summaryrefslogtreecommitdiff
path: root/src/lib/libcrypto/bn/bn_asm.c
diff options
context:
space:
mode:
authorjsing <>2014-05-08 13:20:49 +0000
committerjsing <>2014-05-08 13:20:49 +0000
commit2e8879604fe3abbc2431ca79a4a923f1e87da75e (patch)
tree18398455223278c0cb2bd44f57e4499a4370f665 /src/lib/libcrypto/bn/bn_asm.c
parentf7d9a959949e5f3918c1cf2b27fb4cd7b62d07d5 (diff)
downloadopenbsd-2e8879604fe3abbc2431ca79a4a923f1e87da75e.tar.gz
openbsd-2e8879604fe3abbc2431ca79a4a923f1e87da75e.tar.bz2
openbsd-2e8879604fe3abbc2431ca79a4a923f1e87da75e.zip
Emergency knfectomie requested by tedu@.
Diffstat (limited to 'src/lib/libcrypto/bn/bn_asm.c')
-rw-r--r--src/lib/libcrypto/bn/bn_asm.c1368
1 files changed, 719 insertions, 649 deletions
diff --git a/src/lib/libcrypto/bn/bn_asm.c b/src/lib/libcrypto/bn/bn_asm.c
index f1a0944089..0eebb9824f 100644
--- a/src/lib/libcrypto/bn/bn_asm.c
+++ b/src/lib/libcrypto/bn/bn_asm.c
@@ -5,21 +5,21 @@
5 * This package is an SSL implementation written 5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com). 6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL. 7 * The implementation was written so as to conform with Netscapes SSL.
8 * 8 *
9 * This library is free for commercial and non-commercial use as long as 9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions 10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA, 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 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 13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com). 14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
15 * 15 *
16 * Copyright remains Eric Young's, and as such any Copyright notices in 16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed. 17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution 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. 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 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. 21 * in documentation (online or textual) provided with the package.
22 * 22 *
23 * Redistribution and use in source and binary forms, with or without 23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions 24 * modification, are permitted provided that the following conditions
25 * are met: 25 * are met:
@@ -34,10 +34,10 @@
34 * Eric Young (eay@cryptsoft.com)" 34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library 35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-). 36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from 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: 38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" 39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
40 * 40 *
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND 41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
@@ -49,7 +49,7 @@
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 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 50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
51 * SUCH DAMAGE. 51 * SUCH DAMAGE.
52 * 52 *
53 * The licence and distribution terms for any publically available version or 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 54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence 55 * copied and put under another distribution licence
@@ -68,363 +68,417 @@
68 68
69#if defined(BN_LLONG) || defined(BN_UMULT_HIGH) 69#if defined(BN_LLONG) || defined(BN_UMULT_HIGH)
70 70
71BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) 71BN_ULONG
72 { 72bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
73 BN_ULONG c1=0; 73{
74 BN_ULONG c1 = 0;
74 75
75 assert(num >= 0); 76 assert(num >= 0);
76 if (num <= 0) return(c1); 77 if (num <= 0)
78 return (c1);
77 79
78#ifndef OPENSSL_SMALL_FOOTPRINT 80#ifndef OPENSSL_SMALL_FOOTPRINT
79 while (num&~3) 81 while (num & ~3) {
80 { 82 mul_add(rp[0], ap[0], w, c1);
81 mul_add(rp[0],ap[0],w,c1); 83 mul_add(rp[1], ap[1], w, c1);
82 mul_add(rp[1],ap[1],w,c1); 84 mul_add(rp[2], ap[2], w, c1);
83 mul_add(rp[2],ap[2],w,c1); 85 mul_add(rp[3], ap[3], w, c1);
84 mul_add(rp[3],ap[3],w,c1); 86 ap += 4;
85 ap+=4; rp+=4; num-=4; 87 rp += 4;
86 } 88 num -= 4;
89 }
87#endif 90#endif
88 while (num) 91 while (num) {
89 { 92 mul_add(rp[0], ap[0], w, c1);
90 mul_add(rp[0],ap[0],w,c1); 93 ap++;
91 ap++; rp++; num--; 94 rp++;
92 } 95 num--;
93 96 }
94 return(c1);
95 }
96 97
97BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) 98 return (c1);
98 { 99}
99 BN_ULONG c1=0; 100
101BN_ULONG
102bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
103{
104 BN_ULONG c1 = 0;
100 105
101 assert(num >= 0); 106 assert(num >= 0);
102 if (num <= 0) return(c1); 107 if (num <= 0)
108 return (c1);
103 109
104#ifndef OPENSSL_SMALL_FOOTPRINT 110#ifndef OPENSSL_SMALL_FOOTPRINT
105 while (num&~3) 111 while (num & ~3) {
106 { 112 mul(rp[0], ap[0], w, c1);
107 mul(rp[0],ap[0],w,c1); 113 mul(rp[1], ap[1], w, c1);
108 mul(rp[1],ap[1],w,c1); 114 mul(rp[2], ap[2], w, c1);
109 mul(rp[2],ap[2],w,c1); 115 mul(rp[3], ap[3], w, c1);
110 mul(rp[3],ap[3],w,c1); 116 ap += 4;
111 ap+=4; rp+=4; num-=4; 117 rp += 4;
112 } 118 num -= 4;
119 }
113#endif 120#endif
114 while (num) 121 while (num) {
115 { 122 mul(rp[0], ap[0], w, c1);
116 mul(rp[0],ap[0],w,c1); 123 ap++;
117 ap++; rp++; num--; 124 rp++;
118 } 125 num--;
119 return(c1); 126 }
120 } 127 return (c1);
128}
121 129
122void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n) 130void
123 { 131bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
132{
124 assert(n >= 0); 133 assert(n >= 0);
125 if (n <= 0) return; 134 if (n <= 0)
135 return;
126 136
127#ifndef OPENSSL_SMALL_FOOTPRINT 137#ifndef OPENSSL_SMALL_FOOTPRINT
128 while (n&~3) 138 while (n & ~3) {
129 { 139 sqr(r[0], r[1], a[0]);
130 sqr(r[0],r[1],a[0]); 140 sqr(r[2], r[3], a[1]);
131 sqr(r[2],r[3],a[1]); 141 sqr(r[4], r[5], a[2]);
132 sqr(r[4],r[5],a[2]); 142 sqr(r[6], r[7], a[3]);
133 sqr(r[6],r[7],a[3]); 143 a += 4;
134 a+=4; r+=8; n-=4; 144 r += 8;
135 } 145 n -= 4;
146 }
136#endif 147#endif
137 while (n) 148 while (n) {
138 { 149 sqr(r[0], r[1], a[0]);
139 sqr(r[0],r[1],a[0]); 150 a++;
140 a++; r+=2; n--; 151 r += 2;
141 } 152 n--;
142 } 153 }
154}
143 155
144#else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ 156#else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
145 157
146BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) 158BN_ULONG
147 { 159bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
148 BN_ULONG c=0; 160{
149 BN_ULONG bl,bh; 161 BN_ULONG c = 0;
162 BN_ULONG bl, bh;
150 163
151 assert(num >= 0); 164 assert(num >= 0);
152 if (num <= 0) return((BN_ULONG)0); 165 if (num <= 0)
166 return ((BN_ULONG)0);
153 167
154 bl=LBITS(w); 168 bl = LBITS(w);
155 bh=HBITS(w); 169 bh = HBITS(w);
156 170
157#ifndef OPENSSL_SMALL_FOOTPRINT 171#ifndef OPENSSL_SMALL_FOOTPRINT
158 while (num&~3) 172 while (num & ~3) {
159 { 173 mul_add(rp[0], ap[0], bl, bh, c);
160 mul_add(rp[0],ap[0],bl,bh,c); 174 mul_add(rp[1], ap[1], bl, bh, c);
161 mul_add(rp[1],ap[1],bl,bh,c); 175 mul_add(rp[2], ap[2], bl, bh, c);
162 mul_add(rp[2],ap[2],bl,bh,c); 176 mul_add(rp[3], ap[3], bl, bh, c);
163 mul_add(rp[3],ap[3],bl,bh,c); 177 ap += 4;
164 ap+=4; rp+=4; num-=4; 178 rp += 4;
165 } 179 num -= 4;
180 }
166#endif 181#endif
167 while (num) 182 while (num) {
168 { 183 mul_add(rp[0], ap[0], bl, bh, c);
169 mul_add(rp[0],ap[0],bl,bh,c); 184 ap++;
170 ap++; rp++; num--; 185 rp++;
171 } 186 num--;
172 return(c); 187 }
173 } 188 return (c);
189}
174 190
175BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) 191BN_ULONG
176 { 192bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
177 BN_ULONG carry=0; 193{
178 BN_ULONG bl,bh; 194 BN_ULONG carry = 0;
195 BN_ULONG bl, bh;
179 196
180 assert(num >= 0); 197 assert(num >= 0);
181 if (num <= 0) return((BN_ULONG)0); 198 if (num <= 0)
199 return ((BN_ULONG)0);
182 200
183 bl=LBITS(w); 201 bl = LBITS(w);
184 bh=HBITS(w); 202 bh = HBITS(w);
185 203
186#ifndef OPENSSL_SMALL_FOOTPRINT 204#ifndef OPENSSL_SMALL_FOOTPRINT
187 while (num&~3) 205 while (num & ~3) {
188 { 206 mul(rp[0], ap[0], bl, bh, carry);
189 mul(rp[0],ap[0],bl,bh,carry); 207 mul(rp[1], ap[1], bl, bh, carry);
190 mul(rp[1],ap[1],bl,bh,carry); 208 mul(rp[2], ap[2], bl, bh, carry);
191 mul(rp[2],ap[2],bl,bh,carry); 209 mul(rp[3], ap[3], bl, bh, carry);
192 mul(rp[3],ap[3],bl,bh,carry); 210 ap += 4;
193 ap+=4; rp+=4; num-=4; 211 rp += 4;
194 } 212 num -= 4;
213 }
195#endif 214#endif
196 while (num) 215 while (num) {
197 { 216 mul(rp[0], ap[0], bl, bh, carry);
198 mul(rp[0],ap[0],bl,bh,carry); 217 ap++;
199 ap++; rp++; num--; 218 rp++;
200 } 219 num--;
201 return(carry); 220 }
202 } 221 return (carry);
222}
203 223
204void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n) 224void
205 { 225bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
226{
206 assert(n >= 0); 227 assert(n >= 0);
207 if (n <= 0) return; 228 if (n <= 0)
229 return;
208 230
209#ifndef OPENSSL_SMALL_FOOTPRINT 231#ifndef OPENSSL_SMALL_FOOTPRINT
210 while (n&~3) 232 while (n & ~3) {
211 { 233 sqr64(r[0], r[1], a[0]);
212 sqr64(r[0],r[1],a[0]); 234 sqr64(r[2], r[3], a[1]);
213 sqr64(r[2],r[3],a[1]); 235 sqr64(r[4], r[5], a[2]);
214 sqr64(r[4],r[5],a[2]); 236 sqr64(r[6], r[7], a[3]);
215 sqr64(r[6],r[7],a[3]); 237 a += 4;
216 a+=4; r+=8; n-=4; 238 r += 8;
217 } 239 n -= 4;
240 }
218#endif 241#endif
219 while (n) 242 while (n) {
220 { 243 sqr64(r[0], r[1], a[0]);
221 sqr64(r[0],r[1],a[0]); 244 a++;
222 a++; r+=2; n--; 245 r += 2;
223 } 246 n--;
224 } 247 }
248}
225 249
226#endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ 250#endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
227 251
228#if defined(BN_LLONG) && defined(BN_DIV2W) 252#if defined(BN_LLONG) && defined(BN_DIV2W)
229 253
230BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) 254BN_ULONG
231 { 255bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
232 return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d)); 256{
233 } 257 return ((BN_ULONG)(((((BN_ULLONG)h) << BN_BITS2)|l)/(BN_ULLONG)d));
258}
234 259
235#else 260#else
236 261
237/* Divide h,l by d and return the result. */ 262/* Divide h,l by d and return the result. */
238/* I need to test this some more :-( */ 263/* I need to test this some more :-( */
239BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) 264BN_ULONG
240 { 265bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
241 BN_ULONG dh,dl,q,ret=0,th,tl,t; 266{
242 int i,count=2; 267 BN_ULONG dh, dl, q,ret = 0, th, tl, t;
243 268 int i, count = 2;
244 if (d == 0) return(BN_MASK2); 269
245 270 if (d == 0)
246 i=BN_num_bits_word(d); 271 return (BN_MASK2);
247 assert((i == BN_BITS2) || (h <= (BN_ULONG)1<<i)); 272
248 273 i = BN_num_bits_word(d);
249 i=BN_BITS2-i; 274 assert((i == BN_BITS2) || (h <= (BN_ULONG)1 << i));
250 if (h >= d) h-=d; 275
251 276 i = BN_BITS2 - i;
252 if (i) 277 if (h >= d)
253 { 278 h -= d;
254 d<<=i; 279
255 h=(h<<i)|(l>>(BN_BITS2-i)); 280 if (i) {
256 l<<=i; 281 d <<= i;
257 } 282 h = (h << i) | (l >> (BN_BITS2 - i));
258 dh=(d&BN_MASK2h)>>BN_BITS4; 283 l <<= i;
259 dl=(d&BN_MASK2l); 284 }
260 for (;;) 285 dh = (d & BN_MASK2h) >> BN_BITS4;
261 { 286 dl = (d & BN_MASK2l);
262 if ((h>>BN_BITS4) == dh) 287 for (;;) {
263 q=BN_MASK2l; 288 if ((h >> BN_BITS4) == dh)
289 q = BN_MASK2l;
264 else 290 else
265 q=h/dh; 291 q = h / dh;
266 292
267 th=q*dh; 293 th = q * dh;
268 tl=dl*q; 294 tl = dl * q;
269 for (;;) 295 for (;;) {
270 { 296 t = h - th;
271 t=h-th; 297 if ((t & BN_MASK2h) ||
272 if ((t&BN_MASK2h) || 298 ((tl) <= (
273 ((tl) <= ( 299 (t << BN_BITS4) |
274 (t<<BN_BITS4)| 300 ((l & BN_MASK2h) >> BN_BITS4))))
275 ((l&BN_MASK2h)>>BN_BITS4))))
276 break; 301 break;
277 q--; 302 q--;
278 th-=dh; 303 th -= dh;
279 tl-=dl; 304 tl -= dl;
280 } 305 }
281 t=(tl>>BN_BITS4); 306 t = (tl >> BN_BITS4);
282 tl=(tl<<BN_BITS4)&BN_MASK2h; 307 tl = (tl << BN_BITS4) & BN_MASK2h;
283 th+=t; 308 th += t;
284 309
285 if (l < tl) th++; 310 if (l < tl)
286 l-=tl; 311 th++;
287 if (h < th) 312 l -= tl;
288 { 313 if (h < th) {
289 h+=d; 314 h += d;
290 q--; 315 q--;
291 } 316 }
292 h-=th; 317 h -= th;
293 318
294 if (--count == 0) break; 319 if (--count == 0)
320 break;
295 321
296 ret=q<<BN_BITS4; 322 ret = q << BN_BITS4;
297 h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2; 323 h = ((h << BN_BITS4) | (l >> BN_BITS4)) & BN_MASK2;
298 l=(l&BN_MASK2l)<<BN_BITS4; 324 l = (l & BN_MASK2l) << BN_BITS4;
299 }
300 ret|=q;
301 return(ret);
302 } 325 }
326 ret |= q;
327 return (ret);
328}
303#endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */ 329#endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */
304 330
305#ifdef BN_LLONG 331#ifdef BN_LLONG
306BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n) 332BN_ULONG
307 { 333bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
308 BN_ULLONG ll=0; 334{
335 BN_ULLONG ll = 0;
309 336
310 assert(n >= 0); 337 assert(n >= 0);
311 if (n <= 0) return((BN_ULONG)0); 338 if (n <= 0)
339 return ((BN_ULONG)0);
312 340
313#ifndef OPENSSL_SMALL_FOOTPRINT 341#ifndef OPENSSL_SMALL_FOOTPRINT
314 while (n&~3) 342 while (n & ~3) {
315 { 343 ll += (BN_ULLONG)a[0] + b[0];
316 ll+=(BN_ULLONG)a[0]+b[0]; 344 r[0] = (BN_ULONG)ll & BN_MASK2;
317 r[0]=(BN_ULONG)ll&BN_MASK2; 345 ll >>= BN_BITS2;
318 ll>>=BN_BITS2; 346 ll += (BN_ULLONG)a[1] + b[1];
319 ll+=(BN_ULLONG)a[1]+b[1]; 347 r[1] = (BN_ULONG)ll & BN_MASK2;
320 r[1]=(BN_ULONG)ll&BN_MASK2; 348 ll >>= BN_BITS2;
321 ll>>=BN_BITS2; 349 ll += (BN_ULLONG)a[2] + b[2];
322 ll+=(BN_ULLONG)a[2]+b[2]; 350 r[2] = (BN_ULONG)ll & BN_MASK2;
323 r[2]=(BN_ULONG)ll&BN_MASK2; 351 ll >>= BN_BITS2;
324 ll>>=BN_BITS2; 352 ll += (BN_ULLONG)a[3] + b[3];
325 ll+=(BN_ULLONG)a[3]+b[3]; 353 r[3] = (BN_ULONG)ll & BN_MASK2;
326 r[3]=(BN_ULONG)ll&BN_MASK2; 354 ll >>= BN_BITS2;
327 ll>>=BN_BITS2; 355 a += 4;
328 a+=4; b+=4; r+=4; n-=4; 356 b += 4;
329 } 357 r += 4;
358 n -= 4;
359 }
330#endif 360#endif
331 while (n) 361 while (n) {
332 { 362 ll += (BN_ULLONG)a[0] + b[0];
333 ll+=(BN_ULLONG)a[0]+b[0]; 363 r[0] = (BN_ULONG)ll & BN_MASK2;
334 r[0]=(BN_ULONG)ll&BN_MASK2; 364 ll >>= BN_BITS2;
335 ll>>=BN_BITS2; 365 a++;
336 a++; b++; r++; n--; 366 b++;
337 } 367 r++;
338 return((BN_ULONG)ll); 368 n--;
339 } 369 }
370 return ((BN_ULONG)ll);
371}
340#else /* !BN_LLONG */ 372#else /* !BN_LLONG */
341BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n) 373BN_ULONG
342 { 374bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
343 BN_ULONG c,l,t; 375{
376 BN_ULONG c, l, t;
344 377
345 assert(n >= 0); 378 assert(n >= 0);
346 if (n <= 0) return((BN_ULONG)0); 379 if (n <= 0)
380 return ((BN_ULONG)0);
347 381
348 c=0; 382 c = 0;
349#ifndef OPENSSL_SMALL_FOOTPRINT 383#ifndef OPENSSL_SMALL_FOOTPRINT
350 while (n&~3) 384 while (n & ~3) {
351 { 385 t = a[0];
352 t=a[0]; 386 t = (t + c) & BN_MASK2;
353 t=(t+c)&BN_MASK2; 387 c = (t < c);
354 c=(t < c); 388 l = (t + b[0]) & BN_MASK2;
355 l=(t+b[0])&BN_MASK2; 389 c += (l < t);
356 c+=(l < t); 390 r[0] = l;
357 r[0]=l; 391 t = a[1];
358 t=a[1]; 392 t = (t + c) & BN_MASK2;
359 t=(t+c)&BN_MASK2; 393 c = (t < c);
360 c=(t < c); 394 l = (t + b[1]) & BN_MASK2;
361 l=(t+b[1])&BN_MASK2; 395 c += (l < t);
362 c+=(l < t); 396 r[1] = l;
363 r[1]=l; 397 t = a[2];
364 t=a[2]; 398 t = (t + c) & BN_MASK2;
365 t=(t+c)&BN_MASK2; 399 c = (t < c);
366 c=(t < c); 400 l = (t + b[2]) & BN_MASK2;
367 l=(t+b[2])&BN_MASK2; 401 c += (l < t);
368 c+=(l < t); 402 r[2] = l;
369 r[2]=l; 403 t = a[3];
370 t=a[3]; 404 t = (t + c) & BN_MASK2;
371 t=(t+c)&BN_MASK2; 405 c = (t < c);
372 c=(t < c); 406 l = (t + b[3]) & BN_MASK2;
373 l=(t+b[3])&BN_MASK2; 407 c += (l < t);
374 c+=(l < t); 408 r[3] = l;
375 r[3]=l; 409 a += 4;
376 a+=4; b+=4; r+=4; n-=4; 410 b += 4;
377 } 411 r += 4;
412 n -= 4;
413 }
378#endif 414#endif
379 while(n) 415 while (n) {
380 { 416 t = a[0];
381 t=a[0]; 417 t = (t + c) & BN_MASK2;
382 t=(t+c)&BN_MASK2; 418 c = (t < c);
383 c=(t < c); 419 l = (t + b[0]) & BN_MASK2;
384 l=(t+b[0])&BN_MASK2; 420 c += (l < t);
385 c+=(l < t); 421 r[0] = l;
386 r[0]=l; 422 a++;
387 a++; b++; r++; n--; 423 b++;
388 } 424 r++;
389 return((BN_ULONG)c); 425 n--;
390 } 426 }
427 return ((BN_ULONG)c);
428}
391#endif /* !BN_LLONG */ 429#endif /* !BN_LLONG */
392 430
393BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n) 431BN_ULONG
394 { 432bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
395 BN_ULONG t1,t2; 433{
396 int c=0; 434 BN_ULONG t1, t2;
435 int c = 0;
397 436
398 assert(n >= 0); 437 assert(n >= 0);
399 if (n <= 0) return((BN_ULONG)0); 438 if (n <= 0)
439 return ((BN_ULONG)0);
400 440
401#ifndef OPENSSL_SMALL_FOOTPRINT 441#ifndef OPENSSL_SMALL_FOOTPRINT
402 while (n&~3) 442 while (n&~3) {
403 { 443 t1 = a[0];
404 t1=a[0]; t2=b[0]; 444 t2 = b[0];
405 r[0]=(t1-t2-c)&BN_MASK2; 445 r[0] = (t1 - t2 - c) & BN_MASK2;
406 if (t1 != t2) c=(t1 < t2); 446 if (t1 != t2)
407 t1=a[1]; t2=b[1]; 447 c = (t1 < t2);
408 r[1]=(t1-t2-c)&BN_MASK2; 448 t1 = a[1];
409 if (t1 != t2) c=(t1 < t2); 449 t2 = b[1];
410 t1=a[2]; t2=b[2]; 450 r[1] = (t1 - t2 - c) & BN_MASK2;
411 r[2]=(t1-t2-c)&BN_MASK2; 451 if (t1 != t2)
412 if (t1 != t2) c=(t1 < t2); 452 c = (t1 < t2);
413 t1=a[3]; t2=b[3]; 453 t1 = a[2];
414 r[3]=(t1-t2-c)&BN_MASK2; 454 t2 = b[2];
415 if (t1 != t2) c=(t1 < t2); 455 r[2] = (t1 - t2 - c) & BN_MASK2;
416 a+=4; b+=4; r+=4; n-=4; 456 if (t1 != t2)
417 } 457 c = (t1 < t2);
458 t1 = a[3];
459 t2 = b[3];
460 r[3] = (t1 - t2 - c) & BN_MASK2;
461 if (t1 != t2)
462 c = (t1 < t2);
463 a += 4;
464 b += 4;
465 r += 4;
466 n -= 4;
467 }
418#endif 468#endif
419 while (n) 469 while (n) {
420 { 470 t1 = a[0];
421 t1=a[0]; t2=b[0]; 471 t2 = b[0];
422 r[0]=(t1-t2-c)&BN_MASK2; 472 r[0] = (t1 - t2 - c) & BN_MASK2;
423 if (t1 != t2) c=(t1 < t2); 473 if (t1 != t2)
424 a++; b++; r++; n--; 474 c = (t1 < t2);
425 } 475 a++;
426 return(c); 476 b++;
477 r++;
478 n--;
427 } 479 }
480 return (c);
481}
428 482
429#if defined(BN_MUL_COMBA) && !defined(OPENSSL_SMALL_FOOTPRINT) 483#if defined(BN_MUL_COMBA) && !defined(OPENSSL_SMALL_FOOTPRINT)
430 484
@@ -554,279 +608,283 @@ BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
554 mul_add_c2((a)[i],(a)[j],c0,c1,c2) 608 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
555#endif /* !BN_LLONG */ 609#endif /* !BN_LLONG */
556 610
557void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 611void
558 { 612bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
613{
559#ifdef BN_LLONG 614#ifdef BN_LLONG
560 BN_ULLONG t; 615 BN_ULLONG t;
561#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH) 616#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH)
562 BN_ULONG bl,bh; 617 BN_ULONG bl, bh;
563#endif 618#endif
564 BN_ULONG t1,t2; 619 BN_ULONG t1, t2;
565 BN_ULONG c1,c2,c3; 620 BN_ULONG c1, c2, c3;
566 621
567 c1=0; 622 c1 = 0;
568 c2=0; 623 c2 = 0;
569 c3=0; 624 c3 = 0;
570 mul_add_c(a[0],b[0],c1,c2,c3); 625 mul_add_c(a[0], b[0], c1, c2, c3);
571 r[0]=c1; 626 r[0] = c1;
572 c1=0; 627 c1 = 0;
573 mul_add_c(a[0],b[1],c2,c3,c1); 628 mul_add_c(a[0], b[1], c2, c3, c1);
574 mul_add_c(a[1],b[0],c2,c3,c1); 629 mul_add_c(a[1], b[0], c2, c3, c1);
575 r[1]=c2; 630 r[1] = c2;
576 c2=0; 631 c2 = 0;
577 mul_add_c(a[2],b[0],c3,c1,c2); 632 mul_add_c(a[2], b[0], c3, c1, c2);
578 mul_add_c(a[1],b[1],c3,c1,c2); 633 mul_add_c(a[1], b[1], c3, c1, c2);
579 mul_add_c(a[0],b[2],c3,c1,c2); 634 mul_add_c(a[0], b[2], c3, c1, c2);
580 r[2]=c3; 635 r[2] = c3;
581 c3=0; 636 c3 = 0;
582 mul_add_c(a[0],b[3],c1,c2,c3); 637 mul_add_c(a[0], b[3], c1, c2, c3);
583 mul_add_c(a[1],b[2],c1,c2,c3); 638 mul_add_c(a[1], b[2], c1, c2, c3);
584 mul_add_c(a[2],b[1],c1,c2,c3); 639 mul_add_c(a[2], b[1], c1, c2, c3);
585 mul_add_c(a[3],b[0],c1,c2,c3); 640 mul_add_c(a[3], b[0], c1, c2, c3);
586 r[3]=c1; 641 r[3] = c1;
587 c1=0; 642 c1 = 0;
588 mul_add_c(a[4],b[0],c2,c3,c1); 643 mul_add_c(a[4], b[0], c2, c3, c1);
589 mul_add_c(a[3],b[1],c2,c3,c1); 644 mul_add_c(a[3], b[1], c2, c3, c1);
590 mul_add_c(a[2],b[2],c2,c3,c1); 645 mul_add_c(a[2], b[2], c2, c3, c1);
591 mul_add_c(a[1],b[3],c2,c3,c1); 646 mul_add_c(a[1], b[3], c2, c3, c1);
592 mul_add_c(a[0],b[4],c2,c3,c1); 647 mul_add_c(a[0], b[4], c2, c3, c1);
593 r[4]=c2; 648 r[4] = c2;
594 c2=0; 649 c2 = 0;
595 mul_add_c(a[0],b[5],c3,c1,c2); 650 mul_add_c(a[0], b[5], c3, c1, c2);
596 mul_add_c(a[1],b[4],c3,c1,c2); 651 mul_add_c(a[1], b[4], c3, c1, c2);
597 mul_add_c(a[2],b[3],c3,c1,c2); 652 mul_add_c(a[2], b[3], c3, c1, c2);
598 mul_add_c(a[3],b[2],c3,c1,c2); 653 mul_add_c(a[3], b[2], c3, c1, c2);
599 mul_add_c(a[4],b[1],c3,c1,c2); 654 mul_add_c(a[4], b[1], c3, c1, c2);
600 mul_add_c(a[5],b[0],c3,c1,c2); 655 mul_add_c(a[5], b[0], c3, c1, c2);
601 r[5]=c3; 656 r[5] = c3;
602 c3=0; 657 c3 = 0;
603 mul_add_c(a[6],b[0],c1,c2,c3); 658 mul_add_c(a[6], b[0], c1, c2, c3);
604 mul_add_c(a[5],b[1],c1,c2,c3); 659 mul_add_c(a[5], b[1], c1, c2, c3);
605 mul_add_c(a[4],b[2],c1,c2,c3); 660 mul_add_c(a[4], b[2], c1, c2, c3);
606 mul_add_c(a[3],b[3],c1,c2,c3); 661 mul_add_c(a[3], b[3], c1, c2, c3);
607 mul_add_c(a[2],b[4],c1,c2,c3); 662 mul_add_c(a[2], b[4], c1, c2, c3);
608 mul_add_c(a[1],b[5],c1,c2,c3); 663 mul_add_c(a[1], b[5], c1, c2, c3);
609 mul_add_c(a[0],b[6],c1,c2,c3); 664 mul_add_c(a[0], b[6], c1, c2, c3);
610 r[6]=c1; 665 r[6] = c1;
611 c1=0; 666 c1 = 0;
612 mul_add_c(a[0],b[7],c2,c3,c1); 667 mul_add_c(a[0], b[7], c2, c3, c1);
613 mul_add_c(a[1],b[6],c2,c3,c1); 668 mul_add_c(a[1], b[6], c2, c3, c1);
614 mul_add_c(a[2],b[5],c2,c3,c1); 669 mul_add_c(a[2], b[5], c2, c3, c1);
615 mul_add_c(a[3],b[4],c2,c3,c1); 670 mul_add_c(a[3], b[4], c2, c3, c1);
616 mul_add_c(a[4],b[3],c2,c3,c1); 671 mul_add_c(a[4], b[3], c2, c3, c1);
617 mul_add_c(a[5],b[2],c2,c3,c1); 672 mul_add_c(a[5], b[2], c2, c3, c1);
618 mul_add_c(a[6],b[1],c2,c3,c1); 673 mul_add_c(a[6], b[1], c2, c3, c1);
619 mul_add_c(a[7],b[0],c2,c3,c1); 674 mul_add_c(a[7], b[0], c2, c3, c1);
620 r[7]=c2; 675 r[7] = c2;
621 c2=0; 676 c2 = 0;
622 mul_add_c(a[7],b[1],c3,c1,c2); 677 mul_add_c(a[7], b[1], c3, c1, c2);
623 mul_add_c(a[6],b[2],c3,c1,c2); 678 mul_add_c(a[6], b[2], c3, c1, c2);
624 mul_add_c(a[5],b[3],c3,c1,c2); 679 mul_add_c(a[5], b[3], c3, c1, c2);
625 mul_add_c(a[4],b[4],c3,c1,c2); 680 mul_add_c(a[4], b[4], c3, c1, c2);
626 mul_add_c(a[3],b[5],c3,c1,c2); 681 mul_add_c(a[3], b[5], c3, c1, c2);
627 mul_add_c(a[2],b[6],c3,c1,c2); 682 mul_add_c(a[2], b[6], c3, c1, c2);
628 mul_add_c(a[1],b[7],c3,c1,c2); 683 mul_add_c(a[1], b[7], c3, c1, c2);
629 r[8]=c3; 684 r[8] = c3;
630 c3=0; 685 c3 = 0;
631 mul_add_c(a[2],b[7],c1,c2,c3); 686 mul_add_c(a[2], b[7], c1, c2, c3);
632 mul_add_c(a[3],b[6],c1,c2,c3); 687 mul_add_c(a[3], b[6], c1, c2, c3);
633 mul_add_c(a[4],b[5],c1,c2,c3); 688 mul_add_c(a[4], b[5], c1, c2, c3);
634 mul_add_c(a[5],b[4],c1,c2,c3); 689 mul_add_c(a[5], b[4], c1, c2, c3);
635 mul_add_c(a[6],b[3],c1,c2,c3); 690 mul_add_c(a[6], b[3], c1, c2, c3);
636 mul_add_c(a[7],b[2],c1,c2,c3); 691 mul_add_c(a[7], b[2], c1, c2, c3);
637 r[9]=c1; 692 r[9] = c1;
638 c1=0; 693 c1 = 0;
639 mul_add_c(a[7],b[3],c2,c3,c1); 694 mul_add_c(a[7], b[3], c2, c3, c1);
640 mul_add_c(a[6],b[4],c2,c3,c1); 695 mul_add_c(a[6], b[4], c2, c3, c1);
641 mul_add_c(a[5],b[5],c2,c3,c1); 696 mul_add_c(a[5], b[5], c2, c3, c1);
642 mul_add_c(a[4],b[6],c2,c3,c1); 697 mul_add_c(a[4], b[6], c2, c3, c1);
643 mul_add_c(a[3],b[7],c2,c3,c1); 698 mul_add_c(a[3], b[7], c2, c3, c1);
644 r[10]=c2; 699 r[10] = c2;
645 c2=0; 700 c2 = 0;
646 mul_add_c(a[4],b[7],c3,c1,c2); 701 mul_add_c(a[4], b[7], c3, c1, c2);
647 mul_add_c(a[5],b[6],c3,c1,c2); 702 mul_add_c(a[5], b[6], c3, c1, c2);
648 mul_add_c(a[6],b[5],c3,c1,c2); 703 mul_add_c(a[6], b[5], c3, c1, c2);
649 mul_add_c(a[7],b[4],c3,c1,c2); 704 mul_add_c(a[7], b[4], c3, c1, c2);
650 r[11]=c3; 705 r[11] = c3;
651 c3=0; 706 c3 = 0;
652 mul_add_c(a[7],b[5],c1,c2,c3); 707 mul_add_c(a[7], b[5], c1, c2, c3);
653 mul_add_c(a[6],b[6],c1,c2,c3); 708 mul_add_c(a[6], b[6], c1, c2, c3);
654 mul_add_c(a[5],b[7],c1,c2,c3); 709 mul_add_c(a[5], b[7], c1, c2, c3);
655 r[12]=c1; 710 r[12] = c1;
656 c1=0; 711 c1 = 0;
657 mul_add_c(a[6],b[7],c2,c3,c1); 712 mul_add_c(a[6], b[7], c2, c3, c1);
658 mul_add_c(a[7],b[6],c2,c3,c1); 713 mul_add_c(a[7], b[6], c2, c3, c1);
659 r[13]=c2; 714 r[13] = c2;
660 c2=0; 715 c2 = 0;
661 mul_add_c(a[7],b[7],c3,c1,c2); 716 mul_add_c(a[7], b[7], c3, c1, c2);
662 r[14]=c3; 717 r[14] = c3;
663 r[15]=c1; 718 r[15] = c1;
664 } 719}
665 720
666void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 721void
667 { 722bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
723{
668#ifdef BN_LLONG 724#ifdef BN_LLONG
669 BN_ULLONG t; 725 BN_ULLONG t;
670#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH) 726#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH)
671 BN_ULONG bl,bh; 727 BN_ULONG bl, bh;
672#endif 728#endif
673 BN_ULONG t1,t2; 729 BN_ULONG t1, t2;
674 BN_ULONG c1,c2,c3; 730 BN_ULONG c1, c2, c3;
675 731
676 c1=0; 732 c1 = 0;
677 c2=0; 733 c2 = 0;
678 c3=0; 734 c3 = 0;
679 mul_add_c(a[0],b[0],c1,c2,c3); 735 mul_add_c(a[0], b[0], c1, c2, c3);
680 r[0]=c1; 736 r[0] = c1;
681 c1=0; 737 c1 = 0;
682 mul_add_c(a[0],b[1],c2,c3,c1); 738 mul_add_c(a[0], b[1], c2, c3, c1);
683 mul_add_c(a[1],b[0],c2,c3,c1); 739 mul_add_c(a[1], b[0], c2, c3, c1);
684 r[1]=c2; 740 r[1] = c2;
685 c2=0; 741 c2 = 0;
686 mul_add_c(a[2],b[0],c3,c1,c2); 742 mul_add_c(a[2], b[0], c3, c1, c2);
687 mul_add_c(a[1],b[1],c3,c1,c2); 743 mul_add_c(a[1], b[1], c3, c1, c2);
688 mul_add_c(a[0],b[2],c3,c1,c2); 744 mul_add_c(a[0], b[2], c3, c1, c2);
689 r[2]=c3; 745 r[2] = c3;
690 c3=0; 746 c3 = 0;
691 mul_add_c(a[0],b[3],c1,c2,c3); 747 mul_add_c(a[0], b[3], c1, c2, c3);
692 mul_add_c(a[1],b[2],c1,c2,c3); 748 mul_add_c(a[1], b[2], c1, c2, c3);
693 mul_add_c(a[2],b[1],c1,c2,c3); 749 mul_add_c(a[2], b[1], c1, c2, c3);
694 mul_add_c(a[3],b[0],c1,c2,c3); 750 mul_add_c(a[3], b[0], c1, c2, c3);
695 r[3]=c1; 751 r[3] = c1;
696 c1=0; 752 c1 = 0;
697 mul_add_c(a[3],b[1],c2,c3,c1); 753 mul_add_c(a[3], b[1], c2, c3, c1);
698 mul_add_c(a[2],b[2],c2,c3,c1); 754 mul_add_c(a[2], b[2], c2, c3, c1);
699 mul_add_c(a[1],b[3],c2,c3,c1); 755 mul_add_c(a[1], b[3], c2, c3, c1);
700 r[4]=c2; 756 r[4] = c2;
701 c2=0; 757 c2 = 0;
702 mul_add_c(a[2],b[3],c3,c1,c2); 758 mul_add_c(a[2], b[3], c3, c1, c2);
703 mul_add_c(a[3],b[2],c3,c1,c2); 759 mul_add_c(a[3], b[2], c3, c1, c2);
704 r[5]=c3; 760 r[5] = c3;
705 c3=0; 761 c3 = 0;
706 mul_add_c(a[3],b[3],c1,c2,c3); 762 mul_add_c(a[3], b[3], c1, c2, c3);
707 r[6]=c1; 763 r[6] = c1;
708 r[7]=c2; 764 r[7] = c2;
709 } 765}
710 766
711void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a) 767void
712 { 768bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
769{
713#ifdef BN_LLONG 770#ifdef BN_LLONG
714 BN_ULLONG t,tt; 771 BN_ULLONG t, tt;
715#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH) 772#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH)
716 BN_ULONG bl,bh; 773 BN_ULONG bl, bh;
717#endif 774#endif
718 BN_ULONG t1,t2; 775 BN_ULONG t1, t2;
719 BN_ULONG c1,c2,c3; 776 BN_ULONG c1, c2, c3;
720 777
721 c1=0; 778 c1 = 0;
722 c2=0; 779 c2 = 0;
723 c3=0; 780 c3 = 0;
724 sqr_add_c(a,0,c1,c2,c3); 781 sqr_add_c(a, 0, c1, c2, c3);
725 r[0]=c1; 782 r[0] = c1;
726 c1=0; 783 c1 = 0;
727 sqr_add_c2(a,1,0,c2,c3,c1); 784 sqr_add_c2(a, 1, 0, c2, c3, c1);
728 r[1]=c2; 785 r[1] = c2;
729 c2=0; 786 c2 = 0;
730 sqr_add_c(a,1,c3,c1,c2); 787 sqr_add_c(a, 1, c3, c1, c2);
731 sqr_add_c2(a,2,0,c3,c1,c2); 788 sqr_add_c2(a, 2, 0, c3, c1, c2);
732 r[2]=c3; 789 r[2] = c3;
733 c3=0; 790 c3 = 0;
734 sqr_add_c2(a,3,0,c1,c2,c3); 791 sqr_add_c2(a, 3, 0, c1, c2, c3);
735 sqr_add_c2(a,2,1,c1,c2,c3); 792 sqr_add_c2(a, 2, 1, c1, c2, c3);
736 r[3]=c1; 793 r[3] = c1;
737 c1=0; 794 c1 = 0;
738 sqr_add_c(a,2,c2,c3,c1); 795 sqr_add_c(a, 2, c2, c3, c1);
739 sqr_add_c2(a,3,1,c2,c3,c1); 796 sqr_add_c2(a, 3, 1, c2, c3, c1);
740 sqr_add_c2(a,4,0,c2,c3,c1); 797 sqr_add_c2(a, 4, 0, c2, c3, c1);
741 r[4]=c2; 798 r[4] = c2;
742 c2=0; 799 c2 = 0;
743 sqr_add_c2(a,5,0,c3,c1,c2); 800 sqr_add_c2(a, 5, 0, c3, c1, c2);
744 sqr_add_c2(a,4,1,c3,c1,c2); 801 sqr_add_c2(a, 4, 1, c3, c1, c2);
745 sqr_add_c2(a,3,2,c3,c1,c2); 802 sqr_add_c2(a, 3, 2, c3, c1, c2);
746 r[5]=c3; 803 r[5] = c3;
747 c3=0; 804 c3 = 0;
748 sqr_add_c(a,3,c1,c2,c3); 805 sqr_add_c(a, 3, c1, c2, c3);
749 sqr_add_c2(a,4,2,c1,c2,c3); 806 sqr_add_c2(a, 4, 2, c1, c2, c3);
750 sqr_add_c2(a,5,1,c1,c2,c3); 807 sqr_add_c2(a, 5, 1, c1, c2, c3);
751 sqr_add_c2(a,6,0,c1,c2,c3); 808 sqr_add_c2(a, 6, 0, c1, c2, c3);
752 r[6]=c1; 809 r[6] = c1;
753 c1=0; 810 c1 = 0;
754 sqr_add_c2(a,7,0,c2,c3,c1); 811 sqr_add_c2(a, 7, 0, c2, c3, c1);
755 sqr_add_c2(a,6,1,c2,c3,c1); 812 sqr_add_c2(a, 6, 1, c2, c3, c1);
756 sqr_add_c2(a,5,2,c2,c3,c1); 813 sqr_add_c2(a, 5, 2, c2, c3, c1);
757 sqr_add_c2(a,4,3,c2,c3,c1); 814 sqr_add_c2(a, 4, 3, c2, c3, c1);
758 r[7]=c2; 815 r[7] = c2;
759 c2=0; 816 c2 = 0;
760 sqr_add_c(a,4,c3,c1,c2); 817 sqr_add_c(a, 4, c3, c1, c2);
761 sqr_add_c2(a,5,3,c3,c1,c2); 818 sqr_add_c2(a, 5, 3, c3, c1, c2);
762 sqr_add_c2(a,6,2,c3,c1,c2); 819 sqr_add_c2(a, 6, 2, c3, c1, c2);
763 sqr_add_c2(a,7,1,c3,c1,c2); 820 sqr_add_c2(a, 7, 1, c3, c1, c2);
764 r[8]=c3; 821 r[8] = c3;
765 c3=0; 822 c3 = 0;
766 sqr_add_c2(a,7,2,c1,c2,c3); 823 sqr_add_c2(a, 7, 2, c1, c2, c3);
767 sqr_add_c2(a,6,3,c1,c2,c3); 824 sqr_add_c2(a, 6, 3, c1, c2, c3);
768 sqr_add_c2(a,5,4,c1,c2,c3); 825 sqr_add_c2(a, 5, 4, c1, c2, c3);
769 r[9]=c1; 826 r[9] = c1;
770 c1=0; 827 c1 = 0;
771 sqr_add_c(a,5,c2,c3,c1); 828 sqr_add_c(a, 5, c2, c3, c1);
772 sqr_add_c2(a,6,4,c2,c3,c1); 829 sqr_add_c2(a, 6, 4, c2, c3, c1);
773 sqr_add_c2(a,7,3,c2,c3,c1); 830 sqr_add_c2(a, 7, 3, c2, c3, c1);
774 r[10]=c2; 831 r[10] = c2;
775 c2=0; 832 c2 = 0;
776 sqr_add_c2(a,7,4,c3,c1,c2); 833 sqr_add_c2(a, 7, 4, c3, c1, c2);
777 sqr_add_c2(a,6,5,c3,c1,c2); 834 sqr_add_c2(a, 6, 5, c3, c1, c2);
778 r[11]=c3; 835 r[11] = c3;
779 c3=0; 836 c3 = 0;
780 sqr_add_c(a,6,c1,c2,c3); 837 sqr_add_c(a, 6, c1, c2, c3);
781 sqr_add_c2(a,7,5,c1,c2,c3); 838 sqr_add_c2(a, 7, 5, c1, c2, c3);
782 r[12]=c1; 839 r[12] = c1;
783 c1=0; 840 c1 = 0;
784 sqr_add_c2(a,7,6,c2,c3,c1); 841 sqr_add_c2(a, 7, 6, c2, c3, c1);
785 r[13]=c2; 842 r[13] = c2;
786 c2=0; 843 c2 = 0;
787 sqr_add_c(a,7,c3,c1,c2); 844 sqr_add_c(a, 7, c3, c1, c2);
788 r[14]=c3; 845 r[14] = c3;
789 r[15]=c1; 846 r[15] = c1;
790 } 847}
791 848
792void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a) 849void
793 { 850bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
851{
794#ifdef BN_LLONG 852#ifdef BN_LLONG
795 BN_ULLONG t,tt; 853 BN_ULLONG t, tt;
796#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH) 854#elif !defined(BN_UMULT_LOHI) && !defined(BN_UMULT_HIGH)
797 BN_ULONG bl,bh; 855 BN_ULONG bl, bh;
798#endif 856#endif
799 BN_ULONG t1,t2; 857 BN_ULONG t1, t2;
800 BN_ULONG c1,c2,c3; 858 BN_ULONG c1, c2, c3;
801 859
802 c1=0; 860 c1 = 0;
803 c2=0; 861 c2 = 0;
804 c3=0; 862 c3 = 0;
805 sqr_add_c(a,0,c1,c2,c3); 863 sqr_add_c(a, 0, c1, c2, c3);
806 r[0]=c1; 864 r[0] = c1;
807 c1=0; 865 c1 = 0;
808 sqr_add_c2(a,1,0,c2,c3,c1); 866 sqr_add_c2(a, 1, 0, c2, c3, c1);
809 r[1]=c2; 867 r[1] = c2;
810 c2=0; 868 c2 = 0;
811 sqr_add_c(a,1,c3,c1,c2); 869 sqr_add_c(a, 1, c3, c1, c2);
812 sqr_add_c2(a,2,0,c3,c1,c2); 870 sqr_add_c2(a, 2, 0, c3, c1, c2);
813 r[2]=c3; 871 r[2] = c3;
814 c3=0; 872 c3 = 0;
815 sqr_add_c2(a,3,0,c1,c2,c3); 873 sqr_add_c2(a, 3, 0, c1, c2, c3);
816 sqr_add_c2(a,2,1,c1,c2,c3); 874 sqr_add_c2(a, 2, 1, c1, c2, c3);
817 r[3]=c1; 875 r[3] = c1;
818 c1=0; 876 c1 = 0;
819 sqr_add_c(a,2,c2,c3,c1); 877 sqr_add_c(a, 2, c2, c3, c1);
820 sqr_add_c2(a,3,1,c2,c3,c1); 878 sqr_add_c2(a, 3, 1, c2, c3, c1);
821 r[4]=c2; 879 r[4] = c2;
822 c2=0; 880 c2 = 0;
823 sqr_add_c2(a,3,2,c3,c1,c2); 881 sqr_add_c2(a, 3, 2, c3, c1, c2);
824 r[5]=c3; 882 r[5] = c3;
825 c3=0; 883 c3 = 0;
826 sqr_add_c(a,3,c1,c2,c3); 884 sqr_add_c(a, 3, c1, c2, c3);
827 r[6]=c1; 885 r[6] = c1;
828 r[7]=c2; 886 r[7] = c2;
829 } 887}
830 888
831#ifdef OPENSSL_NO_ASM 889#ifdef OPENSSL_NO_ASM
832#ifdef OPENSSL_BN_ASM_MONT 890#ifdef OPENSSL_BN_ASM_MONT
@@ -844,19 +902,21 @@ void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
844 * versions. Assembler vs. assembler improvement coefficients can 902 * versions. Assembler vs. assembler improvement coefficients can
845 * [and are known to] differ and are to be documented elsewhere. 903 * [and are known to] differ and are to be documented elsewhere.
846 */ 904 */
847int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0p, int num) 905int
848 { 906bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np, const BN_ULONG *n0p, int num)
849 BN_ULONG c0,c1,ml,*tp,n0; 907{
908 BN_ULONG c0, c1, ml, *tp, n0;
850#ifdef mul64 909#ifdef mul64
851 BN_ULONG mh; 910 BN_ULONG mh;
852#endif 911#endif
853 volatile BN_ULONG *vp; 912 volatile BN_ULONG *vp;
854 int i=0,j; 913 int i = 0, j;
855 914
856#if 0 /* template for platform-specific implementation */ 915#if 0 /* template for platform-specific implementation */
857 if (ap==bp) return bn_sqr_mont(rp,ap,np,n0p,num); 916 if (ap == bp)
917 return bn_sqr_mont(rp, ap, np, n0p, num);
858#endif 918#endif
859 vp = tp = alloca((num+2)*sizeof(BN_ULONG)); 919 vp = tp = alloca((num + 2)*sizeof(BN_ULONG));
860 920
861 n0 = *n0p; 921 n0 = *n0p;
862 922
@@ -865,81 +925,80 @@ int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_U
865#ifdef mul64 925#ifdef mul64
866 mh = HBITS(ml); 926 mh = HBITS(ml);
867 ml = LBITS(ml); 927 ml = LBITS(ml);
868 for (j=0;j<num;++j) 928 for (j = 0; j < num; ++j)
869 mul(tp[j],ap[j],ml,mh,c0); 929 mul(tp[j], ap[j], ml, mh, c0);
870#else 930#else
871 for (j=0;j<num;++j) 931 for (j = 0; j < num; ++j)
872 mul(tp[j],ap[j],ml,c0); 932 mul(tp[j], ap[j], ml, c0);
873#endif 933#endif
874 934
875 tp[num] = c0; 935 tp[num] = c0;
876 tp[num+1] = 0; 936 tp[num + 1] = 0;
877 goto enter; 937 goto enter;
878 938
879 for(i=0;i<num;i++) 939 for (i = 0; i < num; i++) {
880 {
881 c0 = 0; 940 c0 = 0;
882 ml = bp[i]; 941 ml = bp[i];
883#ifdef mul64 942#ifdef mul64
884 mh = HBITS(ml); 943 mh = HBITS(ml);
885 ml = LBITS(ml); 944 ml = LBITS(ml);
886 for (j=0;j<num;++j) 945 for (j = 0; j < num; ++j)
887 mul_add(tp[j],ap[j],ml,mh,c0); 946 mul_add(tp[j], ap[j], ml, mh, c0);
888#else 947#else
889 for (j=0;j<num;++j) 948 for (j = 0; j < num; ++j)
890 mul_add(tp[j],ap[j],ml,c0); 949 mul_add(tp[j], ap[j], ml, c0);
891#endif 950#endif
892 c1 = (tp[num] + c0)&BN_MASK2; 951 c1 = (tp[num] + c0) & BN_MASK2;
893 tp[num] = c1; 952 tp[num] = c1;
894 tp[num+1] = (c1<c0?1:0); 953 tp[num + 1] = (c1 < c0 ? 1 : 0);
895 enter: 954enter:
896 c1 = tp[0]; 955 c1 = tp[0];
897 ml = (c1*n0)&BN_MASK2; 956 ml = (c1 * n0) & BN_MASK2;
898 c0 = 0; 957 c0 = 0;
899#ifdef mul64 958#ifdef mul64
900 mh = HBITS(ml); 959 mh = HBITS(ml);
901 ml = LBITS(ml); 960 ml = LBITS(ml);
902 mul_add(c1,np[0],ml,mh,c0); 961 mul_add(c1, np[0], ml, mh, c0);
903#else 962#else
904 mul_add(c1,ml,np[0],c0); 963 mul_add(c1, ml, np[0], c0);
905#endif 964#endif
906 for(j=1;j<num;j++) 965 for (j = 1; j < num; j++) {
907 {
908 c1 = tp[j]; 966 c1 = tp[j];
909#ifdef mul64 967#ifdef mul64
910 mul_add(c1,np[j],ml,mh,c0); 968 mul_add(c1, np[j], ml, mh, c0);
911#else 969#else
912 mul_add(c1,ml,np[j],c0); 970 mul_add(c1, ml, np[j], c0);
913#endif 971#endif
914 tp[j-1] = c1&BN_MASK2; 972 tp[j - 1] = c1 & BN_MASK2;
915 }
916 c1 = (tp[num] + c0)&BN_MASK2;
917 tp[num-1] = c1;
918 tp[num] = tp[num+1] + (c1<c0?1:0);
919 } 973 }
974 c1 = (tp[num] + c0) & BN_MASK2;
975 tp[num - 1] = c1;
976 tp[num] = tp[num + 1] + (c1 < c0 ? 1 : 0);
977 }
920 978
921 if (tp[num]!=0 || tp[num-1]>=np[num-1]) 979 if (tp[num] != 0 || tp[num - 1] >= np[num - 1]) {
922 { 980 c0 = bn_sub_words(rp, tp, np, num);
923 c0 = bn_sub_words(rp,tp,np,num); 981 if (tp[num] != 0 || c0 == 0) {
924 if (tp[num]!=0 || c0==0) 982 for (i = 0; i < num + 2; i++)
925 { 983 vp[i] = 0;
926 for(i=0;i<num+2;i++) vp[i] = 0;
927 return 1; 984 return 1;
928 }
929 } 985 }
930 for(i=0;i<num;i++) rp[i] = tp[i], vp[i] = 0;
931 vp[num] = 0;
932 vp[num+1] = 0;
933 return 1;
934 } 986 }
987 for (i = 0; i < num; i++)
988 rp[i] = tp[i], vp[i] = 0;
989 vp[num] = 0;
990 vp[num + 1] = 0;
991 return 1;
992}
935#else 993#else
936/* 994/*
937 * Return value of 0 indicates that multiplication/convolution was not 995 * Return value of 0 indicates that multiplication/convolution was not
938 * performed to signal the caller to fall down to alternative/original 996 * performed to signal the caller to fall down to alternative/original
939 * code-path. 997 * code-path.
940 */ 998 */
941int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num) 999int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np, const BN_ULONG *n0, int num)
942{ return 0; } 1000 { return 0;
1001}
943#endif /* OPENSSL_BN_ASM_MONT */ 1002#endif /* OPENSSL_BN_ASM_MONT */
944#endif 1003#endif
945 1004
@@ -947,83 +1006,94 @@ int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_U
947 1006
948/* hmm... is it faster just to do a multiply? */ 1007/* hmm... is it faster just to do a multiply? */
949#undef bn_sqr_comba4 1008#undef bn_sqr_comba4
950void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a) 1009void
951 { 1010bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
1011{
952 BN_ULONG t[8]; 1012 BN_ULONG t[8];
953 bn_sqr_normal(r,a,4,t); 1013 bn_sqr_normal(r, a, 4, t);
954 } 1014}
955 1015
956#undef bn_sqr_comba8 1016#undef bn_sqr_comba8
957void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a) 1017void
958 { 1018bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
1019{
959 BN_ULONG t[16]; 1020 BN_ULONG t[16];
960 bn_sqr_normal(r,a,8,t); 1021 bn_sqr_normal(r, a, 8, t);
961 } 1022}
962 1023
963void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 1024void
964 { 1025bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
965 r[4]=bn_mul_words( &(r[0]),a,4,b[0]); 1026{
966 r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]); 1027 r[4] = bn_mul_words(&(r[0]), a, 4, b[0]);
967 r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]); 1028 r[5] = bn_mul_add_words(&(r[1]), a, 4, b[1]);
968 r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]); 1029 r[6] = bn_mul_add_words(&(r[2]), a, 4, b[2]);
969 } 1030 r[7] = bn_mul_add_words(&(r[3]), a, 4, b[3]);
970 1031}
971void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 1032
972 { 1033void
973 r[ 8]=bn_mul_words( &(r[0]),a,8,b[0]); 1034bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
974 r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]); 1035{
975 r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]); 1036 r[8] = bn_mul_words(&(r[0]), a, 8, b[0]);
976 r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]); 1037 r[9] = bn_mul_add_words(&(r[1]), a, 8, b[1]);
977 r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]); 1038 r[10] = bn_mul_add_words(&(r[2]), a, 8, b[2]);
978 r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]); 1039 r[11] = bn_mul_add_words(&(r[3]), a, 8, b[3]);
979 r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]); 1040 r[12] = bn_mul_add_words(&(r[4]), a, 8, b[4]);
980 r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]); 1041 r[13] = bn_mul_add_words(&(r[5]), a, 8, b[5]);
981 } 1042 r[14] = bn_mul_add_words(&(r[6]), a, 8, b[6]);
1043 r[15] = bn_mul_add_words(&(r[7]), a, 8, b[7]);
1044}
982 1045
983#ifdef OPENSSL_NO_ASM 1046#ifdef OPENSSL_NO_ASM
984#ifdef OPENSSL_BN_ASM_MONT 1047#ifdef OPENSSL_BN_ASM_MONT
985#include <alloca.h> 1048#include <alloca.h>
986int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0p, int num) 1049int
987 { 1050bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
988 BN_ULONG c0,c1,*tp,n0=*n0p; 1051 const BN_ULONG *np, const BN_ULONG *n0p, int num)
1052{
1053 BN_ULONG c0, c1, *tp, n0 = *n0p;
989 volatile BN_ULONG *vp; 1054 volatile BN_ULONG *vp;
990 int i=0,j; 1055 int i = 0, j;
991 1056
992 vp = tp = alloca((num+2)*sizeof(BN_ULONG)); 1057 vp = tp = alloca((num + 2) * sizeof(BN_ULONG));
993 1058
994 for(i=0;i<=num;i++) tp[i]=0; 1059 for(i = 0; i <= num; i++)
1060 tp[i] = 0;
995 1061
996 for(i=0;i<num;i++) 1062 for (i = 0; i < num; i++) {
997 { 1063 c0 = bn_mul_add_words(tp, ap, num, bp[i]);
998 c0 = bn_mul_add_words(tp,ap,num,bp[i]); 1064 c1 = (tp[num] + c0) & BN_MASK2;
999 c1 = (tp[num] + c0)&BN_MASK2; 1065 tp[num] = c1;
1000 tp[num] = c1; 1066 tp[num + 1] = (c1 < c0 ? 1 : 0);
1001 tp[num+1] = (c1<c0?1:0);
1002 1067
1003 c0 = bn_mul_add_words(tp,np,num,tp[0]*n0); 1068 c0 = bn_mul_add_words(tp, np, num, tp[0] * n0);
1004 c1 = (tp[num] + c0)&BN_MASK2; 1069 c1 = (tp[num] + c0) & BN_MASK2;
1005 tp[num] = c1; 1070 tp[num] = c1;
1006 tp[num+1] += (c1<c0?1:0); 1071 tp[num + 1] += (c1 < c0 ? 1 : 0);
1007 for(j=0;j<=num;j++) tp[j]=tp[j+1]; 1072 for (j = 0; j <= num; j++)
1008 } 1073 tp[j] = tp[j + 1];
1074 }
1009 1075
1010 if (tp[num]!=0 || tp[num-1]>=np[num-1]) 1076 if (tp[num] != 0 || tp[num - 1] >= np[num - 1]) {
1011 { 1077 c0 = bn_sub_words(rp, tp, np, num);
1012 c0 = bn_sub_words(rp,tp,np,num); 1078 if (tp[num] != 0 || c0 == 0) {
1013 if (tp[num]!=0 || c0==0) 1079 for (i = 0; i < num + 2; i++)
1014 { 1080 vp[i] = 0;
1015 for(i=0;i<num+2;i++) vp[i] = 0;
1016 return 1; 1081 return 1;
1017 }
1018 } 1082 }
1019 for(i=0;i<num;i++) rp[i] = tp[i], vp[i] = 0;
1020 vp[num] = 0;
1021 vp[num+1] = 0;
1022 return 1;
1023 } 1083 }
1084 for (i = 0; i < num; i++)
1085 rp[i] = tp[i], vp[i] = 0;
1086 vp[num] = 0;
1087 vp[num + 1] = 0;
1088 return 1;
1089}
1024#else 1090#else
1025int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num) 1091int
1026{ return 0; } 1092bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
1093 const BN_ULONG *np, const BN_ULONG *n0, int num)
1094{
1095 return 0;
1096}
1027#endif /* OPENSSL_BN_ASM_MONT */ 1097#endif /* OPENSSL_BN_ASM_MONT */
1028#endif 1098#endif
1029 1099