1 /***********************************************************************
2 **
3 ** Implementation of the Skein block functions.
4 **
5 ** Source code author: Doug Whiting, 2008.
6 **
7 ** This algorithm and source code is released to the public domain.
8 **
9 ** Compile-time switches:
10 **
11 ** SKEIN_USE_ASM -- set bits (256/512/1024) to select which
12 ** versions use ASM code for block processing
13 ** [default: use C for all block sizes]
14 **
15 ************************************************************************/
16
17 #include <sys/cdefs.h>
18 __FBSDID("$FreeBSD$");
19
20 #include <sys/endian.h>
21 #include <sys/types.h>
22
23 #ifdef _KERNEL
24 #include <sys/systm.h>
25 #else
26 #include <string.h>
27 #endif
28
29 #include "skein.h"
30
31 #ifndef SKEIN_USE_ASM
32 #define SKEIN_USE_ASM (0) /* default is all C code (no ASM) */
33 #endif
34
35 #ifndef SKEIN_LOOP
36 #define SKEIN_LOOP 001 /* default: unroll 256 and 512, but not 1024 */
37 #endif
38
39 #define BLK_BITS (WCNT*64) /* some useful definitions for code here */
40 #define KW_TWK_BASE (0)
41 #define KW_KEY_BASE (3)
42 #define ks (kw + KW_KEY_BASE)
43 #define ts (kw + KW_TWK_BASE)
44
45 #ifdef SKEIN_DEBUG
46 #define DebugSaveTweak(ctx) { ctx->h.T[0] = ts[0]; ctx->h.T[1] = ts[1]; }
47 #else
48 #define DebugSaveTweak(ctx)
49 #endif
50
51 /*****************************************************************/
52 /* functions to process blkCnt (nonzero) full block(s) of data. */
53 void Skein_256_Process_Block(Skein_256_Ctxt_t *ctx,const u08b_t *blkPtr,size_t blkCnt,size_t byteCntAdd);
54 void Skein_512_Process_Block(Skein_512_Ctxt_t *ctx,const u08b_t *blkPtr,size_t blkCnt,size_t byteCntAdd);
55 void Skein1024_Process_Block(Skein1024_Ctxt_t *ctx,const u08b_t *blkPtr,size_t blkCnt,size_t byteCntAdd);
56
57 /***************************** Skein_256 ******************************/
58 #if !(SKEIN_USE_ASM & 256)
59 void Skein_256_Process_Block(Skein_256_Ctxt_t *ctx,const u08b_t *blkPtr,size_t blkCnt,size_t byteCntAdd)
60 { /* do it in C */
61 enum
62 {
63 WCNT = SKEIN_256_STATE_WORDS
64 };
65 #undef RCNT
66 #define RCNT (SKEIN_256_ROUNDS_TOTAL/8)
67
68 #ifdef SKEIN_LOOP /* configure how much to unroll the loop */
69 #define SKEIN_UNROLL_256 (((SKEIN_LOOP)/100)%10)
70 #else
71 #define SKEIN_UNROLL_256 (0)
72 #endif
73
74 #if SKEIN_UNROLL_256
75 #if (RCNT % SKEIN_UNROLL_256)
76 #error "Invalid SKEIN_UNROLL_256" /* sanity check on unroll count */
77 #endif
78 size_t r;
79 u64b_t kw[WCNT+4+RCNT*2]; /* key schedule words : chaining vars + tweak + "rotation"*/
80 #else
81 u64b_t kw[WCNT+4]; /* key schedule words : chaining vars + tweak */
82 #endif
83 u64b_t X0,X1,X2,X3; /* local copy of context vars, for speed */
84 u64b_t w [WCNT]; /* local copy of input block */
85 #ifdef SKEIN_DEBUG
86 const u64b_t *Xptr[4]; /* use for debugging (help compiler put Xn in registers) */
87 Xptr[0] = &X0; Xptr[1] = &X1; Xptr[2] = &X2; Xptr[3] = &X3;
88 #endif
89 Skein_assert(blkCnt != 0); /* never call with blkCnt == 0! */
90 ts[0] = ctx->h.T[0];
91 ts[1] = ctx->h.T[1];
92 do {
93 /* this implementation only supports 2**64 input bytes (no carry out here) */
94 ts[0] += byteCntAdd; /* update processed length */
95
96 /* precompute the key schedule for this block */
97 ks[0] = ctx->X[0];
98 ks[1] = ctx->X[1];
99 ks[2] = ctx->X[2];
100 ks[3] = ctx->X[3];
101 ks[4] = ks[0] ^ ks[1] ^ ks[2] ^ ks[3] ^ SKEIN_KS_PARITY;
102
103 ts[2] = ts[0] ^ ts[1];
104
105 Skein_Get64_LSB_First(w,blkPtr,WCNT); /* get input block in little-endian format */
106 DebugSaveTweak(ctx);
107 Skein_Show_Block(BLK_BITS,&ctx->h,ctx->X,blkPtr,w,ks,ts);
108
109 X0 = w[0] + ks[0]; /* do the first full key injection */
110 X1 = w[1] + ks[1] + ts[0];
111 X2 = w[2] + ks[2] + ts[1];
112 X3 = w[3] + ks[3];
113
114 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INITIAL,Xptr); /* show starting state values */
115
116 blkPtr += SKEIN_256_BLOCK_BYTES;
117
118 /* run the rounds */
119
120 #define Round256(p0,p1,p2,p3,ROT,rNum) \
121 X##p0 += X##p1; X##p1 = RotL_64(X##p1,ROT##_0); X##p1 ^= X##p0; \
122 X##p2 += X##p3; X##p3 = RotL_64(X##p3,ROT##_1); X##p3 ^= X##p2; \
123
124 #if SKEIN_UNROLL_256 == 0
125 #define R256(p0,p1,p2,p3,ROT,rNum) /* fully unrolled */ \
126 Round256(p0,p1,p2,p3,ROT,rNum) \
127 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,rNum,Xptr);
128
129 #define I256(R) \
130 X0 += ks[((R)+1) % 5]; /* inject the key schedule value */ \
131 X1 += ks[((R)+2) % 5] + ts[((R)+1) % 3]; \
132 X2 += ks[((R)+3) % 5] + ts[((R)+2) % 3]; \
133 X3 += ks[((R)+4) % 5] + (R)+1; \
134 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INJECT,Xptr);
135 #else /* looping version */
136 #define R256(p0,p1,p2,p3,ROT,rNum) \
137 Round256(p0,p1,p2,p3,ROT,rNum) \
138 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,4*(r-1)+rNum,Xptr);
139
140 #define I256(R) \
141 X0 += ks[r+(R)+0]; /* inject the key schedule value */ \
142 X1 += ks[r+(R)+1] + ts[r+(R)+0]; \
143 X2 += ks[r+(R)+2] + ts[r+(R)+1]; \
144 X3 += ks[r+(R)+3] + r+(R) ; \
145 ks[r + (R)+4 ] = ks[r+(R)-1]; /* rotate key schedule */\
146 ts[r + (R)+2 ] = ts[r+(R)-1]; \
147 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INJECT,Xptr);
148
149 for (r=1;r < 2*RCNT;r+=2*SKEIN_UNROLL_256) /* loop thru it */
150 #endif
151 {
152 #define R256_8_rounds(R) \
153 R256(0,1,2,3,R_256_0,8*(R) + 1); \
154 R256(0,3,2,1,R_256_1,8*(R) + 2); \
155 R256(0,1,2,3,R_256_2,8*(R) + 3); \
156 R256(0,3,2,1,R_256_3,8*(R) + 4); \
157 I256(2*(R)); \
158 R256(0,1,2,3,R_256_4,8*(R) + 5); \
159 R256(0,3,2,1,R_256_5,8*(R) + 6); \
160 R256(0,1,2,3,R_256_6,8*(R) + 7); \
161 R256(0,3,2,1,R_256_7,8*(R) + 8); \
162 I256(2*(R)+1);
163
164 R256_8_rounds( 0);
165
166 #define R256_Unroll_R(NN) ((SKEIN_UNROLL_256 == 0 && SKEIN_256_ROUNDS_TOTAL/8 > (NN)) || (SKEIN_UNROLL_256 > (NN)))
167
168 #if R256_Unroll_R( 1)
169 R256_8_rounds( 1);
170 #endif
171 #if R256_Unroll_R( 2)
172 R256_8_rounds( 2);
173 #endif
174 #if R256_Unroll_R( 3)
175 R256_8_rounds( 3);
176 #endif
177 #if R256_Unroll_R( 4)
178 R256_8_rounds( 4);
179 #endif
180 #if R256_Unroll_R( 5)
181 R256_8_rounds( 5);
182 #endif
183 #if R256_Unroll_R( 6)
184 R256_8_rounds( 6);
185 #endif
186 #if R256_Unroll_R( 7)
187 R256_8_rounds( 7);
188 #endif
189 #if R256_Unroll_R( 8)
190 R256_8_rounds( 8);
191 #endif
192 #if R256_Unroll_R( 9)
193 R256_8_rounds( 9);
194 #endif
195 #if R256_Unroll_R(10)
196 R256_8_rounds(10);
197 #endif
198 #if R256_Unroll_R(11)
199 R256_8_rounds(11);
200 #endif
201 #if R256_Unroll_R(12)
202 R256_8_rounds(12);
203 #endif
204 #if R256_Unroll_R(13)
205 R256_8_rounds(13);
206 #endif
207 #if R256_Unroll_R(14)
208 R256_8_rounds(14);
209 #endif
210 #if (SKEIN_UNROLL_256 > 14)
211 #error "need more unrolling in Skein_256_Process_Block"
212 #endif
213 }
214 /* do the final "feedforward" xor, update context chaining vars */
215 ctx->X[0] = X0 ^ w[0];
216 ctx->X[1] = X1 ^ w[1];
217 ctx->X[2] = X2 ^ w[2];
218 ctx->X[3] = X3 ^ w[3];
219
220 Skein_Show_Round(BLK_BITS,&ctx->h,SKEIN_RND_FEED_FWD,ctx->X);
221
222 ts[1] &= ~SKEIN_T1_FLAG_FIRST;
223 }
224 while (--blkCnt);
225 ctx->h.T[0] = ts[0];
226 ctx->h.T[1] = ts[1];
227 }
228
229 #if defined(SKEIN_CODE_SIZE) || defined(SKEIN_PERF)
230 size_t Skein_256_Process_Block_CodeSize(void)
231 {
232 return ((u08b_t *) Skein_256_Process_Block_CodeSize) -
233 ((u08b_t *) Skein_256_Process_Block);
234 }
235 uint_t Skein_256_Unroll_Cnt(void)
236 {
237 return SKEIN_UNROLL_256;
238 }
239 #endif
240 #endif
241
242 /***************************** Skein_512 ******************************/
243 #if !(SKEIN_USE_ASM & 512)
244 void Skein_512_Process_Block(Skein_512_Ctxt_t *ctx,const u08b_t *blkPtr,size_t blkCnt,size_t byteCntAdd)
245 { /* do it in C */
246 enum
247 {
248 WCNT = SKEIN_512_STATE_WORDS
249 };
250 #undef RCNT
251 #define RCNT (SKEIN_512_ROUNDS_TOTAL/8)
252
253 #ifdef SKEIN_LOOP /* configure how much to unroll the loop */
254 #define SKEIN_UNROLL_512 (((SKEIN_LOOP)/10)%10)
255 #else
256 #define SKEIN_UNROLL_512 (0)
257 #endif
258
259 #if SKEIN_UNROLL_512
260 #if (RCNT % SKEIN_UNROLL_512)
261 #error "Invalid SKEIN_UNROLL_512" /* sanity check on unroll count */
262 #endif
263 size_t r;
264 u64b_t kw[WCNT+4+RCNT*2]; /* key schedule words : chaining vars + tweak + "rotation"*/
265 #else
266 u64b_t kw[WCNT+4]; /* key schedule words : chaining vars + tweak */
267 #endif
268 u64b_t X0,X1,X2,X3,X4,X5,X6,X7; /* local copy of vars, for speed */
269 u64b_t w [WCNT]; /* local copy of input block */
270 #ifdef SKEIN_DEBUG
271 const u64b_t *Xptr[8]; /* use for debugging (help compiler put Xn in registers) */
272 Xptr[0] = &X0; Xptr[1] = &X1; Xptr[2] = &X2; Xptr[3] = &X3;
273 Xptr[4] = &X4; Xptr[5] = &X5; Xptr[6] = &X6; Xptr[7] = &X7;
274 #endif
275
276 Skein_assert(blkCnt != 0); /* never call with blkCnt == 0! */
277 ts[0] = ctx->h.T[0];
278 ts[1] = ctx->h.T[1];
279 do {
280 /* this implementation only supports 2**64 input bytes (no carry out here) */
281 ts[0] += byteCntAdd; /* update processed length */
282
283 /* precompute the key schedule for this block */
284 ks[0] = ctx->X[0];
285 ks[1] = ctx->X[1];
286 ks[2] = ctx->X[2];
287 ks[3] = ctx->X[3];
288 ks[4] = ctx->X[4];
289 ks[5] = ctx->X[5];
290 ks[6] = ctx->X[6];
291 ks[7] = ctx->X[7];
292 ks[8] = ks[0] ^ ks[1] ^ ks[2] ^ ks[3] ^
293 ks[4] ^ ks[5] ^ ks[6] ^ ks[7] ^ SKEIN_KS_PARITY;
294
295 ts[2] = ts[0] ^ ts[1];
296
297 Skein_Get64_LSB_First(w,blkPtr,WCNT); /* get input block in little-endian format */
298 DebugSaveTweak(ctx);
299 Skein_Show_Block(BLK_BITS,&ctx->h,ctx->X,blkPtr,w,ks,ts);
300
301 X0 = w[0] + ks[0]; /* do the first full key injection */
302 X1 = w[1] + ks[1];
303 X2 = w[2] + ks[2];
304 X3 = w[3] + ks[3];
305 X4 = w[4] + ks[4];
306 X5 = w[5] + ks[5] + ts[0];
307 X6 = w[6] + ks[6] + ts[1];
308 X7 = w[7] + ks[7];
309
310 blkPtr += SKEIN_512_BLOCK_BYTES;
311
312 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INITIAL,Xptr);
313 /* run the rounds */
314 #define Round512(p0,p1,p2,p3,p4,p5,p6,p7,ROT,rNum) \
315 X##p0 += X##p1; X##p1 = RotL_64(X##p1,ROT##_0); X##p1 ^= X##p0; \
316 X##p2 += X##p3; X##p3 = RotL_64(X##p3,ROT##_1); X##p3 ^= X##p2; \
317 X##p4 += X##p5; X##p5 = RotL_64(X##p5,ROT##_2); X##p5 ^= X##p4; \
318 X##p6 += X##p7; X##p7 = RotL_64(X##p7,ROT##_3); X##p7 ^= X##p6; \
319
320 #if SKEIN_UNROLL_512 == 0
321 #define R512(p0,p1,p2,p3,p4,p5,p6,p7,ROT,rNum) /* unrolled */ \
322 Round512(p0,p1,p2,p3,p4,p5,p6,p7,ROT,rNum) \
323 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,rNum,Xptr);
324
325 #define I512(R) \
326 X0 += ks[((R)+1) % 9]; /* inject the key schedule value */ \
327 X1 += ks[((R)+2) % 9]; \
328 X2 += ks[((R)+3) % 9]; \
329 X3 += ks[((R)+4) % 9]; \
330 X4 += ks[((R)+5) % 9]; \
331 X5 += ks[((R)+6) % 9] + ts[((R)+1) % 3]; \
332 X6 += ks[((R)+7) % 9] + ts[((R)+2) % 3]; \
333 X7 += ks[((R)+8) % 9] + (R)+1; \
334 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INJECT,Xptr);
335 #else /* looping version */
336 #define R512(p0,p1,p2,p3,p4,p5,p6,p7,ROT,rNum) \
337 Round512(p0,p1,p2,p3,p4,p5,p6,p7,ROT,rNum) \
338 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,4*(r-1)+rNum,Xptr);
339
340 #define I512(R) \
341 X0 += ks[r+(R)+0]; /* inject the key schedule value */ \
342 X1 += ks[r+(R)+1]; \
343 X2 += ks[r+(R)+2]; \
344 X3 += ks[r+(R)+3]; \
345 X4 += ks[r+(R)+4]; \
346 X5 += ks[r+(R)+5] + ts[r+(R)+0]; \
347 X6 += ks[r+(R)+6] + ts[r+(R)+1]; \
348 X7 += ks[r+(R)+7] + r+(R) ; \
349 ks[r + (R)+8] = ks[r+(R)-1]; /* rotate key schedule */ \
350 ts[r + (R)+2] = ts[r+(R)-1]; \
351 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INJECT,Xptr);
352
353 for (r=1;r < 2*RCNT;r+=2*SKEIN_UNROLL_512) /* loop thru it */
354 #endif /* end of looped code definitions */
355 {
356 #define R512_8_rounds(R) /* do 8 full rounds */ \
357 R512(0,1,2,3,4,5,6,7,R_512_0,8*(R)+ 1); \
358 R512(2,1,4,7,6,5,0,3,R_512_1,8*(R)+ 2); \
359 R512(4,1,6,3,0,5,2,7,R_512_2,8*(R)+ 3); \
360 R512(6,1,0,7,2,5,4,3,R_512_3,8*(R)+ 4); \
361 I512(2*(R)); \
362 R512(0,1,2,3,4,5,6,7,R_512_4,8*(R)+ 5); \
363 R512(2,1,4,7,6,5,0,3,R_512_5,8*(R)+ 6); \
364 R512(4,1,6,3,0,5,2,7,R_512_6,8*(R)+ 7); \
365 R512(6,1,0,7,2,5,4,3,R_512_7,8*(R)+ 8); \
366 I512(2*(R)+1); /* and key injection */
367
368 R512_8_rounds( 0);
369
370 #define R512_Unroll_R(NN) ((SKEIN_UNROLL_512 == 0 && SKEIN_512_ROUNDS_TOTAL/8 > (NN)) || (SKEIN_UNROLL_512 > (NN)))
371
372 #if R512_Unroll_R( 1)
373 R512_8_rounds( 1);
374 #endif
375 #if R512_Unroll_R( 2)
376 R512_8_rounds( 2);
377 #endif
378 #if R512_Unroll_R( 3)
379 R512_8_rounds( 3);
380 #endif
381 #if R512_Unroll_R( 4)
382 R512_8_rounds( 4);
383 #endif
384 #if R512_Unroll_R( 5)
385 R512_8_rounds( 5);
386 #endif
387 #if R512_Unroll_R( 6)
388 R512_8_rounds( 6);
389 #endif
390 #if R512_Unroll_R( 7)
391 R512_8_rounds( 7);
392 #endif
393 #if R512_Unroll_R( 8)
394 R512_8_rounds( 8);
395 #endif
396 #if R512_Unroll_R( 9)
397 R512_8_rounds( 9);
398 #endif
399 #if R512_Unroll_R(10)
400 R512_8_rounds(10);
401 #endif
402 #if R512_Unroll_R(11)
403 R512_8_rounds(11);
404 #endif
405 #if R512_Unroll_R(12)
406 R512_8_rounds(12);
407 #endif
408 #if R512_Unroll_R(13)
409 R512_8_rounds(13);
410 #endif
411 #if R512_Unroll_R(14)
412 R512_8_rounds(14);
413 #endif
414 #if (SKEIN_UNROLL_512 > 14)
415 #error "need more unrolling in Skein_512_Process_Block"
416 #endif
417 }
418
419 /* do the final "feedforward" xor, update context chaining vars */
420 ctx->X[0] = X0 ^ w[0];
421 ctx->X[1] = X1 ^ w[1];
422 ctx->X[2] = X2 ^ w[2];
423 ctx->X[3] = X3 ^ w[3];
424 ctx->X[4] = X4 ^ w[4];
425 ctx->X[5] = X5 ^ w[5];
426 ctx->X[6] = X6 ^ w[6];
427 ctx->X[7] = X7 ^ w[7];
428 Skein_Show_Round(BLK_BITS,&ctx->h,SKEIN_RND_FEED_FWD,ctx->X);
429
430 ts[1] &= ~SKEIN_T1_FLAG_FIRST;
431 }
432 while (--blkCnt);
433 ctx->h.T[0] = ts[0];
434 ctx->h.T[1] = ts[1];
435 }
436
437 #if defined(SKEIN_CODE_SIZE) || defined(SKEIN_PERF)
438 size_t Skein_512_Process_Block_CodeSize(void)
439 {
440 return ((u08b_t *) Skein_512_Process_Block_CodeSize) -
441 ((u08b_t *) Skein_512_Process_Block);
442 }
443 uint_t Skein_512_Unroll_Cnt(void)
444 {
445 return SKEIN_UNROLL_512;
446 }
447 #endif
448 #endif
449
450 /***************************** Skein1024 ******************************/
451 #if !(SKEIN_USE_ASM & 1024)
452 void Skein1024_Process_Block(Skein1024_Ctxt_t *ctx,const u08b_t *blkPtr,size_t blkCnt,size_t byteCntAdd)
453 { /* do it in C, always looping (unrolled is bigger AND slower!) */
454 enum
455 {
456 WCNT = SKEIN1024_STATE_WORDS
457 };
458 #undef RCNT
459 #define RCNT (SKEIN1024_ROUNDS_TOTAL/8)
460
461 #ifdef SKEIN_LOOP /* configure how much to unroll the loop */
462 #define SKEIN_UNROLL_1024 ((SKEIN_LOOP)%10)
463 #else
464 #define SKEIN_UNROLL_1024 (0)
465 #endif
466
467 #if (SKEIN_UNROLL_1024 != 0)
468 #if (RCNT % SKEIN_UNROLL_1024)
469 #error "Invalid SKEIN_UNROLL_1024" /* sanity check on unroll count */
470 #endif
471 size_t r;
472 u64b_t kw[WCNT+4+RCNT*2]; /* key schedule words : chaining vars + tweak + "rotation"*/
473 #else
474 u64b_t kw[WCNT+4]; /* key schedule words : chaining vars + tweak */
475 #endif
476
477 u64b_t X00,X01,X02,X03,X04,X05,X06,X07, /* local copy of vars, for speed */
478 X08,X09,X10,X11,X12,X13,X14,X15;
479 u64b_t w [WCNT]; /* local copy of input block */
480 #ifdef SKEIN_DEBUG
481 const u64b_t *Xptr[16]; /* use for debugging (help compiler put Xn in registers) */
482 Xptr[ 0] = &X00; Xptr[ 1] = &X01; Xptr[ 2] = &X02; Xptr[ 3] = &X03;
483 Xptr[ 4] = &X04; Xptr[ 5] = &X05; Xptr[ 6] = &X06; Xptr[ 7] = &X07;
484 Xptr[ 8] = &X08; Xptr[ 9] = &X09; Xptr[10] = &X10; Xptr[11] = &X11;
485 Xptr[12] = &X12; Xptr[13] = &X13; Xptr[14] = &X14; Xptr[15] = &X15;
486 #endif
487
488 Skein_assert(blkCnt != 0); /* never call with blkCnt == 0! */
489 ts[0] = ctx->h.T[0];
490 ts[1] = ctx->h.T[1];
491 do {
492 /* this implementation only supports 2**64 input bytes (no carry out here) */
493 ts[0] += byteCntAdd; /* update processed length */
494
495 /* precompute the key schedule for this block */
496 ks[ 0] = ctx->X[ 0];
497 ks[ 1] = ctx->X[ 1];
498 ks[ 2] = ctx->X[ 2];
499 ks[ 3] = ctx->X[ 3];
500 ks[ 4] = ctx->X[ 4];
501 ks[ 5] = ctx->X[ 5];
502 ks[ 6] = ctx->X[ 6];
503 ks[ 7] = ctx->X[ 7];
504 ks[ 8] = ctx->X[ 8];
505 ks[ 9] = ctx->X[ 9];
506 ks[10] = ctx->X[10];
507 ks[11] = ctx->X[11];
508 ks[12] = ctx->X[12];
509 ks[13] = ctx->X[13];
510 ks[14] = ctx->X[14];
511 ks[15] = ctx->X[15];
512 ks[16] = ks[ 0] ^ ks[ 1] ^ ks[ 2] ^ ks[ 3] ^
513 ks[ 4] ^ ks[ 5] ^ ks[ 6] ^ ks[ 7] ^
514 ks[ 8] ^ ks[ 9] ^ ks[10] ^ ks[11] ^
515 ks[12] ^ ks[13] ^ ks[14] ^ ks[15] ^ SKEIN_KS_PARITY;
516
517 ts[2] = ts[0] ^ ts[1];
518
519 Skein_Get64_LSB_First(w,blkPtr,WCNT); /* get input block in little-endian format */
520 DebugSaveTweak(ctx);
521 Skein_Show_Block(BLK_BITS,&ctx->h,ctx->X,blkPtr,w,ks,ts);
522
523 X00 = w[ 0] + ks[ 0]; /* do the first full key injection */
524 X01 = w[ 1] + ks[ 1];
525 X02 = w[ 2] + ks[ 2];
526 X03 = w[ 3] + ks[ 3];
527 X04 = w[ 4] + ks[ 4];
528 X05 = w[ 5] + ks[ 5];
529 X06 = w[ 6] + ks[ 6];
530 X07 = w[ 7] + ks[ 7];
531 X08 = w[ 8] + ks[ 8];
532 X09 = w[ 9] + ks[ 9];
533 X10 = w[10] + ks[10];
534 X11 = w[11] + ks[11];
535 X12 = w[12] + ks[12];
536 X13 = w[13] + ks[13] + ts[0];
537 X14 = w[14] + ks[14] + ts[1];
538 X15 = w[15] + ks[15];
539
540 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INITIAL,Xptr);
541
542 #define Round1024(p0,p1,p2,p3,p4,p5,p6,p7,p8,p9,pA,pB,pC,pD,pE,pF,ROT,rNum) \
543 X##p0 += X##p1; X##p1 = RotL_64(X##p1,ROT##_0); X##p1 ^= X##p0; \
544 X##p2 += X##p3; X##p3 = RotL_64(X##p3,ROT##_1); X##p3 ^= X##p2; \
545 X##p4 += X##p5; X##p5 = RotL_64(X##p5,ROT##_2); X##p5 ^= X##p4; \
546 X##p6 += X##p7; X##p7 = RotL_64(X##p7,ROT##_3); X##p7 ^= X##p6; \
547 X##p8 += X##p9; X##p9 = RotL_64(X##p9,ROT##_4); X##p9 ^= X##p8; \
548 X##pA += X##pB; X##pB = RotL_64(X##pB,ROT##_5); X##pB ^= X##pA; \
549 X##pC += X##pD; X##pD = RotL_64(X##pD,ROT##_6); X##pD ^= X##pC; \
550 X##pE += X##pF; X##pF = RotL_64(X##pF,ROT##_7); X##pF ^= X##pE; \
551
552 #if SKEIN_UNROLL_1024 == 0
553 #define R1024(p0,p1,p2,p3,p4,p5,p6,p7,p8,p9,pA,pB,pC,pD,pE,pF,ROT,rn) \
554 Round1024(p0,p1,p2,p3,p4,p5,p6,p7,p8,p9,pA,pB,pC,pD,pE,pF,ROT,rn) \
555 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,rn,Xptr);
556
557 #define I1024(R) \
558 X00 += ks[((R)+ 1) % 17]; /* inject the key schedule value */ \
559 X01 += ks[((R)+ 2) % 17]; \
560 X02 += ks[((R)+ 3) % 17]; \
561 X03 += ks[((R)+ 4) % 17]; \
562 X04 += ks[((R)+ 5) % 17]; \
563 X05 += ks[((R)+ 6) % 17]; \
564 X06 += ks[((R)+ 7) % 17]; \
565 X07 += ks[((R)+ 8) % 17]; \
566 X08 += ks[((R)+ 9) % 17]; \
567 X09 += ks[((R)+10) % 17]; \
568 X10 += ks[((R)+11) % 17]; \
569 X11 += ks[((R)+12) % 17]; \
570 X12 += ks[((R)+13) % 17]; \
571 X13 += ks[((R)+14) % 17] + ts[((R)+1) % 3]; \
572 X14 += ks[((R)+15) % 17] + ts[((R)+2) % 3]; \
573 X15 += ks[((R)+16) % 17] + (R)+1; \
574 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INJECT,Xptr);
575 #else /* looping version */
576 #define R1024(p0,p1,p2,p3,p4,p5,p6,p7,p8,p9,pA,pB,pC,pD,pE,pF,ROT,rn) \
577 Round1024(p0,p1,p2,p3,p4,p5,p6,p7,p8,p9,pA,pB,pC,pD,pE,pF,ROT,rn) \
578 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,4*(r-1)+rn,Xptr);
579
580 #define I1024(R) \
581 X00 += ks[r+(R)+ 0]; /* inject the key schedule value */ \
582 X01 += ks[r+(R)+ 1]; \
583 X02 += ks[r+(R)+ 2]; \
584 X03 += ks[r+(R)+ 3]; \
585 X04 += ks[r+(R)+ 4]; \
586 X05 += ks[r+(R)+ 5]; \
587 X06 += ks[r+(R)+ 6]; \
588 X07 += ks[r+(R)+ 7]; \
589 X08 += ks[r+(R)+ 8]; \
590 X09 += ks[r+(R)+ 9]; \
591 X10 += ks[r+(R)+10]; \
592 X11 += ks[r+(R)+11]; \
593 X12 += ks[r+(R)+12]; \
594 X13 += ks[r+(R)+13] + ts[r+(R)+0]; \
595 X14 += ks[r+(R)+14] + ts[r+(R)+1]; \
596 X15 += ks[r+(R)+15] + r+(R) ; \
597 ks[r + (R)+16] = ks[r+(R)-1]; /* rotate key schedule */ \
598 ts[r + (R)+ 2] = ts[r+(R)-1]; \
599 Skein_Show_R_Ptr(BLK_BITS,&ctx->h,SKEIN_RND_KEY_INJECT,Xptr);
600
601 for (r=1;r <= 2*RCNT;r+=2*SKEIN_UNROLL_1024) /* loop thru it */
602 #endif
603 {
604 #define R1024_8_rounds(R) /* do 8 full rounds */ \
605 R1024(00,01,02,03,04,05,06,07,08,09,10,11,12,13,14,15,R1024_0,8*(R) + 1); \
606 R1024(00,09,02,13,06,11,04,15,10,07,12,03,14,05,08,01,R1024_1,8*(R) + 2); \
607 R1024(00,07,02,05,04,03,06,01,12,15,14,13,08,11,10,09,R1024_2,8*(R) + 3); \
608 R1024(00,15,02,11,06,13,04,09,14,01,08,05,10,03,12,07,R1024_3,8*(R) + 4); \
609 I1024(2*(R)); \
610 R1024(00,01,02,03,04,05,06,07,08,09,10,11,12,13,14,15,R1024_4,8*(R) + 5); \
611 R1024(00,09,02,13,06,11,04,15,10,07,12,03,14,05,08,01,R1024_5,8*(R) + 6); \
612 R1024(00,07,02,05,04,03,06,01,12,15,14,13,08,11,10,09,R1024_6,8*(R) + 7); \
613 R1024(00,15,02,11,06,13,04,09,14,01,08,05,10,03,12,07,R1024_7,8*(R) + 8); \
614 I1024(2*(R)+1);
615
616 R1024_8_rounds( 0);
617
618 #define R1024_Unroll_R(NN) ((SKEIN_UNROLL_1024 == 0 && SKEIN1024_ROUNDS_TOTAL/8 > (NN)) || (SKEIN_UNROLL_1024 > (NN)))
619
620 #if R1024_Unroll_R( 1)
621 R1024_8_rounds( 1);
622 #endif
623 #if R1024_Unroll_R( 2)
624 R1024_8_rounds( 2);
625 #endif
626 #if R1024_Unroll_R( 3)
627 R1024_8_rounds( 3);
628 #endif
629 #if R1024_Unroll_R( 4)
630 R1024_8_rounds( 4);
631 #endif
632 #if R1024_Unroll_R( 5)
633 R1024_8_rounds( 5);
634 #endif
635 #if R1024_Unroll_R( 6)
636 R1024_8_rounds( 6);
637 #endif
638 #if R1024_Unroll_R( 7)
639 R1024_8_rounds( 7);
640 #endif
641 #if R1024_Unroll_R( 8)
642 R1024_8_rounds( 8);
643 #endif
644 #if R1024_Unroll_R( 9)
645 R1024_8_rounds( 9);
646 #endif
647 #if R1024_Unroll_R(10)
648 R1024_8_rounds(10);
649 #endif
650 #if R1024_Unroll_R(11)
651 R1024_8_rounds(11);
652 #endif
653 #if R1024_Unroll_R(12)
654 R1024_8_rounds(12);
655 #endif
656 #if R1024_Unroll_R(13)
657 R1024_8_rounds(13);
658 #endif
659 #if R1024_Unroll_R(14)
660 R1024_8_rounds(14);
661 #endif
662 #if (SKEIN_UNROLL_1024 > 14)
663 #error "need more unrolling in Skein_1024_Process_Block"
664 #endif
665 }
666 /* do the final "feedforward" xor, update context chaining vars */
667
668 ctx->X[ 0] = X00 ^ w[ 0];
669 ctx->X[ 1] = X01 ^ w[ 1];
670 ctx->X[ 2] = X02 ^ w[ 2];
671 ctx->X[ 3] = X03 ^ w[ 3];
672 ctx->X[ 4] = X04 ^ w[ 4];
673 ctx->X[ 5] = X05 ^ w[ 5];
674 ctx->X[ 6] = X06 ^ w[ 6];
675 ctx->X[ 7] = X07 ^ w[ 7];
676 ctx->X[ 8] = X08 ^ w[ 8];
677 ctx->X[ 9] = X09 ^ w[ 9];
678 ctx->X[10] = X10 ^ w[10];
679 ctx->X[11] = X11 ^ w[11];
680 ctx->X[12] = X12 ^ w[12];
681 ctx->X[13] = X13 ^ w[13];
682 ctx->X[14] = X14 ^ w[14];
683 ctx->X[15] = X15 ^ w[15];
684
685 Skein_Show_Round(BLK_BITS,&ctx->h,SKEIN_RND_FEED_FWD,ctx->X);
686
687 ts[1] &= ~SKEIN_T1_FLAG_FIRST;
688 blkPtr += SKEIN1024_BLOCK_BYTES;
689 }
690 while (--blkCnt);
691 ctx->h.T[0] = ts[0];
692 ctx->h.T[1] = ts[1];
693 }
694
695 #if defined(SKEIN_CODE_SIZE) || defined(SKEIN_PERF)
696 size_t Skein1024_Process_Block_CodeSize(void)
697 {
698 return ((u08b_t *) Skein1024_Process_Block_CodeSize) -
699 ((u08b_t *) Skein1024_Process_Block);
700 }
701 uint_t Skein1024_Unroll_Cnt(void)
702 {
703 return SKEIN_UNROLL_1024;
704 }
705 #endif
706 #endif
Cache object: 26127553ee2d25ffcd51afecd5fdc56b
|