1 /* crypto/sha/sha512.c */
2 /* ====================================================================
3 * Copyright (c) 2004 The OpenSSL Project. All rights reserved
4 * according to the OpenSSL license [found in ../../LICENSE].
5 * ====================================================================
7 #include <openssl/opensslconf.h>
8 #if !defined(OPENSSL_NO_SHA) && !defined(OPENSSL_NO_SHA512)
10 * IMPLEMENTATION NOTES.
12 * As you might have noticed 32-bit hash algorithms:
14 * - permit SHA_LONG to be wider than 32-bit (case on CRAY);
15 * - optimized versions implement two transform functions: one operating
16 * on [aligned] data in host byte order and one - on data in input
18 * - share common byte-order neutral collector and padding function
19 * implementations, ../md32_common.h;
21 * Neither of the above applies to this SHA-512 implementations. Reasons
22 * [in reverse order] are:
24 * - it's the only 64-bit hash algorithm for the moment of this writing,
25 * there is no need for common collector/padding implementation [yet];
26 * - by supporting only one transform function [which operates on
27 * *aligned* data in input stream byte order, big-endian in this case]
28 * we minimize burden of maintenance in two ways: a) collector/padding
29 * function is simpler; b) only one transform function to stare at;
30 * - SHA_LONG64 is required to be exactly 64-bit in order to be able to
31 * apply a number of optimizations to mitigate potential performance
32 * penalties caused by previous design decision;
36 * Implementation relies on the fact that "long long" is 64-bit on
37 * both 32- and 64-bit platforms. If some compiler vendor comes up
38 * with 128-bit long long, adjustment to sha.h would be required.
39 * As this implementation relies on 64-bit integer type, it's totally
40 * inappropriate for platforms which don't support it, most notably
42 * <appro@fy.chalmers.se>
47 #include <openssl/crypto.h>
48 #include <openssl/sha.h>
49 #include <openssl/opensslv.h>
53 const char SHA512_version[]="SHA-512" OPENSSL_VERSION_PTEXT;
55 #if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
56 defined(__x86_64) || defined(_M_AMD64) || defined(_M_X64) || \
57 defined(__s390__) || defined(__s390x__)
58 #define SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA
61 int SHA384_Init (SHA512_CTX *c)
63 c->h[0]=U64(0xcbbb9d5dc1059ed8);
64 c->h[1]=U64(0x629a292a367cd507);
65 c->h[2]=U64(0x9159015a3070dd17);
66 c->h[3]=U64(0x152fecd8f70e5939);
67 c->h[4]=U64(0x67332667ffc00b31);
68 c->h[5]=U64(0x8eb44a8768581511);
69 c->h[6]=U64(0xdb0c2e0d64f98fa7);
70 c->h[7]=U64(0x47b5481dbefa4fa4);
72 c->num=0; c->md_len=SHA384_DIGEST_LENGTH;
76 int SHA512_Init (SHA512_CTX *c)
78 c->h[0]=U64(0x6a09e667f3bcc908);
79 c->h[1]=U64(0xbb67ae8584caa73b);
80 c->h[2]=U64(0x3c6ef372fe94f82b);
81 c->h[3]=U64(0xa54ff53a5f1d36f1);
82 c->h[4]=U64(0x510e527fade682d1);
83 c->h[5]=U64(0x9b05688c2b3e6c1f);
84 c->h[6]=U64(0x1f83d9abfb41bd6b);
85 c->h[7]=U64(0x5be0cd19137e2179);
87 c->num=0; c->md_len=SHA512_DIGEST_LENGTH;
94 void sha512_block_data_order (SHA512_CTX *ctx, const void *in, size_t num);
96 int SHA512_Final (unsigned char *md, SHA512_CTX *c)
98 unsigned char *p=(unsigned char *)c->u.p;
101 p[n]=0x80; /* There always is a room for one */
103 if (n > (sizeof(c->u)-16))
104 memset (p+n,0,sizeof(c->u)-n), n=0,
105 sha512_block_data_order (c,p,1);
107 memset (p+n,0,sizeof(c->u)-16-n);
109 c->u.d[SHA_LBLOCK-2] = c->Nh;
110 c->u.d[SHA_LBLOCK-1] = c->Nl;
112 p[sizeof(c->u)-1] = (unsigned char)(c->Nl);
113 p[sizeof(c->u)-2] = (unsigned char)(c->Nl>>8);
114 p[sizeof(c->u)-3] = (unsigned char)(c->Nl>>16);
115 p[sizeof(c->u)-4] = (unsigned char)(c->Nl>>24);
116 p[sizeof(c->u)-5] = (unsigned char)(c->Nl>>32);
117 p[sizeof(c->u)-6] = (unsigned char)(c->Nl>>40);
118 p[sizeof(c->u)-7] = (unsigned char)(c->Nl>>48);
119 p[sizeof(c->u)-8] = (unsigned char)(c->Nl>>56);
120 p[sizeof(c->u)-9] = (unsigned char)(c->Nh);
121 p[sizeof(c->u)-10] = (unsigned char)(c->Nh>>8);
122 p[sizeof(c->u)-11] = (unsigned char)(c->Nh>>16);
123 p[sizeof(c->u)-12] = (unsigned char)(c->Nh>>24);
124 p[sizeof(c->u)-13] = (unsigned char)(c->Nh>>32);
125 p[sizeof(c->u)-14] = (unsigned char)(c->Nh>>40);
126 p[sizeof(c->u)-15] = (unsigned char)(c->Nh>>48);
127 p[sizeof(c->u)-16] = (unsigned char)(c->Nh>>56);
130 sha512_block_data_order (c,p,1);
136 /* Let compiler decide if it's appropriate to unroll... */
137 case SHA384_DIGEST_LENGTH:
138 for (n=0;n<SHA384_DIGEST_LENGTH/8;n++)
140 SHA_LONG64 t = c->h[n];
142 *(md++) = (unsigned char)(t>>56);
143 *(md++) = (unsigned char)(t>>48);
144 *(md++) = (unsigned char)(t>>40);
145 *(md++) = (unsigned char)(t>>32);
146 *(md++) = (unsigned char)(t>>24);
147 *(md++) = (unsigned char)(t>>16);
148 *(md++) = (unsigned char)(t>>8);
149 *(md++) = (unsigned char)(t);
152 case SHA512_DIGEST_LENGTH:
153 for (n=0;n<SHA512_DIGEST_LENGTH/8;n++)
155 SHA_LONG64 t = c->h[n];
157 *(md++) = (unsigned char)(t>>56);
158 *(md++) = (unsigned char)(t>>48);
159 *(md++) = (unsigned char)(t>>40);
160 *(md++) = (unsigned char)(t>>32);
161 *(md++) = (unsigned char)(t>>24);
162 *(md++) = (unsigned char)(t>>16);
163 *(md++) = (unsigned char)(t>>8);
164 *(md++) = (unsigned char)(t);
167 /* ... as well as make sure md_len is not abused. */
174 int SHA384_Final (unsigned char *md,SHA512_CTX *c)
175 { return SHA512_Final (md,c); }
177 int SHA512_Update (SHA512_CTX *c, const void *_data, size_t len)
180 unsigned char *p=c->u.p;
181 const unsigned char *data=(const unsigned char *)_data;
183 if (len==0) return 1;
185 l = (c->Nl+(((SHA_LONG64)len)<<3))&U64(0xffffffffffffffff);
186 if (l < c->Nl) c->Nh++;
187 if (sizeof(len)>=8) c->Nh+=(((SHA_LONG64)len)>>61);
192 size_t n = sizeof(c->u) - c->num;
196 memcpy (p+c->num,data,len), c->num += len;
200 memcpy (p+c->num,data,n), c->num = 0;
202 sha512_block_data_order (c,p,1);
206 if (len >= sizeof(c->u))
208 #ifndef SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA
209 if ((size_t)data%sizeof(c->u.d[0]) != 0)
210 while (len >= sizeof(c->u))
211 memcpy (p,data,sizeof(c->u)),
212 sha512_block_data_order (c,p,1),
214 data += sizeof(c->u);
217 sha512_block_data_order (c,data,len/sizeof(c->u)),
223 if (len != 0) memcpy (p,data,len), c->num = (int)len;
228 int SHA384_Update (SHA512_CTX *c, const void *data, size_t len)
229 { return SHA512_Update (c,data,len); }
231 void SHA512_Transform (SHA512_CTX *c, const unsigned char *data)
232 { sha512_block_data_order (c,data,1); }
234 unsigned char *SHA384(const unsigned char *d, size_t n, unsigned char *md)
237 static unsigned char m[SHA384_DIGEST_LENGTH];
239 if (md == NULL) md=m;
241 SHA512_Update(&c,d,n);
243 OPENSSL_cleanse(&c,sizeof(c));
247 unsigned char *SHA512(const unsigned char *d, size_t n, unsigned char *md)
250 static unsigned char m[SHA512_DIGEST_LENGTH];
252 if (md == NULL) md=m;
254 SHA512_Update(&c,d,n);
256 OPENSSL_cleanse(&c,sizeof(c));
261 static const SHA_LONG64 K512[80] = {
262 U64(0x428a2f98d728ae22),U64(0x7137449123ef65cd),
263 U64(0xb5c0fbcfec4d3b2f),U64(0xe9b5dba58189dbbc),
264 U64(0x3956c25bf348b538),U64(0x59f111f1b605d019),
265 U64(0x923f82a4af194f9b),U64(0xab1c5ed5da6d8118),
266 U64(0xd807aa98a3030242),U64(0x12835b0145706fbe),
267 U64(0x243185be4ee4b28c),U64(0x550c7dc3d5ffb4e2),
268 U64(0x72be5d74f27b896f),U64(0x80deb1fe3b1696b1),
269 U64(0x9bdc06a725c71235),U64(0xc19bf174cf692694),
270 U64(0xe49b69c19ef14ad2),U64(0xefbe4786384f25e3),
271 U64(0x0fc19dc68b8cd5b5),U64(0x240ca1cc77ac9c65),
272 U64(0x2de92c6f592b0275),U64(0x4a7484aa6ea6e483),
273 U64(0x5cb0a9dcbd41fbd4),U64(0x76f988da831153b5),
274 U64(0x983e5152ee66dfab),U64(0xa831c66d2db43210),
275 U64(0xb00327c898fb213f),U64(0xbf597fc7beef0ee4),
276 U64(0xc6e00bf33da88fc2),U64(0xd5a79147930aa725),
277 U64(0x06ca6351e003826f),U64(0x142929670a0e6e70),
278 U64(0x27b70a8546d22ffc),U64(0x2e1b21385c26c926),
279 U64(0x4d2c6dfc5ac42aed),U64(0x53380d139d95b3df),
280 U64(0x650a73548baf63de),U64(0x766a0abb3c77b2a8),
281 U64(0x81c2c92e47edaee6),U64(0x92722c851482353b),
282 U64(0xa2bfe8a14cf10364),U64(0xa81a664bbc423001),
283 U64(0xc24b8b70d0f89791),U64(0xc76c51a30654be30),
284 U64(0xd192e819d6ef5218),U64(0xd69906245565a910),
285 U64(0xf40e35855771202a),U64(0x106aa07032bbd1b8),
286 U64(0x19a4c116b8d2d0c8),U64(0x1e376c085141ab53),
287 U64(0x2748774cdf8eeb99),U64(0x34b0bcb5e19b48a8),
288 U64(0x391c0cb3c5c95a63),U64(0x4ed8aa4ae3418acb),
289 U64(0x5b9cca4f7763e373),U64(0x682e6ff3d6b2b8a3),
290 U64(0x748f82ee5defb2fc),U64(0x78a5636f43172f60),
291 U64(0x84c87814a1f0ab72),U64(0x8cc702081a6439ec),
292 U64(0x90befffa23631e28),U64(0xa4506cebde82bde9),
293 U64(0xbef9a3f7b2c67915),U64(0xc67178f2e372532b),
294 U64(0xca273eceea26619c),U64(0xd186b8c721c0c207),
295 U64(0xeada7dd6cde0eb1e),U64(0xf57d4f7fee6ed178),
296 U64(0x06f067aa72176fba),U64(0x0a637dc5a2c898a6),
297 U64(0x113f9804bef90dae),U64(0x1b710b35131c471b),
298 U64(0x28db77f523047d84),U64(0x32caab7b40c72493),
299 U64(0x3c9ebe0a15c9bebc),U64(0x431d67c49c100d4c),
300 U64(0x4cc5d4becb3e42b6),U64(0x597f299cfc657e2a),
301 U64(0x5fcb6fab3ad6faec),U64(0x6c44198c4a475817) };
304 # if defined(__GNUC__) && __GNUC__>=2 && !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM)
305 # if defined(__x86_64) || defined(__x86_64__)
306 # define ROTR(a,n) ({ unsigned long ret; \
311 # if !defined(B_ENDIAN)
312 # define PULL64(x) ({ SHA_LONG64 ret=*((const SHA_LONG64 *)(&(x))); \
317 # elif (defined(__i386) || defined(__i386__)) && !defined(B_ENDIAN)
318 # if defined(I386_ONLY)
319 # define PULL64(x) ({ const unsigned int *p=(const unsigned int *)(&(x));\
320 unsigned int hi,lo; \
321 asm("xchgb %%ah,%%al;xchgb %%dh,%%dl;"\
322 "roll $16,%%eax; roll $16,%%edx; "\
323 "xchgb %%ah,%%al;xchgb %%dh,%%dl;" \
324 : "=a"(lo),"=d"(hi) \
325 : "0"(p[1]),"1"(p[0]) : "cc"); \
326 ((SHA_LONG64)hi)<<32|lo; })
328 # define PULL64(x) ({ const unsigned int *p=(const unsigned int *)(&(x));\
329 unsigned int hi,lo; \
330 asm ("bswapl %0; bswapl %1;" \
331 : "=r"(lo),"=r"(hi) \
332 : "0"(p[1]),"1"(p[0])); \
333 ((SHA_LONG64)hi)<<32|lo; })
335 # elif (defined(_ARCH_PPC) && defined(__64BIT__)) || defined(_ARCH_PPC64)
336 # define ROTR(a,n) ({ unsigned long ret; \
337 asm ("rotrdi %0,%1,%2" \
339 : "r"(a),"K"(n)); ret; })
341 # elif defined(_MSC_VER)
342 # if defined(_WIN64) /* applies to both IA-64 and AMD64 */
343 # define ROTR(a,n) _rotr64((a),n)
345 # if defined(_M_IX86) && !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM)
346 # if defined(I386_ONLY)
347 static SHA_LONG64 __fastcall __pull64be(const void *x)
348 { _asm mov edx, [ecx + 0]
349 _asm mov eax, [ecx + 4]
358 static SHA_LONG64 __fastcall __pull64be(const void *x)
359 { _asm mov edx, [ecx + 0]
360 _asm mov eax, [ecx + 4]
365 # define PULL64(x) __pull64be(&(x))
371 #define B(x,j) (((SHA_LONG64)(*(((const unsigned char *)(&x))+j)))<<((7-j)*8))
372 #define PULL64(x) (B(x,0)|B(x,1)|B(x,2)|B(x,3)|B(x,4)|B(x,5)|B(x,6)|B(x,7))
376 #define ROTR(x,s) (((x)>>s) | (x)<<(64-s))
379 #define Sigma0(x) (ROTR((x),28) ^ ROTR((x),34) ^ ROTR((x),39))
380 #define Sigma1(x) (ROTR((x),14) ^ ROTR((x),18) ^ ROTR((x),41))
381 #define sigma0(x) (ROTR((x),1) ^ ROTR((x),8) ^ ((x)>>7))
382 #define sigma1(x) (ROTR((x),19) ^ ROTR((x),61) ^ ((x)>>6))
384 #define Ch(x,y,z) (((x) & (y)) ^ ((~(x)) & (z)))
385 #define Maj(x,y,z) (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))
387 #if defined(OPENSSL_IA32_SSE2) && !defined(OPENSSL_NO_ASM) && !defined(I386_ONLY)
388 #define GO_FOR_SSE2(ctx,in,num) do { \
389 void sha512_block_sse2(void *,const void *,size_t); \
390 if (!(OPENSSL_ia32cap_P & (1<<26))) break; \
391 sha512_block_sse2(ctx->h,in,num); return; \
395 #ifdef OPENSSL_SMALL_FOOTPRINT
397 static void sha512_block_data_order (SHA512_CTX *ctx, const void *in, size_t num)
399 const SHA_LONG64 *W=in;
400 SHA_LONG64 a,b,c,d,e,f,g,h,s0,s1,T1,T2;
405 GO_FOR_SSE2(ctx,in,num);
410 a = ctx->h[0]; b = ctx->h[1]; c = ctx->h[2]; d = ctx->h[3];
411 e = ctx->h[4]; f = ctx->h[5]; g = ctx->h[6]; h = ctx->h[7];
418 T1 = X[i] = PULL64(W[i]);
420 T1 += h + Sigma1(e) + Ch(e,f,g) + K512[i];
421 T2 = Sigma0(a) + Maj(a,b,c);
422 h = g; g = f; f = e; e = d + T1;
423 d = c; c = b; b = a; a = T1 + T2;
428 s0 = X[(i+1)&0x0f]; s0 = sigma0(s0);
429 s1 = X[(i+14)&0x0f]; s1 = sigma1(s1);
431 T1 = X[i&0xf] += s0 + s1 + X[(i+9)&0xf];
432 T1 += h + Sigma1(e) + Ch(e,f,g) + K512[i];
433 T2 = Sigma0(a) + Maj(a,b,c);
434 h = g; g = f; f = e; e = d + T1;
435 d = c; c = b; b = a; a = T1 + T2;
438 ctx->h[0] += a; ctx->h[1] += b; ctx->h[2] += c; ctx->h[3] += d;
439 ctx->h[4] += e; ctx->h[5] += f; ctx->h[6] += g; ctx->h[7] += h;
447 #define ROUND_00_15(i,a,b,c,d,e,f,g,h) do { \
448 T1 += h + Sigma1(e) + Ch(e,f,g) + K512[i]; \
449 h = Sigma0(a) + Maj(a,b,c); \
450 d += T1; h += T1; } while (0)
452 #define ROUND_16_80(i,a,b,c,d,e,f,g,h,X) do { \
453 s0 = X[(i+1)&0x0f]; s0 = sigma0(s0); \
454 s1 = X[(i+14)&0x0f]; s1 = sigma1(s1); \
455 T1 = X[(i)&0x0f] += s0 + s1 + X[(i+9)&0x0f]; \
456 ROUND_00_15(i,a,b,c,d,e,f,g,h); } while (0)
458 static void sha512_block_data_order (SHA512_CTX *ctx, const void *in, size_t num)
460 const SHA_LONG64 *W=in;
461 SHA_LONG64 a,b,c,d,e,f,g,h,s0,s1,T1;
466 GO_FOR_SSE2(ctx,in,num);
471 a = ctx->h[0]; b = ctx->h[1]; c = ctx->h[2]; d = ctx->h[3];
472 e = ctx->h[4]; f = ctx->h[5]; g = ctx->h[6]; h = ctx->h[7];
475 T1 = X[0] = W[0]; ROUND_00_15(0,a,b,c,d,e,f,g,h);
476 T1 = X[1] = W[1]; ROUND_00_15(1,h,a,b,c,d,e,f,g);
477 T1 = X[2] = W[2]; ROUND_00_15(2,g,h,a,b,c,d,e,f);
478 T1 = X[3] = W[3]; ROUND_00_15(3,f,g,h,a,b,c,d,e);
479 T1 = X[4] = W[4]; ROUND_00_15(4,e,f,g,h,a,b,c,d);
480 T1 = X[5] = W[5]; ROUND_00_15(5,d,e,f,g,h,a,b,c);
481 T1 = X[6] = W[6]; ROUND_00_15(6,c,d,e,f,g,h,a,b);
482 T1 = X[7] = W[7]; ROUND_00_15(7,b,c,d,e,f,g,h,a);
483 T1 = X[8] = W[8]; ROUND_00_15(8,a,b,c,d,e,f,g,h);
484 T1 = X[9] = W[9]; ROUND_00_15(9,h,a,b,c,d,e,f,g);
485 T1 = X[10] = W[10]; ROUND_00_15(10,g,h,a,b,c,d,e,f);
486 T1 = X[11] = W[11]; ROUND_00_15(11,f,g,h,a,b,c,d,e);
487 T1 = X[12] = W[12]; ROUND_00_15(12,e,f,g,h,a,b,c,d);
488 T1 = X[13] = W[13]; ROUND_00_15(13,d,e,f,g,h,a,b,c);
489 T1 = X[14] = W[14]; ROUND_00_15(14,c,d,e,f,g,h,a,b);
490 T1 = X[15] = W[15]; ROUND_00_15(15,b,c,d,e,f,g,h,a);
492 T1 = X[0] = PULL64(W[0]); ROUND_00_15(0,a,b,c,d,e,f,g,h);
493 T1 = X[1] = PULL64(W[1]); ROUND_00_15(1,h,a,b,c,d,e,f,g);
494 T1 = X[2] = PULL64(W[2]); ROUND_00_15(2,g,h,a,b,c,d,e,f);
495 T1 = X[3] = PULL64(W[3]); ROUND_00_15(3,f,g,h,a,b,c,d,e);
496 T1 = X[4] = PULL64(W[4]); ROUND_00_15(4,e,f,g,h,a,b,c,d);
497 T1 = X[5] = PULL64(W[5]); ROUND_00_15(5,d,e,f,g,h,a,b,c);
498 T1 = X[6] = PULL64(W[6]); ROUND_00_15(6,c,d,e,f,g,h,a,b);
499 T1 = X[7] = PULL64(W[7]); ROUND_00_15(7,b,c,d,e,f,g,h,a);
500 T1 = X[8] = PULL64(W[8]); ROUND_00_15(8,a,b,c,d,e,f,g,h);
501 T1 = X[9] = PULL64(W[9]); ROUND_00_15(9,h,a,b,c,d,e,f,g);
502 T1 = X[10] = PULL64(W[10]); ROUND_00_15(10,g,h,a,b,c,d,e,f);
503 T1 = X[11] = PULL64(W[11]); ROUND_00_15(11,f,g,h,a,b,c,d,e);
504 T1 = X[12] = PULL64(W[12]); ROUND_00_15(12,e,f,g,h,a,b,c,d);
505 T1 = X[13] = PULL64(W[13]); ROUND_00_15(13,d,e,f,g,h,a,b,c);
506 T1 = X[14] = PULL64(W[14]); ROUND_00_15(14,c,d,e,f,g,h,a,b);
507 T1 = X[15] = PULL64(W[15]); ROUND_00_15(15,b,c,d,e,f,g,h,a);
512 ROUND_16_80(i+0,a,b,c,d,e,f,g,h,X);
513 ROUND_16_80(i+1,h,a,b,c,d,e,f,g,X);
514 ROUND_16_80(i+2,g,h,a,b,c,d,e,f,X);
515 ROUND_16_80(i+3,f,g,h,a,b,c,d,e,X);
516 ROUND_16_80(i+4,e,f,g,h,a,b,c,d,X);
517 ROUND_16_80(i+5,d,e,f,g,h,a,b,c,X);
518 ROUND_16_80(i+6,c,d,e,f,g,h,a,b,X);
519 ROUND_16_80(i+7,b,c,d,e,f,g,h,a,X);
522 ctx->h[0] += a; ctx->h[1] += b; ctx->h[2] += c; ctx->h[3] += d;
523 ctx->h[4] += e; ctx->h[5] += f; ctx->h[6] += g; ctx->h[7] += h;
531 #endif /* SHA512_ASM */
533 #endif /* OPENSSL_NO_SHA512 */