xref: /illumos-gate/usr/src/common/crypto/sha2/sha2.c (revision 4b56a003)
1 /*
2  * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
3  * Use is subject to license terms.
4  */
5 
6 /*
7  * The basic framework for this code came from the reference
8  * implementation for MD5.  That implementation is Copyright (C)
9  * 1991-2, RSA Data Security, Inc. Created 1991. All rights reserved.
10  *
11  * License to copy and use this software is granted provided that it
12  * is identified as the "RSA Data Security, Inc. MD5 Message-Digest
13  * Algorithm" in all material mentioning or referencing this software
14  * or this function.
15  *
16  * License is also granted to make and use derivative works provided
17  * that such works are identified as "derived from the RSA Data
18  * Security, Inc. MD5 Message-Digest Algorithm" in all material
19  * mentioning or referencing the derived work.
20  *
21  * RSA Data Security, Inc. makes no representations concerning either
22  * the merchantability of this software or the suitability of this
23  * software for any particular purpose. It is provided "as is"
24  * without express or implied warranty of any kind.
25  *
26  * These notices must be retained in any copies of any part of this
27  * documentation and/or software.
28  *
29  * NOTE: Cleaned-up and optimized, version of SHA2, based on the FIPS 180-2
30  * standard, available at
31  * http://csrc.nist.gov/publications/fips/fips180-2/fips180-2.pdf
32  * Not as fast as one would like -- further optimizations are encouraged
33  * and appreciated.
34  */
35 
36 #include <sys/types.h>
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/sysmacros.h>
40 #define	_SHA2_IMPL
41 #include <sys/sha2.h>
42 #include <sys/sha2_consts.h>
43 
44 #ifdef _KERNEL
45 #include <sys/cmn_err.h>
46 
47 #else
48 #include <strings.h>
49 #include <stdlib.h>
50 #include <errno.h>
51 
52 #pragma weak SHA256Update = SHA2Update
53 #pragma weak SHA384Update = SHA2Update
54 #pragma weak SHA512Update = SHA2Update
55 
56 #pragma weak SHA256Final = SHA2Final
57 #pragma weak SHA384Final = SHA2Final
58 #pragma weak SHA512Final = SHA2Final
59 
60 #endif	/* _KERNEL */
61 
62 #ifdef _LITTLE_ENDIAN
63 #include <sys/byteorder.h>
64 #define	HAVE_HTONL
65 #endif
66 
67 static void Encode(uint8_t *, uint32_t *, size_t);
68 static void Encode64(uint8_t *, uint64_t *, size_t);
69 
70 #if	defined(__amd64)
71 #define	SHA512Transform(ctx, in) SHA512TransformBlocks((ctx), (in), 1)
72 #define	SHA256Transform(ctx, in) SHA256TransformBlocks((ctx), (in), 1)
73 
74 void SHA512TransformBlocks(SHA2_CTX *ctx, const void *in, size_t num);
75 void SHA256TransformBlocks(SHA2_CTX *ctx, const void *in, size_t num);
76 
77 #else
78 static void SHA256Transform(SHA2_CTX *, const uint8_t *);
79 static void SHA512Transform(SHA2_CTX *, const uint8_t *);
80 #endif	/* __amd64 */
81 
82 static uint8_t PADDING[128] = { 0x80, /* all zeros */ };
83 
84 /* Ch and Maj are the basic SHA2 functions. */
85 #define	Ch(b, c, d)	(((b) & (c)) ^ ((~b) & (d)))
86 #define	Maj(b, c, d)	(((b) & (c)) ^ ((b) & (d)) ^ ((c) & (d)))
87 
88 /* Rotates x right n bits. */
89 #define	ROTR(x, n)	\
90 	(((x) >> (n)) | ((x) << ((sizeof (x) * NBBY)-(n))))
91 
92 /* Shift x right n bits */
93 #define	SHR(x, n)	((x) >> (n))
94 
95 /* SHA256 Functions */
96 #define	BIGSIGMA0_256(x)	(ROTR((x), 2) ^ ROTR((x), 13) ^ ROTR((x), 22))
97 #define	BIGSIGMA1_256(x)	(ROTR((x), 6) ^ ROTR((x), 11) ^ ROTR((x), 25))
98 #define	SIGMA0_256(x)		(ROTR((x), 7) ^ ROTR((x), 18) ^ SHR((x), 3))
99 #define	SIGMA1_256(x)		(ROTR((x), 17) ^ ROTR((x), 19) ^ SHR((x), 10))
100 
101 #define	SHA256ROUND(a, b, c, d, e, f, g, h, i, w)			\
102 	T1 = h + BIGSIGMA1_256(e) + Ch(e, f, g) + SHA256_CONST(i) + w;	\
103 	d += T1;							\
104 	T2 = BIGSIGMA0_256(a) + Maj(a, b, c);				\
105 	h = T1 + T2
106 
107 /* SHA384/512 Functions */
108 #define	BIGSIGMA0(x)	(ROTR((x), 28) ^ ROTR((x), 34) ^ ROTR((x), 39))
109 #define	BIGSIGMA1(x)	(ROTR((x), 14) ^ ROTR((x), 18) ^ ROTR((x), 41))
110 #define	SIGMA0(x)	(ROTR((x), 1) ^ ROTR((x), 8) ^ SHR((x), 7))
111 #define	SIGMA1(x)	(ROTR((x), 19) ^ ROTR((x), 61) ^ SHR((x), 6))
112 #define	SHA512ROUND(a, b, c, d, e, f, g, h, i, w)			\
113 	T1 = h + BIGSIGMA1(e) + Ch(e, f, g) + SHA512_CONST(i) + w;	\
114 	d += T1;							\
115 	T2 = BIGSIGMA0(a) + Maj(a, b, c);				\
116 	h = T1 + T2
117 
118 /*
119  * sparc optimization:
120  *
121  * on the sparc, we can load big endian 32-bit data easily.  note that
122  * special care must be taken to ensure the address is 32-bit aligned.
123  * in the interest of speed, we don't check to make sure, since
124  * careful programming can guarantee this for us.
125  */
126 
127 #if	defined(_BIG_ENDIAN)
128 #define	LOAD_BIG_32(addr)	(*(uint32_t *)(addr))
129 #define	LOAD_BIG_64(addr)	(*(uint64_t *)(addr))
130 
131 #elif	defined(HAVE_HTONL)
132 #define	LOAD_BIG_32(addr) htonl(*((uint32_t *)(addr)))
133 #define	LOAD_BIG_64(addr) htonll(*((uint64_t *)(addr)))
134 
135 #else
136 /* little endian -- will work on big endian, but slowly */
137 #define	LOAD_BIG_32(addr)	\
138 	(((addr)[0] << 24) | ((addr)[1] << 16) | ((addr)[2] << 8) | (addr)[3])
139 #define	LOAD_BIG_64(addr)	\
140 	(((uint64_t)(addr)[0] << 56) | ((uint64_t)(addr)[1] << 48) |	\
141 	    ((uint64_t)(addr)[2] << 40) | ((uint64_t)(addr)[3] << 32) |	\
142 	    ((uint64_t)(addr)[4] << 24) | ((uint64_t)(addr)[5] << 16) |	\
143 	    ((uint64_t)(addr)[6] << 8) | (uint64_t)(addr)[7])
144 #endif	/* _BIG_ENDIAN */
145 
146 
147 #if	!defined(__amd64)
148 /* SHA256 Transform */
149 
150 static void
151 SHA256Transform(SHA2_CTX *ctx, const uint8_t *blk)
152 {
153 	uint32_t a = ctx->state.s32[0];
154 	uint32_t b = ctx->state.s32[1];
155 	uint32_t c = ctx->state.s32[2];
156 	uint32_t d = ctx->state.s32[3];
157 	uint32_t e = ctx->state.s32[4];
158 	uint32_t f = ctx->state.s32[5];
159 	uint32_t g = ctx->state.s32[6];
160 	uint32_t h = ctx->state.s32[7];
161 
162 	uint32_t w0, w1, w2, w3, w4, w5, w6, w7;
163 	uint32_t w8, w9, w10, w11, w12, w13, w14, w15;
164 	uint32_t T1, T2;
165 
166 #if	defined(__sparc)
167 	static const uint32_t sha256_consts[] = {
168 		SHA256_CONST_0, SHA256_CONST_1, SHA256_CONST_2,
169 		SHA256_CONST_3, SHA256_CONST_4, SHA256_CONST_5,
170 		SHA256_CONST_6, SHA256_CONST_7, SHA256_CONST_8,
171 		SHA256_CONST_9, SHA256_CONST_10, SHA256_CONST_11,
172 		SHA256_CONST_12, SHA256_CONST_13, SHA256_CONST_14,
173 		SHA256_CONST_15, SHA256_CONST_16, SHA256_CONST_17,
174 		SHA256_CONST_18, SHA256_CONST_19, SHA256_CONST_20,
175 		SHA256_CONST_21, SHA256_CONST_22, SHA256_CONST_23,
176 		SHA256_CONST_24, SHA256_CONST_25, SHA256_CONST_26,
177 		SHA256_CONST_27, SHA256_CONST_28, SHA256_CONST_29,
178 		SHA256_CONST_30, SHA256_CONST_31, SHA256_CONST_32,
179 		SHA256_CONST_33, SHA256_CONST_34, SHA256_CONST_35,
180 		SHA256_CONST_36, SHA256_CONST_37, SHA256_CONST_38,
181 		SHA256_CONST_39, SHA256_CONST_40, SHA256_CONST_41,
182 		SHA256_CONST_42, SHA256_CONST_43, SHA256_CONST_44,
183 		SHA256_CONST_45, SHA256_CONST_46, SHA256_CONST_47,
184 		SHA256_CONST_48, SHA256_CONST_49, SHA256_CONST_50,
185 		SHA256_CONST_51, SHA256_CONST_52, SHA256_CONST_53,
186 		SHA256_CONST_54, SHA256_CONST_55, SHA256_CONST_56,
187 		SHA256_CONST_57, SHA256_CONST_58, SHA256_CONST_59,
188 		SHA256_CONST_60, SHA256_CONST_61, SHA256_CONST_62,
189 		SHA256_CONST_63
190 	};
191 #endif	/* __sparc */
192 
193 	if ((uintptr_t)blk & 0x3) {		/* not 4-byte aligned? */
194 		bcopy(blk, ctx->buf_un.buf32,  sizeof (ctx->buf_un.buf32));
195 		blk = (uint8_t *)ctx->buf_un.buf32;
196 	}
197 
198 	/* LINTED E_BAD_PTR_CAST_ALIGN */
199 	w0 =  LOAD_BIG_32(blk + 4 * 0);
200 	SHA256ROUND(a, b, c, d, e, f, g, h, 0, w0);
201 	/* LINTED E_BAD_PTR_CAST_ALIGN */
202 	w1 =  LOAD_BIG_32(blk + 4 * 1);
203 	SHA256ROUND(h, a, b, c, d, e, f, g, 1, w1);
204 	/* LINTED E_BAD_PTR_CAST_ALIGN */
205 	w2 =  LOAD_BIG_32(blk + 4 * 2);
206 	SHA256ROUND(g, h, a, b, c, d, e, f, 2, w2);
207 	/* LINTED E_BAD_PTR_CAST_ALIGN */
208 	w3 =  LOAD_BIG_32(blk + 4 * 3);
209 	SHA256ROUND(f, g, h, a, b, c, d, e, 3, w3);
210 	/* LINTED E_BAD_PTR_CAST_ALIGN */
211 	w4 =  LOAD_BIG_32(blk + 4 * 4);
212 	SHA256ROUND(e, f, g, h, a, b, c, d, 4, w4);
213 	/* LINTED E_BAD_PTR_CAST_ALIGN */
214 	w5 =  LOAD_BIG_32(blk + 4 * 5);
215 	SHA256ROUND(d, e, f, g, h, a, b, c, 5, w5);
216 	/* LINTED E_BAD_PTR_CAST_ALIGN */
217 	w6 =  LOAD_BIG_32(blk + 4 * 6);
218 	SHA256ROUND(c, d, e, f, g, h, a, b, 6, w6);
219 	/* LINTED E_BAD_PTR_CAST_ALIGN */
220 	w7 =  LOAD_BIG_32(blk + 4 * 7);
221 	SHA256ROUND(b, c, d, e, f, g, h, a, 7, w7);
222 	/* LINTED E_BAD_PTR_CAST_ALIGN */
223 	w8 =  LOAD_BIG_32(blk + 4 * 8);
224 	SHA256ROUND(a, b, c, d, e, f, g, h, 8, w8);
225 	/* LINTED E_BAD_PTR_CAST_ALIGN */
226 	w9 =  LOAD_BIG_32(blk + 4 * 9);
227 	SHA256ROUND(h, a, b, c, d, e, f, g, 9, w9);
228 	/* LINTED E_BAD_PTR_CAST_ALIGN */
229 	w10 =  LOAD_BIG_32(blk + 4 * 10);
230 	SHA256ROUND(g, h, a, b, c, d, e, f, 10, w10);
231 	/* LINTED E_BAD_PTR_CAST_ALIGN */
232 	w11 =  LOAD_BIG_32(blk + 4 * 11);
233 	SHA256ROUND(f, g, h, a, b, c, d, e, 11, w11);
234 	/* LINTED E_BAD_PTR_CAST_ALIGN */
235 	w12 =  LOAD_BIG_32(blk + 4 * 12);
236 	SHA256ROUND(e, f, g, h, a, b, c, d, 12, w12);
237 	/* LINTED E_BAD_PTR_CAST_ALIGN */
238 	w13 =  LOAD_BIG_32(blk + 4 * 13);
239 	SHA256ROUND(d, e, f, g, h, a, b, c, 13, w13);
240 	/* LINTED E_BAD_PTR_CAST_ALIGN */
241 	w14 =  LOAD_BIG_32(blk + 4 * 14);
242 	SHA256ROUND(c, d, e, f, g, h, a, b, 14, w14);
243 	/* LINTED E_BAD_PTR_CAST_ALIGN */
244 	w15 =  LOAD_BIG_32(blk + 4 * 15);
245 	SHA256ROUND(b, c, d, e, f, g, h, a, 15, w15);
246 
247 	w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
248 	SHA256ROUND(a, b, c, d, e, f, g, h, 16, w0);
249 	w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
250 	SHA256ROUND(h, a, b, c, d, e, f, g, 17, w1);
251 	w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
252 	SHA256ROUND(g, h, a, b, c, d, e, f, 18, w2);
253 	w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
254 	SHA256ROUND(f, g, h, a, b, c, d, e, 19, w3);
255 	w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
256 	SHA256ROUND(e, f, g, h, a, b, c, d, 20, w4);
257 	w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
258 	SHA256ROUND(d, e, f, g, h, a, b, c, 21, w5);
259 	w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
260 	SHA256ROUND(c, d, e, f, g, h, a, b, 22, w6);
261 	w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
262 	SHA256ROUND(b, c, d, e, f, g, h, a, 23, w7);
263 	w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
264 	SHA256ROUND(a, b, c, d, e, f, g, h, 24, w8);
265 	w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
266 	SHA256ROUND(h, a, b, c, d, e, f, g, 25, w9);
267 	w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
268 	SHA256ROUND(g, h, a, b, c, d, e, f, 26, w10);
269 	w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
270 	SHA256ROUND(f, g, h, a, b, c, d, e, 27, w11);
271 	w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
272 	SHA256ROUND(e, f, g, h, a, b, c, d, 28, w12);
273 	w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
274 	SHA256ROUND(d, e, f, g, h, a, b, c, 29, w13);
275 	w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
276 	SHA256ROUND(c, d, e, f, g, h, a, b, 30, w14);
277 	w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
278 	SHA256ROUND(b, c, d, e, f, g, h, a, 31, w15);
279 
280 	w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
281 	SHA256ROUND(a, b, c, d, e, f, g, h, 32, w0);
282 	w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
283 	SHA256ROUND(h, a, b, c, d, e, f, g, 33, w1);
284 	w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
285 	SHA256ROUND(g, h, a, b, c, d, e, f, 34, w2);
286 	w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
287 	SHA256ROUND(f, g, h, a, b, c, d, e, 35, w3);
288 	w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
289 	SHA256ROUND(e, f, g, h, a, b, c, d, 36, w4);
290 	w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
291 	SHA256ROUND(d, e, f, g, h, a, b, c, 37, w5);
292 	w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
293 	SHA256ROUND(c, d, e, f, g, h, a, b, 38, w6);
294 	w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
295 	SHA256ROUND(b, c, d, e, f, g, h, a, 39, w7);
296 	w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
297 	SHA256ROUND(a, b, c, d, e, f, g, h, 40, w8);
298 	w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
299 	SHA256ROUND(h, a, b, c, d, e, f, g, 41, w9);
300 	w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
301 	SHA256ROUND(g, h, a, b, c, d, e, f, 42, w10);
302 	w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
303 	SHA256ROUND(f, g, h, a, b, c, d, e, 43, w11);
304 	w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
305 	SHA256ROUND(e, f, g, h, a, b, c, d, 44, w12);
306 	w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
307 	SHA256ROUND(d, e, f, g, h, a, b, c, 45, w13);
308 	w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
309 	SHA256ROUND(c, d, e, f, g, h, a, b, 46, w14);
310 	w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
311 	SHA256ROUND(b, c, d, e, f, g, h, a, 47, w15);
312 
313 	w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
314 	SHA256ROUND(a, b, c, d, e, f, g, h, 48, w0);
315 	w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
316 	SHA256ROUND(h, a, b, c, d, e, f, g, 49, w1);
317 	w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
318 	SHA256ROUND(g, h, a, b, c, d, e, f, 50, w2);
319 	w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
320 	SHA256ROUND(f, g, h, a, b, c, d, e, 51, w3);
321 	w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
322 	SHA256ROUND(e, f, g, h, a, b, c, d, 52, w4);
323 	w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
324 	SHA256ROUND(d, e, f, g, h, a, b, c, 53, w5);
325 	w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
326 	SHA256ROUND(c, d, e, f, g, h, a, b, 54, w6);
327 	w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
328 	SHA256ROUND(b, c, d, e, f, g, h, a, 55, w7);
329 	w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
330 	SHA256ROUND(a, b, c, d, e, f, g, h, 56, w8);
331 	w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
332 	SHA256ROUND(h, a, b, c, d, e, f, g, 57, w9);
333 	w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
334 	SHA256ROUND(g, h, a, b, c, d, e, f, 58, w10);
335 	w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
336 	SHA256ROUND(f, g, h, a, b, c, d, e, 59, w11);
337 	w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
338 	SHA256ROUND(e, f, g, h, a, b, c, d, 60, w12);
339 	w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
340 	SHA256ROUND(d, e, f, g, h, a, b, c, 61, w13);
341 	w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
342 	SHA256ROUND(c, d, e, f, g, h, a, b, 62, w14);
343 	w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
344 	SHA256ROUND(b, c, d, e, f, g, h, a, 63, w15);
345 
346 	ctx->state.s32[0] += a;
347 	ctx->state.s32[1] += b;
348 	ctx->state.s32[2] += c;
349 	ctx->state.s32[3] += d;
350 	ctx->state.s32[4] += e;
351 	ctx->state.s32[5] += f;
352 	ctx->state.s32[6] += g;
353 	ctx->state.s32[7] += h;
354 }
355 
356 
357 /* SHA384 and SHA512 Transform */
358 
359 static void
360 SHA512Transform(SHA2_CTX *ctx, const uint8_t *blk)
361 {
362 
363 	uint64_t a = ctx->state.s64[0];
364 	uint64_t b = ctx->state.s64[1];
365 	uint64_t c = ctx->state.s64[2];
366 	uint64_t d = ctx->state.s64[3];
367 	uint64_t e = ctx->state.s64[4];
368 	uint64_t f = ctx->state.s64[5];
369 	uint64_t g = ctx->state.s64[6];
370 	uint64_t h = ctx->state.s64[7];
371 
372 	uint64_t w0, w1, w2, w3, w4, w5, w6, w7;
373 	uint64_t w8, w9, w10, w11, w12, w13, w14, w15;
374 	uint64_t T1, T2;
375 
376 #if	defined(__sparc)
377 	static const uint64_t sha512_consts[] = {
378 		SHA512_CONST_0, SHA512_CONST_1, SHA512_CONST_2,
379 		SHA512_CONST_3, SHA512_CONST_4, SHA512_CONST_5,
380 		SHA512_CONST_6, SHA512_CONST_7, SHA512_CONST_8,
381 		SHA512_CONST_9, SHA512_CONST_10, SHA512_CONST_11,
382 		SHA512_CONST_12, SHA512_CONST_13, SHA512_CONST_14,
383 		SHA512_CONST_15, SHA512_CONST_16, SHA512_CONST_17,
384 		SHA512_CONST_18, SHA512_CONST_19, SHA512_CONST_20,
385 		SHA512_CONST_21, SHA512_CONST_22, SHA512_CONST_23,
386 		SHA512_CONST_24, SHA512_CONST_25, SHA512_CONST_26,
387 		SHA512_CONST_27, SHA512_CONST_28, SHA512_CONST_29,
388 		SHA512_CONST_30, SHA512_CONST_31, SHA512_CONST_32,
389 		SHA512_CONST_33, SHA512_CONST_34, SHA512_CONST_35,
390 		SHA512_CONST_36, SHA512_CONST_37, SHA512_CONST_38,
391 		SHA512_CONST_39, SHA512_CONST_40, SHA512_CONST_41,
392 		SHA512_CONST_42, SHA512_CONST_43, SHA512_CONST_44,
393 		SHA512_CONST_45, SHA512_CONST_46, SHA512_CONST_47,
394 		SHA512_CONST_48, SHA512_CONST_49, SHA512_CONST_50,
395 		SHA512_CONST_51, SHA512_CONST_52, SHA512_CONST_53,
396 		SHA512_CONST_54, SHA512_CONST_55, SHA512_CONST_56,
397 		SHA512_CONST_57, SHA512_CONST_58, SHA512_CONST_59,
398 		SHA512_CONST_60, SHA512_CONST_61, SHA512_CONST_62,
399 		SHA512_CONST_63, SHA512_CONST_64, SHA512_CONST_65,
400 		SHA512_CONST_66, SHA512_CONST_67, SHA512_CONST_68,
401 		SHA512_CONST_69, SHA512_CONST_70, SHA512_CONST_71,
402 		SHA512_CONST_72, SHA512_CONST_73, SHA512_CONST_74,
403 		SHA512_CONST_75, SHA512_CONST_76, SHA512_CONST_77,
404 		SHA512_CONST_78, SHA512_CONST_79
405 	};
406 #endif	/* __sparc */
407 
408 
409 	if ((uintptr_t)blk & 0x7) {		/* not 8-byte aligned? */
410 		bcopy(blk, ctx->buf_un.buf64,  sizeof (ctx->buf_un.buf64));
411 		blk = (uint8_t *)ctx->buf_un.buf64;
412 	}
413 
414 	/* LINTED E_BAD_PTR_CAST_ALIGN */
415 	w0 =  LOAD_BIG_64(blk + 8 * 0);
416 	SHA512ROUND(a, b, c, d, e, f, g, h, 0, w0);
417 	/* LINTED E_BAD_PTR_CAST_ALIGN */
418 	w1 =  LOAD_BIG_64(blk + 8 * 1);
419 	SHA512ROUND(h, a, b, c, d, e, f, g, 1, w1);
420 	/* LINTED E_BAD_PTR_CAST_ALIGN */
421 	w2 =  LOAD_BIG_64(blk + 8 * 2);
422 	SHA512ROUND(g, h, a, b, c, d, e, f, 2, w2);
423 	/* LINTED E_BAD_PTR_CAST_ALIGN */
424 	w3 =  LOAD_BIG_64(blk + 8 * 3);
425 	SHA512ROUND(f, g, h, a, b, c, d, e, 3, w3);
426 	/* LINTED E_BAD_PTR_CAST_ALIGN */
427 	w4 =  LOAD_BIG_64(blk + 8 * 4);
428 	SHA512ROUND(e, f, g, h, a, b, c, d, 4, w4);
429 	/* LINTED E_BAD_PTR_CAST_ALIGN */
430 	w5 =  LOAD_BIG_64(blk + 8 * 5);
431 	SHA512ROUND(d, e, f, g, h, a, b, c, 5, w5);
432 	/* LINTED E_BAD_PTR_CAST_ALIGN */
433 	w6 =  LOAD_BIG_64(blk + 8 * 6);
434 	SHA512ROUND(c, d, e, f, g, h, a, b, 6, w6);
435 	/* LINTED E_BAD_PTR_CAST_ALIGN */
436 	w7 =  LOAD_BIG_64(blk + 8 * 7);
437 	SHA512ROUND(b, c, d, e, f, g, h, a, 7, w7);
438 	/* LINTED E_BAD_PTR_CAST_ALIGN */
439 	w8 =  LOAD_BIG_64(blk + 8 * 8);
440 	SHA512ROUND(a, b, c, d, e, f, g, h, 8, w8);
441 	/* LINTED E_BAD_PTR_CAST_ALIGN */
442 	w9 =  LOAD_BIG_64(blk + 8 * 9);
443 	SHA512ROUND(h, a, b, c, d, e, f, g, 9, w9);
444 	/* LINTED E_BAD_PTR_CAST_ALIGN */
445 	w10 =  LOAD_BIG_64(blk + 8 * 10);
446 	SHA512ROUND(g, h, a, b, c, d, e, f, 10, w10);
447 	/* LINTED E_BAD_PTR_CAST_ALIGN */
448 	w11 =  LOAD_BIG_64(blk + 8 * 11);
449 	SHA512ROUND(f, g, h, a, b, c, d, e, 11, w11);
450 	/* LINTED E_BAD_PTR_CAST_ALIGN */
451 	w12 =  LOAD_BIG_64(blk + 8 * 12);
452 	SHA512ROUND(e, f, g, h, a, b, c, d, 12, w12);
453 	/* LINTED E_BAD_PTR_CAST_ALIGN */
454 	w13 =  LOAD_BIG_64(blk + 8 * 13);
455 	SHA512ROUND(d, e, f, g, h, a, b, c, 13, w13);
456 	/* LINTED E_BAD_PTR_CAST_ALIGN */
457 	w14 =  LOAD_BIG_64(blk + 8 * 14);
458 	SHA512ROUND(c, d, e, f, g, h, a, b, 14, w14);
459 	/* LINTED E_BAD_PTR_CAST_ALIGN */
460 	w15 =  LOAD_BIG_64(blk + 8 * 15);
461 	SHA512ROUND(b, c, d, e, f, g, h, a, 15, w15);
462 
463 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
464 	SHA512ROUND(a, b, c, d, e, f, g, h, 16, w0);
465 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
466 	SHA512ROUND(h, a, b, c, d, e, f, g, 17, w1);
467 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
468 	SHA512ROUND(g, h, a, b, c, d, e, f, 18, w2);
469 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
470 	SHA512ROUND(f, g, h, a, b, c, d, e, 19, w3);
471 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
472 	SHA512ROUND(e, f, g, h, a, b, c, d, 20, w4);
473 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
474 	SHA512ROUND(d, e, f, g, h, a, b, c, 21, w5);
475 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
476 	SHA512ROUND(c, d, e, f, g, h, a, b, 22, w6);
477 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
478 	SHA512ROUND(b, c, d, e, f, g, h, a, 23, w7);
479 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
480 	SHA512ROUND(a, b, c, d, e, f, g, h, 24, w8);
481 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
482 	SHA512ROUND(h, a, b, c, d, e, f, g, 25, w9);
483 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
484 	SHA512ROUND(g, h, a, b, c, d, e, f, 26, w10);
485 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
486 	SHA512ROUND(f, g, h, a, b, c, d, e, 27, w11);
487 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
488 	SHA512ROUND(e, f, g, h, a, b, c, d, 28, w12);
489 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
490 	SHA512ROUND(d, e, f, g, h, a, b, c, 29, w13);
491 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
492 	SHA512ROUND(c, d, e, f, g, h, a, b, 30, w14);
493 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
494 	SHA512ROUND(b, c, d, e, f, g, h, a, 31, w15);
495 
496 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
497 	SHA512ROUND(a, b, c, d, e, f, g, h, 32, w0);
498 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
499 	SHA512ROUND(h, a, b, c, d, e, f, g, 33, w1);
500 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
501 	SHA512ROUND(g, h, a, b, c, d, e, f, 34, w2);
502 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
503 	SHA512ROUND(f, g, h, a, b, c, d, e, 35, w3);
504 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
505 	SHA512ROUND(e, f, g, h, a, b, c, d, 36, w4);
506 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
507 	SHA512ROUND(d, e, f, g, h, a, b, c, 37, w5);
508 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
509 	SHA512ROUND(c, d, e, f, g, h, a, b, 38, w6);
510 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
511 	SHA512ROUND(b, c, d, e, f, g, h, a, 39, w7);
512 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
513 	SHA512ROUND(a, b, c, d, e, f, g, h, 40, w8);
514 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
515 	SHA512ROUND(h, a, b, c, d, e, f, g, 41, w9);
516 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
517 	SHA512ROUND(g, h, a, b, c, d, e, f, 42, w10);
518 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
519 	SHA512ROUND(f, g, h, a, b, c, d, e, 43, w11);
520 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
521 	SHA512ROUND(e, f, g, h, a, b, c, d, 44, w12);
522 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
523 	SHA512ROUND(d, e, f, g, h, a, b, c, 45, w13);
524 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
525 	SHA512ROUND(c, d, e, f, g, h, a, b, 46, w14);
526 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
527 	SHA512ROUND(b, c, d, e, f, g, h, a, 47, w15);
528 
529 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
530 	SHA512ROUND(a, b, c, d, e, f, g, h, 48, w0);
531 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
532 	SHA512ROUND(h, a, b, c, d, e, f, g, 49, w1);
533 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
534 	SHA512ROUND(g, h, a, b, c, d, e, f, 50, w2);
535 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
536 	SHA512ROUND(f, g, h, a, b, c, d, e, 51, w3);
537 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
538 	SHA512ROUND(e, f, g, h, a, b, c, d, 52, w4);
539 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
540 	SHA512ROUND(d, e, f, g, h, a, b, c, 53, w5);
541 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
542 	SHA512ROUND(c, d, e, f, g, h, a, b, 54, w6);
543 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
544 	SHA512ROUND(b, c, d, e, f, g, h, a, 55, w7);
545 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
546 	SHA512ROUND(a, b, c, d, e, f, g, h, 56, w8);
547 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
548 	SHA512ROUND(h, a, b, c, d, e, f, g, 57, w9);
549 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
550 	SHA512ROUND(g, h, a, b, c, d, e, f, 58, w10);
551 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
552 	SHA512ROUND(f, g, h, a, b, c, d, e, 59, w11);
553 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
554 	SHA512ROUND(e, f, g, h, a, b, c, d, 60, w12);
555 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
556 	SHA512ROUND(d, e, f, g, h, a, b, c, 61, w13);
557 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
558 	SHA512ROUND(c, d, e, f, g, h, a, b, 62, w14);
559 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
560 	SHA512ROUND(b, c, d, e, f, g, h, a, 63, w15);
561 
562 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
563 	SHA512ROUND(a, b, c, d, e, f, g, h, 64, w0);
564 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
565 	SHA512ROUND(h, a, b, c, d, e, f, g, 65, w1);
566 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
567 	SHA512ROUND(g, h, a, b, c, d, e, f, 66, w2);
568 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
569 	SHA512ROUND(f, g, h, a, b, c, d, e, 67, w3);
570 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
571 	SHA512ROUND(e, f, g, h, a, b, c, d, 68, w4);
572 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
573 	SHA512ROUND(d, e, f, g, h, a, b, c, 69, w5);
574 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
575 	SHA512ROUND(c, d, e, f, g, h, a, b, 70, w6);
576 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
577 	SHA512ROUND(b, c, d, e, f, g, h, a, 71, w7);
578 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
579 	SHA512ROUND(a, b, c, d, e, f, g, h, 72, w8);
580 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
581 	SHA512ROUND(h, a, b, c, d, e, f, g, 73, w9);
582 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
583 	SHA512ROUND(g, h, a, b, c, d, e, f, 74, w10);
584 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
585 	SHA512ROUND(f, g, h, a, b, c, d, e, 75, w11);
586 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
587 	SHA512ROUND(e, f, g, h, a, b, c, d, 76, w12);
588 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
589 	SHA512ROUND(d, e, f, g, h, a, b, c, 77, w13);
590 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
591 	SHA512ROUND(c, d, e, f, g, h, a, b, 78, w14);
592 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
593 	SHA512ROUND(b, c, d, e, f, g, h, a, 79, w15);
594 
595 	ctx->state.s64[0] += a;
596 	ctx->state.s64[1] += b;
597 	ctx->state.s64[2] += c;
598 	ctx->state.s64[3] += d;
599 	ctx->state.s64[4] += e;
600 	ctx->state.s64[5] += f;
601 	ctx->state.s64[6] += g;
602 	ctx->state.s64[7] += h;
603 
604 }
605 #endif	/* !__amd64 */
606 
607 
608 /*
609  * Encode()
610  *
611  * purpose: to convert a list of numbers from little endian to big endian
612  *   input: uint8_t *	: place to store the converted big endian numbers
613  *	    uint32_t *	: place to get numbers to convert from
614  *          size_t	: the length of the input in bytes
615  *  output: void
616  */
617 
618 static void
619 Encode(uint8_t *_RESTRICT_KYWD output, uint32_t *_RESTRICT_KYWD input,
620     size_t len)
621 {
622 	size_t		i, j;
623 
624 #if	defined(__sparc)
625 	if (IS_P2ALIGNED(output, sizeof (uint32_t))) {
626 		for (i = 0, j = 0; j < len; i++, j += 4) {
627 			/* LINTED: pointer alignment */
628 			*((uint32_t *)(output + j)) = input[i];
629 		}
630 	} else {
631 #endif	/* little endian -- will work on big endian, but slowly */
632 		for (i = 0, j = 0; j < len; i++, j += 4) {
633 			output[j]	= (input[i] >> 24) & 0xff;
634 			output[j + 1]	= (input[i] >> 16) & 0xff;
635 			output[j + 2]	= (input[i] >>  8) & 0xff;
636 			output[j + 3]	= input[i] & 0xff;
637 		}
638 #if	defined(__sparc)
639 	}
640 #endif
641 }
642 
643 static void
644 Encode64(uint8_t *_RESTRICT_KYWD output, uint64_t *_RESTRICT_KYWD input,
645     size_t len)
646 {
647 	size_t		i, j;
648 
649 #if	defined(__sparc)
650 	if (IS_P2ALIGNED(output, sizeof (uint64_t))) {
651 		for (i = 0, j = 0; j < len; i++, j += 8) {
652 			/* LINTED: pointer alignment */
653 			*((uint64_t *)(output + j)) = input[i];
654 		}
655 	} else {
656 #endif	/* little endian -- will work on big endian, but slowly */
657 		for (i = 0, j = 0; j < len; i++, j += 8) {
658 
659 			output[j]	= (input[i] >> 56) & 0xff;
660 			output[j + 1]	= (input[i] >> 48) & 0xff;
661 			output[j + 2]	= (input[i] >> 40) & 0xff;
662 			output[j + 3]	= (input[i] >> 32) & 0xff;
663 			output[j + 4]	= (input[i] >> 24) & 0xff;
664 			output[j + 5]	= (input[i] >> 16) & 0xff;
665 			output[j + 6]	= (input[i] >>  8) & 0xff;
666 			output[j + 7]	= input[i] & 0xff;
667 		}
668 #if	defined(__sparc)
669 	}
670 #endif
671 }
672 
673 
674 void
675 SHA2Init(uint64_t mech, SHA2_CTX *ctx)
676 {
677 
678 	switch (mech) {
679 	case SHA256_MECH_INFO_TYPE:
680 	case SHA256_HMAC_MECH_INFO_TYPE:
681 	case SHA256_HMAC_GEN_MECH_INFO_TYPE:
682 		ctx->state.s32[0] = 0x6a09e667U;
683 		ctx->state.s32[1] = 0xbb67ae85U;
684 		ctx->state.s32[2] = 0x3c6ef372U;
685 		ctx->state.s32[3] = 0xa54ff53aU;
686 		ctx->state.s32[4] = 0x510e527fU;
687 		ctx->state.s32[5] = 0x9b05688cU;
688 		ctx->state.s32[6] = 0x1f83d9abU;
689 		ctx->state.s32[7] = 0x5be0cd19U;
690 		break;
691 	case SHA384_MECH_INFO_TYPE:
692 	case SHA384_HMAC_MECH_INFO_TYPE:
693 	case SHA384_HMAC_GEN_MECH_INFO_TYPE:
694 		ctx->state.s64[0] = 0xcbbb9d5dc1059ed8ULL;
695 		ctx->state.s64[1] = 0x629a292a367cd507ULL;
696 		ctx->state.s64[2] = 0x9159015a3070dd17ULL;
697 		ctx->state.s64[3] = 0x152fecd8f70e5939ULL;
698 		ctx->state.s64[4] = 0x67332667ffc00b31ULL;
699 		ctx->state.s64[5] = 0x8eb44a8768581511ULL;
700 		ctx->state.s64[6] = 0xdb0c2e0d64f98fa7ULL;
701 		ctx->state.s64[7] = 0x47b5481dbefa4fa4ULL;
702 		break;
703 	case SHA512_MECH_INFO_TYPE:
704 	case SHA512_HMAC_MECH_INFO_TYPE:
705 	case SHA512_HMAC_GEN_MECH_INFO_TYPE:
706 		ctx->state.s64[0] = 0x6a09e667f3bcc908ULL;
707 		ctx->state.s64[1] = 0xbb67ae8584caa73bULL;
708 		ctx->state.s64[2] = 0x3c6ef372fe94f82bULL;
709 		ctx->state.s64[3] = 0xa54ff53a5f1d36f1ULL;
710 		ctx->state.s64[4] = 0x510e527fade682d1ULL;
711 		ctx->state.s64[5] = 0x9b05688c2b3e6c1fULL;
712 		ctx->state.s64[6] = 0x1f83d9abfb41bd6bULL;
713 		ctx->state.s64[7] = 0x5be0cd19137e2179ULL;
714 		break;
715 #ifdef _KERNEL
716 	default:
717 		cmn_err(CE_PANIC,
718 		    "sha2_init: failed to find a supported algorithm: 0x%x",
719 		    (uint32_t)mech);
720 
721 #endif /* _KERNEL */
722 	}
723 
724 	ctx->algotype = mech;
725 	ctx->count.c64[0] = ctx->count.c64[1] = 0;
726 }
727 
728 #ifndef _KERNEL
729 
730 #pragma inline(SHA256Init, SHA384Init, SHA512Init)
731 void
732 SHA256Init(SHA256_CTX *ctx)
733 {
734 	SHA2Init(SHA256, ctx);
735 }
736 
737 void
738 SHA384Init(SHA384_CTX *ctx)
739 {
740 	SHA2Init(SHA384, ctx);
741 }
742 
743 void
744 SHA512Init(SHA512_CTX *ctx)
745 {
746 	SHA2Init(SHA512, ctx);
747 }
748 
749 #endif /* _KERNEL */
750 
751 /*
752  * SHA2Update()
753  *
754  * purpose: continues an sha2 digest operation, using the message block
755  *          to update the context.
756  *   input: SHA2_CTX *	: the context to update
757  *          void *	: the message block
758  *          size_t      : the length of the message block, in bytes
759  *  output: void
760  */
761 
762 void
763 SHA2Update(SHA2_CTX *ctx, const void *inptr, size_t input_len)
764 {
765 	uint32_t	i, buf_index, buf_len, buf_limit;
766 	const uint8_t	*input = inptr;
767 	uint32_t	algotype = ctx->algotype;
768 #if defined(__amd64)
769 	uint32_t	block_count;
770 #endif	/* !__amd64 */
771 
772 
773 	/* check for noop */
774 	if (input_len == 0)
775 		return;
776 
777 	if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
778 		buf_limit = 64;
779 
780 		/* compute number of bytes mod 64 */
781 		buf_index = (ctx->count.c32[1] >> 3) & 0x3F;
782 
783 		/* update number of bits */
784 		if ((ctx->count.c32[1] += (input_len << 3)) < (input_len << 3))
785 			ctx->count.c32[0]++;
786 
787 		ctx->count.c32[0] += (input_len >> 29);
788 
789 	} else {
790 		buf_limit = 128;
791 
792 		/* compute number of bytes mod 128 */
793 		buf_index = (ctx->count.c64[1] >> 3) & 0x7F;
794 
795 		/* update number of bits */
796 		if ((ctx->count.c64[1] += (input_len << 3)) < (input_len << 3))
797 			ctx->count.c64[0]++;
798 
799 		ctx->count.c64[0] += (input_len >> 29);
800 	}
801 
802 	buf_len = buf_limit - buf_index;
803 
804 	/* transform as many times as possible */
805 	i = 0;
806 	if (input_len >= buf_len) {
807 
808 		/*
809 		 * general optimization:
810 		 *
811 		 * only do initial bcopy() and SHA2Transform() if
812 		 * buf_index != 0.  if buf_index == 0, we're just
813 		 * wasting our time doing the bcopy() since there
814 		 * wasn't any data left over from a previous call to
815 		 * SHA2Update().
816 		 */
817 		if (buf_index) {
818 			bcopy(input, &ctx->buf_un.buf8[buf_index], buf_len);
819 			if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE)
820 				SHA256Transform(ctx, ctx->buf_un.buf8);
821 			else
822 				SHA512Transform(ctx, ctx->buf_un.buf8);
823 
824 			i = buf_len;
825 		}
826 
827 #if !defined(__amd64)
828 		if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
829 			for (; i + buf_limit - 1 < input_len; i += buf_limit) {
830 				SHA256Transform(ctx, &input[i]);
831 			}
832 		} else {
833 			for (; i + buf_limit - 1 < input_len; i += buf_limit) {
834 				SHA512Transform(ctx, &input[i]);
835 			}
836 		}
837 
838 #else
839 		if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
840 			block_count = (input_len - i) >> 6;
841 			if (block_count > 0) {
842 				SHA256TransformBlocks(ctx, &input[i],
843 				    block_count);
844 				i += block_count << 6;
845 			}
846 		} else {
847 			block_count = (input_len - i) >> 7;
848 			if (block_count > 0) {
849 				SHA512TransformBlocks(ctx, &input[i],
850 				    block_count);
851 				i += block_count << 7;
852 			}
853 		}
854 #endif	/* !__amd64 */
855 
856 		/*
857 		 * general optimization:
858 		 *
859 		 * if i and input_len are the same, return now instead
860 		 * of calling bcopy(), since the bcopy() in this case
861 		 * will be an expensive noop.
862 		 */
863 
864 		if (input_len == i)
865 			return;
866 
867 		buf_index = 0;
868 	}
869 
870 	/* buffer remaining input */
871 	bcopy(&input[i], &ctx->buf_un.buf8[buf_index], input_len - i);
872 }
873 
874 
875 /*
876  * SHA2Final()
877  *
878  * purpose: ends an sha2 digest operation, finalizing the message digest and
879  *          zeroing the context.
880  *   input: uchar_t *	: a buffer to store the digest
881  *			: The function actually uses void* because many
882  *			: callers pass things other than uchar_t here.
883  *          SHA2_CTX *  : the context to finalize, save, and zero
884  *  output: void
885  */
886 
887 void
888 SHA2Final(void *digest, SHA2_CTX *ctx)
889 {
890 	uint8_t		bitcount_be[sizeof (ctx->count.c32)];
891 	uint8_t		bitcount_be64[sizeof (ctx->count.c64)];
892 	uint32_t	index;
893 	uint32_t	algotype = ctx->algotype;
894 
895 	if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
896 		index  = (ctx->count.c32[1] >> 3) & 0x3f;
897 		Encode(bitcount_be, ctx->count.c32, sizeof (bitcount_be));
898 		SHA2Update(ctx, PADDING, ((index < 56) ? 56 : 120) - index);
899 		SHA2Update(ctx, bitcount_be, sizeof (bitcount_be));
900 		Encode(digest, ctx->state.s32, sizeof (ctx->state.s32));
901 
902 	} else {
903 		index  = (ctx->count.c64[1] >> 3) & 0x7f;
904 		Encode64(bitcount_be64, ctx->count.c64,
905 		    sizeof (bitcount_be64));
906 		SHA2Update(ctx, PADDING, ((index < 112) ? 112 : 240) - index);
907 		SHA2Update(ctx, bitcount_be64, sizeof (bitcount_be64));
908 		if (algotype <= SHA384_HMAC_GEN_MECH_INFO_TYPE) {
909 			ctx->state.s64[6] = ctx->state.s64[7] = 0;
910 			Encode64(digest, ctx->state.s64,
911 			    sizeof (uint64_t) * 6);
912 		} else
913 			Encode64(digest, ctx->state.s64,
914 			    sizeof (ctx->state.s64));
915 	}
916 
917 	/* zeroize sensitive information */
918 	bzero(ctx, sizeof (*ctx));
919 }
920