcrypto/bn/asm/x86.pl b/src/lib/libcrypto/bn/asm/x86.pl
deleted file mode 100644
index 1bc4f1bb27..0000000000
--- a/src/lib/libcrypto/bn/asm/x86.pl
+++ /dev/null
@@ -1,28 +0,0 @@
1#!/usr/local/bin/perl
2
3push(@INC,"perlasm","../../perlasm");
4require "x86asm.pl";
5
6require("x86/mul_add.pl");
7require("x86/mul.pl");
8require("x86/sqr.pl");
9require("x86/div.pl");
10require("x86/add.pl");
11require("x86/sub.pl");
12require("x86/comba.pl");
13
14&asm_init($ARGV[0],$0);
15
16&bn_mul_add_words("bn_mul_add_words");
17&bn_mul_words("bn_mul_words");
18&bn_sqr_words("bn_sqr_words");
19&bn_div_words("bn_div_words");
20&bn_add_words("bn_add_words");
21&bn_sub_words("bn_sub_words");
22&bn_mul_comba("bn_mul_comba8",8);
23&bn_mul_comba("bn_mul_comba4",4);
24&bn_sqr_comba("bn_sqr_comba8",8);
25&bn_sqr_comba("bn_sqr_comba4",4);
26
27&asm_finish();
28
diff --git a/src/lib/libcrypto/bn/asm/x86/add.pl b/src/lib/libcrypto/bn/asm/x86/add.pl
deleted file mode 100644
index 0b5cf583e3..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/add.pl
+++ /dev/null
@@ -1,76 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub bn_add_words
5 {
6 local($name)=@_;
7
8 &function_begin($name,"");
9
10 &comment("");
11 $a="esi";
12 $b="edi";
13 $c="eax";
14 $r="ebx";
15 $tmp1="ecx";
16 $tmp2="edx";
17 $num="ebp";
18
19 &mov($r,&wparam(0)); # get r
20 &mov($a,&wparam(1)); # get a
21 &mov($b,&wparam(2)); # get b
22 &mov($num,&wparam(3)); # get num
23 &xor($c,$c); # clear carry
24 &and($num,0xfffffff8); # num / 8
25
26 &jz(&label("aw_finish"));
27
28 &set_label("aw_loop",0);
29 for ($i=0; $i<8; $i++)
30 {
31 &comment("Round $i");
32
33 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
34 &mov($tmp2,&DWP($i*4,$b,"",0)); # *b
35 &add($tmp1,$c);
36 &mov($c,0);
37 &adc($c,$c);
38 &add($tmp1,$tmp2);
39 &adc($c,0);
40 &mov(&DWP($i*4,$r,"",0),$tmp1); # *r
41 }
42
43 &comment("");
44 &add($a,32);
45 &add($b,32);
46 &add($r,32);
47 &sub($num,8);
48 &jnz(&label("aw_loop"));
49
50 &set_label("aw_finish",0);
51 &mov($num,&wparam(3)); # get num
52 &and($num,7);
53 &jz(&label("aw_end"));
54
55 for ($i=0; $i<7; $i++)
56 {
57 &comment("Tail Round $i");
58 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
59 &mov($tmp2,&DWP($i*4,$b,"",0));# *b
60 &add($tmp1,$c);
61 &mov($c,0);
62 &adc($c,$c);
63 &add($tmp1,$tmp2);
64 &adc($c,0);
65 &dec($num) if ($i != 6);
66 &mov(&DWP($i*4,$r,"",0),$tmp1); # *a
67 &jz(&label("aw_end")) if ($i != 6);
68 }
69 &set_label("aw_end",0);
70
71# &mov("eax",$c); # $c is "eax"
72
73 &function_end($name);
74 }
75
761;
diff --git a/src/lib/libcrypto/bn/asm/x86/comba.pl b/src/lib/libcrypto/bn/asm/x86/comba.pl
deleted file mode 100644
index 2291253629..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/comba.pl
+++ /dev/null
@@ -1,277 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub mul_add_c
5 {
6 local($a,$ai,$b,$bi,$c0,$c1,$c2,$pos,$i,$na,$nb)=@_;
7
8 # pos == -1 if eax and edx are pre-loaded, 0 to load from next
9 # words, and 1 if load return value
10
11 &comment("mul a[$ai]*b[$bi]");
12
13 # "eax" and "edx" will always be pre-loaded.
14 # &mov("eax",&DWP($ai*4,$a,"",0)) ;
15 # &mov("edx",&DWP($bi*4,$b,"",0));
16
17 &mul("edx");
18 &add($c0,"eax");
19 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 0; # laod next a
20 &mov("eax",&wparam(0)) if $pos > 0; # load r[]
21 ###
22 &adc($c1,"edx");
23 &mov("edx",&DWP(($nb)*4,$b,"",0)) if $pos == 0; # laod next b
24 &mov("edx",&DWP(($nb)*4,$b,"",0)) if $pos == 1; # laod next b
25 ###
26 &adc($c2,0);
27 # is pos > 1, it means it is the last loop
28 &mov(&DWP($i*4,"eax","",0),$c0) if $pos > 0; # save r[];
29 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 1; # laod next a
30 }
31
32sub sqr_add_c
33 {
34 local($r,$a,$ai,$bi,$c0,$c1,$c2,$pos,$i,$na,$nb)=@_;
35
36 # pos == -1 if eax and edx are pre-loaded, 0 to load from next
37 # words, and 1 if load return value
38
39 &comment("sqr a[$ai]*a[$bi]");
40
41 # "eax" and "edx" will always be pre-loaded.
42 # &mov("eax",&DWP($ai*4,$a,"",0)) ;
43 # &mov("edx",&DWP($bi*4,$b,"",0));
44
45 if ($ai == $bi)
46 { &mul("eax");}
47 else
48 { &mul("edx");}
49 &add($c0,"eax");
50 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 0; # load next a
51 ###
52 &adc($c1,"edx");
53 &mov("edx",&DWP(($nb)*4,$a,"",0)) if ($pos == 1) && ($na != $nb);
54 ###
55 &adc($c2,0);
56 # is pos > 1, it means it is the last loop
57 &mov(&DWP($i*4,$r,"",0),$c0) if $pos > 0; # save r[];
58 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 1; # load next b
59 }
60
61sub sqr_add_c2
62 {
63 local($r,$a,$ai,$bi,$c0,$c1,$c2,$pos,$i,$na,$nb)=@_;
64
65 # pos == -1 if eax and edx are pre-loaded, 0 to load from next
66 # words, and 1 if load return value
67
68 &comment("sqr a[$ai]*a[$bi]");
69
70 # "eax" and "edx" will always be pre-loaded.
71 # &mov("eax",&DWP($ai*4,$a,"",0)) ;
72 # &mov("edx",&DWP($bi*4,$a,"",0));
73
74 if ($ai == $bi)
75 { &mul("eax");}
76 else
77 { &mul("edx");}
78 &add("eax","eax");
79 ###
80 &adc("edx","edx");
81 ###
82 &adc($c2,0);
83 &add($c0,"eax");
84 &adc($c1,"edx");
85 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 0; # load next a
86 &mov("eax",&DWP(($na)*4,$a,"",0)) if $pos == 1; # load next b
87 &adc($c2,0);
88 &mov(&DWP($i*4,$r,"",0),$c0) if $pos > 0; # save r[];
89 &mov("edx",&DWP(($nb)*4,$a,"",0)) if ($pos <= 1) && ($na != $nb);
90 ###
91 }
92
93sub bn_mul_comba
94 {
95 local($name,$num)=@_;
96 local($a,$b,$c0,$c1,$c2);
97 local($i,$as,$ae,$bs,$be,$ai,$bi);
98 local($tot,$end);
99
100 &function_begin_B($name,"");
101
102 $c0="ebx";
103 $c1="ecx";
104 $c2="ebp";
105 $a="esi";
106 $b="edi";
107
108 $as=0;
109 $ae=0;
110 $bs=0;
111 $be=0;
112 $tot=$num+$num-1;
113
114 &push("esi");
115 &mov($a,&wparam(1));
116 &push("edi");
117 &mov($b,&wparam(2));
118 &push("ebp");
119 &push("ebx");
120
121 &xor($c0,$c0);
122 &mov("eax",&DWP(0,$a,"",0)); # load the first word
123 &xor($c1,$c1);
124 &mov("edx",&DWP(0,$b,"",0)); # load the first second
125
126 for ($i=0; $i<$tot; $i++)
127 {
128 $ai=$as;
129 $bi=$bs;
130 $end=$be+1;
131
132 &comment("################## Calculate word $i");
133
134 for ($j=$bs; $j<$end; $j++)
135 {
136 &xor($c2,$c2) if ($j == $bs);
137 if (($j+1) == $end)
138 {
139 $v=1;
140 $v=2 if (($i+1) == $tot);
141 }
142 else
143 { $v=0; }
144 if (($j+1) != $end)
145 {
146 $na=($ai-1);
147 $nb=($bi+1);
148 }
149 else
150 {
151 $na=$as+($i < ($num-1));
152 $nb=$bs+($i >= ($num-1));
153 }
154#printf STDERR "[$ai,$bi] -> [$na,$nb]\n";
155 &mul_add_c($a,$ai,$b,$bi,$c0,$c1,$c2,$v,$i,$na,$nb);
156 if ($v)
157 {
158 &comment("saved r[$i]");
159 # &mov("eax",&wparam(0));
160 # &mov(&DWP($i*4,"eax","",0),$c0);
161 ($c0,$c1,$c2)=($c1,$c2,$c0);
162 }
163 $ai--;
164 $bi++;
165 }
166 $as++ if ($i < ($num-1));
167 $ae++ if ($i >= ($num-1));
168
169 $bs++ if ($i >= ($num-1));
170 $be++ if ($i < ($num-1));
171 }
172 &comment("save r[$i]");
173 # &mov("eax",&wparam(0));
174 &mov(&DWP($i*4,"eax","",0),$c0);
175
176 &pop("ebx");
177 &pop("ebp");
178 &pop("edi");
179 &pop("esi");
180 &ret();
181 &function_end_B($name);
182 }
183
184sub bn_sqr_comba
185 {
186 local($name,$num)=@_;
187 local($r,$a,$c0,$c1,$c2)=@_;
188 local($i,$as,$ae,$bs,$be,$ai,$bi);
189 local($b,$tot,$end,$half);
190
191 &function_begin_B($name,"");
192
193 $c0="ebx";
194 $c1="ecx";
195 $c2="ebp";
196 $a="esi";
197 $r="edi";
198
199 &push("esi");
200 &push("edi");
201 &push("ebp");
202 &push("ebx");
203 &mov($r,&wparam(0));
204 &mov($a,&wparam(1));
205 &xor($c0,$c0);
206 &xor($c1,$c1);
207 &mov("eax",&DWP(0,$a,"",0)); # load the first word
208
209 $as=0;
210 $ae=0;
211 $bs=0;
212 $be=0;
213 $tot=$num+$num-1;
214
215 for ($i=0; $i<$tot; $i++)
216 {
217 $ai=$as;
218 $bi=$bs;
219 $end=$be+1;
220
221 &comment("############### Calculate word $i");
222 for ($j=$bs; $j<$end; $j++)
223 {
224 &xor($c2,$c2) if ($j == $bs);
225 if (($ai-1) < ($bi+1))
226 {
227 $v=1;
228 $v=2 if ($i+1) == $tot;
229 }
230 else
231 { $v=0; }
232 if (!$v)
233 {
234 $na=$ai-1;
235 $nb=$bi+1;
236 }
237 else
238 {
239 $na=$as+($i < ($num-1));
240 $nb=$bs+($i >= ($num-1));
241 }
242 if ($ai == $bi)
243 {
244 &sqr_add_c($r,$a,$ai,$bi,
245 $c0,$c1,$c2,$v,$i,$na,$nb);
246 }
247 else
248 {
249 &sqr_add_c2($r,$a,$ai,$bi,
250 $c0,$c1,$c2,$v,$i,$na,$nb);
251 }
252 if ($v)
253 {
254 &comment("saved r[$i]");
255 #&mov(&DWP($i*4,$r,"",0),$c0);
256 ($c0,$c1,$c2)=($c1,$c2,$c0);
257 last;
258 }
259 $ai--;
260 $bi++;
261 }
262 $as++ if ($i < ($num-1));
263 $ae++ if ($i >= ($num-1));
264
265 $bs++ if ($i >= ($num-1));
266 $be++ if ($i < ($num-1));
267 }
268 &mov(&DWP($i*4,$r,"",0),$c0);
269 &pop("ebx");
270 &pop("ebp");
271 &pop("edi");
272 &pop("esi");
273 &ret();
274 &function_end_B($name);
275 }
276
2771;
diff --git a/src/lib/libcrypto/bn/asm/x86/div.pl b/src/lib/libcrypto/bn/asm/x86/div.pl
deleted file mode 100644
index 0e90152caa..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/div.pl
+++ /dev/null
@@ -1,15 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub bn_div_words
5 {
6 local($name)=@_;
7
8 &function_begin($name,"");
9 &mov("edx",&wparam(0)); #
10 &mov("eax",&wparam(1)); #
11 &mov("ebx",&wparam(2)); #
12 &div("ebx");
13 &function_end($name);
14 }
151;
diff --git a/src/lib/libcrypto/bn/asm/x86/mul.pl b/src/lib/libcrypto/bn/asm/x86/mul.pl
deleted file mode 100644
index 674cb9b055..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/mul.pl
+++ /dev/null
@@ -1,77 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub bn_mul_words
5 {
6 local($name)=@_;
7
8 &function_begin($name,"");
9
10 &comment("");
11 $Low="eax";
12 $High="edx";
13 $a="ebx";
14 $w="ecx";
15 $r="edi";
16 $c="esi";
17 $num="ebp";
18
19 &xor($c,$c); # clear carry
20 &mov($r,&wparam(0)); #
21 &mov($a,&wparam(1)); #
22 &mov($num,&wparam(2)); #
23 &mov($w,&wparam(3)); #
24
25 &and($num,0xfffffff8); # num / 8
26 &jz(&label("mw_finish"));
27
28 &set_label("mw_loop",0);
29 for ($i=0; $i<32; $i+=4)
30 {
31 &comment("Round $i");
32
33 &mov("eax",&DWP($i,$a,"",0)); # *a
34 &mul($w); # *a * w
35 &add("eax",$c); # L(t)+=c
36 # XXX
37
38 &adc("edx",0); # H(t)+=carry
39 &mov(&DWP($i,$r,"",0),"eax"); # *r= L(t);
40
41 &mov($c,"edx"); # c= H(t);
42 }
43
44 &comment("");
45 &add($a,32);
46 &add($r,32);
47 &sub($num,8);
48 &jz(&label("mw_finish"));
49 &jmp(&label("mw_loop"));
50
51 &set_label("mw_finish",0);
52 &mov($num,&wparam(2)); # get num
53 &and($num,7);
54 &jnz(&label("mw_finish2"));
55 &jmp(&label("mw_end"));
56
57 &set_label("mw_finish2",1);
58 for ($i=0; $i<7; $i++)
59 {
60 &comment("Tail Round $i");
61 &mov("eax",&DWP($i*4,$a,"",0));# *a
62 &mul($w); # *a * w
63 &add("eax",$c); # L(t)+=c
64 # XXX
65 &adc("edx",0); # H(t)+=carry
66 &mov(&DWP($i*4,$r,"",0),"eax");# *r= L(t);
67 &mov($c,"edx"); # c= H(t);
68 &dec($num) if ($i != 7-1);
69 &jz(&label("mw_end")) if ($i != 7-1);
70 }
71 &set_label("mw_end",0);
72 &mov("eax",$c);
73
74 &function_end($name);
75 }
76
771;
diff --git a/src/lib/libcrypto/bn/asm/x86/mul_add.pl b/src/lib/libcrypto/bn/asm/x86/mul_add.pl
deleted file mode 100644
index 61830d3a90..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/mul_add.pl
+++ /dev/null
@@ -1,87 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub bn_mul_add_words
5 {
6 local($name)=@_;
7
8 &function_begin($name,"");
9
10 &comment("");
11 $Low="eax";
12 $High="edx";
13 $a="ebx";
14 $w="ebp";
15 $r="edi";
16 $c="esi";
17
18 &xor($c,$c); # clear carry
19 &mov($r,&wparam(0)); #
20
21 &mov("ecx",&wparam(2)); #
22 &mov($a,&wparam(1)); #
23
24 &and("ecx",0xfffffff8); # num / 8
25 &mov($w,&wparam(3)); #
26
27 &push("ecx"); # Up the stack for a tmp variable
28
29 &jz(&label("maw_finish"));
30
31 &set_label("maw_loop",0);
32
33 &mov(&swtmp(0),"ecx"); #
34
35 for ($i=0; $i<32; $i+=4)
36 {
37 &comment("Round $i");
38
39 &mov("eax",&DWP($i,$a,"",0)); # *a
40 &mul($w); # *a * w
41 &add("eax",$c); # L(t)+= *r
42 &mov($c,&DWP($i,$r,"",0)); # L(t)+= *r
43 &adc("edx",0); # H(t)+=carry
44 &add("eax",$c); # L(t)+=c
45 &adc("edx",0); # H(t)+=carry
46 &mov(&DWP($i,$r,"",0),"eax"); # *r= L(t);
47 &mov($c,"edx"); # c= H(t);
48 }
49
50 &comment("");
51 &mov("ecx",&swtmp(0)); #
52 &add($a,32);
53 &add($r,32);
54 &sub("ecx",8);
55 &jnz(&label("maw_loop"));
56
57 &set_label("maw_finish",0);
58 &mov("ecx",&wparam(2)); # get num
59 &and("ecx",7);
60 &jnz(&label("maw_finish2")); # helps branch prediction
61 &jmp(&label("maw_end"));
62
63 &set_label("maw_finish2",1);
64 for ($i=0; $i<7; $i++)
65 {
66 &comment("Tail Round $i");
67 &mov("eax",&DWP($i*4,$a,"",0));# *a
68 &mul($w); # *a * w
69 &add("eax",$c); # L(t)+=c
70 &mov($c,&DWP($i*4,$r,"",0)); # L(t)+= *r
71 &adc("edx",0); # H(t)+=carry
72 &add("eax",$c);
73 &adc("edx",0); # H(t)+=carry
74 &dec("ecx") if ($i != 7-1);
75 &mov(&DWP($i*4,$r,"",0),"eax"); # *r= L(t);
76 &mov($c,"edx"); # c= H(t);
77 &jz(&label("maw_end")) if ($i != 7-1);
78 }
79 &set_label("maw_end",0);
80 &mov("eax",$c);
81
82 &pop("ecx"); # clear variable from
83
84 &function_end($name);
85 }
86
871;
diff --git a/src/lib/libcrypto/bn/asm/x86/sqr.pl b/src/lib/libcrypto/bn/asm/x86/sqr.pl
deleted file mode 100644
index 1f90993cf6..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/sqr.pl
+++ /dev/null
@@ -1,60 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub bn_sqr_words
5 {
6 local($name)=@_;
7
8 &function_begin($name,"");
9
10 &comment("");
11 $r="esi";
12 $a="edi";
13 $num="ebx";
14
15 &mov($r,&wparam(0)); #
16 &mov($a,&wparam(1)); #
17 &mov($num,&wparam(2)); #
18
19 &and($num,0xfffffff8); # num / 8
20 &jz(&label("sw_finish"));
21
22 &set_label("sw_loop",0);
23 for ($i=0; $i<32; $i+=4)
24 {
25 &comment("Round $i");
26 &mov("eax",&DWP($i,$a,"",0)); # *a
27 # XXX
28 &mul("eax"); # *a * *a
29 &mov(&DWP($i*2,$r,"",0),"eax"); #
30 &mov(&DWP($i*2+4,$r,"",0),"edx");#
31 }
32
33 &comment("");
34 &add($a,32);
35 &add($r,64);
36 &sub($num,8);
37 &jnz(&label("sw_loop"));
38
39 &set_label("sw_finish",0);
40 &mov($num,&wparam(2)); # get num
41 &and($num,7);
42 &jz(&label("sw_end"));
43
44 for ($i=0; $i<7; $i++)
45 {
46 &comment("Tail Round $i");
47 &mov("eax",&DWP($i*4,$a,"",0)); # *a
48 # XXX
49 &mul("eax"); # *a * *a
50 &mov(&DWP($i*8,$r,"",0),"eax"); #
51 &dec($num) if ($i != 7-1);
52 &mov(&DWP($i*8+4,$r,"",0),"edx");
53 &jz(&label("sw_end")) if ($i != 7-1);
54 }
55 &set_label("sw_end",0);
56
57 &function_end($name);
58 }
59
601;
diff --git a/src/lib/libcrypto/bn/asm/x86/sub.pl b/src/lib/libcrypto/bn/asm/x86/sub.pl
deleted file mode 100644
index 837b0e1b07..0000000000
--- a/src/lib/libcrypto/bn/asm/x86/sub.pl
+++ /dev/null
@@ -1,76 +0,0 @@
1#!/usr/local/bin/perl
2# x86 assember
3
4sub bn_sub_words
5 {
6 local($name)=@_;
7
8 &function_begin($name,"");
9
10 &comment("");
11 $a="esi";
12 $b="edi";
13 $c="eax";
14 $r="ebx";
15 $tmp1="ecx";
16 $tmp2="edx";
17 $num="ebp";
18
19 &mov($r,&wparam(0)); # get r
20 &mov($a,&wparam(1)); # get a
21 &mov($b,&wparam(2)); # get b
22 &mov($num,&wparam(3)); # get num
23 &xor($c,$c); # clear carry
24 &and($num,0xfffffff8); # num / 8
25
26 &jz(&label("aw_finish"));
27
28 &set_label("aw_loop",0);
29 for ($i=0; $i<8; $i++)
30 {
31 &comment("Round $i");
32
33 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
34 &mov($tmp2,&DWP($i*4,$b,"",0)); # *b
35 &sub($tmp1,$c);
36 &mov($c,0);
37 &adc($c,$c);
38 &sub($tmp1,$tmp2);
39 &adc($c,0);
40 &mov(&DWP($i*4,$r,"",0),$tmp1); # *r
41 }
42
43 &comment("");
44 &add($a,32);
45 &add($b,32);
46 &add($r,32);
47 &sub($num,8);
48 &jnz(&label("aw_loop"));
49
50 &set_label("aw_finish",0);
51 &mov($num,&wparam(3)); # get num
52 &and($num,7);
53 &jz(&label("aw_end"));
54
55 for ($i=0; $i<7; $i++)
56 {
57 &comment("Tail Round $i");
58 &mov($tmp1,&DWP($i*4,$a,"",0)); # *a
59 &mov($tmp2,&DWP($i*4,$b,"",0));# *b
60 &sub($tmp1,$c);
61 &mov($c,0);
62 &adc($c,$c);
63 &sub($tmp1,$tmp2);
64 &adc($c,0);
65 &dec($num) if ($i != 6);
66 &mov(&DWP($i*4,$r,"",0),$tmp1); # *a
67 &jz(&label("aw_end")) if ($i != 6);
68 }
69 &set_label("aw_end",0);
70
71# &mov("eax",$c); # $c is "eax"
72
73 &function_end($name);
74 }
75
761;
diff --git a/src/lib/libcrypto/bn/bn.h b/src/lib/libcrypto/bn/bn.h
deleted file mode 100644
index b232c2ceae..0000000000
--- a/src/lib/libcrypto/bn/bn.h
+++ /dev/null
@@ -1,520 +0,0 @@
1/* crypto/bn/bn.h */
2/* Copyright (C) 1995-1997 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#ifndef HEADER_BN_H
60#define HEADER_BN_H
61
62#ifndef NO_FP_API
63#include <stdio.h> /* FILE */
64#endif
65#include <openssl/opensslconf.h>
66
67#ifdef __cplusplus
68extern "C" {
69#endif
70
71#ifdef VMS
72#undef BN_LLONG /* experimental, so far... */
73#endif
74
75#define BN_MUL_COMBA
76#define BN_SQR_COMBA
77#define BN_RECURSION
78#define RECP_MUL_MOD
79#define MONT_MUL_MOD
80
81/* This next option uses the C libraries (2 word)/(1 word) function.
82 * If it is not defined, I use my C version (which is slower).
83 * The reason for this flag is that when the particular C compiler
84 * library routine is used, and the library is linked with a different
85 * compiler, the library is missing. This mostly happens when the
86 * library is built with gcc and then linked using normal cc. This would
87 * be a common occurrence because gcc normally produces code that is
88 * 2 times faster than system compilers for the big number stuff.
89 * For machines with only one compiler (or shared libraries), this should
90 * be on. Again this in only really a problem on machines
91 * using "long long's", are 32bit, and are not using my assembler code. */
92#if defined(MSDOS) || defined(WINDOWS) || defined(WIN32) || defined(linux)
93#define BN_DIV2W
94#endif
95
96/* assuming long is 64bit - this is the DEC Alpha
97 * unsigned long long is only 64 bits :-(, don't define
98 * BN_LLONG for the DEC Alpha */
99#ifdef SIXTY_FOUR_BIT_LONG
100#define BN_ULLONG unsigned long long
101#define BN_ULONG unsigned long
102#define BN_LONG long
103#define BN_BITS 128
104#define BN_BYTES 8
105#define BN_BITS2 64
106#define BN_BITS4 32
107#define BN_MASK (0xffffffffffffffffffffffffffffffffLL)
108#define BN_MASK2 (0xffffffffffffffffL)
109#define BN_MASK2l (0xffffffffL)
110#define BN_MASK2h (0xffffffff00000000L)
111#define BN_MASK2h1 (0xffffffff80000000L)
112#define BN_TBIT (0x8000000000000000L)
113#define BN_DEC_CONV (10000000000000000000UL)
114#define BN_DEC_FMT1 "%lu"
115#define BN_DEC_FMT2 "%019lu"
116#define BN_DEC_NUM 19
117#endif
118
119/* This is where the long long data type is 64 bits, but long is 32.
120 * For machines where there are 64bit registers, this is the mode to use.
121 * IRIX, on R4000 and above should use this mode, along with the relevant
122 * assembler code :-). Do NOT define BN_LLONG.
123 */
124#ifdef SIXTY_FOUR_BIT
125#undef BN_LLONG
126#undef BN_ULLONG
127#define BN_ULONG unsigned long long
128#define BN_LONG long long
129#define BN_BITS 128
130#define BN_BYTES 8
131#define BN_BITS2 64
132#define BN_BITS4 32
133#define BN_MASK2 (0xffffffffffffffffLL)
134#define BN_MASK2l (0xffffffffL)
135#define BN_MASK2h (0xffffffff00000000LL)
136#define BN_MASK2h1 (0xffffffff80000000LL)
137#define BN_TBIT (0x8000000000000000LL)
138#define BN_DEC_CONV (10000000000000000000LL)
139#define BN_DEC_FMT1 "%llu"
140#define BN_DEC_FMT2 "%019llu"
141#define BN_DEC_NUM 19
142#endif
143
144#ifdef THIRTY_TWO_BIT
145#if defined(WIN32) && !defined(__GNUC__)
146#define BN_ULLONG unsigned _int64
147#else
148#define BN_ULLONG unsigned long long
149#endif
150#define BN_ULONG unsigned long
151#define BN_LONG long
152#define BN_BITS 64
153#define BN_BYTES 4
154#define BN_BITS2 32
155#define BN_BITS4 16
156#ifdef WIN32
157/* VC++ doesn't like the LL suffix */
158#define BN_MASK (0xffffffffffffffffL)
159#else
160#define BN_MASK (0xffffffffffffffffLL)
161#endif
162#define BN_MASK2 (0xffffffffL)
163#define BN_MASK2l (0xffff)
164#define BN_MASK2h1 (0xffff8000L)
165#define BN_MASK2h (0xffff0000L)
166#define BN_TBIT (0x80000000L)
167#define BN_DEC_CONV (1000000000L)
168#define BN_DEC_FMT1 "%lu"
169#define BN_DEC_FMT2 "%09lu"
170#define BN_DEC_NUM 9
171#endif
172
173#ifdef SIXTEEN_BIT
174#ifndef BN_DIV2W
175#define BN_DIV2W
176#endif
177#define BN_ULLONG unsigned long
178#define BN_ULONG unsigned short
179#define BN_LONG short
180#define BN_BITS 32
181#define BN_BYTES 2
182#define BN_BITS2 16
183#define BN_BITS4 8
184#define BN_MASK (0xffffffff)
185#define BN_MASK2 (0xffff)
186#define BN_MASK2l (0xff)
187#define BN_MASK2h1 (0xff80)
188#define BN_MASK2h (0xff00)
189#define BN_TBIT (0x8000)
190#define BN_DEC_CONV (100000)
191#define BN_DEC_FMT1 "%u"
192#define BN_DEC_FMT2 "%05u"
193#define BN_DEC_NUM 5
194#endif
195
196#ifdef EIGHT_BIT
197#ifndef BN_DIV2W
198#define BN_DIV2W
199#endif
200#define BN_ULLONG unsigned short
201#define BN_ULONG unsigned char
202#define BN_LONG char
203#define BN_BITS 16
204#define BN_BYTES 1
205#define BN_BITS2 8
206#define BN_BITS4 4
207#define BN_MASK (0xffff)
208#define BN_MASK2 (0xff)
209#define BN_MASK2l (0xf)
210#define BN_MASK2h1 (0xf8)
211#define BN_MASK2h (0xf0)
212#define BN_TBIT (0x80)
213#define BN_DEC_CONV (100)
214#define BN_DEC_FMT1 "%u"
215#define BN_DEC_FMT2 "%02u"
216#define BN_DEC_NUM 2
217#endif
218
219#define BN_DEFAULT_BITS 1280
220
221#ifdef BIGNUM
222#undef BIGNUM
223#endif
224
225#define BN_FLG_MALLOCED 0x01
226#define BN_FLG_STATIC_DATA 0x02
227#define BN_FLG_FREE 0x8000 /* used for debuging */
228#define BN_set_flags(b,n) ((b)->flags|=(n))
229#define BN_get_flags(b,n) ((b)->flags&(n))
230
231typedef struct bignum_st
232 {
233 BN_ULONG *d; /* Pointer to an array of 'BN_BITS2' bit chunks. */
234 int top; /* Index of last used d +1. */
235 /* The next are internal book keeping for bn_expand. */
236 int dmax; /* Size of the d array. */
237 int neg; /* one if the number is negative */
238 int flags;
239 } BIGNUM;
240
241/* Used for temp variables */
242#define BN_CTX_NUM 16
243#define BN_CTX_NUM_POS 12
244typedef struct bignum_ctx
245 {
246 int tos;
247 BIGNUM bn[BN_CTX_NUM];
248 int flags;
249 int depth;
250 int pos[BN_CTX_NUM_POS];
251 int too_many;
252 } BN_CTX;
253
254typedef struct bn_blinding_st
255 {
256 int init;
257 BIGNUM *A;
258 BIGNUM *Ai;
259 BIGNUM *mod; /* just a reference */
260 } BN_BLINDING;
261
262/* Used for montgomery multiplication */
263typedef struct bn_mont_ctx_st
264 {
265 int ri; /* number of bits in R */
266 BIGNUM RR; /* used to convert to montgomery form */
267 BIGNUM N; /* The modulus */
268 BIGNUM Ni; /* R*(1/R mod N) - N*Ni = 1
269 * (Ni is only stored for bignum algorithm) */
270 BN_ULONG n0; /* least significant word of Ni */
271 int flags;
272 } BN_MONT_CTX;
273
274/* Used for reciprocal division/mod functions
275 * It cannot be shared between threads
276 */
277typedef struct bn_recp_ctx_st
278 {
279 BIGNUM N; /* the divisor */
280 BIGNUM Nr; /* the reciprocal */
281 int num_bits;
282 int shift;
283 int flags;
284 } BN_RECP_CTX;
285
286#define BN_to_montgomery(r,a,mont,ctx) BN_mod_mul_montgomery(\
287 r,a,&((mont)->RR),(mont),ctx)
288
289#define BN_prime_checks 0 /* default: select number of iterations
290 based on the size of the number */
291
292/* number of Miller-Rabin iterations for an error rate of less than 2^-80
293 * for random 'b'-bit input, b >= 100 (taken from table 4.4 in the Handbook
294 * of Applied Cryptography [Menezes, van Oorschot, Vanstone; CRC Press 1996];
295 * original paper: Damgaard, Landrock, Pomerance: Average case error estimates
296 * for the strong probable prime test. -- Math. Comp. 61 (1993) 177-194) */
297#define BN_prime_checks_for_size(b) ((b) >= 1300 ? 2 : \
298 (b) >= 850 ? 3 : \
299 (b) >= 650 ? 4 : \
300 (b) >= 550 ? 5 : \
301 (b) >= 450 ? 6 : \
302 (b) >= 400 ? 7 : \
303 (b) >= 350 ? 8 : \
304 (b) >= 300 ? 9 : \
305 (b) >= 250 ? 12 : \
306 (b) >= 200 ? 15 : \
307 (b) >= 150 ? 18 : \
308 /* b >= 100 */ 27)
309
310#define BN_num_bytes(a) ((BN_num_bits(a)+7)/8)
311#define BN_is_word(a,w) (((a)->top == 1) && ((a)->d[0] == (BN_ULONG)(w)))
312#define BN_is_zero(a) (((a)->top == 0) || BN_is_word(a,0))
313#define BN_is_one(a) (BN_is_word((a),1))
314#define BN_is_odd(a) (((a)->top > 0) && ((a)->d[0] & 1))
315#define BN_one(a) (BN_set_word((a),1))
316#define BN_zero(a) (BN_set_word((a),0))
317
318/*#define BN_ascii2bn(a) BN_hex2bn(a) */
319/*#define BN_bn2ascii(a) BN_bn2hex(a) */
320
321BIGNUM *BN_value_one(void);
322char * BN_options(void);
323BN_CTX *BN_CTX_new(void);
324void BN_CTX_init(BN_CTX *c);
325void BN_CTX_free(BN_CTX *c);
326void BN_CTX_start(BN_CTX *ctx);
327BIGNUM *BN_CTX_get(BN_CTX *ctx);
328void BN_CTX_end(BN_CTX *ctx);
329int BN_rand(BIGNUM *rnd, int bits, int top,int bottom);
330int BN_pseudo_rand(BIGNUM *rnd, int bits, int top,int bottom);
331int BN_rand_range(BIGNUM *rnd, BIGNUM *range);
332int BN_num_bits(const BIGNUM *a);
333int BN_num_bits_word(BN_ULONG);
334BIGNUM *BN_new(void);
335void BN_init(BIGNUM *);
336void BN_clear_free(BIGNUM *a);
337BIGNUM *BN_copy(BIGNUM *a, const BIGNUM *b);
338BIGNUM *BN_bin2bn(const unsigned char *s,int len,BIGNUM *ret);
339int BN_bn2bin(const BIGNUM *a, unsigned char *to);
340BIGNUM *BN_mpi2bn(unsigned char *s,int len,BIGNUM *ret);
341int BN_bn2mpi(const BIGNUM *a, unsigned char *to);
342int BN_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b);
343int BN_usub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b);
344int BN_uadd(BIGNUM *r, const BIGNUM *a, const BIGNUM *b);
345int BN_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b);
346int BN_mod(BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, BN_CTX *ctx);
347int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d,
348 BN_CTX *ctx);
349int BN_mul(BIGNUM *r, BIGNUM *a, BIGNUM *b, BN_CTX *ctx);
350int BN_sqr(BIGNUM *r, BIGNUM *a,BN_CTX *ctx);
351BN_ULONG BN_mod_word(const BIGNUM *a, BN_ULONG w);
352BN_ULONG BN_div_word(BIGNUM *a, BN_ULONG w);
353int BN_mul_word(BIGNUM *a, BN_ULONG w);
354int BN_add_word(BIGNUM *a, BN_ULONG w);
355int BN_sub_word(BIGNUM *a, BN_ULONG w);
356int BN_set_word(BIGNUM *a, BN_ULONG w);
357BN_ULONG BN_get_word(BIGNUM *a);
358int BN_cmp(const BIGNUM *a, const BIGNUM *b);
359void BN_free(BIGNUM *a);
360int BN_is_bit_set(const BIGNUM *a, int n);
361int BN_lshift(BIGNUM *r, const BIGNUM *a, int n);
362int BN_lshift1(BIGNUM *r, BIGNUM *a);
363int BN_exp(BIGNUM *r, BIGNUM *a, BIGNUM *p,BN_CTX *ctx);
364int BN_mod_exp(BIGNUM *r, BIGNUM *a, const BIGNUM *p,
365 const BIGNUM *m,BN_CTX *ctx);
366int BN_mod_exp_mont(BIGNUM *r, BIGNUM *a, const BIGNUM *p,
367 const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx);
368int BN_mod_exp_mont_word(BIGNUM *r, BN_ULONG a, const BIGNUM *p,
369 const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx);
370int BN_mod_exp2_mont(BIGNUM *r, BIGNUM *a1, BIGNUM *p1,BIGNUM *a2,
371 BIGNUM *p2,BIGNUM *m,BN_CTX *ctx,BN_MONT_CTX *m_ctx);
372int BN_mod_exp_simple(BIGNUM *r, BIGNUM *a, BIGNUM *p,
373 BIGNUM *m,BN_CTX *ctx);
374int BN_mask_bits(BIGNUM *a,int n);
375int BN_mod_mul(BIGNUM *ret, BIGNUM *a, BIGNUM *b, const BIGNUM *m, BN_CTX *ctx);
376#ifndef NO_FP_API
377int BN_print_fp(FILE *fp, const BIGNUM *a);
378#endif
379#ifdef HEADER_BIO_H
380int BN_print(BIO *fp, const BIGNUM *a);
381#else
382int BN_print(void *fp, const BIGNUM *a);
383#endif
384int BN_reciprocal(BIGNUM *r, BIGNUM *m, int len, BN_CTX *ctx);
385int BN_rshift(BIGNUM *r, BIGNUM *a, int n);
386int BN_rshift1(BIGNUM *r, BIGNUM *a);
387void BN_clear(BIGNUM *a);
388BIGNUM *BN_dup(const BIGNUM *a);
389int BN_ucmp(const BIGNUM *a, const BIGNUM *b);
390int BN_set_bit(BIGNUM *a, int n);
391int BN_clear_bit(BIGNUM *a, int n);
392char * BN_bn2hex(const BIGNUM *a);
393char * BN_bn2dec(const BIGNUM *a);
394int BN_hex2bn(BIGNUM **a, const char *str);
395int BN_dec2bn(BIGNUM **a, const char *str);
396int BN_gcd(BIGNUM *r,BIGNUM *in_a,BIGNUM *in_b,BN_CTX *ctx);
397BIGNUM *BN_mod_inverse(BIGNUM *ret,BIGNUM *a, const BIGNUM *n,BN_CTX *ctx);
398BIGNUM *BN_generate_prime(BIGNUM *ret,int bits,int safe,BIGNUM *add,
399 BIGNUM *rem,void (*callback)(int,int,void *),void *cb_arg);
400int BN_is_prime(const BIGNUM *p,int nchecks,
401 void (*callback)(int,int,void *),
402 BN_CTX *ctx,void *cb_arg);
403int BN_is_prime_fasttest(const BIGNUM *p,int nchecks,
404 void (*callback)(int,int,void *),BN_CTX *ctx,void *cb_arg,
405 int do_trial_division);
406void ERR_load_BN_strings(void );
407
408BN_MONT_CTX *BN_MONT_CTX_new(void );
409void BN_MONT_CTX_init(BN_MONT_CTX *ctx);
410int BN_mod_mul_montgomery(BIGNUM *r,BIGNUM *a,BIGNUM *b,BN_MONT_CTX *mont,
411 BN_CTX *ctx);
412int BN_from_montgomery(BIGNUM *r,BIGNUM *a,BN_MONT_CTX *mont,BN_CTX *ctx);
413void BN_MONT_CTX_free(BN_MONT_CTX *mont);
414int BN_MONT_CTX_set(BN_MONT_CTX *mont,const BIGNUM *modulus,BN_CTX *ctx);
415BN_MONT_CTX *BN_MONT_CTX_copy(BN_MONT_CTX *to,BN_MONT_CTX *from);
416
417BN_BLINDING *BN_BLINDING_new(BIGNUM *A,BIGNUM *Ai,BIGNUM *mod);
418void BN_BLINDING_free(BN_BLINDING *b);
419int BN_BLINDING_update(BN_BLINDING *b,BN_CTX *ctx);
420int BN_BLINDING_convert(BIGNUM *n, BN_BLINDING *r, BN_CTX *ctx);
421int BN_BLINDING_invert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx);
422
423void BN_set_params(int mul,int high,int low,int mont);
424int BN_get_params(int which); /* 0, mul, 1 high, 2 low, 3 mont */
425
426void BN_RECP_CTX_init(BN_RECP_CTX *recp);
427BN_RECP_CTX *BN_RECP_CTX_new(void);
428void BN_RECP_CTX_free(BN_RECP_CTX *recp);
429int BN_RECP_CTX_set(BN_RECP_CTX *recp,const BIGNUM *rdiv,BN_CTX *ctx);
430int BN_mod_mul_reciprocal(BIGNUM *r, BIGNUM *x, BIGNUM *y,
431 BN_RECP_CTX *recp,BN_CTX *ctx);
432int BN_mod_exp_recp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p,
433 const BIGNUM *m, BN_CTX *ctx);
434int BN_div_recp(BIGNUM *dv, BIGNUM *rem, BIGNUM *m,
435 BN_RECP_CTX *recp, BN_CTX *ctx);
436
437/* library internal functions */
438
439#define bn_expand(a,bits) ((((((bits+BN_BITS2-1))/BN_BITS2)) <= (a)->dmax)?\
440 (a):bn_expand2((a),(bits)/BN_BITS2+1))
441#define bn_wexpand(a,words) (((words) <= (a)->dmax)?(a):bn_expand2((a),(words)))
442BIGNUM *bn_expand2(BIGNUM *a, int words);
443
444#define bn_fix_top(a) \
445 { \
446 BN_ULONG *ftl; \
447 if ((a)->top > 0) \
448 { \
449 for (ftl= &((a)->d[(a)->top-1]); (a)->top > 0; (a)->top--) \
450 if (*(ftl--)) break; \
451 } \
452 }
453
454BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w);
455BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w);
456void bn_sqr_words(BN_ULONG *rp, BN_ULONG *ap, int num);
457BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d);
458BN_ULONG bn_add_words(BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int num);
459BN_ULONG bn_sub_words(BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int num);
460
461#ifdef BN_DEBUG
462 void bn_dump1(FILE *o, const char *a, BN_ULONG *b,int n);
463# define bn_print(a) {fprintf(stderr, #a "="); BN_print_fp(stderr,a); \
464 fprintf(stderr,"\n");}
465# define bn_dump(a,n) bn_dump1(stderr,#a,a,n);
466#else
467# define bn_print(a)
468# define bn_dump(a,b)
469#endif
470
471int BN_bntest_rand(BIGNUM *rnd, int bits, int top,int bottom);
472
473/* BEGIN ERROR CODES */
474/* The following lines are auto generated by the script mkerr.pl. Any changes
475 * made after this point may be overwritten when the script is next run.
476 */
477
478/* Error codes for the BN functions. */
479
480/* Function codes. */
481#define BN_F_BN_BLINDING_CONVERT 100
482#define BN_F_BN_BLINDING_INVERT 101
483#define BN_F_BN_BLINDING_NEW 102
484#define BN_F_BN_BLINDING_UPDATE 103
485#define BN_F_BN_BN2DEC 104
486#define BN_F_BN_BN2HEX 105
487#define BN_F_BN_CTX_GET 116
488#define BN_F_BN_CTX_NEW 106
489#define BN_F_BN_DIV 107
490#define BN_F_BN_EXPAND2 108
491#define BN_F_BN_MOD_EXP2_MONT 118
492#define BN_F_BN_MOD_EXP_MONT 109
493#define BN_F_BN_MOD_EXP_MONT_WORD 117
494#define BN_F_BN_MOD_INVERSE 110
495#define BN_F_BN_MOD_MUL_RECIPROCAL 111
496#define BN_F_BN_MPI2BN 112
497#define BN_F_BN_NEW 113
498#define BN_F_BN_RAND 114
499#define BN_F_BN_RAND_RANGE 122
500#define BN_F_BN_USUB 115
501
502/* Reason codes. */
503#define BN_R_ARG2_LT_ARG3 100
504#define BN_R_BAD_RECIPROCAL 101
505#define BN_R_BIGNUM_TOO_LONG 114
506#define BN_R_CALLED_WITH_EVEN_MODULUS 102
507#define BN_R_DIV_BY_ZERO 103
508#define BN_R_ENCODING_ERROR 104
509#define BN_R_EXPAND_ON_STATIC_BIGNUM_DATA 105
510#define BN_R_INVALID_LENGTH 106
511#define BN_R_INVALID_RANGE 115
512#define BN_R_NOT_INITIALIZED 107
513#define BN_R_NO_INVERSE 108
514#define BN_R_TOO_MANY_TEMPORARY_VARIABLES 109
515
516#ifdef __cplusplus
517}
518#endif
519#endif
520
diff --git a/src/lib/libcrypto/bn/bn_add.c b/src/lib/libcrypto/bn/bn_add.c
deleted file mode 100644
index 5d24691233..0000000000
--- a/src/lib/libcrypto/bn/bn_add.c
+++ /dev/null
@@ -1,307 +0,0 @@
1/* crypto/bn/bn_add.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63/* r can == a or b */
64int BN_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b)
65 {
66 const BIGNUM *tmp;
67
68 bn_check_top(a);
69 bn_check_top(b);
70
71 /* a + b a+b
72 * a + -b a-b
73 * -a + b b-a
74 * -a + -b -(a+b)
75 */
76 if (a->neg ^ b->neg)
77 {
78 /* only one is negative */
79 if (a->neg)
80 { tmp=a; a=b; b=tmp; }
81
82 /* we are now a - b */
83
84 if (BN_ucmp(a,b) < 0)
85 {
86 if (!BN_usub(r,b,a)) return(0);
87 r->neg=1;
88 }
89 else
90 {
91 if (!BN_usub(r,a,b)) return(0);
92 r->neg=0;
93 }
94 return(1);
95 }
96
97 if (a->neg) /* both are neg */
98 r->neg=1;
99 else
100 r->neg=0;
101
102 if (!BN_uadd(r,a,b)) return(0);
103 return(1);
104 }
105
106/* unsigned add of b to a, r must be large enough */
107int BN_uadd(BIGNUM *r, const BIGNUM *a, const BIGNUM *b)
108 {
109 register int i;
110 int max,min;
111 BN_ULONG *ap,*bp,*rp,carry,t1;
112 const BIGNUM *tmp;
113
114 bn_check_top(a);
115 bn_check_top(b);
116
117 if (a->top < b->top)
118 { tmp=a; a=b; b=tmp; }
119 max=a->top;
120 min=b->top;
121
122 if (bn_wexpand(r,max+1) == NULL)
123 return(0);
124
125 r->top=max;
126
127
128 ap=a->d;
129 bp=b->d;
130 rp=r->d;
131 carry=0;
132
133 carry=bn_add_words(rp,ap,bp,min);
134 rp+=min;
135 ap+=min;
136 bp+=min;
137 i=min;
138
139 if (carry)
140 {
141 while (i < max)
142 {
143 i++;
144 t1= *(ap++);
145 if ((*(rp++)=(t1+1)&BN_MASK2) >= t1)
146 {
147 carry=0;
148 break;
149 }
150 }
151 if ((i >= max) && carry)
152 {
153 *(rp++)=1;
154 r->top++;
155 }
156 }
157 if (rp != ap)
158 {
159 for (; i<max; i++)
160 *(rp++)= *(ap++);
161 }
162 /* memcpy(rp,ap,sizeof(*ap)*(max-i));*/
163 return(1);
164 }
165
166/* unsigned subtraction of b from a, a must be larger than b. */
167int BN_usub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b)
168 {
169 int max,min;
170 register BN_ULONG t1,t2,*ap,*bp,*rp;
171 int i,carry;
172#if defined(IRIX_CC_BUG) && !defined(LINT)
173 int dummy;
174#endif
175
176 bn_check_top(a);
177 bn_check_top(b);
178
179 if (a->top < b->top) /* hmm... should not be happening */
180 {
181 BNerr(BN_F_BN_USUB,BN_R_ARG2_LT_ARG3);
182 return(0);
183 }
184
185 max=a->top;
186 min=b->top;
187 if (bn_wexpand(r,max) == NULL) return(0);
188
189 ap=a->d;
190 bp=b->d;
191 rp=r->d;
192
193#if 1
194 carry=0;
195 for (i=0; i<min; i++)
196 {
197 t1= *(ap++);
198 t2= *(bp++);
199 if (carry)
200 {
201 carry=(t1 <= t2);
202 t1=(t1-t2-1)&BN_MASK2;
203 }
204 else
205 {
206 carry=(t1 < t2);
207 t1=(t1-t2)&BN_MASK2;
208 }
209#if defined(IRIX_CC_BUG) && !defined(LINT)
210 dummy=t1;
211#endif
212 *(rp++)=t1&BN_MASK2;
213 }
214#else
215 carry=bn_sub_words(rp,ap,bp,min);
216 ap+=min;
217 bp+=min;
218 rp+=min;
219 i=min;
220#endif
221 if (carry) /* subtracted */
222 {
223 while (i < max)
224 {
225 i++;
226 t1= *(ap++);
227 t2=(t1-1)&BN_MASK2;
228 *(rp++)=t2;
229 if (t1 > t2) break;
230 }
231 }
232#if 0
233 memcpy(rp,ap,sizeof(*rp)*(max-i));
234#else
235 if (rp != ap)
236 {
237 for (;;)
238 {
239 if (i++ >= max) break;
240 rp[0]=ap[0];
241 if (i++ >= max) break;
242 rp[1]=ap[1];
243 if (i++ >= max) break;
244 rp[2]=ap[2];
245 if (i++ >= max) break;
246 rp[3]=ap[3];
247 rp+=4;
248 ap+=4;
249 }
250 }
251#endif
252
253 r->top=max;
254 bn_fix_top(r);
255 return(1);
256 }
257
258int BN_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b)
259 {
260 int max;
261 int add=0,neg=0;
262 const BIGNUM *tmp;
263
264 bn_check_top(a);
265 bn_check_top(b);
266
267 /* a - b a-b
268 * a - -b a+b
269 * -a - b -(a+b)
270 * -a - -b b-a
271 */
272 if (a->neg)
273 {
274 if (b->neg)
275 { tmp=a; a=b; b=tmp; }
276 else
277 { add=1; neg=1; }
278 }
279 else
280 {
281 if (b->neg) { add=1; neg=0; }
282 }
283
284 if (add)
285 {
286 if (!BN_uadd(r,a,b)) return(0);
287 r->neg=neg;
288 return(1);
289 }
290
291 /* We are actually doing a - b :-) */
292
293 max=(a->top > b->top)?a->top:b->top;
294 if (bn_wexpand(r,max) == NULL) return(0);
295 if (BN_ucmp(a,b) < 0)
296 {
297 if (!BN_usub(r,b,a)) return(0);
298 r->neg=1;
299 }
300 else
301 {
302 if (!BN_usub(r,a,b)) return(0);
303 r->neg=0;
304 }
305 return(1);
306 }
307
diff --git a/src/lib/libcrypto/bn/bn_asm.c b/src/lib/libcrypto/bn/bn_asm.c
deleted file mode 100644
index 44e52a40db..0000000000
--- a/src/lib/libcrypto/bn/bn_asm.c
+++ /dev/null
@@ -1,832 +0,0 @@
1/* crypto/bn/bn_asm.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#ifndef BN_DEBUG
60# undef NDEBUG /* avoid conflicting definitions */
61# define NDEBUG
62#endif
63
64#include <stdio.h>
65#include <assert.h>
66#include "cryptlib.h"
67#include "bn_lcl.h"
68
69#if defined(BN_LLONG) || defined(BN_UMULT_HIGH)
70
71BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
72 {
73 BN_ULONG c1=0;
74
75 assert(num >= 0);
76 if (num <= 0) return(c1);
77
78 while (num&~3)
79 {
80 mul_add(rp[0],ap[0],w,c1);
81 mul_add(rp[1],ap[1],w,c1);
82 mul_add(rp[2],ap[2],w,c1);
83 mul_add(rp[3],ap[3],w,c1);
84 ap+=4; rp+=4; num-=4;
85 }
86 if (num)
87 {
88 mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1;
89 mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1;
90 mul_add(rp[2],ap[2],w,c1); return c1;
91 }
92
93 return(c1);
94 }
95
96BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
97 {
98 BN_ULONG c1=0;
99
100 assert(num >= 0);
101 if (num <= 0) return(c1);
102
103 while (num&~3)
104 {
105 mul(rp[0],ap[0],w,c1);
106 mul(rp[1],ap[1],w,c1);
107 mul(rp[2],ap[2],w,c1);
108 mul(rp[3],ap[3],w,c1);
109 ap+=4; rp+=4; num-=4;
110 }
111 if (num)
112 {
113 mul(rp[0],ap[0],w,c1); if (--num == 0) return c1;
114 mul(rp[1],ap[1],w,c1); if (--num == 0) return c1;
115 mul(rp[2],ap[2],w,c1);
116 }
117 return(c1);
118 }
119
120void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n)
121 {
122 assert(n >= 0);
123 if (n <= 0) return;
124 while (n&~3)
125 {
126 sqr(r[0],r[1],a[0]);
127 sqr(r[2],r[3],a[1]);
128 sqr(r[4],r[5],a[2]);
129 sqr(r[6],r[7],a[3]);
130 a+=4; r+=8; n-=4;
131 }
132 if (n)
133 {
134 sqr(r[0],r[1],a[0]); if (--n == 0) return;
135 sqr(r[2],r[3],a[1]); if (--n == 0) return;
136 sqr(r[4],r[5],a[2]);
137 }
138 }
139
140#else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
141
142BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
143 {
144 BN_ULONG c=0;
145 BN_ULONG bl,bh;
146
147 assert(num >= 0);
148 if (num <= 0) return((BN_ULONG)0);
149
150 bl=LBITS(w);
151 bh=HBITS(w);
152
153 for (;;)
154 {
155 mul_add(rp[0],ap[0],bl,bh,c);
156 if (--num == 0) break;
157 mul_add(rp[1],ap[1],bl,bh,c);
158 if (--num == 0) break;
159 mul_add(rp[2],ap[2],bl,bh,c);
160 if (--num == 0) break;
161 mul_add(rp[3],ap[3],bl,bh,c);
162 if (--num == 0) break;
163 ap+=4;
164 rp+=4;
165 }
166 return(c);
167 }
168
169BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
170 {
171 BN_ULONG carry=0;
172 BN_ULONG bl,bh;
173
174 assert(num >= 0);
175 if (num <= 0) return((BN_ULONG)0);
176
177 bl=LBITS(w);
178 bh=HBITS(w);
179
180 for (;;)
181 {
182 mul(rp[0],ap[0],bl,bh,carry);
183 if (--num == 0) break;
184 mul(rp[1],ap[1],bl,bh,carry);
185 if (--num == 0) break;
186 mul(rp[2],ap[2],bl,bh,carry);
187 if (--num == 0) break;
188 mul(rp[3],ap[3],bl,bh,carry);
189 if (--num == 0) break;
190 ap+=4;
191 rp+=4;
192 }
193 return(carry);
194 }
195
196void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n)
197 {
198 assert(n >= 0);
199 if (n <= 0) return;
200 for (;;)
201 {
202 sqr64(r[0],r[1],a[0]);
203 if (--n == 0) break;
204
205 sqr64(r[2],r[3],a[1]);
206 if (--n == 0) break;
207
208 sqr64(r[4],r[5],a[2]);
209 if (--n == 0) break;
210
211 sqr64(r[6],r[7],a[3]);
212 if (--n == 0) break;
213
214 a+=4;
215 r+=8;
216 }
217 }
218
219#endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
220
221#if defined(BN_LLONG) && defined(BN_DIV2W)
222
223BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
224 {
225 return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d));
226 }
227
228#else
229
230/* Divide h,l by d and return the result. */
231/* I need to test this some more :-( */
232BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
233 {
234 BN_ULONG dh,dl,q,ret=0,th,tl,t;
235 int i,count=2;
236
237 if (d == 0) return(BN_MASK2);
238
239 i=BN_num_bits_word(d);
240 assert((i == BN_BITS2) || (h > (BN_ULONG)1<<i));
241
242 i=BN_BITS2-i;
243 if (h >= d) h-=d;
244
245 if (i)
246 {
247 d<<=i;
248 h=(h<<i)|(l>>(BN_BITS2-i));
249 l<<=i;
250 }
251 dh=(d&BN_MASK2h)>>BN_BITS4;
252 dl=(d&BN_MASK2l);
253 for (;;)
254 {
255 if ((h>>BN_BITS4) == dh)
256 q=BN_MASK2l;
257 else
258 q=h/dh;
259
260 th=q*dh;
261 tl=dl*q;
262 for (;;)
263 {
264 t=h-th;
265 if ((t&BN_MASK2h) ||
266 ((tl) <= (
267 (t<<BN_BITS4)|
268 ((l&BN_MASK2h)>>BN_BITS4))))
269 break;
270 q--;
271 th-=dh;
272 tl-=dl;
273 }
274 t=(tl>>BN_BITS4);
275 tl=(tl<<BN_BITS4)&BN_MASK2h;
276 th+=t;
277
278 if (l < tl) th++;
279 l-=tl;
280 if (h < th)
281 {
282 h+=d;
283 q--;
284 }
285 h-=th;
286
287 if (--count == 0) break;
288
289 ret=q<<BN_BITS4;
290 h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2;
291 l=(l&BN_MASK2l)<<BN_BITS4;
292 }
293 ret|=q;
294 return(ret);
295 }
296#endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */
297
298#ifdef BN_LLONG
299BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
300 {
301 BN_ULLONG ll=0;
302
303 assert(n >= 0);
304 if (n <= 0) return((BN_ULONG)0);
305
306 for (;;)
307 {
308 ll+=(BN_ULLONG)a[0]+b[0];
309 r[0]=(BN_ULONG)ll&BN_MASK2;
310 ll>>=BN_BITS2;
311 if (--n <= 0) break;
312
313 ll+=(BN_ULLONG)a[1]+b[1];
314 r[1]=(BN_ULONG)ll&BN_MASK2;
315 ll>>=BN_BITS2;
316 if (--n <= 0) break;
317
318 ll+=(BN_ULLONG)a[2]+b[2];
319 r[2]=(BN_ULONG)ll&BN_MASK2;
320 ll>>=BN_BITS2;
321 if (--n <= 0) break;
322
323 ll+=(BN_ULLONG)a[3]+b[3];
324 r[3]=(BN_ULONG)ll&BN_MASK2;
325 ll>>=BN_BITS2;
326 if (--n <= 0) break;
327
328 a+=4;
329 b+=4;
330 r+=4;
331 }
332 return((BN_ULONG)ll);
333 }
334#else /* !BN_LLONG */
335BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
336 {
337 BN_ULONG c,l,t;
338
339 assert(n >= 0);
340 if (n <= 0) return((BN_ULONG)0);
341
342 c=0;
343 for (;;)
344 {
345 t=a[0];
346 t=(t+c)&BN_MASK2;
347 c=(t < c);
348 l=(t+b[0])&BN_MASK2;
349 c+=(l < t);
350 r[0]=l;
351 if (--n <= 0) break;
352
353 t=a[1];
354 t=(t+c)&BN_MASK2;
355 c=(t < c);
356 l=(t+b[1])&BN_MASK2;
357 c+=(l < t);
358 r[1]=l;
359 if (--n <= 0) break;
360
361 t=a[2];
362 t=(t+c)&BN_MASK2;
363 c=(t < c);
364 l=(t+b[2])&BN_MASK2;
365 c+=(l < t);
366 r[2]=l;
367 if (--n <= 0) break;
368
369 t=a[3];
370 t=(t+c)&BN_MASK2;
371 c=(t < c);
372 l=(t+b[3])&BN_MASK2;
373 c+=(l < t);
374 r[3]=l;
375 if (--n <= 0) break;
376
377 a+=4;
378 b+=4;
379 r+=4;
380 }
381 return((BN_ULONG)c);
382 }
383#endif /* !BN_LLONG */
384
385BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
386 {
387 BN_ULONG t1,t2;
388 int c=0;
389
390 assert(n >= 0);
391 if (n <= 0) return((BN_ULONG)0);
392
393 for (;;)
394 {
395 t1=a[0]; t2=b[0];
396 r[0]=(t1-t2-c)&BN_MASK2;
397 if (t1 != t2) c=(t1 < t2);
398 if (--n <= 0) break;
399
400 t1=a[1]; t2=b[1];
401 r[1]=(t1-t2-c)&BN_MASK2;
402 if (t1 != t2) c=(t1 < t2);
403 if (--n <= 0) break;
404
405 t1=a[2]; t2=b[2];
406 r[2]=(t1-t2-c)&BN_MASK2;
407 if (t1 != t2) c=(t1 < t2);
408 if (--n <= 0) break;
409
410 t1=a[3]; t2=b[3];
411 r[3]=(t1-t2-c)&BN_MASK2;
412 if (t1 != t2) c=(t1 < t2);
413 if (--n <= 0) break;
414
415 a+=4;
416 b+=4;
417 r+=4;
418 }
419 return(c);
420 }
421
422#ifdef BN_MUL_COMBA
423
424#undef bn_mul_comba8
425#undef bn_mul_comba4
426#undef bn_sqr_comba8
427#undef bn_sqr_comba4
428
429/* mul_add_c(a,b,c0,c1,c2) -- c+=a*b for three word number c=(c2,c1,c0) */
430/* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */
431/* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */
432/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */
433
434#ifdef BN_LLONG
435#define mul_add_c(a,b,c0,c1,c2) \
436 t=(BN_ULLONG)a*b; \
437 t1=(BN_ULONG)Lw(t); \
438 t2=(BN_ULONG)Hw(t); \
439 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
440 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
441
442#define mul_add_c2(a,b,c0,c1,c2) \
443 t=(BN_ULLONG)a*b; \
444 tt=(t+t)&BN_MASK; \
445 if (tt < t) c2++; \
446 t1=(BN_ULONG)Lw(tt); \
447 t2=(BN_ULONG)Hw(tt); \
448 c0=(c0+t1)&BN_MASK2; \
449 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
450 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
451
452#define sqr_add_c(a,i,c0,c1,c2) \
453 t=(BN_ULLONG)a[i]*a[i]; \
454 t1=(BN_ULONG)Lw(t); \
455 t2=(BN_ULONG)Hw(t); \
456 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
457 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
458
459#define sqr_add_c2(a,i,j,c0,c1,c2) \
460 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
461
462#elif defined(BN_UMULT_HIGH)
463
464#define mul_add_c(a,b,c0,c1,c2) { \
465 BN_ULONG ta=(a),tb=(b); \
466 t1 = ta * tb; \
467 t2 = BN_UMULT_HIGH(ta,tb); \
468 c0 += t1; t2 += (c0<t1)?1:0; \
469 c1 += t2; c2 += (c1<t2)?1:0; \
470 }
471
472#define mul_add_c2(a,b,c0,c1,c2) { \
473 BN_ULONG ta=(a),tb=(b),t0; \
474 t1 = BN_UMULT_HIGH(ta,tb); \
475 t0 = ta * tb; \
476 t2 = t1+t1; c2 += (t2<t1)?1:0; \
477 t1 = t0+t0; t2 += (t1<t0)?1:0; \
478 c0 += t1; t2 += (c0<t1)?1:0; \
479 c1 += t2; c2 += (c1<t2)?1:0; \
480 }
481
482#define sqr_add_c(a,i,c0,c1,c2) { \
483 BN_ULONG ta=(a)[i]; \
484 t1 = ta * ta; \
485 t2 = BN_UMULT_HIGH(ta,ta); \
486 c0 += t1; t2 += (c0<t1)?1:0; \
487 c1 += t2; c2 += (c1<t2)?1:0; \
488 }
489
490#define sqr_add_c2(a,i,j,c0,c1,c2) \
491 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
492
493#else /* !BN_LLONG */
494#define mul_add_c(a,b,c0,c1,c2) \
495 t1=LBITS(a); t2=HBITS(a); \
496 bl=LBITS(b); bh=HBITS(b); \
497 mul64(t1,t2,bl,bh); \
498 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
499 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
500
501#define mul_add_c2(a,b,c0,c1,c2) \
502 t1=LBITS(a); t2=HBITS(a); \
503 bl=LBITS(b); bh=HBITS(b); \
504 mul64(t1,t2,bl,bh); \
505 if (t2 & BN_TBIT) c2++; \
506 t2=(t2+t2)&BN_MASK2; \
507 if (t1 & BN_TBIT) t2++; \
508 t1=(t1+t1)&BN_MASK2; \
509 c0=(c0+t1)&BN_MASK2; \
510 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
511 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
512
513#define sqr_add_c(a,i,c0,c1,c2) \
514 sqr64(t1,t2,(a)[i]); \
515 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
516 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
517
518#define sqr_add_c2(a,i,j,c0,c1,c2) \
519 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
520#endif /* !BN_LLONG */
521
522void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
523 {
524#ifdef BN_LLONG
525 BN_ULLONG t;
526#else
527 BN_ULONG bl,bh;
528#endif
529 BN_ULONG t1,t2;
530 BN_ULONG c1,c2,c3;
531
532 c1=0;
533 c2=0;
534 c3=0;
535 mul_add_c(a[0],b[0],c1,c2,c3);
536 r[0]=c1;
537 c1=0;
538 mul_add_c(a[0],b[1],c2,c3,c1);
539 mul_add_c(a[1],b[0],c2,c3,c1);
540 r[1]=c2;
541 c2=0;
542 mul_add_c(a[2],b[0],c3,c1,c2);
543 mul_add_c(a[1],b[1],c3,c1,c2);
544 mul_add_c(a[0],b[2],c3,c1,c2);
545 r[2]=c3;
546 c3=0;
547 mul_add_c(a[0],b[3],c1,c2,c3);
548 mul_add_c(a[1],b[2],c1,c2,c3);
549 mul_add_c(a[2],b[1],c1,c2,c3);
550 mul_add_c(a[3],b[0],c1,c2,c3);
551 r[3]=c1;
552 c1=0;
553 mul_add_c(a[4],b[0],c2,c3,c1);
554 mul_add_c(a[3],b[1],c2,c3,c1);
555 mul_add_c(a[2],b[2],c2,c3,c1);
556 mul_add_c(a[1],b[3],c2,c3,c1);
557 mul_add_c(a[0],b[4],c2,c3,c1);
558 r[4]=c2;
559 c2=0;
560 mul_add_c(a[0],b[5],c3,c1,c2);
561 mul_add_c(a[1],b[4],c3,c1,c2);
562 mul_add_c(a[2],b[3],c3,c1,c2);
563 mul_add_c(a[3],b[2],c3,c1,c2);
564 mul_add_c(a[4],b[1],c3,c1,c2);
565 mul_add_c(a[5],b[0],c3,c1,c2);
566 r[5]=c3;
567 c3=0;
568 mul_add_c(a[6],b[0],c1,c2,c3);
569 mul_add_c(a[5],b[1],c1,c2,c3);
570 mul_add_c(a[4],b[2],c1,c2,c3);
571 mul_add_c(a[3],b[3],c1,c2,c3);
572 mul_add_c(a[2],b[4],c1,c2,c3);
573 mul_add_c(a[1],b[5],c1,c2,c3);
574 mul_add_c(a[0],b[6],c1,c2,c3);
575 r[6]=c1;
576 c1=0;
577 mul_add_c(a[0],b[7],c2,c3,c1);
578 mul_add_c(a[1],b[6],c2,c3,c1);
579 mul_add_c(a[2],b[5],c2,c3,c1);
580 mul_add_c(a[3],b[4],c2,c3,c1);
581 mul_add_c(a[4],b[3],c2,c3,c1);
582 mul_add_c(a[5],b[2],c2,c3,c1);
583 mul_add_c(a[6],b[1],c2,c3,c1);
584 mul_add_c(a[7],b[0],c2,c3,c1);
585 r[7]=c2;
586 c2=0;
587 mul_add_c(a[7],b[1],c3,c1,c2);
588 mul_add_c(a[6],b[2],c3,c1,c2);
589 mul_add_c(a[5],b[3],c3,c1,c2);
590 mul_add_c(a[4],b[4],c3,c1,c2);
591 mul_add_c(a[3],b[5],c3,c1,c2);
592 mul_add_c(a[2],b[6],c3,c1,c2);
593 mul_add_c(a[1],b[7],c3,c1,c2);
594 r[8]=c3;
595 c3=0;
596 mul_add_c(a[2],b[7],c1,c2,c3);
597 mul_add_c(a[3],b[6],c1,c2,c3);
598 mul_add_c(a[4],b[5],c1,c2,c3);
599 mul_add_c(a[5],b[4],c1,c2,c3);
600 mul_add_c(a[6],b[3],c1,c2,c3);
601 mul_add_c(a[7],b[2],c1,c2,c3);
602 r[9]=c1;
603 c1=0;
604 mul_add_c(a[7],b[3],c2,c3,c1);
605 mul_add_c(a[6],b[4],c2,c3,c1);
606 mul_add_c(a[5],b[5],c2,c3,c1);
607 mul_add_c(a[4],b[6],c2,c3,c1);
608 mul_add_c(a[3],b[7],c2,c3,c1);
609 r[10]=c2;
610 c2=0;
611 mul_add_c(a[4],b[7],c3,c1,c2);
612 mul_add_c(a[5],b[6],c3,c1,c2);
613 mul_add_c(a[6],b[5],c3,c1,c2);
614 mul_add_c(a[7],b[4],c3,c1,c2);
615 r[11]=c3;
616 c3=0;
617 mul_add_c(a[7],b[5],c1,c2,c3);
618 mul_add_c(a[6],b[6],c1,c2,c3);
619 mul_add_c(a[5],b[7],c1,c2,c3);
620 r[12]=c1;
621 c1=0;
622 mul_add_c(a[6],b[7],c2,c3,c1);
623 mul_add_c(a[7],b[6],c2,c3,c1);
624 r[13]=c2;
625 c2=0;
626 mul_add_c(a[7],b[7],c3,c1,c2);
627 r[14]=c3;
628 r[15]=c1;
629 }
630
631void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
632 {
633#ifdef BN_LLONG
634 BN_ULLONG t;
635#else
636 BN_ULONG bl,bh;
637#endif
638 BN_ULONG t1,t2;
639 BN_ULONG c1,c2,c3;
640
641 c1=0;
642 c2=0;
643 c3=0;
644 mul_add_c(a[0],b[0],c1,c2,c3);
645 r[0]=c1;
646 c1=0;
647 mul_add_c(a[0],b[1],c2,c3,c1);
648 mul_add_c(a[1],b[0],c2,c3,c1);
649 r[1]=c2;
650 c2=0;
651 mul_add_c(a[2],b[0],c3,c1,c2);
652 mul_add_c(a[1],b[1],c3,c1,c2);
653 mul_add_c(a[0],b[2],c3,c1,c2);
654 r[2]=c3;
655 c3=0;
656 mul_add_c(a[0],b[3],c1,c2,c3);
657 mul_add_c(a[1],b[2],c1,c2,c3);
658 mul_add_c(a[2],b[1],c1,c2,c3);
659 mul_add_c(a[3],b[0],c1,c2,c3);
660 r[3]=c1;
661 c1=0;
662 mul_add_c(a[3],b[1],c2,c3,c1);
663 mul_add_c(a[2],b[2],c2,c3,c1);
664 mul_add_c(a[1],b[3],c2,c3,c1);
665 r[4]=c2;
666 c2=0;
667 mul_add_c(a[2],b[3],c3,c1,c2);
668 mul_add_c(a[3],b[2],c3,c1,c2);
669 r[5]=c3;
670 c3=0;
671 mul_add_c(a[3],b[3],c1,c2,c3);
672 r[6]=c1;
673 r[7]=c2;
674 }
675
676void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
677 {
678#ifdef BN_LLONG
679 BN_ULLONG t,tt;
680#else
681 BN_ULONG bl,bh;
682#endif
683 BN_ULONG t1,t2;
684 BN_ULONG c1,c2,c3;
685
686 c1=0;
687 c2=0;
688 c3=0;
689 sqr_add_c(a,0,c1,c2,c3);
690 r[0]=c1;
691 c1=0;
692 sqr_add_c2(a,1,0,c2,c3,c1);
693 r[1]=c2;
694 c2=0;
695 sqr_add_c(a,1,c3,c1,c2);
696 sqr_add_c2(a,2,0,c3,c1,c2);
697 r[2]=c3;
698 c3=0;
699 sqr_add_c2(a,3,0,c1,c2,c3);
700 sqr_add_c2(a,2,1,c1,c2,c3);
701 r[3]=c1;
702 c1=0;
703 sqr_add_c(a,2,c2,c3,c1);
704 sqr_add_c2(a,3,1,c2,c3,c1);
705 sqr_add_c2(a,4,0,c2,c3,c1);
706 r[4]=c2;
707 c2=0;
708 sqr_add_c2(a,5,0,c3,c1,c2);
709 sqr_add_c2(a,4,1,c3,c1,c2);
710 sqr_add_c2(a,3,2,c3,c1,c2);
711 r[5]=c3;
712 c3=0;
713 sqr_add_c(a,3,c1,c2,c3);
714 sqr_add_c2(a,4,2,c1,c2,c3);
715 sqr_add_c2(a,5,1,c1,c2,c3);
716 sqr_add_c2(a,6,0,c1,c2,c3);
717 r[6]=c1;
718 c1=0;
719 sqr_add_c2(a,7,0,c2,c3,c1);
720 sqr_add_c2(a,6,1,c2,c3,c1);
721 sqr_add_c2(a,5,2,c2,c3,c1);
722 sqr_add_c2(a,4,3,c2,c3,c1);
723 r[7]=c2;
724 c2=0;
725 sqr_add_c(a,4,c3,c1,c2);
726 sqr_add_c2(a,5,3,c3,c1,c2);
727 sqr_add_c2(a,6,2,c3,c1,c2);
728 sqr_add_c2(a,7,1,c3,c1,c2);
729 r[8]=c3;
730 c3=0;
731 sqr_add_c2(a,7,2,c1,c2,c3);
732 sqr_add_c2(a,6,3,c1,c2,c3);
733 sqr_add_c2(a,5,4,c1,c2,c3);
734 r[9]=c1;
735 c1=0;
736 sqr_add_c(a,5,c2,c3,c1);
737 sqr_add_c2(a,6,4,c2,c3,c1);
738 sqr_add_c2(a,7,3,c2,c3,c1);
739 r[10]=c2;
740 c2=0;
741 sqr_add_c2(a,7,4,c3,c1,c2);
742 sqr_add_c2(a,6,5,c3,c1,c2);
743 r[11]=c3;
744 c3=0;
745 sqr_add_c(a,6,c1,c2,c3);
746 sqr_add_c2(a,7,5,c1,c2,c3);
747 r[12]=c1;
748 c1=0;
749 sqr_add_c2(a,7,6,c2,c3,c1);
750 r[13]=c2;
751 c2=0;
752 sqr_add_c(a,7,c3,c1,c2);
753 r[14]=c3;
754 r[15]=c1;
755 }
756
757void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
758 {
759#ifdef BN_LLONG
760 BN_ULLONG t,tt;
761#else
762 BN_ULONG bl,bh;
763#endif
764 BN_ULONG t1,t2;
765 BN_ULONG c1,c2,c3;
766
767 c1=0;
768 c2=0;
769 c3=0;
770 sqr_add_c(a,0,c1,c2,c3);
771 r[0]=c1;
772 c1=0;
773 sqr_add_c2(a,1,0,c2,c3,c1);
774 r[1]=c2;
775 c2=0;
776 sqr_add_c(a,1,c3,c1,c2);
777 sqr_add_c2(a,2,0,c3,c1,c2);
778 r[2]=c3;
779 c3=0;
780 sqr_add_c2(a,3,0,c1,c2,c3);
781 sqr_add_c2(a,2,1,c1,c2,c3);
782 r[3]=c1;
783 c1=0;
784 sqr_add_c(a,2,c2,c3,c1);
785 sqr_add_c2(a,3,1,c2,c3,c1);
786 r[4]=c2;
787 c2=0;
788 sqr_add_c2(a,3,2,c3,c1,c2);
789 r[5]=c3;
790 c3=0;
791 sqr_add_c(a,3,c1,c2,c3);
792 r[6]=c1;
793 r[7]=c2;
794 }
795#else /* !BN_MUL_COMBA */
796
797/* hmm... is it faster just to do a multiply? */
798#undef bn_sqr_comba4
799void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
800 {
801 BN_ULONG t[8];
802 bn_sqr_normal(r,a,4,t);
803 }
804
805#undef bn_sqr_comba8
806void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
807 {
808 BN_ULONG t[16];
809 bn_sqr_normal(r,a,8,t);
810 }
811
812void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
813 {
814 r[4]=bn_mul_words( &(r[0]),a,4,b[0]);
815 r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]);
816 r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]);
817 r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]);
818 }
819
820void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
821 {
822 r[ 8]=bn_mul_words( &(r[0]),a,8,b[0]);
823 r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]);
824 r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]);
825 r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]);
826 r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]);
827 r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]);
828 r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]);
829 r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]);
830 }
831
832#endif /* !BN_MUL_COMBA */
diff --git a/src/lib/libcrypto/bn/bn_blind.c b/src/lib/libcrypto/bn/bn_blind.c
deleted file mode 100644
index 2d287e6d1b..0000000000
--- a/src/lib/libcrypto/bn/bn_blind.c
+++ /dev/null
@@ -1,144 +0,0 @@
1/* crypto/bn/bn_blind.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63BN_BLINDING *BN_BLINDING_new(BIGNUM *A, BIGNUM *Ai, BIGNUM *mod)
64 {
65 BN_BLINDING *ret=NULL;
66
67 bn_check_top(Ai);
68 bn_check_top(mod);
69
70 if ((ret=(BN_BLINDING *)OPENSSL_malloc(sizeof(BN_BLINDING))) == NULL)
71 {
72 BNerr(BN_F_BN_BLINDING_NEW,ERR_R_MALLOC_FAILURE);
73 return(NULL);
74 }
75 memset(ret,0,sizeof(BN_BLINDING));
76 if ((ret->A=BN_new()) == NULL) goto err;
77 if ((ret->Ai=BN_new()) == NULL) goto err;
78 if (!BN_copy(ret->A,A)) goto err;
79 if (!BN_copy(ret->Ai,Ai)) goto err;
80 ret->mod=mod;
81 return(ret);
82err:
83 if (ret != NULL) BN_BLINDING_free(ret);
84 return(NULL);
85 }
86
87void BN_BLINDING_free(BN_BLINDING *r)
88 {
89 if(r == NULL)
90 return;
91
92 if (r->A != NULL) BN_free(r->A );
93 if (r->Ai != NULL) BN_free(r->Ai);
94 OPENSSL_free(r);
95 }
96
97int BN_BLINDING_update(BN_BLINDING *b, BN_CTX *ctx)
98 {
99 int ret=0;
100
101 if ((b->A == NULL) || (b->Ai == NULL))
102 {
103 BNerr(BN_F_BN_BLINDING_UPDATE,BN_R_NOT_INITIALIZED);
104 goto err;
105 }
106
107 if (!BN_mod_mul(b->A,b->A,b->A,b->mod,ctx)) goto err;
108 if (!BN_mod_mul(b->Ai,b->Ai,b->Ai,b->mod,ctx)) goto err;
109
110 ret=1;
111err:
112 return(ret);
113 }
114
115int BN_BLINDING_convert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx)
116 {
117 bn_check_top(n);
118
119 if ((b->A == NULL) || (b->Ai == NULL))
120 {
121 BNerr(BN_F_BN_BLINDING_CONVERT,BN_R_NOT_INITIALIZED);
122 return(0);
123 }
124 return(BN_mod_mul(n,n,b->A,b->mod,ctx));
125 }
126
127int BN_BLINDING_invert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx)
128 {
129 int ret;
130
131 bn_check_top(n);
132 if ((b->A == NULL) || (b->Ai == NULL))
133 {
134 BNerr(BN_F_BN_BLINDING_INVERT,BN_R_NOT_INITIALIZED);
135 return(0);
136 }
137 if ((ret=BN_mod_mul(n,n,b->Ai,b->mod,ctx)) >= 0)
138 {
139 if (!BN_BLINDING_update(b,ctx))
140 return(0);
141 }
142 return(ret);
143 }
144
diff --git a/src/lib/libcrypto/bn/bn_ctx.c b/src/lib/libcrypto/bn/bn_ctx.c
deleted file mode 100644
index b1a8d7571e..0000000000
--- a/src/lib/libcrypto/bn/bn_ctx.c
+++ /dev/null
@@ -1,144 +0,0 @@
1/* crypto/bn/bn_ctx.c */
2/* Written by Ulf Moeller for the OpenSSL project. */
3/* ====================================================================
4 * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 *
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 *
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
16 * distribution.
17 *
18 * 3. All advertising materials mentioning features or use of this
19 * software must display the following acknowledgment:
20 * "This product includes software developed by the OpenSSL Project
21 * for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
22 *
23 * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
24 * endorse or promote products derived from this software without
25 * prior written permission. For written permission, please contact
26 * openssl-core@openssl.org.
27 *
28 * 5. Products derived from this software may not be called "OpenSSL"
29 * nor may "OpenSSL" appear in their names without prior written
30 * permission of the OpenSSL Project.
31 *
32 * 6. Redistributions of any form whatsoever must retain the following
33 * acknowledgment:
34 * "This product includes software developed by the OpenSSL Project
35 * for use in the OpenSSL Toolkit (http://www.openssl.org/)"
36 *
37 * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
38 * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
39 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
40 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
41 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
42 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
43 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
44 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
45 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
46 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
47 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
48 * OF THE POSSIBILITY OF SUCH DAMAGE.
49 * ====================================================================
50 *
51 * This product includes cryptographic software written by Eric Young
52 * (eay@cryptsoft.com). This product includes software written by Tim
53 * Hudson (tjh@cryptsoft.com).
54 *
55 */
56
57#ifndef BN_CTX_DEBUG
58# undef NDEBUG /* avoid conflicting definitions */
59# define NDEBUG
60#endif
61
62#include <stdio.h>
63#include <assert.h>
64#include "cryptlib.h"
65#include <openssl/bn.h>
66
67
68BN_CTX *BN_CTX_new(void)
69 {
70 BN_CTX *ret;
71
72 ret=(BN_CTX *)OPENSSL_malloc(sizeof(BN_CTX));
73 if (ret == NULL)
74 {
75 BNerr(BN_F_BN_CTX_NEW,ERR_R_MALLOC_FAILURE);
76 return(NULL);
77 }
78
79 BN_CTX_init(ret);
80 ret->flags=BN_FLG_MALLOCED;
81 return(ret);
82 }
83
84void BN_CTX_init(BN_CTX *ctx)
85 {
86 int i;
87 ctx->tos = 0;
88 ctx->flags = 0;
89 ctx->depth = 0;
90 ctx->too_many = 0;
91 for (i = 0; i < BN_CTX_NUM; i++)
92 BN_init(&(ctx->bn[i]));
93 }
94
95void BN_CTX_free(BN_CTX *ctx)
96 {
97 int i;
98
99 if (ctx == NULL) return;
100 assert(ctx->depth == 0);
101
102 for (i=0; i < BN_CTX_NUM; i++)
103 BN_clear_free(&(ctx->bn[i]));
104 if (ctx->flags & BN_FLG_MALLOCED)
105 OPENSSL_free(ctx);
106 }
107
108void BN_CTX_start(BN_CTX *ctx)
109 {
110 if (ctx->depth < BN_CTX_NUM_POS)
111 ctx->pos[ctx->depth] = ctx->tos;
112 ctx->depth++;
113 }
114
115BIGNUM *BN_CTX_get(BN_CTX *ctx)
116 {
117 if (ctx->depth > BN_CTX_NUM_POS || ctx->tos >= BN_CTX_NUM)
118 {
119 if (!ctx->too_many)
120 {
121 BNerr(BN_F_BN_CTX_GET,BN_R_TOO_MANY_TEMPORARY_VARIABLES);
122 /* disable error code until BN_CTX_end is called: */
123 ctx->too_many = 1;
124 }
125 return NULL;
126 }
127 return (&(ctx->bn[ctx->tos++]));
128 }
129
130void BN_CTX_end(BN_CTX *ctx)
131 {
132 if (ctx == NULL) return;
133 assert(ctx->depth > 0);
134 if (ctx->depth == 0)
135 /* should never happen, but we can tolerate it if not in
136 * debug mode (could be a 'goto err' in the calling function
137 * before BN_CTX_start was reached) */
138 BN_CTX_start(ctx);
139
140 ctx->too_many = 0;
141 ctx->depth--;
142 if (ctx->depth < BN_CTX_NUM_POS)
143 ctx->tos = ctx->pos[ctx->depth];
144 }
diff --git a/src/lib/libcrypto/bn/bn_div.c b/src/lib/libcrypto/bn/bn_div.c
deleted file mode 100644
index c328b5b411..0000000000
--- a/src/lib/libcrypto/bn/bn_div.c
+++ /dev/null
@@ -1,381 +0,0 @@
1/* crypto/bn/bn_div.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include <openssl/bn.h>
61#include "cryptlib.h"
62#include "bn_lcl.h"
63
64/* The old slow way */
65#if 0
66int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d,
67 BN_CTX *ctx)
68 {
69 int i,nm,nd;
70 int ret = 0;
71 BIGNUM *D;
72
73 bn_check_top(m);
74 bn_check_top(d);
75 if (BN_is_zero(d))
76 {
77 BNerr(BN_F_BN_DIV,BN_R_DIV_BY_ZERO);
78 return(0);
79 }
80
81 if (BN_ucmp(m,d) < 0)
82 {
83 if (rem != NULL)
84 { if (BN_copy(rem,m) == NULL) return(0); }
85 if (dv != NULL) BN_zero(dv);
86 return(1);
87 }
88
89 BN_CTX_start(ctx);
90 D = BN_CTX_get(ctx);
91 if (dv == NULL) dv = BN_CTX_get(ctx);
92 if (rem == NULL) rem = BN_CTX_get(ctx);
93 if (D == NULL || dv == NULL || rem == NULL)
94 goto end;
95
96 nd=BN_num_bits(d);
97 nm=BN_num_bits(m);
98 if (BN_copy(D,d) == NULL) goto end;
99 if (BN_copy(rem,m) == NULL) goto end;
100
101 /* The next 2 are needed so we can do a dv->d[0]|=1 later
102 * since BN_lshift1 will only work once there is a value :-) */
103 BN_zero(dv);
104 bn_wexpand(dv,1);
105 dv->top=1;
106
107 if (!BN_lshift(D,D,nm-nd)) goto end;
108 for (i=nm-nd; i>=0; i--)
109 {
110 if (!BN_lshift1(dv,dv)) goto end;
111 if (BN_ucmp(rem,D) >= 0)
112 {
113 dv->d[0]|=1;
114 if (!BN_usub(rem,rem,D)) goto end;
115 }
116/* CAN IMPROVE (and have now :=) */
117 if (!BN_rshift1(D,D)) goto end;
118 }
119 rem->neg=BN_is_zero(rem)?0:m->neg;
120 dv->neg=m->neg^d->neg;
121 ret = 1;
122 end:
123 BN_CTX_end(ctx);
124 return(ret);
125 }
126
127#else
128
129#if !defined(NO_ASM) && !defined(NO_INLINE_ASM) && !defined(PEDANTIC) && !defined(BN_DIV3W)
130# if defined(__GNUC__) && __GNUC__>=2
131# if defined(__i386)
132 /*
133 * There were two reasons for implementing this template:
134 * - GNU C generates a call to a function (__udivdi3 to be exact)
135 * in reply to ((((BN_ULLONG)n0)<<BN_BITS2)|n1)/d0 (I fail to
136 * understand why...);
137 * - divl doesn't only calculate quotient, but also leaves
138 * remainder in %edx which we can definitely use here:-)
139 *
140 * <appro@fy.chalmers.se>
141 */
142# define bn_div_words(n0,n1,d0) \
143 ({ asm volatile ( \
144 "divl %4" \
145 : "=a"(q), "=d"(rem) \
146 : "a"(n1), "d"(n0), "g"(d0) \
147 : "cc"); \
148 q; \
149 })
150# define REMAINDER_IS_ALREADY_CALCULATED
151# endif /* __<cpu> */
152# endif /* __GNUC__ */
153#endif /* NO_ASM */
154
155int BN_div(BIGNUM *dv, BIGNUM *rm, const BIGNUM *num, const BIGNUM *divisor,
156 BN_CTX *ctx)
157 {
158 int norm_shift,i,j,loop;
159 BIGNUM *tmp,wnum,*snum,*sdiv,*res;
160 BN_ULONG *resp,*wnump;
161 BN_ULONG d0,d1;
162 int num_n,div_n;
163
164 bn_check_top(num);
165 bn_check_top(divisor);
166
167 if (BN_is_zero(divisor))
168 {
169 BNerr(BN_F_BN_DIV,BN_R_DIV_BY_ZERO);
170 return(0);
171 }
172
173 if (BN_ucmp(num,divisor) < 0)
174 {
175 if (rm != NULL)
176 { if (BN_copy(rm,num) == NULL) return(0); }
177 if (dv != NULL) BN_zero(dv);
178 return(1);
179 }
180
181 BN_CTX_start(ctx);
182 tmp=BN_CTX_get(ctx);
183 snum=BN_CTX_get(ctx);
184 sdiv=BN_CTX_get(ctx);
185 if (dv == NULL)
186 res=BN_CTX_get(ctx);
187 else res=dv;
188 if (sdiv==NULL || res == NULL) goto err;
189 tmp->neg=0;
190
191 /* First we normalise the numbers */
192 norm_shift=BN_BITS2-((BN_num_bits(divisor))%BN_BITS2);
193 BN_lshift(sdiv,divisor,norm_shift);
194 sdiv->neg=0;
195 norm_shift+=BN_BITS2;
196 BN_lshift(snum,num,norm_shift);
197 snum->neg=0;
198 div_n=sdiv->top;
199 num_n=snum->top;
200 loop=num_n-div_n;
201
202 /* Lets setup a 'window' into snum
203 * This is the part that corresponds to the current
204 * 'area' being divided */
205 BN_init(&wnum);
206 wnum.d= &(snum->d[loop]);
207 wnum.top= div_n;
208 wnum.dmax= snum->dmax+1; /* a bit of a lie */
209
210 /* Get the top 2 words of sdiv */
211 /* i=sdiv->top; */
212 d0=sdiv->d[div_n-1];
213 d1=(div_n == 1)?0:sdiv->d[div_n-2];
214
215 /* pointer to the 'top' of snum */
216 wnump= &(snum->d[num_n-1]);
217
218 /* Setup to 'res' */
219 res->neg= (num->neg^divisor->neg);
220 if (!bn_wexpand(res,(loop+1))) goto err;
221 res->top=loop;
222 resp= &(res->d[loop-1]);
223
224 /* space for temp */
225 if (!bn_wexpand(tmp,(div_n+1))) goto err;
226
227 if (BN_ucmp(&wnum,sdiv) >= 0)
228 {
229 if (!BN_usub(&wnum,&wnum,sdiv)) goto err;
230 *resp=1;
231 res->d[res->top-1]=1;
232 }
233 else
234 res->top--;
235 resp--;
236
237 for (i=0; i<loop-1; i++)
238 {
239 BN_ULONG q,l0;
240#if defined(BN_DIV3W) && !defined(NO_ASM)
241 BN_ULONG bn_div_3_words(BN_ULONG*,BN_ULONG,BN_ULONG);
242 q=bn_div_3_words(wnump,d1,d0);
243#else
244 BN_ULONG n0,n1,rem=0;
245
246 n0=wnump[0];
247 n1=wnump[-1];
248 if (n0 == d0)
249 q=BN_MASK2;
250 else /* n0 < d0 */
251 {
252#ifdef BN_LLONG
253 BN_ULLONG t2;
254
255#if defined(BN_LLONG) && defined(BN_DIV2W) && !defined(bn_div_words)
256 q=(BN_ULONG)(((((BN_ULLONG)n0)<<BN_BITS2)|n1)/d0);
257#else
258 q=bn_div_words(n0,n1,d0);
259#endif
260
261#ifndef REMAINDER_IS_ALREADY_CALCULATED
262 /*
263 * rem doesn't have to be BN_ULLONG. The least we
264 * know it's less that d0, isn't it?
265 */
266 rem=(n1-q*d0)&BN_MASK2;
267#endif
268 t2=(BN_ULLONG)d1*q;
269
270 for (;;)
271 {
272 if (t2 <= ((((BN_ULLONG)rem)<<BN_BITS2)|wnump[-2]))
273 break;
274 q--;
275 rem += d0;
276 if (rem < d0) break; /* don't let rem overflow */
277 t2 -= d1;
278 }
279#else /* !BN_LLONG */
280 BN_ULONG t2l,t2h,ql,qh;
281
282 q=bn_div_words(n0,n1,d0);
283#ifndef REMAINDER_IS_ALREADY_CALCULATED
284 rem=(n1-q*d0)&BN_MASK2;
285#endif
286
287#ifdef BN_UMULT_HIGH
288 t2l = d1 * q;
289 t2h = BN_UMULT_HIGH(d1,q);
290#else
291 t2l=LBITS(d1); t2h=HBITS(d1);
292 ql =LBITS(q); qh =HBITS(q);
293 mul64(t2l,t2h,ql,qh); /* t2=(BN_ULLONG)d1*q; */
294#endif
295
296 for (;;)
297 {
298 if ((t2h < rem) ||
299 ((t2h == rem) && (t2l <= wnump[-2])))
300 break;
301 q--;
302 rem += d0;
303 if (rem < d0) break; /* don't let rem overflow */
304 if (t2l < d1) t2h--; t2l -= d1;
305 }
306#endif /* !BN_LLONG */
307 }
308#endif /* !BN_DIV3W */
309
310 l0=bn_mul_words(tmp->d,sdiv->d,div_n,q);
311 wnum.d--; wnum.top++;
312 tmp->d[div_n]=l0;
313 for (j=div_n+1; j>0; j--)
314 if (tmp->d[j-1]) break;
315 tmp->top=j;
316
317 j=wnum.top;
318 BN_sub(&wnum,&wnum,tmp);
319
320 snum->top=snum->top+wnum.top-j;
321
322 if (wnum.neg)
323 {
324 q--;
325 j=wnum.top;
326 BN_add(&wnum,&wnum,sdiv);
327 snum->top+=wnum.top-j;
328 }
329 *(resp--)=q;
330 wnump--;
331 }
332 if (rm != NULL)
333 {
334 BN_rshift(rm,snum,norm_shift);
335 rm->neg=num->neg;
336 }
337 BN_CTX_end(ctx);
338 return(1);
339err:
340 BN_CTX_end(ctx);
341 return(0);
342 }
343
344#endif
345
346/* rem != m */
347int BN_mod(BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, BN_CTX *ctx)
348 {
349#if 0 /* The old slow way */
350 int i,nm,nd;
351 BIGNUM *dv;
352
353 if (BN_ucmp(m,d) < 0)
354 return((BN_copy(rem,m) == NULL)?0:1);
355
356 BN_CTX_start(ctx);
357 dv=BN_CTX_get(ctx);
358
359 if (!BN_copy(rem,m)) goto err;
360
361 nm=BN_num_bits(rem);
362 nd=BN_num_bits(d);
363 if (!BN_lshift(dv,d,nm-nd)) goto err;
364 for (i=nm-nd; i>=0; i--)
365 {
366 if (BN_cmp(rem,dv) >= 0)
367 {
368 if (!BN_sub(rem,rem,dv)) goto err;
369 }
370 if (!BN_rshift1(dv,dv)) goto err;
371 }
372 BN_CTX_end(ctx);
373 return(1);
374 err:
375 BN_CTX_end(ctx);
376 return(0);
377#else
378 return(BN_div(NULL,rem,m,d,ctx));
379#endif
380 }
381
diff --git a/src/lib/libcrypto/bn/bn_err.c b/src/lib/libcrypto/bn/bn_err.c
deleted file mode 100644
index adc6a214fc..0000000000
--- a/src/lib/libcrypto/bn/bn_err.c
+++ /dev/null
@@ -1,124 +0,0 @@
1/* crypto/bn/bn_err.c */
2/* ====================================================================
3 * Copyright (c) 1999 The OpenSSL Project. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 *
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 *
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in
14 * the documentation and/or other materials provided with the
15 * distribution.
16 *
17 * 3. All advertising materials mentioning features or use of this
18 * software must display the following acknowledgment:
19 * "This product includes software developed by the OpenSSL Project
20 * for use in the OpenSSL Toolkit. (http://www.OpenSSL.org/)"
21 *
22 * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
23 * endorse or promote products derived from this software without
24 * prior written permission. For written permission, please contact
25 * openssl-core@OpenSSL.org.
26 *
27 * 5. Products derived from this software may not be called "OpenSSL"
28 * nor may "OpenSSL" appear in their names without prior written
29 * permission of the OpenSSL Project.
30 *
31 * 6. Redistributions of any form whatsoever must retain the following
32 * acknowledgment:
33 * "This product includes software developed by the OpenSSL Project
34 * for use in the OpenSSL Toolkit (http://www.OpenSSL.org/)"
35 *
36 * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
37 * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
38 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
39 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
40 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
41 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
42 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
43 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
44 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
45 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
46 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
47 * OF THE POSSIBILITY OF SUCH DAMAGE.
48 * ====================================================================
49 *
50 * This product includes cryptographic software written by Eric Young
51 * (eay@cryptsoft.com). This product includes software written by Tim
52 * Hudson (tjh@cryptsoft.com).
53 *
54 */
55
56/* NOTE: this file was auto generated by the mkerr.pl script: any changes
57 * made to it will be overwritten when the script next updates this file,
58 * only reason strings will be preserved.
59 */
60
61#include <stdio.h>
62#include <openssl/err.h>
63#include <openssl/bn.h>
64
65/* BEGIN ERROR CODES */
66#ifndef NO_ERR
67static ERR_STRING_DATA BN_str_functs[]=
68 {
69{ERR_PACK(0,BN_F_BN_BLINDING_CONVERT,0), "BN_BLINDING_convert"},
70{ERR_PACK(0,BN_F_BN_BLINDING_INVERT,0), "BN_BLINDING_invert"},
71{ERR_PACK(0,BN_F_BN_BLINDING_NEW,0), "BN_BLINDING_new"},
72{ERR_PACK(0,BN_F_BN_BLINDING_UPDATE,0), "BN_BLINDING_update"},
73{ERR_PACK(0,BN_F_BN_BN2DEC,0), "BN_bn2dec"},
74{ERR_PACK(0,BN_F_BN_BN2HEX,0), "BN_bn2hex"},
75{ERR_PACK(0,BN_F_BN_CTX_GET,0), "BN_CTX_get"},
76{ERR_PACK(0,BN_F_BN_CTX_NEW,0), "BN_CTX_new"},
77{ERR_PACK(0,BN_F_BN_DIV,0), "BN_div"},
78{ERR_PACK(0,BN_F_BN_EXPAND2,0), "bn_expand2"},
79{ERR_PACK(0,BN_F_BN_MOD_EXP2_MONT,0), "BN_mod_exp2_mont"},
80{ERR_PACK(0,BN_F_BN_MOD_EXP_MONT,0), "BN_mod_exp_mont"},
81{ERR_PACK(0,BN_F_BN_MOD_EXP_MONT_WORD,0), "BN_mod_exp_mont_word"},
82{ERR_PACK(0,BN_F_BN_MOD_INVERSE,0), "BN_mod_inverse"},
83{ERR_PACK(0,BN_F_BN_MOD_MUL_RECIPROCAL,0), "BN_mod_mul_reciprocal"},
84{ERR_PACK(0,BN_F_BN_MPI2BN,0), "BN_mpi2bn"},
85{ERR_PACK(0,BN_F_BN_NEW,0), "BN_new"},
86{ERR_PACK(0,BN_F_BN_RAND,0), "BN_rand"},
87{ERR_PACK(0,BN_F_BN_RAND_RANGE,0), "BN_rand_range"},
88{ERR_PACK(0,BN_F_BN_USUB,0), "BN_usub"},
89{0,NULL}
90 };
91
92static ERR_STRING_DATA BN_str_reasons[]=
93 {
94{BN_R_ARG2_LT_ARG3 ,"arg2 lt arg3"},
95{BN_R_BAD_RECIPROCAL ,"bad reciprocal"},
96{BN_R_BIGNUM_TOO_LONG ,"bignum too long"},
97{BN_R_CALLED_WITH_EVEN_MODULUS ,"called with even modulus"},
98{BN_R_DIV_BY_ZERO ,"div by zero"},
99{BN_R_ENCODING_ERROR ,"encoding error"},
100{BN_R_EXPAND_ON_STATIC_BIGNUM_DATA ,"expand on static bignum data"},
101{BN_R_INVALID_LENGTH ,"invalid length"},
102{BN_R_INVALID_RANGE ,"invalid range"},
103{BN_R_NOT_INITIALIZED ,"not initialized"},
104{BN_R_NO_INVERSE ,"no inverse"},
105{BN_R_TOO_MANY_TEMPORARY_VARIABLES ,"too many temporary variables"},
106{0,NULL}
107 };
108
109#endif
110
111void ERR_load_BN_strings(void)
112 {
113 static int init=1;
114
115 if (init)
116 {
117 init=0;
118#ifndef NO_ERR
119 ERR_load_strings(ERR_LIB_BN,BN_str_functs);
120 ERR_load_strings(ERR_LIB_BN,BN_str_reasons);
121#endif
122
123 }
124 }
diff --git a/src/lib/libcrypto/bn/bn_exp.c b/src/lib/libcrypto/bn/bn_exp.c
deleted file mode 100644
index d2c91628ac..0000000000
--- a/src/lib/libcrypto/bn/bn_exp.c
+++ /dev/null
@@ -1,696 +0,0 @@
1/* crypto/bn/bn_exp.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58/* ====================================================================
59 * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved.
60 *
61 * Redistribution and use in source and binary forms, with or without
62 * modification, are permitted provided that the following conditions
63 * are met:
64 *
65 * 1. Redistributions of source code must retain the above copyright
66 * notice, this list of conditions and the following disclaimer.
67 *
68 * 2. Redistributions in binary form must reproduce the above copyright
69 * notice, this list of conditions and the following disclaimer in
70 * the documentation and/or other materials provided with the
71 * distribution.
72 *
73 * 3. All advertising materials mentioning features or use of this
74 * software must display the following acknowledgment:
75 * "This product includes software developed by the OpenSSL Project
76 * for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
77 *
78 * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
79 * endorse or promote products derived from this software without
80 * prior written permission. For written permission, please contact
81 * openssl-core@openssl.org.
82 *
83 * 5. Products derived from this software may not be called "OpenSSL"
84 * nor may "OpenSSL" appear in their names without prior written
85 * permission of the OpenSSL Project.
86 *
87 * 6. Redistributions of any form whatsoever must retain the following
88 * acknowledgment:
89 * "This product includes software developed by the OpenSSL Project
90 * for use in the OpenSSL Toolkit (http://www.openssl.org/)"
91 *
92 * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
93 * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
94 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
95 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
96 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
97 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
98 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
99 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
100 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
101 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
102 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
103 * OF THE POSSIBILITY OF SUCH DAMAGE.
104 * ====================================================================
105 *
106 * This product includes cryptographic software written by Eric Young
107 * (eay@cryptsoft.com). This product includes software written by Tim
108 * Hudson (tjh@cryptsoft.com).
109 *
110 */
111
112
113#include <stdio.h>
114#include "cryptlib.h"
115#include "bn_lcl.h"
116
117#define TABLE_SIZE 32
118
119/* slow but works */
120int BN_mod_mul(BIGNUM *ret, BIGNUM *a, BIGNUM *b, const BIGNUM *m, BN_CTX *ctx)
121 {
122 BIGNUM *t;
123 int r=0;
124
125 bn_check_top(a);
126 bn_check_top(b);
127 bn_check_top(m);
128
129 BN_CTX_start(ctx);
130 if ((t = BN_CTX_get(ctx)) == NULL) goto err;
131 if (a == b)
132 { if (!BN_sqr(t,a,ctx)) goto err; }
133 else
134 { if (!BN_mul(t,a,b,ctx)) goto err; }
135 if (!BN_mod(ret,t,m,ctx)) goto err;
136 r=1;
137err:
138 BN_CTX_end(ctx);
139 return(r);
140 }
141
142
143/* this one works - simple but works */
144int BN_exp(BIGNUM *r, BIGNUM *a, BIGNUM *p, BN_CTX *ctx)
145 {
146 int i,bits,ret=0;
147 BIGNUM *v,*rr;
148
149 BN_CTX_start(ctx);
150 if ((r == a) || (r == p))
151 rr = BN_CTX_get(ctx);
152 else
153 rr = r;
154 if ((v = BN_CTX_get(ctx)) == NULL) goto err;
155
156 if (BN_copy(v,a) == NULL) goto err;
157 bits=BN_num_bits(p);
158
159 if (BN_is_odd(p))
160 { if (BN_copy(rr,a) == NULL) goto err; }
161 else { if (!BN_one(rr)) goto err; }
162
163 for (i=1; i<bits; i++)
164 {
165 if (!BN_sqr(v,v,ctx)) goto err;
166 if (BN_is_bit_set(p,i))
167 {
168 if (!BN_mul(rr,rr,v,ctx)) goto err;
169 }
170 }
171 ret=1;
172err:
173 if (r != rr) BN_copy(r,rr);
174 BN_CTX_end(ctx);
175 return(ret);
176 }
177
178
179int BN_mod_exp(BIGNUM *r, BIGNUM *a, const BIGNUM *p, const BIGNUM *m,
180 BN_CTX *ctx)
181 {
182 int ret;
183
184 bn_check_top(a);
185 bn_check_top(p);
186 bn_check_top(m);
187
188#ifdef MONT_MUL_MOD
189 /* I have finally been able to take out this pre-condition of
190 * the top bit being set. It was caused by an error in BN_div
191 * with negatives. There was also another problem when for a^b%m
192 * a >= m. eay 07-May-97 */
193/* if ((m->d[m->top-1]&BN_TBIT) && BN_is_odd(m)) */
194
195 if (BN_is_odd(m))
196 {
197 if (a->top == 1)
198 {
199 BN_ULONG A = a->d[0];
200 ret=BN_mod_exp_mont_word(r,A,p,m,ctx,NULL);
201 }
202 else
203 ret=BN_mod_exp_mont(r,a,p,m,ctx,NULL);
204 }
205 else
206#endif
207#ifdef RECP_MUL_MOD
208 { ret=BN_mod_exp_recp(r,a,p,m,ctx); }
209#else
210 { ret=BN_mod_exp_simple(r,a,p,m,ctx); }
211#endif
212
213 return(ret);
214 }
215
216
217int BN_mod_exp_recp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p,
218 const BIGNUM *m, BN_CTX *ctx)
219 {
220 int i,j,bits,ret=0,wstart,wend,window,wvalue;
221 int start=1,ts=0;
222 BIGNUM *aa;
223 BIGNUM val[TABLE_SIZE];
224 BN_RECP_CTX recp;
225
226 bits=BN_num_bits(p);
227
228 if (bits == 0)
229 {
230 BN_one(r);
231 return(1);
232 }
233
234 BN_CTX_start(ctx);
235 if ((aa = BN_CTX_get(ctx)) == NULL) goto err;
236
237 BN_RECP_CTX_init(&recp);
238 if (BN_RECP_CTX_set(&recp,m,ctx) <= 0) goto err;
239
240 BN_init(&(val[0]));
241 ts=1;
242
243 if (!BN_mod(&(val[0]),a,m,ctx)) goto err; /* 1 */
244
245 window = BN_window_bits_for_exponent_size(bits);
246 if (window > 1)
247 {
248 if (!BN_mod_mul_reciprocal(aa,&(val[0]),&(val[0]),&recp,ctx))
249 goto err; /* 2 */
250 j=1<<(window-1);
251 for (i=1; i<j; i++)
252 {
253 BN_init(&val[i]);
254 if (!BN_mod_mul_reciprocal(&(val[i]),&(val[i-1]),aa,&recp,ctx))
255 goto err;
256 }
257 ts=i;
258 }
259
260 start=1; /* This is used to avoid multiplication etc
261 * when there is only the value '1' in the
262 * buffer. */
263 wvalue=0; /* The 'value' of the window */
264 wstart=bits-1; /* The top bit of the window */
265 wend=0; /* The bottom bit of the window */
266
267 if (!BN_one(r)) goto err;
268
269 for (;;)
270 {
271 if (BN_is_bit_set(p,wstart) == 0)
272 {
273 if (!start)
274 if (!BN_mod_mul_reciprocal(r,r,r,&recp,ctx))
275 goto err;
276 if (wstart == 0) break;
277 wstart--;
278 continue;
279 }
280 /* We now have wstart on a 'set' bit, we now need to work out
281 * how bit a window to do. To do this we need to scan
282 * forward until the last set bit before the end of the
283 * window */
284 j=wstart;
285 wvalue=1;
286 wend=0;
287 for (i=1; i<window; i++)
288 {
289 if (wstart-i < 0) break;
290 if (BN_is_bit_set(p,wstart-i))
291 {
292 wvalue<<=(i-wend);
293 wvalue|=1;
294 wend=i;
295 }
296 }
297
298 /* wend is the size of the current window */
299 j=wend+1;
300 /* add the 'bytes above' */
301 if (!start)
302 for (i=0; i<j; i++)
303 {
304 if (!BN_mod_mul_reciprocal(r,r,r,&recp,ctx))
305 goto err;
306 }
307
308 /* wvalue will be an odd number < 2^window */
309 if (!BN_mod_mul_reciprocal(r,r,&(val[wvalue>>1]),&recp,ctx))
310 goto err;
311
312 /* move the 'window' down further */
313 wstart-=wend+1;
314 wvalue=0;
315 start=0;
316 if (wstart < 0) break;
317 }
318 ret=1;
319err:
320 BN_CTX_end(ctx);
321 for (i=0; i<ts; i++)
322 BN_clear_free(&(val[i]));
323 BN_RECP_CTX_free(&recp);
324 return(ret);
325 }
326
327
328int BN_mod_exp_mont(BIGNUM *rr, BIGNUM *a, const BIGNUM *p,
329 const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont)
330 {
331 int i,j,bits,ret=0,wstart,wend,window,wvalue;
332 int start=1,ts=0;
333 BIGNUM *d,*r;
334 BIGNUM *aa;
335 BIGNUM val[TABLE_SIZE];
336 BN_MONT_CTX *mont=NULL;
337
338 bn_check_top(a);
339 bn_check_top(p);
340 bn_check_top(m);
341
342 if (!(m->d[0] & 1))
343 {
344 BNerr(BN_F_BN_MOD_EXP_MONT,BN_R_CALLED_WITH_EVEN_MODULUS);
345 return(0);
346 }
347 bits=BN_num_bits(p);
348 if (bits == 0)
349 {
350 BN_one(rr);
351 return(1);
352 }
353 BN_CTX_start(ctx);
354 d = BN_CTX_get(ctx);
355 r = BN_CTX_get(ctx);
356 if (d == NULL || r == NULL) goto err;
357
358 /* If this is not done, things will break in the montgomery
359 * part */
360
361 if (in_mont != NULL)
362 mont=in_mont;
363 else
364 {
365 if ((mont=BN_MONT_CTX_new()) == NULL) goto err;
366 if (!BN_MONT_CTX_set(mont,m,ctx)) goto err;
367 }
368
369 BN_init(&val[0]);
370 ts=1;
371 if (BN_ucmp(a,m) >= 0)
372 {
373 if (!BN_mod(&(val[0]),a,m,ctx))
374 goto err;
375 aa= &(val[0]);
376 }
377 else
378 aa=a;
379 if (!BN_to_montgomery(&(val[0]),aa,mont,ctx)) goto err; /* 1 */
380
381 window = BN_window_bits_for_exponent_size(bits);
382 if (window > 1)
383 {
384 if (!BN_mod_mul_montgomery(d,&(val[0]),&(val[0]),mont,ctx)) goto err; /* 2 */
385 j=1<<(window-1);
386 for (i=1; i<j; i++)
387 {
388 BN_init(&(val[i]));
389 if (!BN_mod_mul_montgomery(&(val[i]),&(val[i-1]),d,mont,ctx))
390 goto err;
391 }
392 ts=i;
393 }
394
395 start=1; /* This is used to avoid multiplication etc
396 * when there is only the value '1' in the
397 * buffer. */
398 wvalue=0; /* The 'value' of the window */
399 wstart=bits-1; /* The top bit of the window */
400 wend=0; /* The bottom bit of the window */
401
402 if (!BN_to_montgomery(r,BN_value_one(),mont,ctx)) goto err;
403 for (;;)
404 {
405 if (BN_is_bit_set(p,wstart) == 0)
406 {
407 if (!start)
408 {
409 if (!BN_mod_mul_montgomery(r,r,r,mont,ctx))
410 goto err;
411 }
412 if (wstart == 0) break;
413 wstart--;
414 continue;
415 }
416 /* We now have wstart on a 'set' bit, we now need to work out
417 * how bit a window to do. To do this we need to scan
418 * forward until the last set bit before the end of the
419 * window */
420 j=wstart;
421 wvalue=1;
422 wend=0;
423 for (i=1; i<window; i++)
424 {
425 if (wstart-i < 0) break;
426 if (BN_is_bit_set(p,wstart-i))
427 {
428 wvalue<<=(i-wend);
429 wvalue|=1;
430 wend=i;
431 }
432 }
433
434 /* wend is the size of the current window */
435 j=wend+1;
436 /* add the 'bytes above' */
437 if (!start)
438 for (i=0; i<j; i++)
439 {
440 if (!BN_mod_mul_montgomery(r,r,r,mont,ctx))
441 goto err;
442 }
443
444 /* wvalue will be an odd number < 2^window */
445 if (!BN_mod_mul_montgomery(r,r,&(val[wvalue>>1]),mont,ctx))
446 goto err;
447
448 /* move the 'window' down further */
449 wstart-=wend+1;
450 wvalue=0;
451 start=0;
452 if (wstart < 0) break;
453 }
454 if (!BN_from_montgomery(rr,r,mont,ctx)) goto err;
455 ret=1;
456err:
457 if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont);
458 BN_CTX_end(ctx);
459 for (i=0; i<ts; i++)
460 BN_clear_free(&(val[i]));
461 return(ret);
462 }
463
464int BN_mod_exp_mont_word(BIGNUM *rr, BN_ULONG a, const BIGNUM *p,
465 const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont)
466 {
467 BN_MONT_CTX *mont = NULL;
468 int b, bits, ret=0;
469 int r_is_one;
470 BN_ULONG w, next_w;
471 BIGNUM *d, *r, *t;
472 BIGNUM *swap_tmp;
473#define BN_MOD_MUL_WORD(r, w, m) \
474 (BN_mul_word(r, (w)) && \
475 (/* BN_ucmp(r, (m)) < 0 ? 1 :*/ \
476 (BN_mod(t, r, m, ctx) && (swap_tmp = r, r = t, t = swap_tmp, 1))))
477 /* BN_MOD_MUL_WORD is only used with 'w' large,
478 * so the BN_ucmp test is probably more overhead
479 * than always using BN_mod (which uses BN_copy if
480 * a similar test returns true). */
481#define BN_TO_MONTGOMERY_WORD(r, w, mont) \
482 (BN_set_word(r, (w)) && BN_to_montgomery(r, r, (mont), ctx))
483
484 bn_check_top(p);
485 bn_check_top(m);
486
487 if (!(m->d[0] & 1))
488 {
489 BNerr(BN_F_BN_MOD_EXP_MONT_WORD,BN_R_CALLED_WITH_EVEN_MODULUS);
490 return(0);
491 }
492 bits = BN_num_bits(p);
493 if (bits == 0)
494 {
495 BN_one(rr);
496 return(1);
497 }
498 BN_CTX_start(ctx);
499 d = BN_CTX_get(ctx);
500 r = BN_CTX_get(ctx);
501 t = BN_CTX_get(ctx);
502 if (d == NULL || r == NULL || t == NULL) goto err;
503
504 if (in_mont != NULL)
505 mont=in_mont;
506 else
507 {
508 if ((mont = BN_MONT_CTX_new()) == NULL) goto err;
509 if (!BN_MONT_CTX_set(mont, m, ctx)) goto err;
510 }
511
512 r_is_one = 1; /* except for Montgomery factor */
513
514 /* bits-1 >= 0 */
515
516 /* The result is accumulated in the product r*w. */
517 w = a; /* bit 'bits-1' of 'p' is always set */
518 for (b = bits-2; b >= 0; b--)
519 {
520 /* First, square r*w. */
521 next_w = w*w;
522 if ((next_w/w) != w) /* overflow */
523 {
524 if (r_is_one)
525 {
526 if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) goto err;
527 r_is_one = 0;
528 }
529 else
530 {
531 if (!BN_MOD_MUL_WORD(r, w, m)) goto err;
532 }
533 next_w = 1;
534 }
535 w = next_w;
536 if (!r_is_one)
537 {
538 if (!BN_mod_mul_montgomery(r, r, r, mont, ctx)) goto err;
539 }
540
541 /* Second, multiply r*w by 'a' if exponent bit is set. */
542 if (BN_is_bit_set(p, b))
543 {
544 next_w = w*a;
545 if ((next_w/a) != w) /* overflow */
546 {
547 if (r_is_one)
548 {
549 if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) goto err;
550 r_is_one = 0;
551 }
552 else
553 {
554 if (!BN_MOD_MUL_WORD(r, w, m)) goto err;
555 }
556 next_w = a;
557 }
558 w = next_w;
559 }
560 }
561
562 /* Finally, set r:=r*w. */
563 if (w != 1)
564 {
565 if (r_is_one)
566 {
567 if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) goto err;
568 r_is_one = 0;
569 }
570 else
571 {
572 if (!BN_MOD_MUL_WORD(r, w, m)) goto err;
573 }
574 }
575
576 if (r_is_one) /* can happen only if a == 1*/
577 {
578 if (!BN_one(rr)) goto err;
579 }
580 else
581 {
582 if (!BN_from_montgomery(rr, r, mont, ctx)) goto err;
583 }
584 ret = 1;
585err:
586 if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont);
587 BN_CTX_end(ctx);
588 return(ret);
589 }
590
591
592/* The old fallback, simple version :-) */
593int BN_mod_exp_simple(BIGNUM *r, BIGNUM *a, BIGNUM *p, BIGNUM *m,
594 BN_CTX *ctx)
595 {
596 int i,j,bits,ret=0,wstart,wend,window,wvalue,ts=0;
597 int start=1;
598 BIGNUM *d;
599 BIGNUM val[TABLE_SIZE];
600
601 bits=BN_num_bits(p);
602
603 if (bits == 0)
604 {
605 BN_one(r);
606 return(1);
607 }
608
609 BN_CTX_start(ctx);
610 if ((d = BN_CTX_get(ctx)) == NULL) goto err;
611
612 BN_init(&(val[0]));
613 ts=1;
614 if (!BN_mod(&(val[0]),a,m,ctx)) goto err; /* 1 */
615
616 window = BN_window_bits_for_exponent_size(bits);
617 if (window > 1)
618 {
619 if (!BN_mod_mul(d,&(val[0]),&(val[0]),m,ctx))
620 goto err; /* 2 */
621 j=1<<(window-1);
622 for (i=1; i<j; i++)
623 {
624 BN_init(&(val[i]));
625 if (!BN_mod_mul(&(val[i]),&(val[i-1]),d,m,ctx))
626 goto err;
627 }
628 ts=i;
629 }
630
631 start=1; /* This is used to avoid multiplication etc
632 * when there is only the value '1' in the
633 * buffer. */
634 wvalue=0; /* The 'value' of the window */
635 wstart=bits-1; /* The top bit of the window */
636 wend=0; /* The bottom bit of the window */
637
638 if (!BN_one(r)) goto err;
639
640 for (;;)
641 {
642 if (BN_is_bit_set(p,wstart) == 0)
643 {
644 if (!start)
645 if (!BN_mod_mul(r,r,r,m,ctx))
646 goto err;
647 if (wstart == 0) break;
648 wstart--;
649 continue;
650 }
651 /* We now have wstart on a 'set' bit, we now need to work out
652 * how bit a window to do. To do this we need to scan
653 * forward until the last set bit before the end of the
654 * window */
655 j=wstart;
656 wvalue=1;
657 wend=0;
658 for (i=1; i<window; i++)
659 {
660 if (wstart-i < 0) break;
661 if (BN_is_bit_set(p,wstart-i))
662 {
663 wvalue<<=(i-wend);
664 wvalue|=1;
665 wend=i;
666 }
667 }
668
669 /* wend is the size of the current window */
670 j=wend+1;
671 /* add the 'bytes above' */
672 if (!start)
673 for (i=0; i<j; i++)
674 {
675 if (!BN_mod_mul(r,r,r,m,ctx))
676 goto err;
677 }
678
679 /* wvalue will be an odd number < 2^window */
680 if (!BN_mod_mul(r,r,&(val[wvalue>>1]),m,ctx))
681 goto err;
682
683 /* move the 'window' down further */
684 wstart-=wend+1;
685 wvalue=0;
686 start=0;
687 if (wstart < 0) break;
688 }
689 ret=1;
690err:
691 BN_CTX_end(ctx);
692 for (i=0; i<ts; i++)
693 BN_clear_free(&(val[i]));
694 return(ret);
695 }
696
diff --git a/src/lib/libcrypto/bn/bn_exp2.c b/src/lib/libcrypto/bn/bn_exp2.c
deleted file mode 100644
index 29029f4c72..0000000000
--- a/src/lib/libcrypto/bn/bn_exp2.c
+++ /dev/null
@@ -1,300 +0,0 @@
1/* crypto/bn/bn_exp2.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58/* ====================================================================
59 * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved.
60 *
61 * Redistribution and use in source and binary forms, with or without
62 * modification, are permitted provided that the following conditions
63 * are met:
64 *
65 * 1. Redistributions of source code must retain the above copyright
66 * notice, this list of conditions and the following disclaimer.
67 *
68 * 2. Redistributions in binary form must reproduce the above copyright
69 * notice, this list of conditions and the following disclaimer in
70 * the documentation and/or other materials provided with the
71 * distribution.
72 *
73 * 3. All advertising materials mentioning features or use of this
74 * software must display the following acknowledgment:
75 * "This product includes software developed by the OpenSSL Project
76 * for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
77 *
78 * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
79 * endorse or promote products derived from this software without
80 * prior written permission. For written permission, please contact
81 * openssl-core@openssl.org.
82 *
83 * 5. Products derived from this software may not be called "OpenSSL"
84 * nor may "OpenSSL" appear in their names without prior written
85 * permission of the OpenSSL Project.
86 *
87 * 6. Redistributions of any form whatsoever must retain the following
88 * acknowledgment:
89 * "This product includes software developed by the OpenSSL Project
90 * for use in the OpenSSL Toolkit (http://www.openssl.org/)"
91 *
92 * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
93 * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
94 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
95 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
96 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
97 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
98 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
99 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
100 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
101 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
102 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
103 * OF THE POSSIBILITY OF SUCH DAMAGE.
104 * ====================================================================
105 *
106 * This product includes cryptographic software written by Eric Young
107 * (eay@cryptsoft.com). This product includes software written by Tim
108 * Hudson (tjh@cryptsoft.com).
109 *
110 */
111
112#include <stdio.h>
113#include "cryptlib.h"
114#include "bn_lcl.h"
115
116#define TABLE_SIZE 32
117
118int BN_mod_exp2_mont(BIGNUM *rr, BIGNUM *a1, BIGNUM *p1, BIGNUM *a2,
119 BIGNUM *p2, BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont)
120 {
121 int i,j,bits,b,bits1,bits2,ret=0,wpos1,wpos2,window1,window2,wvalue1,wvalue2;
122 int r_is_one=1,ts1=0,ts2=0;
123 BIGNUM *d,*r;
124 BIGNUM *a_mod_m;
125 BIGNUM val1[TABLE_SIZE], val2[TABLE_SIZE];
126 BN_MONT_CTX *mont=NULL;
127
128 bn_check_top(a1);
129 bn_check_top(p1);
130 bn_check_top(a2);
131 bn_check_top(p2);
132 bn_check_top(m);
133
134 if (!(m->d[0] & 1))
135 {
136 BNerr(BN_F_BN_MOD_EXP2_MONT,BN_R_CALLED_WITH_EVEN_MODULUS);
137 return(0);
138 }
139 bits1=BN_num_bits(p1);
140 bits2=BN_num_bits(p2);
141 if ((bits1 == 0) && (bits2 == 0))
142 {
143 BN_one(rr);
144 return(1);
145 }
146 bits=(bits1 > bits2)?bits1:bits2;
147
148 BN_CTX_start(ctx);
149 d = BN_CTX_get(ctx);
150 r = BN_CTX_get(ctx);
151 if (d == NULL || r == NULL) goto err;
152
153 if (in_mont != NULL)
154 mont=in_mont;
155 else
156 {
157 if ((mont=BN_MONT_CTX_new()) == NULL) goto err;
158 if (!BN_MONT_CTX_set(mont,m,ctx)) goto err;
159 }
160
161 window1 = BN_window_bits_for_exponent_size(bits1);
162 window2 = BN_window_bits_for_exponent_size(bits2);
163
164 /*
165 * Build table for a1: val1[i] := a1^(2*i + 1) mod m for i = 0 .. 2^(window1-1)
166 */
167 BN_init(&val1[0]);
168 ts1=1;
169 if (BN_ucmp(a1,m) >= 0)
170 {
171 if (!BN_mod(&(val1[0]),a1,m,ctx))
172 goto err;
173 a_mod_m = &(val1[0]);
174 }
175 else
176 a_mod_m = a1;
177 if (!BN_to_montgomery(&(val1[0]),a_mod_m,mont,ctx)) goto err;
178 if (window1 > 1)
179 {
180 if (!BN_mod_mul_montgomery(d,&(val1[0]),&(val1[0]),mont,ctx)) goto err;
181
182 j=1<<(window1-1);
183 for (i=1; i<j; i++)
184 {
185 BN_init(&(val1[i]));
186 if (!BN_mod_mul_montgomery(&(val1[i]),&(val1[i-1]),d,mont,ctx))
187 goto err;
188 }
189 ts1=i;
190 }
191
192
193 /*
194 * Build table for a2: val2[i] := a2^(2*i + 1) mod m for i = 0 .. 2^(window2-1)
195 */
196 BN_init(&val2[0]);
197 ts2=1;
198 if (BN_ucmp(a2,m) >= 0)
199 {
200 if (!BN_mod(&(val2[0]),a2,m,ctx))
201 goto err;
202 a_mod_m = &(val2[0]);
203 }
204 else
205 a_mod_m = a2;
206 if (!BN_to_montgomery(&(val2[0]),a_mod_m,mont,ctx)) goto err;
207 if (window2 > 1)
208 {
209 if (!BN_mod_mul_montgomery(d,&(val2[0]),&(val2[0]),mont,ctx)) goto err;
210
211 j=1<<(window2-1);
212 for (i=1; i<j; i++)
213 {
214 BN_init(&(val2[i]));
215 if (!BN_mod_mul_montgomery(&(val2[i]),&(val2[i-1]),d,mont,ctx))
216 goto err;
217 }
218 ts2=i;
219 }
220
221
222 /* Now compute the power product, using independent windows. */
223 r_is_one=1;
224 wvalue1=0; /* The 'value' of the first window */
225 wvalue2=0; /* The 'value' of the second window */
226 wpos1=0; /* If wvalue1 > 0, the bottom bit of the first window */
227 wpos2=0; /* If wvalue2 > 0, the bottom bit of the second window */
228
229 if (!BN_to_montgomery(r,BN_value_one(),mont,ctx)) goto err;
230 for (b=bits-1; b>=0; b--)
231 {
232 if (!r_is_one)
233 {
234 if (!BN_mod_mul_montgomery(r,r,r,mont,ctx))
235 goto err;
236 }
237
238 if (!wvalue1)
239 if (BN_is_bit_set(p1, b))
240 {
241 /* consider bits b-window1+1 .. b for this window */
242 i = b-window1+1;
243 while (!BN_is_bit_set(p1, i)) /* works for i<0 */
244 i++;
245 wpos1 = i;
246 wvalue1 = 1;
247 for (i = b-1; i >= wpos1; i--)
248 {
249 wvalue1 <<= 1;
250 if (BN_is_bit_set(p1, i))
251 wvalue1++;
252 }
253 }
254
255 if (!wvalue2)
256 if (BN_is_bit_set(p2, b))
257 {
258 /* consider bits b-window2+1 .. b for this window */
259 i = b-window2+1;
260 while (!BN_is_bit_set(p2, i))
261 i++;
262 wpos2 = i;
263 wvalue2 = 1;
264 for (i = b-1; i >= wpos2; i--)
265 {
266 wvalue2 <<= 1;
267 if (BN_is_bit_set(p2, i))
268 wvalue2++;
269 }
270 }
271
272 if (wvalue1 && b == wpos1)
273 {
274 /* wvalue1 is odd and < 2^window1 */
275 if (!BN_mod_mul_montgomery(r,r,&(val1[wvalue1>>1]),mont,ctx))
276 goto err;
277 wvalue1 = 0;
278 r_is_one = 0;
279 }
280
281 if (wvalue2 && b == wpos2)
282 {
283 /* wvalue2 is odd and < 2^window2 */
284 if (!BN_mod_mul_montgomery(r,r,&(val2[wvalue2>>1]),mont,ctx))
285 goto err;
286 wvalue2 = 0;
287 r_is_one = 0;
288 }
289 }
290 BN_from_montgomery(rr,r,mont,ctx);
291 ret=1;
292err:
293 if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont);
294 BN_CTX_end(ctx);
295 for (i=0; i<ts1; i++)
296 BN_clear_free(&(val1[i]));
297 for (i=0; i<ts2; i++)
298 BN_clear_free(&(val2[i]));
299 return(ret);
300 }
diff --git a/src/lib/libcrypto/bn/bn_gcd.c b/src/lib/libcrypto/bn/bn_gcd.c
deleted file mode 100644
index 398207196b..0000000000
--- a/src/lib/libcrypto/bn/bn_gcd.c
+++ /dev/null
@@ -1,210 +0,0 @@
1/* crypto/bn/bn_gcd.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63static BIGNUM *euclid(BIGNUM *a, BIGNUM *b);
64
65int BN_gcd(BIGNUM *r, BIGNUM *in_a, BIGNUM *in_b, BN_CTX *ctx)
66 {
67 BIGNUM *a,*b,*t;
68 int ret=0;
69
70 bn_check_top(in_a);
71 bn_check_top(in_b);
72
73 BN_CTX_start(ctx);
74 a = BN_CTX_get(ctx);
75 b = BN_CTX_get(ctx);
76 if (a == NULL || b == NULL) goto err;
77
78 if (BN_copy(a,in_a) == NULL) goto err;
79 if (BN_copy(b,in_b) == NULL) goto err;
80
81 if (BN_cmp(a,b) < 0) { t=a; a=b; b=t; }
82 t=euclid(a,b);
83 if (t == NULL) goto err;
84
85 if (BN_copy(r,t) == NULL) goto err;
86 ret=1;
87err:
88 BN_CTX_end(ctx);
89 return(ret);
90 }
91
92static BIGNUM *euclid(BIGNUM *a, BIGNUM *b)
93 {
94 BIGNUM *t;
95 int shifts=0;
96
97 bn_check_top(a);
98 bn_check_top(b);
99
100 for (;;)
101 {
102 if (BN_is_zero(b))
103 break;
104
105 if (BN_is_odd(a))
106 {
107 if (BN_is_odd(b))
108 {
109 if (!BN_sub(a,a,b)) goto err;
110 if (!BN_rshift1(a,a)) goto err;
111 if (BN_cmp(a,b) < 0)
112 { t=a; a=b; b=t; }
113 }
114 else /* a odd - b even */
115 {
116 if (!BN_rshift1(b,b)) goto err;
117 if (BN_cmp(a,b) < 0)
118 { t=a; a=b; b=t; }
119 }
120 }
121 else /* a is even */
122 {
123 if (BN_is_odd(b))
124 {
125 if (!BN_rshift1(a,a)) goto err;
126 if (BN_cmp(a,b) < 0)
127 { t=a; a=b; b=t; }
128 }
129 else /* a even - b even */
130 {
131 if (!BN_rshift1(a,a)) goto err;
132 if (!BN_rshift1(b,b)) goto err;
133 shifts++;
134 }
135 }
136 }
137 if (shifts)
138 {
139 if (!BN_lshift(a,a,shifts)) goto err;
140 }
141 return(a);
142err:
143 return(NULL);
144 }
145
146/* solves ax == 1 (mod n) */
147BIGNUM *BN_mod_inverse(BIGNUM *in, BIGNUM *a, const BIGNUM *n, BN_CTX *ctx)
148 {
149 BIGNUM *A,*B,*X,*Y,*M,*D,*R=NULL;
150 BIGNUM *T,*ret=NULL;
151 int sign;
152
153 bn_check_top(a);
154 bn_check_top(n);
155
156 BN_CTX_start(ctx);
157 A = BN_CTX_get(ctx);
158 B = BN_CTX_get(ctx);
159 X = BN_CTX_get(ctx);
160 D = BN_CTX_get(ctx);
161 M = BN_CTX_get(ctx);
162 Y = BN_CTX_get(ctx);
163 if (Y == NULL) goto err;
164
165 if (in == NULL)
166 R=BN_new();
167 else
168 R=in;
169 if (R == NULL) goto err;
170
171 BN_zero(X);
172 BN_one(Y);
173 if (BN_copy(A,a) == NULL) goto err;
174 if (BN_copy(B,n) == NULL) goto err;
175 sign=1;
176
177 while (!BN_is_zero(B))
178 {
179 if (!BN_div(D,M,A,B,ctx)) goto err;
180 T=A;
181 A=B;
182 B=M;
183 /* T has a struct, M does not */
184
185 if (!BN_mul(T,D,X,ctx)) goto err;
186 if (!BN_add(T,T,Y)) goto err;
187 M=Y;
188 Y=X;
189 X=T;
190 sign= -sign;
191 }
192 if (sign < 0)
193 {
194 if (!BN_sub(Y,n,Y)) goto err;
195 }
196
197 if (BN_is_one(A))
198 { if (!BN_mod(R,Y,n,ctx)) goto err; }
199 else
200 {
201 BNerr(BN_F_BN_MOD_INVERSE,BN_R_NO_INVERSE);
202 goto err;
203 }
204 ret=R;
205err:
206 if ((ret == NULL) && (in == NULL)) BN_free(R);
207 BN_CTX_end(ctx);
208 return(ret);
209 }
210
diff --git a/src/lib/libcrypto/bn/bn_lcl.h b/src/lib/libcrypto/bn/bn_lcl.h
deleted file mode 100644
index 9c959921b4..0000000000
--- a/src/lib/libcrypto/bn/bn_lcl.h
+++ /dev/null
@@ -1,419 +0,0 @@
1/* crypto/bn/bn_lcl.h */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58/* ====================================================================
59 * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved.
60 *
61 * Redistribution and use in source and binary forms, with or without
62 * modification, are permitted provided that the following conditions
63 * are met:
64 *
65 * 1. Redistributions of source code must retain the above copyright
66 * notice, this list of conditions and the following disclaimer.
67 *
68 * 2. Redistributions in binary form must reproduce the above copyright
69 * notice, this list of conditions and the following disclaimer in
70 * the documentation and/or other materials provided with the
71 * distribution.
72 *
73 * 3. All advertising materials mentioning features or use of this
74 * software must display the following acknowledgment:
75 * "This product includes software developed by the OpenSSL Project
76 * for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
77 *
78 * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
79 * endorse or promote products derived from this software without
80 * prior written permission. For written permission, please contact
81 * openssl-core@openssl.org.
82 *
83 * 5. Products derived from this software may not be called "OpenSSL"
84 * nor may "OpenSSL" appear in their names without prior written
85 * permission of the OpenSSL Project.
86 *
87 * 6. Redistributions of any form whatsoever must retain the following
88 * acknowledgment:
89 * "This product includes software developed by the OpenSSL Project
90 * for use in the OpenSSL Toolkit (http://www.openssl.org/)"
91 *
92 * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
93 * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
94 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
95 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
96 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
97 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
98 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
99 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
100 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
101 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
102 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
103 * OF THE POSSIBILITY OF SUCH DAMAGE.
104 * ====================================================================
105 *
106 * This product includes cryptographic software written by Eric Young
107 * (eay@cryptsoft.com). This product includes software written by Tim
108 * Hudson (tjh@cryptsoft.com).
109 *
110 */
111
112#ifndef HEADER_BN_LCL_H
113#define HEADER_BN_LCL_H
114
115#include <openssl/bn.h>
116
117#ifdef __cplusplus
118extern "C" {
119#endif
120
121
122/*
123 * BN_window_bits_for_exponent_size -- macro for sliding window mod_exp functions
124 *
125 *
126 * For window size 'w' (w >= 2) and a random 'b' bits exponent,
127 * the number of multiplications is a constant plus on average
128 *
129 * 2^(w-1) + (b-w)/(w+1);
130 *
131 * here 2^(w-1) is for precomputing the table (we actually need
132 * entries only for windows that have the lowest bit set), and
133 * (b-w)/(w+1) is an approximation for the expected number of
134 * w-bit windows, not counting the first one.
135 *
136 * Thus we should use
137 *
138 * w >= 6 if b > 671
139 * w = 5 if 671 > b > 239
140 * w = 4 if 239 > b > 79
141 * w = 3 if 79 > b > 23
142 * w <= 2 if 23 > b
143 *
144 * (with draws in between). Very small exponents are often selected
145 * with low Hamming weight, so we use w = 1 for b <= 23.
146 */
147#if 1
148#define BN_window_bits_for_exponent_size(b) \
149 ((b) > 671 ? 6 : \
150 (b) > 239 ? 5 : \
151 (b) > 79 ? 4 : \
152 (b) > 23 ? 3 : 1)
153#else
154/* Old SSLeay/OpenSSL table.
155 * Maximum window size was 5, so this table differs for b==1024;
156 * but it coincides for other interesting values (b==160, b==512).
157 */
158#define BN_window_bits_for_exponent_size(b) \
159 ((b) > 255 ? 5 : \
160 (b) > 127 ? 4 : \
161 (b) > 17 ? 3 : 1)
162#endif
163
164
165
166/* Pentium pro 16,16,16,32,64 */
167/* Alpha 16,16,16,16.64 */
168#define BN_MULL_SIZE_NORMAL (16) /* 32 */
169#define BN_MUL_RECURSIVE_SIZE_NORMAL (16) /* 32 less than */
170#define BN_SQR_RECURSIVE_SIZE_NORMAL (16) /* 32 */
171#define BN_MUL_LOW_RECURSIVE_SIZE_NORMAL (32) /* 32 */
172#define BN_MONT_CTX_SET_SIZE_WORD (64) /* 32 */
173
174#if !defined(NO_ASM) && !defined(NO_INLINE_ASM) && !defined(PEDANTIC)
175/*
176 * BN_UMULT_HIGH section.
177 *
178 * No, I'm not trying to overwhelm you when stating that the
179 * product of N-bit numbers is 2*N bits wide:-) No, I don't expect
180 * you to be impressed when I say that if the compiler doesn't
181 * support 2*N integer type, then you have to replace every N*N
182 * multiplication with 4 (N/2)*(N/2) accompanied by some shifts
183 * and additions which unavoidably results in severe performance
184 * penalties. Of course provided that the hardware is capable of
185 * producing 2*N result... That's when you normally start
186 * considering assembler implementation. However! It should be
187 * pointed out that some CPUs (most notably Alpha, PowerPC and
188 * upcoming IA-64 family:-) provide *separate* instruction
189 * calculating the upper half of the product placing the result
190 * into a general purpose register. Now *if* the compiler supports
191 * inline assembler, then it's not impossible to implement the
192 * "bignum" routines (and have the compiler optimize 'em)
193 * exhibiting "native" performance in C. That's what BN_UMULT_HIGH
194 * macro is about:-)
195 *
196 * <appro@fy.chalmers.se>
197 */
198# if defined(__alpha) && (defined(SIXTY_FOUR_BIT_LONG) || defined(SIXTY_FOUR_BIT))
199# if defined(__DECC)
200# include <c_asm.h>
201# define BN_UMULT_HIGH(a,b) (BN_ULONG)asm("umulh %a0,%a1,%v0",(a),(b))
202# elif defined(__GNUC__)
203# define BN_UMULT_HIGH(a,b) ({ \
204 register BN_ULONG ret; \
205 asm ("umulh %1,%2,%0" \
206 : "=r"(ret) \
207 : "r"(a), "r"(b)); \
208 ret; })
209# endif /* compiler */
210# elif defined(_ARCH_PPC) && defined(__64BIT__) && defined(SIXTY_FOUR_BIT_LONG)
211# if defined(__GNUC__)
212# define BN_UMULT_HIGH(a,b) ({ \
213 register BN_ULONG ret; \
214 asm ("mulhdu %0,%1,%2" \
215 : "=r"(ret) \
216 : "r"(a), "r"(b)); \
217 ret; })
218# endif /* compiler */
219# endif /* cpu */
220#endif /* NO_ASM */
221
222/*************************************************************
223 * Using the long long type
224 */
225#define Lw(t) (((BN_ULONG)(t))&BN_MASK2)
226#define Hw(t) (((BN_ULONG)((t)>>BN_BITS2))&BN_MASK2)
227
228/* This is used for internal error checking and is not normally used */
229#ifdef BN_DEBUG
230# include <assert.h>
231# define bn_check_top(a) assert ((a)->top >= 0 && (a)->top <= (a)->dmax);
232#else
233# define bn_check_top(a)
234#endif
235
236/* This macro is to add extra stuff for development checking */
237#ifdef BN_DEBUG
238#define bn_set_max(r) ((r)->max=(r)->top,BN_set_flags((r),BN_FLG_STATIC_DATA))
239#else
240#define bn_set_max(r)
241#endif
242
243/* These macros are used to 'take' a section of a bignum for read only use */
244#define bn_set_low(r,a,n) \
245 { \
246 (r)->top=((a)->top > (n))?(n):(a)->top; \
247 (r)->d=(a)->d; \
248 (r)->neg=(a)->neg; \
249 (r)->flags|=BN_FLG_STATIC_DATA; \
250 bn_set_max(r); \
251 }
252
253#define bn_set_high(r,a,n) \
254 { \
255 if ((a)->top > (n)) \
256 { \
257 (r)->top=(a)->top-n; \
258 (r)->d= &((a)->d[n]); \
259 } \
260 else \
261 (r)->top=0; \
262 (r)->neg=(a)->neg; \
263 (r)->flags|=BN_FLG_STATIC_DATA; \
264 bn_set_max(r); \
265 }
266
267#ifdef BN_LLONG
268#define mul_add(r,a,w,c) { \
269 BN_ULLONG t; \
270 t=(BN_ULLONG)w * (a) + (r) + (c); \
271 (r)= Lw(t); \
272 (c)= Hw(t); \
273 }
274
275#define mul(r,a,w,c) { \
276 BN_ULLONG t; \
277 t=(BN_ULLONG)w * (a) + (c); \
278 (r)= Lw(t); \
279 (c)= Hw(t); \
280 }
281
282#define sqr(r0,r1,a) { \
283 BN_ULLONG t; \
284 t=(BN_ULLONG)(a)*(a); \
285 (r0)=Lw(t); \
286 (r1)=Hw(t); \
287 }
288
289#elif defined(BN_UMULT_HIGH)
290#define mul_add(r,a,w,c) { \
291 BN_ULONG high,low,ret,tmp=(a); \
292 ret = (r); \
293 high= BN_UMULT_HIGH(w,tmp); \
294 ret += (c); \
295 low = (w) * tmp; \
296 (c) = (ret<(c))?1:0; \
297 (c) += high; \
298 ret += low; \
299 (c) += (ret<low)?1:0; \
300 (r) = ret; \
301 }
302
303#define mul(r,a,w,c) { \
304 BN_ULONG high,low,ret,ta=(a); \
305 low = (w) * ta; \
306 high= BN_UMULT_HIGH(w,ta); \
307 ret = low + (c); \
308 (c) = high; \
309 (c) += (ret<low)?1:0; \
310 (r) = ret; \
311 }
312
313#define sqr(r0,r1,a) { \
314 BN_ULONG tmp=(a); \
315 (r0) = tmp * tmp; \
316 (r1) = BN_UMULT_HIGH(tmp,tmp); \
317 }
318
319#else
320/*************************************************************
321 * No long long type
322 */
323
324#define LBITS(a) ((a)&BN_MASK2l)
325#define HBITS(a) (((a)>>BN_BITS4)&BN_MASK2l)
326#define L2HBITS(a) ((BN_ULONG)((a)&BN_MASK2l)<<BN_BITS4)
327
328#define LLBITS(a) ((a)&BN_MASKl)
329#define LHBITS(a) (((a)>>BN_BITS2)&BN_MASKl)
330#define LL2HBITS(a) ((BN_ULLONG)((a)&BN_MASKl)<<BN_BITS2)
331
332#define mul64(l,h,bl,bh) \
333 { \
334 BN_ULONG m,m1,lt,ht; \
335 \
336 lt=l; \
337 ht=h; \
338 m =(bh)*(lt); \
339 lt=(bl)*(lt); \
340 m1=(bl)*(ht); \
341 ht =(bh)*(ht); \
342 m=(m+m1)&BN_MASK2; if (m < m1) ht+=L2HBITS(1L); \
343 ht+=HBITS(m); \
344 m1=L2HBITS(m); \
345 lt=(lt+m1)&BN_MASK2; if (lt < m1) ht++; \
346 (l)=lt; \
347 (h)=ht; \
348 }
349
350#define sqr64(lo,ho,in) \
351 { \
352 BN_ULONG l,h,m; \
353 \
354 h=(in); \
355 l=LBITS(h); \
356 h=HBITS(h); \
357 m =(l)*(h); \
358 l*=l; \
359 h*=h; \
360 h+=(m&BN_MASK2h1)>>(BN_BITS4-1); \
361 m =(m&BN_MASK2l)<<(BN_BITS4+1); \
362 l=(l+m)&BN_MASK2; if (l < m) h++; \
363 (lo)=l; \
364 (ho)=h; \
365 }
366
367#define mul_add(r,a,bl,bh,c) { \
368 BN_ULONG l,h; \
369 \
370 h= (a); \
371 l=LBITS(h); \
372 h=HBITS(h); \
373 mul64(l,h,(bl),(bh)); \
374 \
375 /* non-multiply part */ \
376 l=(l+(c))&BN_MASK2; if (l < (c)) h++; \
377 (c)=(r); \
378 l=(l+(c))&BN_MASK2; if (l < (c)) h++; \
379 (c)=h&BN_MASK2; \
380 (r)=l; \
381 }
382
383#define mul(r,a,bl,bh,c) { \
384 BN_ULONG l,h; \
385 \
386 h= (a); \
387 l=LBITS(h); \
388 h=HBITS(h); \
389 mul64(l,h,(bl),(bh)); \
390 \
391 /* non-multiply part */ \
392 l+=(c); if ((l&BN_MASK2) < (c)) h++; \
393 (c)=h&BN_MASK2; \
394 (r)=l&BN_MASK2; \
395 }
396#endif /* !BN_LLONG */
397
398void bn_mul_normal(BN_ULONG *r,BN_ULONG *a,int na,BN_ULONG *b,int nb);
399void bn_mul_comba8(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b);
400void bn_mul_comba4(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b);
401void bn_sqr_normal(BN_ULONG *r, BN_ULONG *a, int n, BN_ULONG *tmp);
402void bn_sqr_comba8(BN_ULONG *r,BN_ULONG *a);
403void bn_sqr_comba4(BN_ULONG *r,BN_ULONG *a);
404int bn_cmp_words(BN_ULONG *a,BN_ULONG *b,int n);
405void bn_mul_recursive(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,int n2,BN_ULONG *t);
406void bn_mul_part_recursive(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,
407 int tn, int n,BN_ULONG *t);
408void bn_sqr_recursive(BN_ULONG *r,BN_ULONG *a, int n2, BN_ULONG *t);
409void bn_mul_low_normal(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b, int n);
410void bn_mul_low_recursive(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,int n2,
411 BN_ULONG *t);
412void bn_mul_high(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,BN_ULONG *l,int n2,
413 BN_ULONG *t);
414
415#ifdef __cplusplus
416}
417#endif
418
419#endif
diff --git a/src/lib/libcrypto/bn/bn_lib.c b/src/lib/libcrypto/bn/bn_lib.c
deleted file mode 100644
index 7767d65170..0000000000
--- a/src/lib/libcrypto/bn/bn_lib.c
+++ /dev/null
@@ -1,762 +0,0 @@
1/* crypto/bn/bn_lib.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#ifndef BN_DEBUG
60# undef NDEBUG /* avoid conflicting definitions */
61# define NDEBUG
62#endif
63
64#include <assert.h>
65#include <limits.h>
66#include <stdio.h>
67#include "cryptlib.h"
68#include "bn_lcl.h"
69
70const char *BN_version="Big Number" OPENSSL_VERSION_PTEXT;
71
72/* For a 32 bit machine
73 * 2 - 4 == 128
74 * 3 - 8 == 256
75 * 4 - 16 == 512
76 * 5 - 32 == 1024
77 * 6 - 64 == 2048
78 * 7 - 128 == 4096
79 * 8 - 256 == 8192
80 */
81static int bn_limit_bits=0;
82static int bn_limit_num=8; /* (1<<bn_limit_bits) */
83static int bn_limit_bits_low=0;
84static int bn_limit_num_low=8; /* (1<<bn_limit_bits_low) */
85static int bn_limit_bits_high=0;
86static int bn_limit_num_high=8; /* (1<<bn_limit_bits_high) */
87static int bn_limit_bits_mont=0;
88static int bn_limit_num_mont=8; /* (1<<bn_limit_bits_mont) */
89
90void BN_set_params(int mult, int high, int low, int mont)
91 {
92 if (mult >= 0)
93 {
94 if (mult > (sizeof(int)*8)-1)
95 mult=sizeof(int)*8-1;
96 bn_limit_bits=mult;
97 bn_limit_num=1<<mult;
98 }
99 if (high >= 0)
100 {
101 if (high > (sizeof(int)*8)-1)
102 high=sizeof(int)*8-1;
103 bn_limit_bits_high=high;
104 bn_limit_num_high=1<<high;
105 }
106 if (low >= 0)
107 {
108 if (low > (sizeof(int)*8)-1)
109 low=sizeof(int)*8-1;
110 bn_limit_bits_low=low;
111 bn_limit_num_low=1<<low;
112 }
113 if (mont >= 0)
114 {
115 if (mont > (sizeof(int)*8)-1)
116 mont=sizeof(int)*8-1;
117 bn_limit_bits_mont=mont;
118 bn_limit_num_mont=1<<mont;
119 }
120 }
121
122int BN_get_params(int which)
123 {
124 if (which == 0) return(bn_limit_bits);
125 else if (which == 1) return(bn_limit_bits_high);
126 else if (which == 2) return(bn_limit_bits_low);
127 else if (which == 3) return(bn_limit_bits_mont);
128 else return(0);
129 }
130
131BIGNUM *BN_value_one(void)
132 {
133 static BN_ULONG data_one=1L;
134 static BIGNUM const_one={&data_one,1,1,0};
135
136 return(&const_one);
137 }
138
139char *BN_options(void)
140 {
141 static int init=0;
142 static char data[16];
143
144 if (!init)
145 {
146 init++;
147#ifdef BN_LLONG
148 sprintf(data,"bn(%d,%d)",(int)sizeof(BN_ULLONG)*8,
149 (int)sizeof(BN_ULONG)*8);
150#else
151 sprintf(data,"bn(%d,%d)",(int)sizeof(BN_ULONG)*8,
152 (int)sizeof(BN_ULONG)*8);
153#endif
154 }
155 return(data);
156 }
157
158int BN_num_bits_word(BN_ULONG l)
159 {
160 static const char bits[256]={
161 0,1,2,2,3,3,3,3,4,4,4,4,4,4,4,4,
162 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,
163 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
164 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
165 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
166 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
167 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
168 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
169 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
170 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
171 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
172 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
173 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
174 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
175 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
176 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
177 };
178
179#if defined(SIXTY_FOUR_BIT_LONG)
180 if (l & 0xffffffff00000000L)
181 {
182 if (l & 0xffff000000000000L)
183 {
184 if (l & 0xff00000000000000L)
185 {
186 return(bits[(int)(l>>56)]+56);
187 }
188 else return(bits[(int)(l>>48)]+48);
189 }
190 else
191 {
192 if (l & 0x0000ff0000000000L)
193 {
194 return(bits[(int)(l>>40)]+40);
195 }
196 else return(bits[(int)(l>>32)]+32);
197 }
198 }
199 else
200#else
201#ifdef SIXTY_FOUR_BIT
202 if (l & 0xffffffff00000000LL)
203 {
204 if (l & 0xffff000000000000LL)
205 {
206 if (l & 0xff00000000000000LL)
207 {
208 return(bits[(int)(l>>56)]+56);
209 }
210 else return(bits[(int)(l>>48)]+48);
211 }
212 else
213 {
214 if (l & 0x0000ff0000000000LL)
215 {
216 return(bits[(int)(l>>40)]+40);
217 }
218 else return(bits[(int)(l>>32)]+32);
219 }
220 }
221 else
222#endif
223#endif
224 {
225#if defined(THIRTY_TWO_BIT) || defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG)
226 if (l & 0xffff0000L)
227 {
228 if (l & 0xff000000L)
229 return(bits[(int)(l>>24L)]+24);
230 else return(bits[(int)(l>>16L)]+16);
231 }
232 else
233#endif
234 {
235#if defined(SIXTEEN_BIT) || defined(THIRTY_TWO_BIT) || defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG)
236 if (l & 0xff00L)
237 return(bits[(int)(l>>8)]+8);
238 else
239#endif
240 return(bits[(int)(l )] );
241 }
242 }
243 }
244
245int BN_num_bits(const BIGNUM *a)
246 {
247 BN_ULONG l;
248 int i;
249
250 bn_check_top(a);
251
252 if (a->top == 0) return(0);
253 l=a->d[a->top-1];
254 assert(l != 0);
255 i=(a->top-1)*BN_BITS2;
256 return(i+BN_num_bits_word(l));
257 }
258
259void BN_clear_free(BIGNUM *a)
260 {
261 int i;
262
263 if (a == NULL) return;
264 if (a->d != NULL)
265 {
266 memset(a->d,0,a->dmax*sizeof(a->d[0]));
267 if (!(BN_get_flags(a,BN_FLG_STATIC_DATA)))
268 OPENSSL_free(a->d);
269 }
270 i=BN_get_flags(a,BN_FLG_MALLOCED);
271 memset(a,0,sizeof(BIGNUM));
272 if (i)
273 OPENSSL_free(a);
274 }
275
276void BN_free(BIGNUM *a)
277 {
278 if (a == NULL) return;
279 if ((a->d != NULL) && !(BN_get_flags(a,BN_FLG_STATIC_DATA)))
280 OPENSSL_free(a->d);
281 a->flags|=BN_FLG_FREE; /* REMOVE? */
282 if (a->flags & BN_FLG_MALLOCED)
283 OPENSSL_free(a);
284 }
285
286void BN_init(BIGNUM *a)
287 {
288 memset(a,0,sizeof(BIGNUM));
289 }
290
291BIGNUM *BN_new(void)
292 {
293 BIGNUM *ret;
294
295 if ((ret=(BIGNUM *)OPENSSL_malloc(sizeof(BIGNUM))) == NULL)
296 {
297 BNerr(BN_F_BN_NEW,ERR_R_MALLOC_FAILURE);
298 return(NULL);
299 }
300 ret->flags=BN_FLG_MALLOCED;
301 ret->top=0;
302 ret->neg=0;
303 ret->dmax=0;
304 ret->d=NULL;
305 return(ret);
306 }
307
308/* This is an internal function that should not be used in applications.
309 * It ensures that 'b' has enough room for a 'words' word number number.
310 * It is mostly used by the various BIGNUM routines. If there is an error,
311 * NULL is returned. If not, 'b' is returned. */
312
313BIGNUM *bn_expand2(BIGNUM *b, int words)
314 {
315 BN_ULONG *A,*a;
316 const BN_ULONG *B;
317 int i;
318
319 bn_check_top(b);
320
321 if (words > b->dmax)
322 {
323 if (words > (INT_MAX/(4*BN_BITS2)))
324 {
325 BNerr(BN_F_BN_EXPAND2,BN_R_BIGNUM_TOO_LONG);
326 return NULL;
327 }
328
329 bn_check_top(b);
330 if (BN_get_flags(b,BN_FLG_STATIC_DATA))
331 {
332 BNerr(BN_F_BN_EXPAND2,BN_R_EXPAND_ON_STATIC_BIGNUM_DATA);
333 return(NULL);
334 }
335 a=A=(BN_ULONG *)OPENSSL_malloc(sizeof(BN_ULONG)*(words+1));
336 if (A == NULL)
337 {
338 BNerr(BN_F_BN_EXPAND2,ERR_R_MALLOC_FAILURE);
339 return(NULL);
340 }
341#if 1
342 B=b->d;
343 /* Check if the previous number needs to be copied */
344 if (B != NULL)
345 {
346#if 0
347 /* This lot is an unrolled loop to copy b->top
348 * BN_ULONGs from B to A
349 */
350/*
351 * I have nothing against unrolling but it's usually done for
352 * several reasons, namely:
353 * - minimize percentage of decision making code, i.e. branches;
354 * - avoid cache trashing;
355 * - make it possible to schedule loads earlier;
356 * Now let's examine the code below. The cornerstone of C is
357 * "programmer is always right" and that's what we love it for:-)
358 * For this very reason C compilers have to be paranoid when it
359 * comes to data aliasing and assume the worst. Yeah, but what
360 * does it mean in real life? This means that loop body below will
361 * be compiled to sequence of loads immediately followed by stores
362 * as compiler assumes the worst, something in A==B+1 style. As a
363 * result CPU pipeline is going to starve for incoming data. Secondly
364 * if A and B happen to share same cache line such code is going to
365 * cause severe cache trashing. Both factors have severe impact on
366 * performance of modern CPUs and this is the reason why this
367 * particular piece of code is #ifdefed away and replaced by more
368 * "friendly" version found in #else section below. This comment
369 * also applies to BN_copy function.
370 *
371 * <appro@fy.chalmers.se>
372 */
373 for (i=b->top&(~7); i>0; i-=8)
374 {
375 A[0]=B[0]; A[1]=B[1]; A[2]=B[2]; A[3]=B[3];
376 A[4]=B[4]; A[5]=B[5]; A[6]=B[6]; A[7]=B[7];
377 A+=8;
378 B+=8;
379 }
380 switch (b->top&7)
381 {
382 case 7:
383 A[6]=B[6];
384 case 6:
385 A[5]=B[5];
386 case 5:
387 A[4]=B[4];
388 case 4:
389 A[3]=B[3];
390 case 3:
391 A[2]=B[2];
392 case 2:
393 A[1]=B[1];
394 case 1:
395 A[0]=B[0];
396 case 0:
397 /* I need the 'case 0' entry for utrix cc.
398 * If the optimizer is turned on, it does the
399 * switch table by doing
400 * a=top&7
401 * a--;
402 * goto jump_table[a];
403 * If top is 0, this makes us jump to 0xffffffc
404 * which is rather bad :-(.
405 * eric 23-Apr-1998
406 */
407 ;
408 }
409#else
410 for (i=b->top>>2; i>0; i--,A+=4,B+=4)
411 {
412 /*
413 * The fact that the loop is unrolled
414 * 4-wise is a tribute to Intel. It's
415 * the one that doesn't have enough
416 * registers to accomodate more data.
417 * I'd unroll it 8-wise otherwise:-)
418 *
419 * <appro@fy.chalmers.se>
420 */
421 BN_ULONG a0,a1,a2,a3;
422 a0=B[0]; a1=B[1]; a2=B[2]; a3=B[3];
423 A[0]=a0; A[1]=a1; A[2]=a2; A[3]=a3;
424 }
425 switch (b->top&3)
426 {
427 case 3: A[2]=B[2];
428 case 2: A[1]=B[1];
429 case 1: A[0]=B[0];
430 case 0: ; /* ultrix cc workaround, see above */
431 }
432#endif
433 OPENSSL_free(b->d);
434 }
435
436 b->d=a;
437 b->dmax=words;
438
439 /* Now need to zero any data between b->top and b->max */
440
441 A= &(b->d[b->top]);
442 for (i=(b->dmax - b->top)>>3; i>0; i--,A+=8)
443 {
444 A[0]=0; A[1]=0; A[2]=0; A[3]=0;
445 A[4]=0; A[5]=0; A[6]=0; A[7]=0;
446 }
447 for (i=(b->dmax - b->top)&7; i>0; i--,A++)
448 A[0]=0;
449#else
450 memset(A,0,sizeof(BN_ULONG)*(words+1));
451 memcpy(A,b->d,sizeof(b->d[0])*b->top);
452 b->d=a;
453 b->max=words;
454#endif
455
456/* memset(&(p[b->max]),0,((words+1)-b->max)*sizeof(BN_ULONG)); */
457/* { int i; for (i=b->max; i<words+1; i++) p[i]=i;} */
458
459 }
460 return(b);
461 }
462
463BIGNUM *BN_dup(const BIGNUM *a)
464 {
465 BIGNUM *r;
466
467 if (a == NULL) return NULL;
468
469 bn_check_top(a);
470
471 r=BN_new();
472 if (r == NULL) return(NULL);
473 return((BIGNUM *)BN_copy(r,a));
474 }
475
476BIGNUM *BN_copy(BIGNUM *a, const BIGNUM *b)
477 {
478 int i;
479 BN_ULONG *A;
480 const BN_ULONG *B;
481
482 bn_check_top(b);
483
484 if (a == b) return(a);
485 if (bn_wexpand(a,b->top) == NULL) return(NULL);
486
487#if 1
488 A=a->d;
489 B=b->d;
490 for (i=b->top>>2; i>0; i--,A+=4,B+=4)
491 {
492 BN_ULONG a0,a1,a2,a3;
493 a0=B[0]; a1=B[1]; a2=B[2]; a3=B[3];
494 A[0]=a0; A[1]=a1; A[2]=a2; A[3]=a3;
495 }
496 switch (b->top&3)
497 {
498 case 3: A[2]=B[2];
499 case 2: A[1]=B[1];
500 case 1: A[0]=B[0];
501 case 0: ; /* ultrix cc workaround, see comments in bn_expand2 */
502 }
503#else
504 memcpy(a->d,b->d,sizeof(b->d[0])*b->top);
505#endif
506
507/* memset(&(a->d[b->top]),0,sizeof(a->d[0])*(a->max-b->top));*/
508 a->top=b->top;
509 if ((a->top == 0) && (a->d != NULL))
510 a->d[0]=0;
511 a->neg=b->neg;
512 return(a);
513 }
514
515void BN_clear(BIGNUM *a)
516 {
517 if (a->d != NULL)
518 memset(a->d,0,a->dmax*sizeof(a->d[0]));
519 a->top=0;
520 a->neg=0;
521 }
522
523BN_ULONG BN_get_word(BIGNUM *a)
524 {
525 int i,n;
526 BN_ULONG ret=0;
527
528 n=BN_num_bytes(a);
529 if (n > sizeof(BN_ULONG))
530 return(BN_MASK2);
531 for (i=a->top-1; i>=0; i--)
532 {
533#ifndef SIXTY_FOUR_BIT /* the data item > unsigned long */
534 ret<<=BN_BITS4; /* stops the compiler complaining */
535 ret<<=BN_BITS4;
536#else
537 ret=0;
538#endif
539 ret|=a->d[i];
540 }
541 return(ret);
542 }
543
544int BN_set_word(BIGNUM *a, BN_ULONG w)
545 {
546 int i,n;
547 if (bn_expand(a,sizeof(BN_ULONG)*8) == NULL) return(0);
548
549 n=sizeof(BN_ULONG)/BN_BYTES;
550 a->neg=0;
551 a->top=0;
552 a->d[0]=(BN_ULONG)w&BN_MASK2;
553 if (a->d[0] != 0) a->top=1;
554 for (i=1; i<n; i++)
555 {
556 /* the following is done instead of
557 * w>>=BN_BITS2 so compilers don't complain
558 * on builds where sizeof(long) == BN_TYPES */
559#ifndef SIXTY_FOUR_BIT /* the data item > unsigned long */
560 w>>=BN_BITS4;
561 w>>=BN_BITS4;
562#else
563 w=0;
564#endif
565 a->d[i]=(BN_ULONG)w&BN_MASK2;
566 if (a->d[i] != 0) a->top=i+1;
567 }
568 return(1);
569 }
570
571/* ignore negative */
572BIGNUM *BN_bin2bn(const unsigned char *s, int len, BIGNUM *ret)
573 {
574 unsigned int i,m;
575 unsigned int n;
576 BN_ULONG l;
577
578 if (ret == NULL) ret=BN_new();
579 if (ret == NULL) return(NULL);
580 l=0;
581 n=len;
582 if (n == 0)
583 {
584 ret->top=0;
585 return(ret);
586 }
587 if (bn_expand(ret,(int)(n+2)*8) == NULL)
588 return(NULL);
589 i=((n-1)/BN_BYTES)+1;
590 m=((n-1)%(BN_BYTES));
591 ret->top=i;
592 while (n-- > 0)
593 {
594 l=(l<<8L)| *(s++);
595 if (m-- == 0)
596 {
597 ret->d[--i]=l;
598 l=0;
599 m=BN_BYTES-1;
600 }
601 }
602 /* need to call this due to clear byte at top if avoiding
603 * having the top bit set (-ve number) */
604 bn_fix_top(ret);
605 return(ret);
606 }
607
608/* ignore negative */
609int BN_bn2bin(const BIGNUM *a, unsigned char *to)
610 {
611 int n,i;
612 BN_ULONG l;
613
614 n=i=BN_num_bytes(a);
615 while (i-- > 0)
616 {
617 l=a->d[i/BN_BYTES];
618 *(to++)=(unsigned char)(l>>(8*(i%BN_BYTES)))&0xff;
619 }
620 return(n);
621 }
622
623int BN_ucmp(const BIGNUM *a, const BIGNUM *b)
624 {
625 int i;
626 BN_ULONG t1,t2,*ap,*bp;
627
628 bn_check_top(a);
629 bn_check_top(b);
630
631 i=a->top-b->top;
632 if (i != 0) return(i);
633 ap=a->d;
634 bp=b->d;
635 for (i=a->top-1; i>=0; i--)
636 {
637 t1= ap[i];
638 t2= bp[i];
639 if (t1 != t2)
640 return(t1 > t2?1:-1);
641 }
642 return(0);
643 }
644
645int BN_cmp(const BIGNUM *a, const BIGNUM *b)
646 {
647 int i;
648 int gt,lt;
649 BN_ULONG t1,t2;
650
651 if ((a == NULL) || (b == NULL))
652 {
653 if (a != NULL)
654 return(-1);
655 else if (b != NULL)
656 return(1);
657 else
658 return(0);
659 }
660
661 bn_check_top(a);
662 bn_check_top(b);
663
664 if (a->neg != b->neg)
665 {
666 if (a->neg)
667 return(-1);
668 else return(1);
669 }
670 if (a->neg == 0)
671 { gt=1; lt= -1; }
672 else { gt= -1; lt=1; }
673
674 if (a->top > b->top) return(gt);
675 if (a->top < b->top) return(lt);
676 for (i=a->top-1; i>=0; i--)
677 {
678 t1=a->d[i];
679 t2=b->d[i];
680 if (t1 > t2) return(gt);
681 if (t1 < t2) return(lt);
682 }
683 return(0);
684 }
685
686int BN_set_bit(BIGNUM *a, int n)
687 {
688 int i,j,k;
689
690 i=n/BN_BITS2;
691 j=n%BN_BITS2;
692 if (a->top <= i)
693 {
694 if (bn_wexpand(a,i+1) == NULL) return(0);
695 for(k=a->top; k<i+1; k++)
696 a->d[k]=0;
697 a->top=i+1;
698 }
699
700 a->d[i]|=(((BN_ULONG)1)<<j);
701 return(1);
702 }
703
704int BN_clear_bit(BIGNUM *a, int n)
705 {
706 int i,j;
707
708 i=n/BN_BITS2;
709 j=n%BN_BITS2;
710 if (a->top <= i) return(0);
711
712 a->d[i]&=(~(((BN_ULONG)1)<<j));
713 bn_fix_top(a);
714 return(1);
715 }
716
717int BN_is_bit_set(const BIGNUM *a, int n)
718 {
719 int i,j;
720
721 if (n < 0) return(0);
722 i=n/BN_BITS2;
723 j=n%BN_BITS2;
724 if (a->top <= i) return(0);
725 return((a->d[i]&(((BN_ULONG)1)<<j))?1:0);
726 }
727
728int BN_mask_bits(BIGNUM *a, int n)
729 {
730 int b,w;
731
732 w=n/BN_BITS2;
733 b=n%BN_BITS2;
734 if (w >= a->top) return(0);
735 if (b == 0)
736 a->top=w;
737 else
738 {
739 a->top=w+1;
740 a->d[w]&= ~(BN_MASK2<<b);
741 }
742 bn_fix_top(a);
743 return(1);
744 }
745
746int bn_cmp_words(BN_ULONG *a, BN_ULONG *b, int n)
747 {
748 int i;
749 BN_ULONG aa,bb;
750
751 aa=a[n-1];
752 bb=b[n-1];
753 if (aa != bb) return((aa > bb)?1:-1);
754 for (i=n-2; i>=0; i--)
755 {
756 aa=a[i];
757 bb=b[i];
758 if (aa != bb) return((aa > bb)?1:-1);
759 }
760 return(0);
761 }
762
diff --git a/src/lib/libcrypto/bn/bn_mont.c b/src/lib/libcrypto/bn/bn_mont.c
deleted file mode 100644
index 8cf1febacc..0000000000
--- a/src/lib/libcrypto/bn/bn_mont.c
+++ /dev/null
@@ -1,346 +0,0 @@
1/* crypto/bn/bn_mont.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59/*
60 * Details about Montgomery multiplication algorithms can be found at
61 * http://security.ece.orst.edu/publications.html, e.g.
62 * http://security.ece.orst.edu/koc/papers/j37acmon.pdf and
63 * sections 3.8 and 4.2 in http://security.ece.orst.edu/koc/papers/r01rsasw.pdf
64 */
65
66#include <stdio.h>
67#include "cryptlib.h"
68#include "bn_lcl.h"
69
70#define MONT_WORD /* use the faster word-based algorithm */
71
72int BN_mod_mul_montgomery(BIGNUM *r, BIGNUM *a, BIGNUM *b,
73 BN_MONT_CTX *mont, BN_CTX *ctx)
74 {
75 BIGNUM *tmp,*tmp2;
76 int ret=0;
77
78 BN_CTX_start(ctx);
79 tmp = BN_CTX_get(ctx);
80 tmp2 = BN_CTX_get(ctx);
81 if (tmp == NULL || tmp2 == NULL) goto err;
82
83 bn_check_top(tmp);
84 bn_check_top(tmp2);
85
86 if (a == b)
87 {
88 if (!BN_sqr(tmp,a,ctx)) goto err;
89 }
90 else
91 {
92 if (!BN_mul(tmp,a,b,ctx)) goto err;
93 }
94 /* reduce from aRR to aR */
95 if (!BN_from_montgomery(r,tmp,mont,ctx)) goto err;
96 ret=1;
97err:
98 BN_CTX_end(ctx);
99 return(ret);
100 }
101
102int BN_from_montgomery(BIGNUM *ret, BIGNUM *a, BN_MONT_CTX *mont,
103 BN_CTX *ctx)
104 {
105 int retn=0;
106
107#ifdef MONT_WORD
108 BIGNUM *n,*r;
109 BN_ULONG *ap,*np,*rp,n0,v,*nrp;
110 int al,nl,max,i,x,ri;
111
112 BN_CTX_start(ctx);
113 if ((r = BN_CTX_get(ctx)) == NULL) goto err;
114
115 if (!BN_copy(r,a)) goto err;
116 n= &(mont->N);
117
118 ap=a->d;
119 /* mont->ri is the size of mont->N in bits (rounded up
120 to the word size) */
121 al=ri=mont->ri/BN_BITS2;
122
123 nl=n->top;
124 if ((al == 0) || (nl == 0)) { r->top=0; return(1); }
125
126 max=(nl+al+1); /* allow for overflow (no?) XXX */
127 if (bn_wexpand(r,max) == NULL) goto err;
128 if (bn_wexpand(ret,max) == NULL) goto err;
129
130 r->neg=a->neg^n->neg;
131 np=n->d;
132 rp=r->d;
133 nrp= &(r->d[nl]);
134
135 /* clear the top words of T */
136#if 1
137 for (i=r->top; i<max; i++) /* memset? XXX */
138 r->d[i]=0;
139#else
140 memset(&(r->d[r->top]),0,(max-r->top)*sizeof(BN_ULONG));
141#endif
142
143 r->top=max;
144 n0=mont->n0;
145
146#ifdef BN_COUNT
147 printf("word BN_from_montgomery %d * %d\n",nl,nl);
148#endif
149 for (i=0; i<nl; i++)
150 {
151#ifdef __TANDEM
152 {
153 long long t1;
154 long long t2;
155 long long t3;
156 t1 = rp[0] * (n0 & 0177777);
157 t2 = 037777600000l;
158 t2 = n0 & t2;
159 t3 = rp[0] & 0177777;
160 t2 = (t3 * t2) & BN_MASK2;
161 t1 = t1 + t2;
162 v=bn_mul_add_words(rp,np,nl,(BN_ULONG) t1);
163 }
164#else
165 v=bn_mul_add_words(rp,np,nl,(rp[0]*n0)&BN_MASK2);
166#endif
167 nrp++;
168 rp++;
169 if (((nrp[-1]+=v)&BN_MASK2) >= v)
170 continue;
171 else
172 {
173 if (((++nrp[0])&BN_MASK2) != 0) continue;
174 if (((++nrp[1])&BN_MASK2) != 0) continue;
175 for (x=2; (((++nrp[x])&BN_MASK2) == 0); x++) ;
176 }
177 }
178 bn_fix_top(r);
179
180 /* mont->ri will be a multiple of the word size */
181#if 0
182 BN_rshift(ret,r,mont->ri);
183#else
184 ret->neg = r->neg;
185 x=ri;
186 rp=ret->d;
187 ap= &(r->d[x]);
188 if (r->top < x)
189 al=0;
190 else
191 al=r->top-x;
192 ret->top=al;
193 al-=4;
194 for (i=0; i<al; i+=4)
195 {
196 BN_ULONG t1,t2,t3,t4;
197
198 t1=ap[i+0];
199 t2=ap[i+1];
200 t3=ap[i+2];
201 t4=ap[i+3];
202 rp[i+0]=t1;
203 rp[i+1]=t2;
204 rp[i+2]=t3;
205 rp[i+3]=t4;
206 }
207 al+=4;
208 for (; i<al; i++)
209 rp[i]=ap[i];
210#endif
211#else /* !MONT_WORD */
212 BIGNUM *t1,*t2;
213
214 BN_CTX_start(ctx);
215 t1 = BN_CTX_get(ctx);
216 t2 = BN_CTX_get(ctx);
217 if (t1 == NULL || t2 == NULL) goto err;
218
219 if (!BN_copy(t1,a)) goto err;
220 BN_mask_bits(t1,mont->ri);
221
222 if (!BN_mul(t2,t1,&mont->Ni,ctx)) goto err;
223 BN_mask_bits(t2,mont->ri);
224
225 if (!BN_mul(t1,t2,&mont->N,ctx)) goto err;
226 if (!BN_add(t2,a,t1)) goto err;
227 BN_rshift(ret,t2,mont->ri);
228#endif /* MONT_WORD */
229
230 if (BN_ucmp(ret, &(mont->N)) >= 0)
231 {
232 BN_usub(ret,ret,&(mont->N));
233 }
234 retn=1;
235 err:
236 BN_CTX_end(ctx);
237 return(retn);
238 }
239
240BN_MONT_CTX *BN_MONT_CTX_new(void)
241 {
242 BN_MONT_CTX *ret;
243
244 if ((ret=(BN_MONT_CTX *)OPENSSL_malloc(sizeof(BN_MONT_CTX))) == NULL)
245 return(NULL);
246
247 BN_MONT_CTX_init(ret);
248 ret->flags=BN_FLG_MALLOCED;
249 return(ret);
250 }
251
252void BN_MONT_CTX_init(BN_MONT_CTX *ctx)
253 {
254 ctx->ri=0;
255 BN_init(&(ctx->RR));
256 BN_init(&(ctx->N));
257 BN_init(&(ctx->Ni));
258 ctx->flags=0;
259 }
260
261void BN_MONT_CTX_free(BN_MONT_CTX *mont)
262 {
263 if(mont == NULL)
264 return;
265
266 BN_free(&(mont->RR));
267 BN_free(&(mont->N));
268 BN_free(&(mont->Ni));
269 if (mont->flags & BN_FLG_MALLOCED)
270 OPENSSL_free(mont);
271 }
272
273int BN_MONT_CTX_set(BN_MONT_CTX *mont, const BIGNUM *mod, BN_CTX *ctx)
274 {
275 BIGNUM Ri,*R;
276
277 BN_init(&Ri);
278 R= &(mont->RR); /* grab RR as a temp */
279 BN_copy(&(mont->N),mod); /* Set N */
280
281#ifdef MONT_WORD
282 {
283 BIGNUM tmod;
284 BN_ULONG buf[2];
285
286 mont->ri=(BN_num_bits(mod)+(BN_BITS2-1))/BN_BITS2*BN_BITS2;
287 BN_zero(R);
288 BN_set_bit(R,BN_BITS2); /* R */
289
290 buf[0]=mod->d[0]; /* tmod = N mod word size */
291 buf[1]=0;
292 tmod.d=buf;
293 tmod.top=1;
294 tmod.dmax=2;
295 tmod.neg=mod->neg;
296 /* Ri = R^-1 mod N*/
297 if ((BN_mod_inverse(&Ri,R,&tmod,ctx)) == NULL)
298 goto err;
299 BN_lshift(&Ri,&Ri,BN_BITS2); /* R*Ri */
300 if (!BN_is_zero(&Ri))
301 BN_sub_word(&Ri,1);
302 else /* if N mod word size == 1 */
303 BN_set_word(&Ri,BN_MASK2); /* Ri-- (mod word size) */
304 BN_div(&Ri,NULL,&Ri,&tmod,ctx); /* Ni = (R*Ri-1)/N,
305 * keep only least significant word: */
306 mont->n0=Ri.d[0];
307 BN_free(&Ri);
308 }
309#else /* !MONT_WORD */
310 { /* bignum version */
311 mont->ri=BN_num_bits(mod);
312 BN_zero(R);
313 BN_set_bit(R,mont->ri); /* R = 2^ri */
314 /* Ri = R^-1 mod N*/
315 if ((BN_mod_inverse(&Ri,R,mod,ctx)) == NULL)
316 goto err;
317 BN_lshift(&Ri,&Ri,mont->ri); /* R*Ri */
318 BN_sub_word(&Ri,1);
319 /* Ni = (R*Ri-1) / N */
320 BN_div(&(mont->Ni),NULL,&Ri,mod,ctx);
321 BN_free(&Ri);
322 }
323#endif
324
325 /* setup RR for conversions */
326 BN_zero(&(mont->RR));
327 BN_set_bit(&(mont->RR),mont->ri*2);
328 BN_mod(&(mont->RR),&(mont->RR),&(mont->N),ctx);
329
330 return(1);
331err:
332 return(0);
333 }
334
335BN_MONT_CTX *BN_MONT_CTX_copy(BN_MONT_CTX *to, BN_MONT_CTX *from)
336 {
337 if (to == from) return(to);
338
339 BN_copy(&(to->RR),&(from->RR));
340 BN_copy(&(to->N),&(from->N));
341 BN_copy(&(to->Ni),&(from->Ni));
342 to->ri=from->ri;
343 to->n0=from->n0;
344 return(to);
345 }
346
diff --git a/src/lib/libcrypto/bn/bn_mpi.c b/src/lib/libcrypto/bn/bn_mpi.c
deleted file mode 100644
index 80e1dca6b7..0000000000
--- a/src/lib/libcrypto/bn/bn_mpi.c
+++ /dev/null
@@ -1,129 +0,0 @@
1/* crypto/bn/bn_mpi.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63int BN_bn2mpi(const BIGNUM *a, unsigned char *d)
64 {
65 int bits;
66 int num=0;
67 int ext=0;
68 long l;
69
70 bits=BN_num_bits(a);
71 num=(bits+7)/8;
72 if (bits > 0)
73 {
74 ext=((bits & 0x07) == 0);
75 }
76 if (d == NULL)
77 return(num+4+ext);
78
79 l=num+ext;
80 d[0]=(unsigned char)(l>>24)&0xff;
81 d[1]=(unsigned char)(l>>16)&0xff;
82 d[2]=(unsigned char)(l>> 8)&0xff;
83 d[3]=(unsigned char)(l )&0xff;
84 if (ext) d[4]=0;
85 num=BN_bn2bin(a,&(d[4+ext]));
86 if (a->neg)
87 d[4]|=0x80;
88 return(num+4+ext);
89 }
90
91BIGNUM *BN_mpi2bn(unsigned char *d, int n, BIGNUM *a)
92 {
93 long len;
94 int neg=0;
95
96 if (n < 4)
97 {
98 BNerr(BN_F_BN_MPI2BN,BN_R_INVALID_LENGTH);
99 return(NULL);
100 }
101 len=((long)d[0]<<24)|((long)d[1]<<16)|((int)d[2]<<8)|(int)d[3];
102 if ((len+4) != n)
103 {
104 BNerr(BN_F_BN_MPI2BN,BN_R_ENCODING_ERROR);
105 return(NULL);
106 }
107
108 if (a == NULL) a=BN_new();
109 if (a == NULL) return(NULL);
110
111 if (len == 0)
112 {
113 a->neg=0;
114 a->top=0;
115 return(a);
116 }
117 d+=4;
118 if ((*d) & 0x80)
119 neg=1;
120 if (BN_bin2bn(d,(int)len,a) == NULL)
121 return(NULL);
122 a->neg=neg;
123 if (neg)
124 {
125 BN_clear_bit(a,BN_num_bits(a)-1);
126 }
127 return(a);
128 }
129
diff --git a/src/lib/libcrypto/bn/bn_mul.c b/src/lib/libcrypto/bn/bn_mul.c
deleted file mode 100644
index 3e8d8b9567..0000000000
--- a/src/lib/libcrypto/bn/bn_mul.c
+++ /dev/null
@@ -1,794 +0,0 @@
1/* crypto/bn/bn_mul.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63#ifdef BN_RECURSION
64/* Karatsuba recursive multiplication algorithm
65 * (cf. Knuth, The Art of Computer Programming, Vol. 2) */
66
67/* r is 2*n2 words in size,
68 * a and b are both n2 words in size.
69 * n2 must be a power of 2.
70 * We multiply and return the result.
71 * t must be 2*n2 words in size
72 * We calculate
73 * a[0]*b[0]
74 * a[0]*b[0]+a[1]*b[1]+(a[0]-a[1])*(b[1]-b[0])
75 * a[1]*b[1]
76 */
77void bn_mul_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2,
78 BN_ULONG *t)
79 {
80 int n=n2/2,c1,c2;
81 unsigned int neg,zero;
82 BN_ULONG ln,lo,*p;
83
84# ifdef BN_COUNT
85 printf(" bn_mul_recursive %d * %d\n",n2,n2);
86# endif
87# ifdef BN_MUL_COMBA
88# if 0
89 if (n2 == 4)
90 {
91 bn_mul_comba4(r,a,b);
92 return;
93 }
94# endif
95 if (n2 == 8)
96 {
97 bn_mul_comba8(r,a,b);
98 return;
99 }
100# endif /* BN_MUL_COMBA */
101 if (n2 < BN_MUL_RECURSIVE_SIZE_NORMAL)
102 {
103 /* This should not happen */
104 bn_mul_normal(r,a,n2,b,n2);
105 return;
106 }
107 /* r=(a[0]-a[1])*(b[1]-b[0]) */
108 c1=bn_cmp_words(a,&(a[n]),n);
109 c2=bn_cmp_words(&(b[n]),b,n);
110 zero=neg=0;
111 switch (c1*3+c2)
112 {
113 case -4:
114 bn_sub_words(t, &(a[n]),a, n); /* - */
115 bn_sub_words(&(t[n]),b, &(b[n]),n); /* - */
116 break;
117 case -3:
118 zero=1;
119 break;
120 case -2:
121 bn_sub_words(t, &(a[n]),a, n); /* - */
122 bn_sub_words(&(t[n]),&(b[n]),b, n); /* + */
123 neg=1;
124 break;
125 case -1:
126 case 0:
127 case 1:
128 zero=1;
129 break;
130 case 2:
131 bn_sub_words(t, a, &(a[n]),n); /* + */
132 bn_sub_words(&(t[n]),b, &(b[n]),n); /* - */
133 neg=1;
134 break;
135 case 3:
136 zero=1;
137 break;
138 case 4:
139 bn_sub_words(t, a, &(a[n]),n);
140 bn_sub_words(&(t[n]),&(b[n]),b, n);
141 break;
142 }
143
144# ifdef BN_MUL_COMBA
145 if (n == 4)
146 {
147 if (!zero)
148 bn_mul_comba4(&(t[n2]),t,&(t[n]));
149 else
150 memset(&(t[n2]),0,8*sizeof(BN_ULONG));
151
152 bn_mul_comba4(r,a,b);
153 bn_mul_comba4(&(r[n2]),&(a[n]),&(b[n]));
154 }
155 else if (n == 8)
156 {
157 if (!zero)
158 bn_mul_comba8(&(t[n2]),t,&(t[n]));
159 else
160 memset(&(t[n2]),0,16*sizeof(BN_ULONG));
161
162 bn_mul_comba8(r,a,b);
163 bn_mul_comba8(&(r[n2]),&(a[n]),&(b[n]));
164 }
165 else
166# endif /* BN_MUL_COMBA */
167 {
168 p= &(t[n2*2]);
169 if (!zero)
170 bn_mul_recursive(&(t[n2]),t,&(t[n]),n,p);
171 else
172 memset(&(t[n2]),0,n2*sizeof(BN_ULONG));
173 bn_mul_recursive(r,a,b,n,p);
174 bn_mul_recursive(&(r[n2]),&(a[n]),&(b[n]),n,p);
175 }
176
177 /* t[32] holds (a[0]-a[1])*(b[1]-b[0]), c1 is the sign
178 * r[10] holds (a[0]*b[0])
179 * r[32] holds (b[1]*b[1])
180 */
181
182 c1=(int)(bn_add_words(t,r,&(r[n2]),n2));
183
184 if (neg) /* if t[32] is negative */
185 {
186 c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2));
187 }
188 else
189 {
190 /* Might have a carry */
191 c1+=(int)(bn_add_words(&(t[n2]),&(t[n2]),t,n2));
192 }
193
194 /* t[32] holds (a[0]-a[1])*(b[1]-b[0])+(a[0]*b[0])+(a[1]*b[1])
195 * r[10] holds (a[0]*b[0])
196 * r[32] holds (b[1]*b[1])
197 * c1 holds the carry bits
198 */
199 c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2));
200 if (c1)
201 {
202 p= &(r[n+n2]);
203 lo= *p;
204 ln=(lo+c1)&BN_MASK2;
205 *p=ln;
206
207 /* The overflow will stop before we over write
208 * words we should not overwrite */
209 if (ln < (BN_ULONG)c1)
210 {
211 do {
212 p++;
213 lo= *p;
214 ln=(lo+1)&BN_MASK2;
215 *p=ln;
216 } while (ln == 0);
217 }
218 }
219 }
220
221/* n+tn is the word length
222 * t needs to be n*4 is size, as does r */
223void bn_mul_part_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int tn,
224 int n, BN_ULONG *t)
225 {
226 int i,j,n2=n*2;
227 unsigned int c1,c2,neg,zero;
228 BN_ULONG ln,lo,*p;
229
230# ifdef BN_COUNT
231 printf(" bn_mul_part_recursive %d * %d\n",tn+n,tn+n);
232# endif
233 if (n < 8)
234 {
235 i=tn+n;
236 bn_mul_normal(r,a,i,b,i);
237 return;
238 }
239
240 /* r=(a[0]-a[1])*(b[1]-b[0]) */
241 c1=bn_cmp_words(a,&(a[n]),n);
242 c2=bn_cmp_words(&(b[n]),b,n);
243 zero=neg=0;
244 switch (c1*3+c2)
245 {
246 case -4:
247 bn_sub_words(t, &(a[n]),a, n); /* - */
248 bn_sub_words(&(t[n]),b, &(b[n]),n); /* - */
249 break;
250 case -3:
251 zero=1;
252 /* break; */
253 case -2:
254 bn_sub_words(t, &(a[n]),a, n); /* - */
255 bn_sub_words(&(t[n]),&(b[n]),b, n); /* + */
256 neg=1;
257 break;
258 case -1:
259 case 0:
260 case 1:
261 zero=1;
262 /* break; */
263 case 2:
264 bn_sub_words(t, a, &(a[n]),n); /* + */
265 bn_sub_words(&(t[n]),b, &(b[n]),n); /* - */
266 neg=1;
267 break;
268 case 3:
269 zero=1;
270 /* break; */
271 case 4:
272 bn_sub_words(t, a, &(a[n]),n);
273 bn_sub_words(&(t[n]),&(b[n]),b, n);
274 break;
275 }
276 /* The zero case isn't yet implemented here. The speedup
277 would probably be negligible. */
278# if 0
279 if (n == 4)
280 {
281 bn_mul_comba4(&(t[n2]),t,&(t[n]));
282 bn_mul_comba4(r,a,b);
283 bn_mul_normal(&(r[n2]),&(a[n]),tn,&(b[n]),tn);
284 memset(&(r[n2+tn*2]),0,sizeof(BN_ULONG)*(n2-tn*2));
285 }
286 else
287# endif
288 if (n == 8)
289 {
290 bn_mul_comba8(&(t[n2]),t,&(t[n]));
291 bn_mul_comba8(r,a,b);
292 bn_mul_normal(&(r[n2]),&(a[n]),tn,&(b[n]),tn);
293 memset(&(r[n2+tn*2]),0,sizeof(BN_ULONG)*(n2-tn*2));
294 }
295 else
296 {
297 p= &(t[n2*2]);
298 bn_mul_recursive(&(t[n2]),t,&(t[n]),n,p);
299 bn_mul_recursive(r,a,b,n,p);
300 i=n/2;
301 /* If there is only a bottom half to the number,
302 * just do it */
303 j=tn-i;
304 if (j == 0)
305 {
306 bn_mul_recursive(&(r[n2]),&(a[n]),&(b[n]),i,p);
307 memset(&(r[n2+i*2]),0,sizeof(BN_ULONG)*(n2-i*2));
308 }
309 else if (j > 0) /* eg, n == 16, i == 8 and tn == 11 */
310 {
311 bn_mul_part_recursive(&(r[n2]),&(a[n]),&(b[n]),
312 j,i,p);
313 memset(&(r[n2+tn*2]),0,
314 sizeof(BN_ULONG)*(n2-tn*2));
315 }
316 else /* (j < 0) eg, n == 16, i == 8 and tn == 5 */
317 {
318 memset(&(r[n2]),0,sizeof(BN_ULONG)*n2);
319 if (tn < BN_MUL_RECURSIVE_SIZE_NORMAL)
320 {
321 bn_mul_normal(&(r[n2]),&(a[n]),tn,&(b[n]),tn);
322 }
323 else
324 {
325 for (;;)
326 {
327 i/=2;
328 if (i < tn)
329 {
330 bn_mul_part_recursive(&(r[n2]),
331 &(a[n]),&(b[n]),
332 tn-i,i,p);
333 break;
334 }
335 else if (i == tn)
336 {
337 bn_mul_recursive(&(r[n2]),
338 &(a[n]),&(b[n]),
339 i,p);
340 break;
341 }
342 }
343 }
344 }
345 }
346
347 /* t[32] holds (a[0]-a[1])*(b[1]-b[0]), c1 is the sign
348 * r[10] holds (a[0]*b[0])
349 * r[32] holds (b[1]*b[1])
350 */
351
352 c1=(int)(bn_add_words(t,r,&(r[n2]),n2));
353
354 if (neg) /* if t[32] is negative */
355 {
356 c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2));
357 }
358 else
359 {
360 /* Might have a carry */
361 c1+=(int)(bn_add_words(&(t[n2]),&(t[n2]),t,n2));
362 }
363
364 /* t[32] holds (a[0]-a[1])*(b[1]-b[0])+(a[0]*b[0])+(a[1]*b[1])
365 * r[10] holds (a[0]*b[0])
366 * r[32] holds (b[1]*b[1])
367 * c1 holds the carry bits
368 */
369 c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2));
370 if (c1)
371 {
372 p= &(r[n+n2]);
373 lo= *p;
374 ln=(lo+c1)&BN_MASK2;
375 *p=ln;
376
377 /* The overflow will stop before we over write
378 * words we should not overwrite */
379 if (ln < c1)
380 {
381 do {
382 p++;
383 lo= *p;
384 ln=(lo+1)&BN_MASK2;
385 *p=ln;
386 } while (ln == 0);
387 }
388 }
389 }
390
391/* a and b must be the same size, which is n2.
392 * r needs to be n2 words and t needs to be n2*2
393 */
394void bn_mul_low_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2,
395 BN_ULONG *t)
396 {
397 int n=n2/2;
398
399# ifdef BN_COUNT
400 printf(" bn_mul_low_recursive %d * %d\n",n2,n2);
401# endif
402
403 bn_mul_recursive(r,a,b,n,&(t[0]));
404 if (n >= BN_MUL_LOW_RECURSIVE_SIZE_NORMAL)
405 {
406 bn_mul_low_recursive(&(t[0]),&(a[0]),&(b[n]),n,&(t[n2]));
407 bn_add_words(&(r[n]),&(r[n]),&(t[0]),n);
408 bn_mul_low_recursive(&(t[0]),&(a[n]),&(b[0]),n,&(t[n2]));
409 bn_add_words(&(r[n]),&(r[n]),&(t[0]),n);
410 }
411 else
412 {
413 bn_mul_low_normal(&(t[0]),&(a[0]),&(b[n]),n);
414 bn_mul_low_normal(&(t[n]),&(a[n]),&(b[0]),n);
415 bn_add_words(&(r[n]),&(r[n]),&(t[0]),n);
416 bn_add_words(&(r[n]),&(r[n]),&(t[n]),n);
417 }
418 }
419
420/* a and b must be the same size, which is n2.
421 * r needs to be n2 words and t needs to be n2*2
422 * l is the low words of the output.
423 * t needs to be n2*3
424 */
425void bn_mul_high(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, BN_ULONG *l, int n2,
426 BN_ULONG *t)
427 {
428 int i,n;
429 int c1,c2;
430 int neg,oneg,zero;
431 BN_ULONG ll,lc,*lp,*mp;
432
433# ifdef BN_COUNT
434 printf(" bn_mul_high %d * %d\n",n2,n2);
435# endif
436 n=n2/2;
437
438 /* Calculate (al-ah)*(bh-bl) */
439 neg=zero=0;
440 c1=bn_cmp_words(&(a[0]),&(a[n]),n);
441 c2=bn_cmp_words(&(b[n]),&(b[0]),n);
442 switch (c1*3+c2)
443 {
444 case -4:
445 bn_sub_words(&(r[0]),&(a[n]),&(a[0]),n);
446 bn_sub_words(&(r[n]),&(b[0]),&(b[n]),n);
447 break;
448 case -3:
449 zero=1;
450 break;
451 case -2:
452 bn_sub_words(&(r[0]),&(a[n]),&(a[0]),n);
453 bn_sub_words(&(r[n]),&(b[n]),&(b[0]),n);
454 neg=1;
455 break;
456 case -1:
457 case 0:
458 case 1:
459 zero=1;
460 break;
461 case 2:
462 bn_sub_words(&(r[0]),&(a[0]),&(a[n]),n);
463 bn_sub_words(&(r[n]),&(b[0]),&(b[n]),n);
464 neg=1;
465 break;
466 case 3:
467 zero=1;
468 break;
469 case 4:
470 bn_sub_words(&(r[0]),&(a[0]),&(a[n]),n);
471 bn_sub_words(&(r[n]),&(b[n]),&(b[0]),n);
472 break;
473 }
474
475 oneg=neg;
476 /* t[10] = (a[0]-a[1])*(b[1]-b[0]) */
477 /* r[10] = (a[1]*b[1]) */
478# ifdef BN_MUL_COMBA
479 if (n == 8)
480 {
481 bn_mul_comba8(&(t[0]),&(r[0]),&(r[n]));
482 bn_mul_comba8(r,&(a[n]),&(b[n]));
483 }
484 else
485# endif
486 {
487 bn_mul_recursive(&(t[0]),&(r[0]),&(r[n]),n,&(t[n2]));
488 bn_mul_recursive(r,&(a[n]),&(b[n]),n,&(t[n2]));
489 }
490
491 /* s0 == low(al*bl)
492 * s1 == low(ah*bh)+low((al-ah)*(bh-bl))+low(al*bl)+high(al*bl)
493 * We know s0 and s1 so the only unknown is high(al*bl)
494 * high(al*bl) == s1 - low(ah*bh+s0+(al-ah)*(bh-bl))
495 * high(al*bl) == s1 - (r[0]+l[0]+t[0])
496 */
497 if (l != NULL)
498 {
499 lp= &(t[n2+n]);
500 c1=(int)(bn_add_words(lp,&(r[0]),&(l[0]),n));
501 }
502 else
503 {
504 c1=0;
505 lp= &(r[0]);
506 }
507
508 if (neg)
509 neg=(int)(bn_sub_words(&(t[n2]),lp,&(t[0]),n));
510 else
511 {
512 bn_add_words(&(t[n2]),lp,&(t[0]),n);
513 neg=0;
514 }
515
516 if (l != NULL)
517 {
518 bn_sub_words(&(t[n2+n]),&(l[n]),&(t[n2]),n);
519 }
520 else
521 {
522 lp= &(t[n2+n]);
523 mp= &(t[n2]);
524 for (i=0; i<n; i++)
525 lp[i]=((~mp[i])+1)&BN_MASK2;
526 }
527
528 /* s[0] = low(al*bl)
529 * t[3] = high(al*bl)
530 * t[10] = (a[0]-a[1])*(b[1]-b[0]) neg is the sign
531 * r[10] = (a[1]*b[1])
532 */
533 /* R[10] = al*bl
534 * R[21] = al*bl + ah*bh + (a[0]-a[1])*(b[1]-b[0])
535 * R[32] = ah*bh
536 */
537 /* R[1]=t[3]+l[0]+r[0](+-)t[0] (have carry/borrow)
538 * R[2]=r[0]+t[3]+r[1](+-)t[1] (have carry/borrow)
539 * R[3]=r[1]+(carry/borrow)
540 */
541 if (l != NULL)
542 {
543 lp= &(t[n2]);
544 c1= (int)(bn_add_words(lp,&(t[n2+n]),&(l[0]),n));
545 }
546 else
547 {
548 lp= &(t[n2+n]);
549 c1=0;
550 }
551 c1+=(int)(bn_add_words(&(t[n2]),lp, &(r[0]),n));
552 if (oneg)
553 c1-=(int)(bn_sub_words(&(t[n2]),&(t[n2]),&(t[0]),n));
554 else
555 c1+=(int)(bn_add_words(&(t[n2]),&(t[n2]),&(t[0]),n));
556
557 c2 =(int)(bn_add_words(&(r[0]),&(r[0]),&(t[n2+n]),n));
558 c2+=(int)(bn_add_words(&(r[0]),&(r[0]),&(r[n]),n));
559 if (oneg)
560 c2-=(int)(bn_sub_words(&(r[0]),&(r[0]),&(t[n]),n));
561 else
562 c2+=(int)(bn_add_words(&(r[0]),&(r[0]),&(t[n]),n));
563
564 if (c1 != 0) /* Add starting at r[0], could be +ve or -ve */
565 {
566 i=0;
567 if (c1 > 0)
568 {
569 lc=c1;
570 do {
571 ll=(r[i]+lc)&BN_MASK2;
572 r[i++]=ll;
573 lc=(lc > ll);
574 } while (lc);
575 }
576 else
577 {
578 lc= -c1;
579 do {
580 ll=r[i];
581 r[i++]=(ll-lc)&BN_MASK2;
582 lc=(lc > ll);
583 } while (lc);
584 }
585 }
586 if (c2 != 0) /* Add starting at r[1] */
587 {
588 i=n;
589 if (c2 > 0)
590 {
591 lc=c2;
592 do {
593 ll=(r[i]+lc)&BN_MASK2;
594 r[i++]=ll;
595 lc=(lc > ll);
596 } while (lc);
597 }
598 else
599 {
600 lc= -c2;
601 do {
602 ll=r[i];
603 r[i++]=(ll-lc)&BN_MASK2;
604 lc=(lc > ll);
605 } while (lc);
606 }
607 }
608 }
609#endif /* BN_RECURSION */
610
611int BN_mul(BIGNUM *r, BIGNUM *a, BIGNUM *b, BN_CTX *ctx)
612 {
613 int top,al,bl;
614 BIGNUM *rr;
615 int ret = 0;
616#if defined(BN_MUL_COMBA) || defined(BN_RECURSION)
617 int i;
618#endif
619#ifdef BN_RECURSION
620 BIGNUM *t;
621 int j,k;
622#endif
623
624#ifdef BN_COUNT
625 printf("BN_mul %d * %d\n",a->top,b->top);
626#endif
627
628 bn_check_top(a);
629 bn_check_top(b);
630 bn_check_top(r);
631
632 al=a->top;
633 bl=b->top;
634
635 if ((al == 0) || (bl == 0))
636 {
637 BN_zero(r);
638 return(1);
639 }
640 top=al+bl;
641
642 BN_CTX_start(ctx);
643 if ((r == a) || (r == b))
644 {
645 if ((rr = BN_CTX_get(ctx)) == NULL) goto err;
646 }
647 else
648 rr = r;
649 rr->neg=a->neg^b->neg;
650
651#if defined(BN_MUL_COMBA) || defined(BN_RECURSION)
652 i = al-bl;
653#endif
654#ifdef BN_MUL_COMBA
655 if (i == 0)
656 {
657# if 0
658 if (al == 4)
659 {
660 if (bn_wexpand(rr,8) == NULL) goto err;
661 rr->top=8;
662 bn_mul_comba4(rr->d,a->d,b->d);
663 goto end;
664 }
665# endif
666 if (al == 8)
667 {
668 if (bn_wexpand(rr,16) == NULL) goto err;
669 rr->top=16;
670 bn_mul_comba8(rr->d,a->d,b->d);
671 goto end;
672 }
673 }
674#endif /* BN_MUL_COMBA */
675#ifdef BN_RECURSION
676 if ((al >= BN_MULL_SIZE_NORMAL) && (bl >= BN_MULL_SIZE_NORMAL))
677 {
678 if (i == 1 && !BN_get_flags(b,BN_FLG_STATIC_DATA))
679 {
680 bn_wexpand(b,al);
681 b->d[bl]=0;
682 bl++;
683 i--;
684 }
685 else if (i == -1 && !BN_get_flags(a,BN_FLG_STATIC_DATA))
686 {
687 bn_wexpand(a,bl);
688 a->d[al]=0;
689 al++;
690 i++;
691 }
692 if (i == 0)
693 {
694 /* symmetric and > 4 */
695 /* 16 or larger */
696 j=BN_num_bits_word((BN_ULONG)al);
697 j=1<<(j-1);
698 k=j+j;
699 t = BN_CTX_get(ctx);
700 if (al == j) /* exact multiple */
701 {
702 bn_wexpand(t,k*2);
703 bn_wexpand(rr,k*2);
704 bn_mul_recursive(rr->d,a->d,b->d,al,t->d);
705 }
706 else
707 {
708 bn_wexpand(a,k);
709 bn_wexpand(b,k);
710 bn_wexpand(t,k*4);
711 bn_wexpand(rr,k*4);
712 for (i=a->top; i<k; i++)
713 a->d[i]=0;
714 for (i=b->top; i<k; i++)
715 b->d[i]=0;
716 bn_mul_part_recursive(rr->d,a->d,b->d,al-j,j,t->d);
717 }
718 rr->top=top;
719 goto end;
720 }
721 }
722#endif /* BN_RECURSION */
723 if (bn_wexpand(rr,top) == NULL) goto err;
724 rr->top=top;
725 bn_mul_normal(rr->d,a->d,al,b->d,bl);
726
727#if defined(BN_MUL_COMBA) || defined(BN_RECURSION)
728end:
729#endif
730 bn_fix_top(rr);
731 if (r != rr) BN_copy(r,rr);
732 ret=1;
733err:
734 BN_CTX_end(ctx);
735 return(ret);
736 }
737
738void bn_mul_normal(BN_ULONG *r, BN_ULONG *a, int na, BN_ULONG *b, int nb)
739 {
740 BN_ULONG *rr;
741
742#ifdef BN_COUNT
743 printf(" bn_mul_normal %d * %d\n",na,nb);
744#endif
745
746 if (na < nb)
747 {
748 int itmp;
749 BN_ULONG *ltmp;
750
751 itmp=na; na=nb; nb=itmp;
752 ltmp=a; a=b; b=ltmp;
753
754 }
755 rr= &(r[na]);
756 rr[0]=bn_mul_words(r,a,na,b[0]);
757
758 for (;;)
759 {
760 if (--nb <= 0) return;
761 rr[1]=bn_mul_add_words(&(r[1]),a,na,b[1]);
762 if (--nb <= 0) return;
763 rr[2]=bn_mul_add_words(&(r[2]),a,na,b[2]);
764 if (--nb <= 0) return;
765 rr[3]=bn_mul_add_words(&(r[3]),a,na,b[3]);
766 if (--nb <= 0) return;
767 rr[4]=bn_mul_add_words(&(r[4]),a,na,b[4]);
768 rr+=4;
769 r+=4;
770 b+=4;
771 }
772 }
773
774void bn_mul_low_normal(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
775 {
776#ifdef BN_COUNT
777 printf(" bn_mul_low_normal %d * %d\n",n,n);
778#endif
779 bn_mul_words(r,a,n,b[0]);
780
781 for (;;)
782 {
783 if (--n <= 0) return;
784 bn_mul_add_words(&(r[1]),a,n,b[1]);
785 if (--n <= 0) return;
786 bn_mul_add_words(&(r[2]),a,n,b[2]);
787 if (--n <= 0) return;
788 bn_mul_add_words(&(r[3]),a,n,b[3]);
789 if (--n <= 0) return;
790 bn_mul_add_words(&(r[4]),a,n,b[4]);
791 r+=4;
792 b+=4;
793 }
794 }
diff --git a/src/lib/libcrypto/bn/bn_prime.c b/src/lib/libcrypto/bn/bn_prime.c
deleted file mode 100644
index a5f01b92eb..0000000000
--- a/src/lib/libcrypto/bn/bn_prime.c
+++ /dev/null
@@ -1,465 +0,0 @@
1/* crypto/bn/bn_prime.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58/* ====================================================================
59 * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved.
60 *
61 * Redistribution and use in source and binary forms, with or without
62 * modification, are permitted provided that the following conditions
63 * are met:
64 *
65 * 1. Redistributions of source code must retain the above copyright
66 * notice, this list of conditions and the following disclaimer.
67 *
68 * 2. Redistributions in binary form must reproduce the above copyright
69 * notice, this list of conditions and the following disclaimer in
70 * the documentation and/or other materials provided with the
71 * distribution.
72 *
73 * 3. All advertising materials mentioning features or use of this
74 * software must display the following acknowledgment:
75 * "This product includes software developed by the OpenSSL Project
76 * for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
77 *
78 * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
79 * endorse or promote products derived from this software without
80 * prior written permission. For written permission, please contact
81 * openssl-core@openssl.org.
82 *
83 * 5. Products derived from this software may not be called "OpenSSL"
84 * nor may "OpenSSL" appear in their names without prior written
85 * permission of the OpenSSL Project.
86 *
87 * 6. Redistributions of any form whatsoever must retain the following
88 * acknowledgment:
89 * "This product includes software developed by the OpenSSL Project
90 * for use in the OpenSSL Toolkit (http://www.openssl.org/)"
91 *
92 * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
93 * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
94 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
95 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
96 * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
97 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
98 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
99 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
100 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
101 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
102 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
103 * OF THE POSSIBILITY OF SUCH DAMAGE.
104 * ====================================================================
105 *
106 * This product includes cryptographic software written by Eric Young
107 * (eay@cryptsoft.com). This product includes software written by Tim
108 * Hudson (tjh@cryptsoft.com).
109 *
110 */
111
112#include <stdio.h>
113#include <time.h>
114#include "cryptlib.h"
115#include "bn_lcl.h"
116#include <openssl/rand.h>
117
118/* The quick sieve algorithm approach to weeding out primes is
119 * Philip Zimmermann's, as implemented in PGP. I have had a read of
120 * his comments and implemented my own version.
121 */
122#include "bn_prime.h"
123
124static int witness(BIGNUM *w, const BIGNUM *a, const BIGNUM *a1,
125 const BIGNUM *a1_odd, int k, BN_CTX *ctx, BN_MONT_CTX *mont);
126static int probable_prime(BIGNUM *rnd, int bits);
127static int probable_prime_dh(BIGNUM *rnd, int bits,
128 BIGNUM *add, BIGNUM *rem, BN_CTX *ctx);
129static int probable_prime_dh_safe(BIGNUM *rnd, int bits,
130 BIGNUM *add, BIGNUM *rem, BN_CTX *ctx);
131
132BIGNUM *BN_generate_prime(BIGNUM *ret, int bits, int safe, BIGNUM *add,
133 BIGNUM *rem, void (*callback)(int,int,void *), void *cb_arg)
134 {
135 BIGNUM *rnd=NULL;
136 BIGNUM t;
137 int found=0;
138 int i,j,c1=0;
139 BN_CTX *ctx;
140 int checks = BN_prime_checks_for_size(bits);
141
142 ctx=BN_CTX_new();
143 if (ctx == NULL) goto err;
144 if (ret == NULL)
145 {
146 if ((rnd=BN_new()) == NULL) goto err;
147 }
148 else
149 rnd=ret;
150 BN_init(&t);
151loop:
152 /* make a random number and set the top and bottom bits */
153 if (add == NULL)
154 {
155 if (!probable_prime(rnd,bits)) goto err;
156 }
157 else
158 {
159 if (safe)
160 {
161 if (!probable_prime_dh_safe(rnd,bits,add,rem,ctx))
162 goto err;
163 }
164 else
165 {
166 if (!probable_prime_dh(rnd,bits,add,rem,ctx))
167 goto err;
168 }
169 }
170 /* if (BN_mod_word(rnd,(BN_ULONG)3) == 1) goto loop; */
171 if (callback != NULL) callback(0,c1++,cb_arg);
172
173 if (!safe)
174 {
175 i=BN_is_prime_fasttest(rnd,checks,callback,ctx,cb_arg,0);
176 if (i == -1) goto err;
177 if (i == 0) goto loop;
178 }
179 else
180 {
181 /* for "safe prime" generation,
182 * check that (p-1)/2 is prime.
183 * Since a prime is odd, We just
184 * need to divide by 2 */
185 if (!BN_rshift1(&t,rnd)) goto err;
186
187 for (i=0; i<checks; i++)
188 {
189 j=BN_is_prime_fasttest(rnd,1,callback,ctx,cb_arg,0);
190 if (j == -1) goto err;
191 if (j == 0) goto loop;
192
193 j=BN_is_prime_fasttest(&t,1,callback,ctx,cb_arg,0);
194 if (j == -1) goto err;
195 if (j == 0) goto loop;
196
197 if (callback != NULL) callback(2,c1-1,cb_arg);
198 /* We have a safe prime test pass */
199 }
200 }
201 /* we have a prime :-) */
202 found = 1;
203err:
204 if (!found && (ret == NULL) && (rnd != NULL)) BN_free(rnd);
205 BN_free(&t);
206 if (ctx != NULL) BN_CTX_free(ctx);
207 return(found ? rnd : NULL);
208 }
209
210int BN_is_prime(const BIGNUM *a, int checks, void (*callback)(int,int,void *),
211 BN_CTX *ctx_passed, void *cb_arg)
212 {
213 return BN_is_prime_fasttest(a, checks, callback, ctx_passed, cb_arg, 0);
214 }
215
216int BN_is_prime_fasttest(const BIGNUM *a, int checks,
217 void (*callback)(int,int,void *),
218 BN_CTX *ctx_passed, void *cb_arg,
219 int do_trial_division)
220 {
221 int i, j, ret = -1;
222 int k;
223 BN_CTX *ctx = NULL;
224 BIGNUM *A1, *A1_odd, *check; /* taken from ctx */
225 BN_MONT_CTX *mont = NULL;
226 const BIGNUM *A = NULL;
227
228 if (checks == BN_prime_checks)
229 checks = BN_prime_checks_for_size(BN_num_bits(a));
230
231 /* first look for small factors */
232 if (!BN_is_odd(a))
233 return(0);
234 if (do_trial_division)
235 {
236 for (i = 1; i < NUMPRIMES; i++)
237 if (BN_mod_word(a, primes[i]) == 0)
238 return 0;
239 if (callback != NULL) callback(1, -1, cb_arg);
240 }
241
242 if (ctx_passed != NULL)
243 ctx = ctx_passed;
244 else
245 if ((ctx=BN_CTX_new()) == NULL)
246 goto err;
247 BN_CTX_start(ctx);
248
249 /* A := abs(a) */
250 if (a->neg)
251 {
252 BIGNUM *t;
253 if ((t = BN_CTX_get(ctx)) == NULL) goto err;
254 BN_copy(t, a);
255 t->neg = 0;
256 A = t;
257 }
258 else
259 A = a;
260 A1 = BN_CTX_get(ctx);
261 A1_odd = BN_CTX_get(ctx);
262 check = BN_CTX_get(ctx);
263 if (check == NULL) goto err;
264
265 /* compute A1 := A - 1 */
266 if (!BN_copy(A1, A))
267 goto err;
268 if (!BN_sub_word(A1, 1))
269 goto err;
270 if (BN_is_zero(A1))
271 {
272 ret = 0;
273 goto err;
274 }
275
276 /* write A1 as A1_odd * 2^k */
277 k = 1;
278 while (!BN_is_bit_set(A1, k))
279 k++;
280 if (!BN_rshift(A1_odd, A1, k))
281 goto err;
282
283 /* Montgomery setup for computations mod A */
284 mont = BN_MONT_CTX_new();
285 if (mont == NULL)
286 goto err;
287 if (!BN_MONT_CTX_set(mont, A, ctx))
288 goto err;
289
290 for (i = 0; i < checks; i++)
291 {
292 if (!BN_pseudo_rand(check, BN_num_bits(A1), 0, 0))
293 goto err;
294 if (BN_cmp(check, A1) >= 0)
295 if (!BN_sub(check, check, A1))
296 goto err;
297 if (!BN_add_word(check, 1))
298 goto err;
299 /* now 1 <= check < A */
300
301 j = witness(check, A, A1, A1_odd, k, ctx, mont);
302 if (j == -1) goto err;
303 if (j)
304 {
305 ret=0;
306 goto err;
307 }
308 if (callback != NULL) callback(1,i,cb_arg);
309 }
310 ret=1;
311err:
312 if (ctx != NULL)
313 {
314 BN_CTX_end(ctx);
315 if (ctx_passed == NULL)
316 BN_CTX_free(ctx);
317 }
318 if (mont != NULL)
319 BN_MONT_CTX_free(mont);
320
321 return(ret);
322 }
323
324static int witness(BIGNUM *w, const BIGNUM *a, const BIGNUM *a1,
325 const BIGNUM *a1_odd, int k, BN_CTX *ctx, BN_MONT_CTX *mont)
326 {
327 if (!BN_mod_exp_mont(w, w, a1_odd, a, ctx, mont)) /* w := w^a1_odd mod a */
328 return -1;
329 if (BN_is_one(w))
330 return 0; /* probably prime */
331 if (BN_cmp(w, a1) == 0)
332 return 0; /* w == -1 (mod a), 'a' is probably prime */
333 while (--k)
334 {
335 if (!BN_mod_mul(w, w, w, a, ctx)) /* w := w^2 mod a */
336 return -1;
337 if (BN_is_one(w))
338 return 1; /* 'a' is composite, otherwise a previous 'w' would
339 * have been == -1 (mod 'a') */
340 if (BN_cmp(w, a1) == 0)
341 return 0; /* w == -1 (mod a), 'a' is probably prime */
342 }
343 /* If we get here, 'w' is the (a-1)/2-th power of the original 'w',
344 * and it is neither -1 nor +1 -- so 'a' cannot be prime */
345 return 1;
346 }
347
348static int probable_prime(BIGNUM *rnd, int bits)
349 {
350 int i;
351 BN_ULONG mods[NUMPRIMES];
352 BN_ULONG delta,d;
353
354again:
355 if (!BN_rand(rnd,bits,1,1)) return(0);
356 /* we now have a random number 'rand' to test. */
357 for (i=1; i<NUMPRIMES; i++)
358 mods[i]=BN_mod_word(rnd,(BN_ULONG)primes[i]);
359 delta=0;
360 loop: for (i=1; i<NUMPRIMES; i++)
361 {
362 /* check that rnd is not a prime and also
363 * that gcd(rnd-1,primes) == 1 (except for 2) */
364 if (((mods[i]+delta)%primes[i]) <= 1)
365 {
366 d=delta;
367 delta+=2;
368 /* perhaps need to check for overflow of
369 * delta (but delta can be up to 2^32)
370 * 21-May-98 eay - added overflow check */
371 if (delta < d) goto again;
372 goto loop;
373 }
374 }
375 if (!BN_add_word(rnd,delta)) return(0);
376 return(1);
377 }
378
379static int probable_prime_dh(BIGNUM *rnd, int bits, BIGNUM *add, BIGNUM *rem,
380 BN_CTX *ctx)
381 {
382 int i,ret=0;
383 BIGNUM *t1;
384
385 BN_CTX_start(ctx);
386 if ((t1 = BN_CTX_get(ctx)) == NULL) goto err;
387
388 if (!BN_rand(rnd,bits,0,1)) goto err;
389
390 /* we need ((rnd-rem) % add) == 0 */
391
392 if (!BN_mod(t1,rnd,add,ctx)) goto err;
393 if (!BN_sub(rnd,rnd,t1)) goto err;
394 if (rem == NULL)
395 { if (!BN_add_word(rnd,1)) goto err; }
396 else
397 { if (!BN_add(rnd,rnd,rem)) goto err; }
398
399 /* we now have a random number 'rand' to test. */
400
401 loop: for (i=1; i<NUMPRIMES; i++)
402 {
403 /* check that rnd is a prime */
404 if (BN_mod_word(rnd,(BN_ULONG)primes[i]) <= 1)
405 {
406 if (!BN_add(rnd,rnd,add)) goto err;
407 goto loop;
408 }
409 }
410 ret=1;
411err:
412 BN_CTX_end(ctx);
413 return(ret);
414 }
415
416static int probable_prime_dh_safe(BIGNUM *p, int bits, BIGNUM *padd,
417 BIGNUM *rem, BN_CTX *ctx)
418 {
419 int i,ret=0;
420 BIGNUM *t1,*qadd,*q;
421
422 bits--;
423 BN_CTX_start(ctx);
424 t1 = BN_CTX_get(ctx);
425 q = BN_CTX_get(ctx);
426 qadd = BN_CTX_get(ctx);
427 if (qadd == NULL) goto err;
428
429 if (!BN_rshift1(qadd,padd)) goto err;
430
431 if (!BN_rand(q,bits,0,1)) goto err;
432
433 /* we need ((rnd-rem) % add) == 0 */
434 if (!BN_mod(t1,q,qadd,ctx)) goto err;
435 if (!BN_sub(q,q,t1)) goto err;
436 if (rem == NULL)
437 { if (!BN_add_word(q,1)) goto err; }
438 else
439 {
440 if (!BN_rshift1(t1,rem)) goto err;
441 if (!BN_add(q,q,t1)) goto err;
442 }
443
444 /* we now have a random number 'rand' to test. */
445 if (!BN_lshift1(p,q)) goto err;
446 if (!BN_add_word(p,1)) goto err;
447
448 loop: for (i=1; i<NUMPRIMES; i++)
449 {
450 /* check that p and q are prime */
451 /* check that for p and q
452 * gcd(p-1,primes) == 1 (except for 2) */
453 if ( (BN_mod_word(p,(BN_ULONG)primes[i]) == 0) ||
454 (BN_mod_word(q,(BN_ULONG)primes[i]) == 0))
455 {
456 if (!BN_add(p,p,padd)) goto err;
457 if (!BN_add(q,q,qadd)) goto err;
458 goto loop;
459 }
460 }
461 ret=1;
462err:
463 BN_CTX_end(ctx);
464 return(ret);
465 }
diff --git a/src/lib/libcrypto/bn/bn_prime.h b/src/lib/libcrypto/bn/bn_prime.h
deleted file mode 100644
index b7cf9a9bfe..0000000000
--- a/src/lib/libcrypto/bn/bn_prime.h
+++ /dev/null
@@ -1,325 +0,0 @@
1/* Auto generated by bn_prime.pl */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#ifndef EIGHT_BIT
60#define NUMPRIMES 2048
61#else
62#define NUMPRIMES 54
63#endif
64static const unsigned int primes[NUMPRIMES]=
65 {
66 2, 3, 5, 7, 11, 13, 17, 19,
67 23, 29, 31, 37, 41, 43, 47, 53,
68 59, 61, 67, 71, 73, 79, 83, 89,
69 97, 101, 103, 107, 109, 113, 127, 131,
70 137, 139, 149, 151, 157, 163, 167, 173,
71 179, 181, 191, 193, 197, 199, 211, 223,
72 227, 229, 233, 239, 241, 251,
73#ifndef EIGHT_BIT
74 257, 263,
75 269, 271, 277, 281, 283, 293, 307, 311,
76 313, 317, 331, 337, 347, 349, 353, 359,
77 367, 373, 379, 383, 389, 397, 401, 409,
78 419, 421, 431, 433, 439, 443, 449, 457,
79 461, 463, 467, 479, 487, 491, 499, 503,
80 509, 521, 523, 541, 547, 557, 563, 569,
81 571, 577, 587, 593, 599, 601, 607, 613,
82 617, 619, 631, 641, 643, 647, 653, 659,
83 661, 673, 677, 683, 691, 701, 709, 719,
84 727, 733, 739, 743, 751, 757, 761, 769,
85 773, 787, 797, 809, 811, 821, 823, 827,
86 829, 839, 853, 857, 859, 863, 877, 881,
87 883, 887, 907, 911, 919, 929, 937, 941,
88 947, 953, 967, 971, 977, 983, 991, 997,
89 1009,1013,1019,1021,1031,1033,1039,1049,
90 1051,1061,1063,1069,1087,1091,1093,1097,
91 1103,1109,1117,1123,1129,1151,1153,1163,
92 1171,1181,1187,1193,1201,1213,1217,1223,
93 1229,1231,1237,1249,1259,1277,1279,1283,
94 1289,1291,1297,1301,1303,1307,1319,1321,
95 1327,1361,1367,1373,1381,1399,1409,1423,
96 1427,1429,1433,1439,1447,1451,1453,1459,
97 1471,1481,1483,1487,1489,1493,1499,1511,
98 1523,1531,1543,1549,1553,1559,1567,1571,
99 1579,1583,1597,1601,1607,1609,1613,1619,
100 1621,1627,1637,1657,1663,1667,1669,1693,
101 1697,1699,1709,1721,1723,1733,1741,1747,
102 1753,1759,1777,1783,1787,1789,1801,1811,
103 1823,1831,1847,1861,1867,1871,1873,1877,
104 1879,1889,1901,1907,1913,1931,1933,1949,
105 1951,1973,1979,1987,1993,1997,1999,2003,
106 2011,2017,2027,2029,2039,2053,2063,2069,
107 2081,2083,2087,2089,2099,2111,2113,2129,
108 2131,2137,2141,2143,2153,2161,2179,2203,
109 2207,2213,2221,2237,2239,2243,2251,2267,
110 2269,2273,2281,2287,2293,2297,2309,2311,
111 2333,2339,2341,2347,2351,2357,2371,2377,
112 2381,2383,2389,2393,2399,2411,2417,2423,
113 2437,2441,2447,2459,2467,2473,2477,2503,
114 2521,2531,2539,2543,2549,2551,2557,2579,
115 2591,2593,2609,2617,2621,2633,2647,2657,
116 2659,2663,2671,2677,2683,2687,2689,2693,
117 2699,2707,2711,2713,2719,2729,2731,2741,
118 2749,2753,2767,2777,2789,2791,2797,2801,
119 2803,2819,2833,2837,2843,2851,2857,2861,
120 2879,2887,2897,2903,2909,2917,2927,2939,
121 2953,2957,2963,2969,2971,2999,3001,3011,
122 3019,3023,3037,3041,3049,3061,3067,3079,
123 3083,3089,3109,3119,3121,3137,3163,3167,
124 3169,3181,3187,3191,3203,3209,3217,3221,
125 3229,3251,3253,3257,3259,3271,3299,3301,
126 3307,3313,3319,3323,3329,3331,3343,3347,
127 3359,3361,3371,3373,3389,3391,3407,3413,
128 3433,3449,3457,3461,3463,3467,3469,3491,
129 3499,3511,3517,3527,3529,3533,3539,3541,
130 3547,3557,3559,3571,3581,3583,3593,3607,
131 3613,3617,3623,3631,3637,3643,3659,3671,
132 3673,3677,3691,3697,3701,3709,3719,3727,
133 3733,3739,3761,3767,3769,3779,3793,3797,
134 3803,3821,3823,3833,3847,3851,3853,3863,
135 3877,3881,3889,3907,3911,3917,3919,3923,
136 3929,3931,3943,3947,3967,3989,4001,4003,
137 4007,4013,4019,4021,4027,4049,4051,4057,
138 4073,4079,4091,4093,4099,4111,4127,4129,
139 4133,4139,4153,4157,4159,4177,4201,4211,
140 4217,4219,4229,4231,4241,4243,4253,4259,
141 4261,4271,4273,4283,4289,4297,4327,4337,
142 4339,4349,4357,4363,4373,4391,4397,4409,
143 4421,4423,4441,4447,4451,4457,4463,4481,
144 4483,4493,4507,4513,4517,4519,4523,4547,
145 4549,4561,4567,4583,4591,4597,4603,4621,
146 4637,4639,4643,4649,4651,4657,4663,4673,
147 4679,4691,4703,4721,4723,4729,4733,4751,
148 4759,4783,4787,4789,4793,4799,4801,4813,
149 4817,4831,4861,4871,4877,4889,4903,4909,
150 4919,4931,4933,4937,4943,4951,4957,4967,
151 4969,4973,4987,4993,4999,5003,5009,5011,
152 5021,5023,5039,5051,5059,5077,5081,5087,
153 5099,5101,5107,5113,5119,5147,5153,5167,
154 5171,5179,5189,5197,5209,5227,5231,5233,
155 5237,5261,5273,5279,5281,5297,5303,5309,
156 5323,5333,5347,5351,5381,5387,5393,5399,
157 5407,5413,5417,5419,5431,5437,5441,5443,
158 5449,5471,5477,5479,5483,5501,5503,5507,
159 5519,5521,5527,5531,5557,5563,5569,5573,
160 5581,5591,5623,5639,5641,5647,5651,5653,
161 5657,5659,5669,5683,5689,5693,5701,5711,
162 5717,5737,5741,5743,5749,5779,5783,5791,
163 5801,5807,5813,5821,5827,5839,5843,5849,
164 5851,5857,5861,5867,5869,5879,5881,5897,
165 5903,5923,5927,5939,5953,5981,5987,6007,
166 6011,6029,6037,6043,6047,6053,6067,6073,
167 6079,6089,6091,6101,6113,6121,6131,6133,
168 6143,6151,6163,6173,6197,6199,6203,6211,
169 6217,6221,6229,6247,6257,6263,6269,6271,
170 6277,6287,6299,6301,6311,6317,6323,6329,
171 6337,6343,6353,6359,6361,6367,6373,6379,
172 6389,6397,6421,6427,6449,6451,6469,6473,
173 6481,6491,6521,6529,6547,6551,6553,6563,
174 6569,6571,6577,6581,6599,6607,6619,6637,
175 6653,6659,6661,6673,6679,6689,6691,6701,
176 6703,6709,6719,6733,6737,6761,6763,6779,
177 6781,6791,6793,6803,6823,6827,6829,6833,
178 6841,6857,6863,6869,6871,6883,6899,6907,
179 6911,6917,6947,6949,6959,6961,6967,6971,
180 6977,6983,6991,6997,7001,7013,7019,7027,
181 7039,7043,7057,7069,7079,7103,7109,7121,
182 7127,7129,7151,7159,7177,7187,7193,7207,
183 7211,7213,7219,7229,7237,7243,7247,7253,
184 7283,7297,7307,7309,7321,7331,7333,7349,
185 7351,7369,7393,7411,7417,7433,7451,7457,
186 7459,7477,7481,7487,7489,7499,7507,7517,
187 7523,7529,7537,7541,7547,7549,7559,7561,
188 7573,7577,7583,7589,7591,7603,7607,7621,
189 7639,7643,7649,7669,7673,7681,7687,7691,
190 7699,7703,7717,7723,7727,7741,7753,7757,
191 7759,7789,7793,7817,7823,7829,7841,7853,
192 7867,7873,7877,7879,7883,7901,7907,7919,
193 7927,7933,7937,7949,7951,7963,7993,8009,
194 8011,8017,8039,8053,8059,8069,8081,8087,
195 8089,8093,8101,8111,8117,8123,8147,8161,
196 8167,8171,8179,8191,8209,8219,8221,8231,
197 8233,8237,8243,8263,8269,8273,8287,8291,
198 8293,8297,8311,8317,8329,8353,8363,8369,
199 8377,8387,8389,8419,8423,8429,8431,8443,
200 8447,8461,8467,8501,8513,8521,8527,8537,
201 8539,8543,8563,8573,8581,8597,8599,8609,
202 8623,8627,8629,8641,8647,8663,8669,8677,
203 8681,8689,8693,8699,8707,8713,8719,8731,
204 8737,8741,8747,8753,8761,8779,8783,8803,
205 8807,8819,8821,8831,8837,8839,8849,8861,
206 8863,8867,8887,8893,8923,8929,8933,8941,
207 8951,8963,8969,8971,8999,9001,9007,9011,
208 9013,9029,9041,9043,9049,9059,9067,9091,
209 9103,9109,9127,9133,9137,9151,9157,9161,
210 9173,9181,9187,9199,9203,9209,9221,9227,
211 9239,9241,9257,9277,9281,9283,9293,9311,
212 9319,9323,9337,9341,9343,9349,9371,9377,
213 9391,9397,9403,9413,9419,9421,9431,9433,
214 9437,9439,9461,9463,9467,9473,9479,9491,
215 9497,9511,9521,9533,9539,9547,9551,9587,
216 9601,9613,9619,9623,9629,9631,9643,9649,
217 9661,9677,9679,9689,9697,9719,9721,9733,
218 9739,9743,9749,9767,9769,9781,9787,9791,
219 9803,9811,9817,9829,9833,9839,9851,9857,
220 9859,9871,9883,9887,9901,9907,9923,9929,
221 9931,9941,9949,9967,9973,10007,10009,10037,
222 10039,10061,10067,10069,10079,10091,10093,10099,
223 10103,10111,10133,10139,10141,10151,10159,10163,
224 10169,10177,10181,10193,10211,10223,10243,10247,
225 10253,10259,10267,10271,10273,10289,10301,10303,
226 10313,10321,10331,10333,10337,10343,10357,10369,
227 10391,10399,10427,10429,10433,10453,10457,10459,
228 10463,10477,10487,10499,10501,10513,10529,10531,
229 10559,10567,10589,10597,10601,10607,10613,10627,
230 10631,10639,10651,10657,10663,10667,10687,10691,
231 10709,10711,10723,10729,10733,10739,10753,10771,
232 10781,10789,10799,10831,10837,10847,10853,10859,
233 10861,10867,10883,10889,10891,10903,10909,10937,
234 10939,10949,10957,10973,10979,10987,10993,11003,
235 11027,11047,11057,11059,11069,11071,11083,11087,
236 11093,11113,11117,11119,11131,11149,11159,11161,
237 11171,11173,11177,11197,11213,11239,11243,11251,
238 11257,11261,11273,11279,11287,11299,11311,11317,
239 11321,11329,11351,11353,11369,11383,11393,11399,
240 11411,11423,11437,11443,11447,11467,11471,11483,
241 11489,11491,11497,11503,11519,11527,11549,11551,
242 11579,11587,11593,11597,11617,11621,11633,11657,
243 11677,11681,11689,11699,11701,11717,11719,11731,
244 11743,11777,11779,11783,11789,11801,11807,11813,
245 11821,11827,11831,11833,11839,11863,11867,11887,
246 11897,11903,11909,11923,11927,11933,11939,11941,
247 11953,11959,11969,11971,11981,11987,12007,12011,
248 12037,12041,12043,12049,12071,12073,12097,12101,
249 12107,12109,12113,12119,12143,12149,12157,12161,
250 12163,12197,12203,12211,12227,12239,12241,12251,
251 12253,12263,12269,12277,12281,12289,12301,12323,
252 12329,12343,12347,12373,12377,12379,12391,12401,
253 12409,12413,12421,12433,12437,12451,12457,12473,
254 12479,12487,12491,12497,12503,12511,12517,12527,
255 12539,12541,12547,12553,12569,12577,12583,12589,
256 12601,12611,12613,12619,12637,12641,12647,12653,
257 12659,12671,12689,12697,12703,12713,12721,12739,
258 12743,12757,12763,12781,12791,12799,12809,12821,
259 12823,12829,12841,12853,12889,12893,12899,12907,
260 12911,12917,12919,12923,12941,12953,12959,12967,
261 12973,12979,12983,13001,13003,13007,13009,13033,
262 13037,13043,13049,13063,13093,13099,13103,13109,
263 13121,13127,13147,13151,13159,13163,13171,13177,
264 13183,13187,13217,13219,13229,13241,13249,13259,
265 13267,13291,13297,13309,13313,13327,13331,13337,
266 13339,13367,13381,13397,13399,13411,13417,13421,
267 13441,13451,13457,13463,13469,13477,13487,13499,
268 13513,13523,13537,13553,13567,13577,13591,13597,
269 13613,13619,13627,13633,13649,13669,13679,13681,
270 13687,13691,13693,13697,13709,13711,13721,13723,
271 13729,13751,13757,13759,13763,13781,13789,13799,
272 13807,13829,13831,13841,13859,13873,13877,13879,
273 13883,13901,13903,13907,13913,13921,13931,13933,
274 13963,13967,13997,13999,14009,14011,14029,14033,
275 14051,14057,14071,14081,14083,14087,14107,14143,
276 14149,14153,14159,14173,14177,14197,14207,14221,
277 14243,14249,14251,14281,14293,14303,14321,14323,
278 14327,14341,14347,14369,14387,14389,14401,14407,
279 14411,14419,14423,14431,14437,14447,14449,14461,
280 14479,14489,14503,14519,14533,14537,14543,14549,
281 14551,14557,14561,14563,14591,14593,14621,14627,
282 14629,14633,14639,14653,14657,14669,14683,14699,
283 14713,14717,14723,14731,14737,14741,14747,14753,
284 14759,14767,14771,14779,14783,14797,14813,14821,
285 14827,14831,14843,14851,14867,14869,14879,14887,
286 14891,14897,14923,14929,14939,14947,14951,14957,
287 14969,14983,15013,15017,15031,15053,15061,15073,
288 15077,15083,15091,15101,15107,15121,15131,15137,
289 15139,15149,15161,15173,15187,15193,15199,15217,
290 15227,15233,15241,15259,15263,15269,15271,15277,
291 15287,15289,15299,15307,15313,15319,15329,15331,
292 15349,15359,15361,15373,15377,15383,15391,15401,
293 15413,15427,15439,15443,15451,15461,15467,15473,
294 15493,15497,15511,15527,15541,15551,15559,15569,
295 15581,15583,15601,15607,15619,15629,15641,15643,
296 15647,15649,15661,15667,15671,15679,15683,15727,
297 15731,15733,15737,15739,15749,15761,15767,15773,
298 15787,15791,15797,15803,15809,15817,15823,15859,
299 15877,15881,15887,15889,15901,15907,15913,15919,
300 15923,15937,15959,15971,15973,15991,16001,16007,
301 16033,16057,16061,16063,16067,16069,16073,16087,
302 16091,16097,16103,16111,16127,16139,16141,16183,
303 16187,16189,16193,16217,16223,16229,16231,16249,
304 16253,16267,16273,16301,16319,16333,16339,16349,
305 16361,16363,16369,16381,16411,16417,16421,16427,
306 16433,16447,16451,16453,16477,16481,16487,16493,
307 16519,16529,16547,16553,16561,16567,16573,16603,
308 16607,16619,16631,16633,16649,16651,16657,16661,
309 16673,16691,16693,16699,16703,16729,16741,16747,
310 16759,16763,16787,16811,16823,16829,16831,16843,
311 16871,16879,16883,16889,16901,16903,16921,16927,
312 16931,16937,16943,16963,16979,16981,16987,16993,
313 17011,17021,17027,17029,17033,17041,17047,17053,
314 17077,17093,17099,17107,17117,17123,17137,17159,
315 17167,17183,17189,17191,17203,17207,17209,17231,
316 17239,17257,17291,17293,17299,17317,17321,17327,
317 17333,17341,17351,17359,17377,17383,17387,17389,
318 17393,17401,17417,17419,17431,17443,17449,17467,
319 17471,17477,17483,17489,17491,17497,17509,17519,
320 17539,17551,17569,17573,17579,17581,17597,17599,
321 17609,17623,17627,17657,17659,17669,17681,17683,
322 17707,17713,17729,17737,17747,17749,17761,17783,
323 17789,17791,17807,17827,17837,17839,17851,17863,
324#endif
325 };
diff --git a/src/lib/libcrypto/bn/bn_prime.pl b/src/lib/libcrypto/bn/bn_prime.pl
deleted file mode 100644
index 9fc3765486..0000000000
--- a/src/lib/libcrypto/bn/bn_prime.pl
+++ /dev/null
@@ -1,117 +0,0 @@
1#!/usr/local/bin/perl
2# bn_prime.pl
3
4$num=2048;
5$num=$ARGV[0] if ($#ARGV >= 0);
6
7push(@primes,2);
8$p=1;
9loop: while ($#primes < $num-1)
10 {
11 $p+=2;
12 $s=int(sqrt($p));
13
14 for ($i=0; $primes[$i]<=$s; $i++)
15 {
16 next loop if (($p%$primes[$i]) == 0);
17 }
18 push(@primes,$p);
19 }
20
21# print <<"EOF";
22# /* Auto generated by bn_prime.pl */
23# /* Copyright (C) 1995-1997 Eric Young (eay\@mincom.oz.au).
24# * All rights reserved.
25# * Copyright remains Eric Young's, and as such any Copyright notices in
26# * the code are not to be removed.
27# * See the COPYRIGHT file in the SSLeay distribution for more details.
28# */
29#
30# EOF
31
32print <<\EOF;
33/* Auto generated by bn_prime.pl */
34/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
35 * All rights reserved.
36 *
37 * This package is an SSL implementation written
38 * by Eric Young (eay@cryptsoft.com).
39 * The implementation was written so as to conform with Netscapes SSL.
40 *
41 * This library is free for commercial and non-commercial use as long as
42 * the following conditions are aheared to. The following conditions
43 * apply to all code found in this distribution, be it the RC4, RSA,
44 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
45 * included with this distribution is covered by the same copyright terms
46 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
47 *
48 * Copyright remains Eric Young's, and as such any Copyright notices in
49 * the code are not to be removed.
50 * If this package is used in a product, Eric Young should be given attribution
51 * as the author of the parts of the library used.
52 * This can be in the form of a textual message at program startup or
53 * in documentation (online or textual) provided with the package.
54 *
55 * Redistribution and use in source and binary forms, with or without
56 * modification, are permitted provided that the following conditions
57 * are met:
58 * 1. Redistributions of source code must retain the copyright
59 * notice, this list of conditions and the following disclaimer.
60 * 2. Redistributions in binary form must reproduce the above copyright
61 * notice, this list of conditions and the following disclaimer in the
62 * documentation and/or other materials provided with the distribution.
63 * 3. All advertising materials mentioning features or use of this software
64 * must display the following acknowledgement:
65 * "This product includes cryptographic software written by
66 * Eric Young (eay@cryptsoft.com)"
67 * The word 'cryptographic' can be left out if the rouines from the library
68 * being used are not cryptographic related :-).
69 * 4. If you include any Windows specific code (or a derivative thereof) from
70 * the apps directory (application code) you must include an acknowledgement:
71 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
72 *
73 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
74 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
75 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
76 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
77 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
78 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
79 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
80 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
81 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
82 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
83 * SUCH DAMAGE.
84 *
85 * The licence and distribution terms for any publically available version or
86 * derivative of this code cannot be changed. i.e. this code cannot simply be
87 * copied and put under another distribution licence
88 * [including the GNU Public Licence.]
89 */
90
91EOF
92
93for ($i=0; $i <= $#primes; $i++)
94 {
95 if ($primes[$i] > 256)
96 {
97 $eight=$i;
98 last;
99 }
100 }
101
102printf "#ifndef EIGHT_BIT\n";
103printf "#define NUMPRIMES %d\n",$num;
104printf "#else\n";
105printf "#define NUMPRIMES %d\n",$eight;
106printf "#endif\n";
107print "static const unsigned int primes[NUMPRIMES]=\n\t{\n\t";
108$init=0;
109for ($i=0; $i <= $#primes; $i++)
110 {
111 printf "\n#ifndef EIGHT_BIT\n\t" if ($primes[$i] > 256) && !($init++);
112 printf("\n\t") if (($i%8) == 0) && ($i != 0);
113 printf("%4d,",$primes[$i]);
114 }
115print "\n#endif\n\t};\n";
116
117
diff --git a/src/lib/libcrypto/bn/bn_print.c b/src/lib/libcrypto/bn/bn_print.c
deleted file mode 100644
index 532e66bcc3..0000000000
--- a/src/lib/libcrypto/bn/bn_print.c
+++ /dev/null
@@ -1,332 +0,0 @@
1/* crypto/bn/bn_print.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include <ctype.h>
61#include "cryptlib.h"
62#include <openssl/buffer.h>
63#include "bn_lcl.h"
64
65static const char *Hex="0123456789ABCDEF";
66
67/* Must 'OPENSSL_free' the returned data */
68char *BN_bn2hex(const BIGNUM *a)
69 {
70 int i,j,v,z=0;
71 char *buf;
72 char *p;
73
74 buf=(char *)OPENSSL_malloc(a->top*BN_BYTES*2+2);
75 if (buf == NULL)
76 {
77 BNerr(BN_F_BN_BN2HEX,ERR_R_MALLOC_FAILURE);
78 goto err;
79 }
80 p=buf;
81 if (a->neg) *(p++)='-';
82 if (a->top == 0) *(p++)='0';
83 for (i=a->top-1; i >=0; i--)
84 {
85 for (j=BN_BITS2-8; j >= 0; j-=8)
86 {
87 /* strip leading zeros */
88 v=((int)(a->d[i]>>(long)j))&0xff;
89 if (z || (v != 0))
90 {
91 *(p++)=Hex[v>>4];
92 *(p++)=Hex[v&0x0f];
93 z=1;
94 }
95 }
96 }
97 *p='\0';
98err:
99 return(buf);
100 }
101
102/* Must 'OPENSSL_free' the returned data */
103char *BN_bn2dec(const BIGNUM *a)
104 {
105 int i=0,num;
106 char *buf=NULL;
107 char *p;
108 BIGNUM *t=NULL;
109 BN_ULONG *bn_data=NULL,*lp;
110
111 i=BN_num_bits(a)*3;
112 num=(i/10+i/1000+3)+1;
113 bn_data=(BN_ULONG *)OPENSSL_malloc((num/BN_DEC_NUM+1)*sizeof(BN_ULONG));
114 buf=(char *)OPENSSL_malloc(num+3);
115 if ((buf == NULL) || (bn_data == NULL))
116 {
117 BNerr(BN_F_BN_BN2DEC,ERR_R_MALLOC_FAILURE);
118 goto err;
119 }
120 if ((t=BN_dup(a)) == NULL) goto err;
121
122 p=buf;
123 lp=bn_data;
124 if (t->neg) *(p++)='-';
125 if (t->top == 0)
126 {
127 *(p++)='0';
128 *(p++)='\0';
129 }
130 else
131 {
132 i=0;
133 while (!BN_is_zero(t))
134 {
135 *lp=BN_div_word(t,BN_DEC_CONV);
136 lp++;
137 }
138 lp--;
139 /* We now have a series of blocks, BN_DEC_NUM chars
140 * in length, where the last one needs truncation.
141 * The blocks need to be reversed in order. */
142 sprintf(p,BN_DEC_FMT1,*lp);
143 while (*p) p++;
144 while (lp != bn_data)
145 {
146 lp--;
147 sprintf(p,BN_DEC_FMT2,*lp);
148 while (*p) p++;
149 }
150 }
151err:
152 if (bn_data != NULL) OPENSSL_free(bn_data);
153 if (t != NULL) BN_free(t);
154 return(buf);
155 }
156
157int BN_hex2bn(BIGNUM **bn, const char *a)
158 {
159 BIGNUM *ret=NULL;
160 BN_ULONG l=0;
161 int neg=0,h,m,i,j,k,c;
162 int num;
163
164 if ((a == NULL) || (*a == '\0')) return(0);
165
166 if (*a == '-') { neg=1; a++; }
167
168 for (i=0; isxdigit((unsigned char) a[i]); i++)
169 ;
170
171 num=i+neg;
172 if (bn == NULL) return(num);
173
174 /* a is the start of the hex digits, and it is 'i' long */
175 if (*bn == NULL)
176 {
177 if ((ret=BN_new()) == NULL) return(0);
178 }
179 else
180 {
181 ret= *bn;
182 BN_zero(ret);
183 }
184
185 /* i is the number of hex digests; */
186 if (bn_expand(ret,i*4) == NULL) goto err;
187
188 j=i; /* least significant 'hex' */
189 m=0;
190 h=0;
191 while (j > 0)
192 {
193 m=((BN_BYTES*2) <= j)?(BN_BYTES*2):j;
194 l=0;
195 for (;;)
196 {
197 c=a[j-m];
198 if ((c >= '0') && (c <= '9')) k=c-'0';
199 else if ((c >= 'a') && (c <= 'f')) k=c-'a'+10;
200 else if ((c >= 'A') && (c <= 'F')) k=c-'A'+10;
201 else k=0; /* paranoia */
202 l=(l<<4)|k;
203
204 if (--m <= 0)
205 {
206 ret->d[h++]=l;
207 break;
208 }
209 }
210 j-=(BN_BYTES*2);
211 }
212 ret->top=h;
213 bn_fix_top(ret);
214 ret->neg=neg;
215
216 *bn=ret;
217 return(num);
218err:
219 if (*bn == NULL) BN_free(ret);
220 return(0);
221 }
222
223int BN_dec2bn(BIGNUM **bn, const char *a)
224 {
225 BIGNUM *ret=NULL;
226 BN_ULONG l=0;
227 int neg=0,i,j;
228 int num;
229
230 if ((a == NULL) || (*a == '\0')) return(0);
231 if (*a == '-') { neg=1; a++; }
232
233 for (i=0; isdigit((unsigned char) a[i]); i++)
234 ;
235
236 num=i+neg;
237 if (bn == NULL) return(num);
238
239 /* a is the start of the digits, and it is 'i' long.
240 * We chop it into BN_DEC_NUM digits at a time */
241 if (*bn == NULL)
242 {
243 if ((ret=BN_new()) == NULL) return(0);
244 }
245 else
246 {
247 ret= *bn;
248 BN_zero(ret);
249 }
250
251 /* i is the number of digests, a bit of an over expand; */
252 if (bn_expand(ret,i*4) == NULL) goto err;
253
254 j=BN_DEC_NUM-(i%BN_DEC_NUM);
255 if (j == BN_DEC_NUM) j=0;
256 l=0;
257 while (*a)
258 {
259 l*=10;
260 l+= *a-'0';
261 a++;
262 if (++j == BN_DEC_NUM)
263 {
264 BN_mul_word(ret,BN_DEC_CONV);
265 BN_add_word(ret,l);
266 l=0;
267 j=0;
268 }
269 }
270 ret->neg=neg;
271
272 bn_fix_top(ret);
273 *bn=ret;
274 return(num);
275err:
276 if (*bn == NULL) BN_free(ret);
277 return(0);
278 }
279
280#ifndef NO_BIO
281#ifndef NO_FP_API
282int BN_print_fp(FILE *fp, const BIGNUM *a)
283 {
284 BIO *b;
285 int ret;
286
287 if ((b=BIO_new(BIO_s_file())) == NULL)
288 return(0);
289 BIO_set_fp(b,fp,BIO_NOCLOSE);
290 ret=BN_print(b,a);
291 BIO_free(b);
292 return(ret);
293 }
294#endif
295
296int BN_print(BIO *bp, const BIGNUM *a)
297 {
298 int i,j,v,z=0;
299 int ret=0;
300
301 if ((a->neg) && (BIO_write(bp,"-",1) != 1)) goto end;
302 if ((a->top == 0) && (BIO_write(bp,"0",1) != 1)) goto end;
303 for (i=a->top-1; i >=0; i--)
304 {
305 for (j=BN_BITS2-4; j >= 0; j-=4)
306 {
307 /* strip leading zeros */
308 v=((int)(a->d[i]>>(long)j))&0x0f;
309 if (z || (v != 0))
310 {
311 if (BIO_write(bp,&(Hex[v]),1) != 1)
312 goto end;
313 z=1;
314 }
315 }
316 }
317 ret=1;
318end:
319 return(ret);
320 }
321#endif
322
323#ifdef BN_DEBUG
324void bn_dump1(FILE *o, const char *a, BN_ULONG *b,int n)
325 {
326 int i;
327 fprintf(o, "%s=", a);
328 for (i=n-1;i>=0;i--)
329 fprintf(o, "%08lX", b[i]); /* assumes 32-bit BN_ULONG */
330 fprintf(o, "\n");
331 }
332#endif
diff --git a/src/lib/libcrypto/bn/bn_rand.c b/src/lib/libcrypto/bn/bn_rand.c
deleted file mode 100644
index acd0619921..0000000000
--- a/src/lib/libcrypto/bn/bn_rand.c
+++ /dev/null
@@ -1,223 +0,0 @@
1/* crypto/bn/bn_rand.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include <time.h>
61#include "cryptlib.h"
62#include "bn_lcl.h"
63#include <openssl/rand.h>
64
65static int bnrand(int pseudorand, BIGNUM *rnd, int bits, int top, int bottom)
66 {
67 unsigned char *buf=NULL;
68 int ret=0,bit,bytes,mask;
69 time_t tim;
70
71 if (bits == 0)
72 {
73 BN_zero(rnd);
74 return 1;
75 }
76
77 bytes=(bits+7)/8;
78 bit=(bits-1)%8;
79 mask=0xff<<(bit+1);
80
81 buf=(unsigned char *)OPENSSL_malloc(bytes);
82 if (buf == NULL)
83 {
84 BNerr(BN_F_BN_RAND,ERR_R_MALLOC_FAILURE);
85 goto err;
86 }
87
88 /* make a random number and set the top and bottom bits */
89 time(&tim);
90 RAND_add(&tim,sizeof(tim),0);
91
92 if (pseudorand)
93 {
94 if (RAND_pseudo_bytes(buf, bytes) == -1)
95 goto err;
96 }
97 else
98 {
99 if (RAND_bytes(buf, bytes) <= 0)
100 goto err;
101 }
102
103#if 1
104 if (pseudorand == 2)
105 {
106 /* generate patterns that are more likely to trigger BN
107 library bugs */
108 int i;
109 unsigned char c;
110
111 for (i = 0; i < bytes; i++)
112 {
113 RAND_pseudo_bytes(&c, 1);
114 if (c >= 128 && i > 0)
115 buf[i] = buf[i-1];
116 else if (c < 42)
117 buf[i] = 0;
118 else if (c < 84)
119 buf[i] = 255;
120 }
121 }
122#endif
123
124 if (top != -1)
125 {
126 if (top)
127 {
128 if (bit == 0)
129 {
130 buf[0]=1;
131 buf[1]|=0x80;
132 }
133 else
134 {
135 buf[0]|=(3<<(bit-1));
136 }
137 }
138 else
139 {
140 buf[0]|=(1<<bit);
141 }
142 }
143 buf[0] &= ~mask;
144 if (bottom) /* set bottom bit if requested */
145 buf[bytes-1]|=1;
146 if (!BN_bin2bn(buf,bytes,rnd)) goto err;
147 ret=1;
148err:
149 if (buf != NULL)
150 {
151 memset(buf,0,bytes);
152 OPENSSL_free(buf);
153 }
154 return(ret);
155 }
156
157int BN_rand(BIGNUM *rnd, int bits, int top, int bottom)
158 {
159 return bnrand(0, rnd, bits, top, bottom);
160 }
161
162int BN_pseudo_rand(BIGNUM *rnd, int bits, int top, int bottom)
163 {
164 return bnrand(1, rnd, bits, top, bottom);
165 }
166
167#if 1
168int BN_bntest_rand(BIGNUM *rnd, int bits, int top, int bottom)
169 {
170 return bnrand(2, rnd, bits, top, bottom);
171 }
172#endif
173
174/* random number r: 0 <= r < range */
175int BN_rand_range(BIGNUM *r, BIGNUM *range)
176 {
177 int n;
178
179 if (range->neg || BN_is_zero(range))
180 {
181 BNerr(BN_F_BN_RAND_RANGE, BN_R_INVALID_RANGE);
182 return 0;
183 }
184
185 n = BN_num_bits(range); /* n > 0 */
186
187 if (n == 1)
188 {
189 if (!BN_zero(r)) return 0;
190 }
191 else if (BN_is_bit_set(range, n - 2))
192 {
193 do
194 {
195 /* range = 11..._2, so each iteration succeeds with probability >= .75 */
196 if (!BN_rand(r, n, -1, 0)) return 0;
197 }
198 while (BN_cmp(r, range) >= 0);
199 }
200 else
201 {
202 /* range = 10..._2,
203 * so 3*range (= 11..._2) is exactly one bit longer than range */
204 do
205 {
206 if (!BN_rand(r, n + 1, -1, 0)) return 0;
207 /* If r < 3*range, use r := r MOD range
208 * (which is either r, r - range, or r - 2*range).
209 * Otherwise, iterate once more.
210 * Since 3*range = 11..._2, each iteration succeeds with
211 * probability >= .75. */
212 if (BN_cmp(r ,range) >= 0)
213 {
214 if (!BN_sub(r, r, range)) return 0;
215 if (BN_cmp(r, range) >= 0)
216 if (!BN_sub(r, r, range)) return 0;
217 }
218 }
219 while (BN_cmp(r, range) >= 0);
220 }
221
222 return 1;
223 }
diff --git a/src/lib/libcrypto/bn/bn_recp.c b/src/lib/libcrypto/bn/bn_recp.c
deleted file mode 100644
index d019941d6b..0000000000
--- a/src/lib/libcrypto/bn/bn_recp.c
+++ /dev/null
@@ -1,220 +0,0 @@
1/* crypto/bn/bn_recp.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63void BN_RECP_CTX_init(BN_RECP_CTX *recp)
64 {
65 BN_init(&(recp->N));
66 BN_init(&(recp->Nr));
67 recp->num_bits=0;
68 recp->flags=0;
69 }
70
71BN_RECP_CTX *BN_RECP_CTX_new(void)
72 {
73 BN_RECP_CTX *ret;
74
75 if ((ret=(BN_RECP_CTX *)OPENSSL_malloc(sizeof(BN_RECP_CTX))) == NULL)
76 return(NULL);
77
78 BN_RECP_CTX_init(ret);
79 ret->flags=BN_FLG_MALLOCED;
80 return(ret);
81 }
82
83void BN_RECP_CTX_free(BN_RECP_CTX *recp)
84 {
85 if(recp == NULL)
86 return;
87
88 BN_free(&(recp->N));
89 BN_free(&(recp->Nr));
90 if (recp->flags & BN_FLG_MALLOCED)
91 OPENSSL_free(recp);
92 }
93
94int BN_RECP_CTX_set(BN_RECP_CTX *recp, const BIGNUM *d, BN_CTX *ctx)
95 {
96 BN_copy(&(recp->N),d);
97 BN_zero(&(recp->Nr));
98 recp->num_bits=BN_num_bits(d);
99 recp->shift=0;
100 return(1);
101 }
102
103int BN_mod_mul_reciprocal(BIGNUM *r, BIGNUM *x, BIGNUM *y, BN_RECP_CTX *recp,
104 BN_CTX *ctx)
105 {
106 int ret=0;
107 BIGNUM *a;
108
109 BN_CTX_start(ctx);
110 if ((a = BN_CTX_get(ctx)) == NULL) goto err;
111 if (y != NULL)
112 {
113 if (x == y)
114 { if (!BN_sqr(a,x,ctx)) goto err; }
115 else
116 { if (!BN_mul(a,x,y,ctx)) goto err; }
117 }
118 else
119 a=x; /* Just do the mod */
120
121 BN_div_recp(NULL,r,a,recp,ctx);
122 ret=1;
123err:
124 BN_CTX_end(ctx);
125 return(ret);
126 }
127
128int BN_div_recp(BIGNUM *dv, BIGNUM *rem, BIGNUM *m, BN_RECP_CTX *recp,
129 BN_CTX *ctx)
130 {
131 int i,j,ret=0;
132 BIGNUM *a,*b,*d,*r;
133
134 BN_CTX_start(ctx);
135 a=BN_CTX_get(ctx);
136 b=BN_CTX_get(ctx);
137 if (dv != NULL)
138 d=dv;
139 else
140 d=BN_CTX_get(ctx);
141 if (rem != NULL)
142 r=rem;
143 else
144 r=BN_CTX_get(ctx);
145 if (a == NULL || b == NULL || d == NULL || r == NULL) goto err;
146
147 if (BN_ucmp(m,&(recp->N)) < 0)
148 {
149 BN_zero(d);
150 BN_copy(r,m);
151 BN_CTX_end(ctx);
152 return(1);
153 }
154
155 /* We want the remainder
156 * Given input of ABCDEF / ab
157 * we need multiply ABCDEF by 3 digests of the reciprocal of ab
158 *
159 */
160 i=BN_num_bits(m);
161
162 j=recp->num_bits<<1;
163 if (j>i) i=j;
164 j>>=1;
165
166 if (i != recp->shift)
167 recp->shift=BN_reciprocal(&(recp->Nr),&(recp->N),
168 i,ctx);
169
170 if (!BN_rshift(a,m,j)) goto err;
171 if (!BN_mul(b,a,&(recp->Nr),ctx)) goto err;
172 if (!BN_rshift(d,b,i-j)) goto err;
173 d->neg=0;
174 if (!BN_mul(b,&(recp->N),d,ctx)) goto err;
175 if (!BN_usub(r,m,b)) goto err;
176 r->neg=0;
177
178#if 1
179 j=0;
180 while (BN_ucmp(r,&(recp->N)) >= 0)
181 {
182 if (j++ > 2)
183 {
184 BNerr(BN_F_BN_MOD_MUL_RECIPROCAL,BN_R_BAD_RECIPROCAL);
185 goto err;
186 }
187 if (!BN_usub(r,r,&(recp->N))) goto err;
188 if (!BN_add_word(d,1)) goto err;
189 }
190#endif
191
192 r->neg=BN_is_zero(r)?0:m->neg;
193 d->neg=m->neg^recp->N.neg;
194 ret=1;
195err:
196 BN_CTX_end(ctx);
197 return(ret);
198 }
199
200/* len is the expected size of the result
201 * We actually calculate with an extra word of precision, so
202 * we can do faster division if the remainder is not required.
203 */
204int BN_reciprocal(BIGNUM *r, BIGNUM *m, int len, BN_CTX *ctx)
205 {
206 int ret= -1;
207 BIGNUM t;
208
209 BN_init(&t);
210
211 BN_zero(&t);
212 if (!BN_set_bit(&t,len)) goto err;
213
214 if (!BN_div(r,NULL,&t,m,ctx)) goto err;
215 ret=len;
216err:
217 BN_free(&t);
218 return(ret);
219 }
220
diff --git a/src/lib/libcrypto/bn/bn_shift.c b/src/lib/libcrypto/bn/bn_shift.c
deleted file mode 100644
index c2608f9f4a..0000000000
--- a/src/lib/libcrypto/bn/bn_shift.c
+++ /dev/null
@@ -1,205 +0,0 @@
1/* crypto/bn/bn_shift.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63int BN_lshift1(BIGNUM *r, BIGNUM *a)
64 {
65 register BN_ULONG *ap,*rp,t,c;
66 int i;
67
68 if (r != a)
69 {
70 r->neg=a->neg;
71 if (bn_wexpand(r,a->top+1) == NULL) return(0);
72 r->top=a->top;
73 }
74 else
75 {
76 if (bn_wexpand(r,a->top+1) == NULL) return(0);
77 }
78 ap=a->d;
79 rp=r->d;
80 c=0;
81 for (i=0; i<a->top; i++)
82 {
83 t= *(ap++);
84 *(rp++)=((t<<1)|c)&BN_MASK2;
85 c=(t & BN_TBIT)?1:0;
86 }
87 if (c)
88 {
89 *rp=1;
90 r->top++;
91 }
92 return(1);
93 }
94
95int BN_rshift1(BIGNUM *r, BIGNUM *a)
96 {
97 BN_ULONG *ap,*rp,t,c;
98 int i;
99
100 if (BN_is_zero(a))
101 {
102 BN_zero(r);
103 return(1);
104 }
105 if (a != r)
106 {
107 if (bn_wexpand(r,a->top) == NULL) return(0);
108 r->top=a->top;
109 r->neg=a->neg;
110 }
111 ap=a->d;
112 rp=r->d;
113 c=0;
114 for (i=a->top-1; i>=0; i--)
115 {
116 t=ap[i];
117 rp[i]=((t>>1)&BN_MASK2)|c;
118 c=(t&1)?BN_TBIT:0;
119 }
120 bn_fix_top(r);
121 return(1);
122 }
123
124int BN_lshift(BIGNUM *r, const BIGNUM *a, int n)
125 {
126 int i,nw,lb,rb;
127 BN_ULONG *t,*f;
128 BN_ULONG l;
129
130 r->neg=a->neg;
131 if (bn_wexpand(r,a->top+(n/BN_BITS2)+1) == NULL) return(0);
132 nw=n/BN_BITS2;
133 lb=n%BN_BITS2;
134 rb=BN_BITS2-lb;
135 f=a->d;
136 t=r->d;
137 t[a->top+nw]=0;
138 if (lb == 0)
139 for (i=a->top-1; i>=0; i--)
140 t[nw+i]=f[i];
141 else
142 for (i=a->top-1; i>=0; i--)
143 {
144 l=f[i];
145 t[nw+i+1]|=(l>>rb)&BN_MASK2;
146 t[nw+i]=(l<<lb)&BN_MASK2;
147 }
148 memset(t,0,nw*sizeof(t[0]));
149/* for (i=0; i<nw; i++)
150 t[i]=0;*/
151 r->top=a->top+nw+1;
152 bn_fix_top(r);
153 return(1);
154 }
155
156int BN_rshift(BIGNUM *r, BIGNUM *a, int n)
157 {
158 int i,j,nw,lb,rb;
159 BN_ULONG *t,*f;
160 BN_ULONG l,tmp;
161
162 nw=n/BN_BITS2;
163 rb=n%BN_BITS2;
164 lb=BN_BITS2-rb;
165 if (nw > a->top || a->top == 0)
166 {
167 BN_zero(r);
168 return(1);
169 }
170 if (r != a)
171 {
172 r->neg=a->neg;
173 if (bn_wexpand(r,a->top-nw+1) == NULL) return(0);
174 }
175 else
176 {
177 if (n == 0)
178 return 1; /* or the copying loop will go berserk */
179 }
180
181 f= &(a->d[nw]);
182 t=r->d;
183 j=a->top-nw;
184 r->top=j;
185
186 if (rb == 0)
187 {
188 for (i=j+1; i > 0; i--)
189 *(t++)= *(f++);
190 }
191 else
192 {
193 l= *(f++);
194 for (i=1; i<j; i++)
195 {
196 tmp =(l>>rb)&BN_MASK2;
197 l= *(f++);
198 *(t++) =(tmp|(l<<lb))&BN_MASK2;
199 }
200 *(t++) =(l>>rb)&BN_MASK2;
201 }
202 *t=0;
203 bn_fix_top(r);
204 return(1);
205 }
diff --git a/src/lib/libcrypto/bn/bn_sqr.c b/src/lib/libcrypto/bn/bn_sqr.c
deleted file mode 100644
index 75f4f38392..0000000000
--- a/src/lib/libcrypto/bn/bn_sqr.c
+++ /dev/null
@@ -1,288 +0,0 @@
1/* crypto/bn/bn_sqr.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63/* r must not be a */
64/* I've just gone over this and it is now %20 faster on x86 - eay - 27 Jun 96 */
65int BN_sqr(BIGNUM *r, BIGNUM *a, BN_CTX *ctx)
66 {
67 int max,al;
68 int ret = 0;
69 BIGNUM *tmp,*rr;
70
71#ifdef BN_COUNT
72printf("BN_sqr %d * %d\n",a->top,a->top);
73#endif
74 bn_check_top(a);
75
76 al=a->top;
77 if (al <= 0)
78 {
79 r->top=0;
80 return(1);
81 }
82
83 BN_CTX_start(ctx);
84 rr=(a != r) ? r : BN_CTX_get(ctx);
85 tmp=BN_CTX_get(ctx);
86 if (tmp == NULL) goto err;
87
88 max=(al+al);
89 if (bn_wexpand(rr,max+1) == NULL) goto err;
90
91 r->neg=0;
92 if (al == 4)
93 {
94#ifndef BN_SQR_COMBA
95 BN_ULONG t[8];
96 bn_sqr_normal(rr->d,a->d,4,t);
97#else
98 bn_sqr_comba4(rr->d,a->d);
99#endif
100 }
101 else if (al == 8)
102 {
103#ifndef BN_SQR_COMBA
104 BN_ULONG t[16];
105 bn_sqr_normal(rr->d,a->d,8,t);
106#else
107 bn_sqr_comba8(rr->d,a->d);
108#endif
109 }
110 else
111 {
112#if defined(BN_RECURSION)
113 if (al < BN_SQR_RECURSIVE_SIZE_NORMAL)
114 {
115 BN_ULONG t[BN_SQR_RECURSIVE_SIZE_NORMAL*2];
116 bn_sqr_normal(rr->d,a->d,al,t);
117 }
118 else
119 {
120 int j,k;
121
122 j=BN_num_bits_word((BN_ULONG)al);
123 j=1<<(j-1);
124 k=j+j;
125 if (al == j)
126 {
127 if (bn_wexpand(a,k*2) == NULL) goto err;
128 if (bn_wexpand(tmp,k*2) == NULL) goto err;
129 bn_sqr_recursive(rr->d,a->d,al,tmp->d);
130 }
131 else
132 {
133 if (bn_wexpand(tmp,max) == NULL) goto err;
134 bn_sqr_normal(rr->d,a->d,al,tmp->d);
135 }
136 }
137#else
138 if (bn_wexpand(tmp,max) == NULL) goto err;
139 bn_sqr_normal(rr->d,a->d,al,tmp->d);
140#endif
141 }
142
143 rr->top=max;
144 if ((max > 0) && (rr->d[max-1] == 0)) rr->top--;
145 if (rr != r) BN_copy(r,rr);
146 ret = 1;
147 err:
148 BN_CTX_end(ctx);
149 return(ret);
150 }
151
152/* tmp must have 2*n words */
153void bn_sqr_normal(BN_ULONG *r, BN_ULONG *a, int n, BN_ULONG *tmp)
154 {
155 int i,j,max;
156 BN_ULONG *ap,*rp;
157
158 max=n*2;
159 ap=a;
160 rp=r;
161 rp[0]=rp[max-1]=0;
162 rp++;
163 j=n;
164
165 if (--j > 0)
166 {
167 ap++;
168 rp[j]=bn_mul_words(rp,ap,j,ap[-1]);
169 rp+=2;
170 }
171
172 for (i=n-2; i>0; i--)
173 {
174 j--;
175 ap++;
176 rp[j]=bn_mul_add_words(rp,ap,j,ap[-1]);
177 rp+=2;
178 }
179
180 bn_add_words(r,r,r,max);
181
182 /* There will not be a carry */
183
184 bn_sqr_words(tmp,a,n);
185
186 bn_add_words(r,r,tmp,max);
187 }
188
189#ifdef BN_RECURSION
190/* r is 2*n words in size,
191 * a and b are both n words in size. (There's not actually a 'b' here ...)
192 * n must be a power of 2.
193 * We multiply and return the result.
194 * t must be 2*n words in size
195 * We calculate
196 * a[0]*b[0]
197 * a[0]*b[0]+a[1]*b[1]+(a[0]-a[1])*(b[1]-b[0])
198 * a[1]*b[1]
199 */
200void bn_sqr_recursive(BN_ULONG *r, BN_ULONG *a, int n2, BN_ULONG *t)
201 {
202 int n=n2/2;
203 int zero,c1;
204 BN_ULONG ln,lo,*p;
205
206#ifdef BN_COUNT
207printf(" bn_sqr_recursive %d * %d\n",n2,n2);
208#endif
209 if (n2 == 4)
210 {
211#ifndef BN_SQR_COMBA
212 bn_sqr_normal(r,a,4,t);
213#else
214 bn_sqr_comba4(r,a);
215#endif
216 return;
217 }
218 else if (n2 == 8)
219 {
220#ifndef BN_SQR_COMBA
221 bn_sqr_normal(r,a,8,t);
222#else
223 bn_sqr_comba8(r,a);
224#endif
225 return;
226 }
227 if (n2 < BN_SQR_RECURSIVE_SIZE_NORMAL)
228 {
229 bn_sqr_normal(r,a,n2,t);
230 return;
231 }
232 /* r=(a[0]-a[1])*(a[1]-a[0]) */
233 c1=bn_cmp_words(a,&(a[n]),n);
234 zero=0;
235 if (c1 > 0)
236 bn_sub_words(t,a,&(a[n]),n);
237 else if (c1 < 0)
238 bn_sub_words(t,&(a[n]),a,n);
239 else
240 zero=1;
241
242 /* The result will always be negative unless it is zero */
243 p= &(t[n2*2]);
244
245 if (!zero)
246 bn_sqr_recursive(&(t[n2]),t,n,p);
247 else
248 memset(&(t[n2]),0,n*sizeof(BN_ULONG));
249 bn_sqr_recursive(r,a,n,p);
250 bn_sqr_recursive(&(r[n2]),&(a[n]),n,p);
251
252 /* t[32] holds (a[0]-a[1])*(a[1]-a[0]), it is negative or zero
253 * r[10] holds (a[0]*b[0])
254 * r[32] holds (b[1]*b[1])
255 */
256
257 c1=(int)(bn_add_words(t,r,&(r[n2]),n2));
258
259 /* t[32] is negative */
260 c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2));
261
262 /* t[32] holds (a[0]-a[1])*(a[1]-a[0])+(a[0]*a[0])+(a[1]*a[1])
263 * r[10] holds (a[0]*a[0])
264 * r[32] holds (a[1]*a[1])
265 * c1 holds the carry bits
266 */
267 c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2));
268 if (c1)
269 {
270 p= &(r[n+n2]);
271 lo= *p;
272 ln=(lo+c1)&BN_MASK2;
273 *p=ln;
274
275 /* The overflow will stop before we over write
276 * words we should not overwrite */
277 if (ln < (BN_ULONG)c1)
278 {
279 do {
280 p++;
281 lo= *p;
282 ln=(lo+1)&BN_MASK2;
283 *p=ln;
284 } while (ln == 0);
285 }
286 }
287 }
288#endif
diff --git a/src/lib/libcrypto/bn/bn_word.c b/src/lib/libcrypto/bn/bn_word.c
deleted file mode 100644
index cd59baa2c4..0000000000
--- a/src/lib/libcrypto/bn/bn_word.c
+++ /dev/null
@@ -1,199 +0,0 @@
1/* crypto/bn/bn_word.c */
2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
3 * All rights reserved.
4 *
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
8 *
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
22 *
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
25 * are met:
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE.
52 *
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
57 */
58
59#include <stdio.h>
60#include "cryptlib.h"
61#include "bn_lcl.h"
62
63BN_ULONG BN_mod_word(const BIGNUM *a, BN_ULONG w)
64 {
65#ifndef BN_LLONG
66 BN_ULONG ret=0;
67#else
68 BN_ULLONG ret=0;
69#endif
70 int i;
71
72 w&=BN_MASK2;
73 for (i=a->top-1; i>=0; i--)
74 {
75#ifndef BN_LLONG
76 ret=((ret<<BN_BITS4)|((a->d[i]>>BN_BITS4)&BN_MASK2l))%w;
77 ret=((ret<<BN_BITS4)|(a->d[i]&BN_MASK2l))%w;
78#else
79 ret=(BN_ULLONG)(((ret<<(BN_ULLONG)BN_BITS2)|a->d[i])%
80 (BN_ULLONG)w);
81#endif
82 }
83 return((BN_ULONG)ret);
84 }
85
86BN_ULONG BN_div_word(BIGNUM *a, BN_ULONG w)
87 {
88 BN_ULONG ret;
89 int i;
90
91 if (a->top == 0) return(0);
92 ret=0;
93 w&=BN_MASK2;
94 for (i=a->top-1; i>=0; i--)
95 {
96 BN_ULONG l,d;
97
98 l=a->d[i];
99 d=bn_div_words(ret,l,w);
100 ret=(l-((d*w)&BN_MASK2))&BN_MASK2;
101 a->d[i]=d;
102 }
103 if ((a->top > 0) && (a->d[a->top-1] == 0))
104 a->top--;
105 return(ret);
106 }
107
108int BN_add_word(BIGNUM *a, BN_ULONG w)
109 {
110 BN_ULONG l;
111 int i;
112
113 if (a->neg)
114 {
115 a->neg=0;
116 i=BN_sub_word(a,w);
117 if (!BN_is_zero(a))
118 a->neg=!(a->neg);
119 return(i);
120 }
121 w&=BN_MASK2;
122 if (bn_wexpand(a,a->top+1) == NULL) return(0);
123 i=0;
124 for (;;)
125 {
126 l=(a->d[i]+(BN_ULONG)w)&BN_MASK2;
127 a->d[i]=l;
128 if (w > l)
129 w=1;
130 else
131 break;
132 i++;
133 }
134 if (i >= a->top)
135 a->top++;
136 return(1);
137 }
138
139int BN_sub_word(BIGNUM *a, BN_ULONG w)
140 {
141 int i;
142
143 if (BN_is_zero(a) || a->neg)
144 {
145 a->neg=0;
146 i=BN_add_word(a,w);
147 a->neg=1;
148 return(i);
149 }
150
151 w&=BN_MASK2;
152 if ((a->top == 1) && (a->d[0] < w))
153 {
154 a->d[0]=w-a->d[0];
155 a->neg=1;
156 return(1);
157 }
158 i=0;
159 for (;;)
160 {
161 if (a->d[i] >= w)
162 {
163 a->d[i]-=w;
164 break;
165 }
166 else
167 {
168 a->d[i]=(a->d[i]-w)&BN_MASK2;
169 i++;
170 w=1;
171 }
172 }
173 if ((a->d[i] == 0) && (i == (a->top-1)))
174 a->top--;
175 return(1);
176 }
177
178int BN_mul_word(BIGNUM *a, BN_ULONG w)
179 {
180 BN_ULONG ll;
181
182 w&=BN_MASK2;
183 if (a->top)
184 {
185 if (w == 0)
186 BN_zero(a);
187 else
188 {
189 ll=bn_mul_words(a->d,a->d,a->top,w);
190 if (ll)
191 {
192 if (bn_wexpand(a,a->top+1) == NULL) return(0);
193 a->d[a->top++]=ll;
194 }
195 }
196 }
197 return(1);
198 }
199
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-21 08:23:10 +0000