xref: /illumos-gate/usr/src/common/crypto/sha2/sha2.c (revision eae2e508a8e70b1ec407b10bd068c080651bbe5c)
1 /*
2  * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
3  * Use is subject to license terms.
4  */
5 
6 #pragma ident	"%Z%%M%	%I%	%E% SMI"
7 
8 /*
9  * The basic framework for this code came from the reference
10  * implementation for MD5.  That implementation is Copyright (C)
11  * 1991-2, RSA Data Security, Inc. Created 1991. All rights reserved.
12  *
13  * License to copy and use this software is granted provided that it
14  * is identified as the "RSA Data Security, Inc. MD5 Message-Digest
15  * Algorithm" in all material mentioning or referencing this software
16  * or this function.
17  *
18  * License is also granted to make and use derivative works provided
19  * that such works are identified as "derived from the RSA Data
20  * Security, Inc. MD5 Message-Digest Algorithm" in all material
21  * mentioning or referencing the derived work.
22  *
23  * RSA Data Security, Inc. makes no representations concerning either
24  * the merchantability of this software or the suitability of this
25  * software for any particular purpose. It is provided "as is"
26  * without express or implied warranty of any kind.
27  *
28  * These notices must be retained in any copies of any part of this
29  * documentation and/or software.
30  *
31  * NOTE: Cleaned-up and optimized, version of SHA2, based on the FIPS 180-2
32  * standard, available at http://www.itl.nist.gov/div897/pubs/fip180-2.htm
33  * Not as fast as one would like -- further optimizations are encouraged
34  * and appreciated.
35  */
36 
37 #include <sys/types.h>
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/sysmacros.h>
41 #define	_SHA2_IMPL
42 #include <sys/sha2.h>
43 #include <sys/sha2_consts.h>
44 
45 #ifdef _KERNEL
46 #include <sys/cmn_err.h>
47 
48 #else
49 #include <strings.h>
50 #include <stdlib.h>
51 #include <errno.h>
52 
53 #pragma weak SHA256Update = SHA2Update
54 #pragma weak SHA384Update = SHA2Update
55 #pragma weak SHA512Update = SHA2Update
56 
57 #pragma weak SHA256Final = SHA2Final
58 #pragma weak SHA384Final = SHA2Final
59 #pragma weak SHA512Final = SHA2Final
60 
61 #endif	/* _KERNEL */
62 
63 static void Encode(uint8_t *, uint32_t *, size_t);
64 static void Encode64(uint8_t *, uint64_t *, size_t);
65 
66 #if	defined(__amd64)
67 #define	SHA512Transform(ctx, in) SHA512TransformBlocks((ctx), (in), 1)
68 #define	SHA256Transform(ctx, in) SHA256TransformBlocks((ctx), (in), 1)
69 
70 void SHA512TransformBlocks(SHA2_CTX *ctx, const void *in, size_t num);
71 void SHA256TransformBlocks(SHA2_CTX *ctx, const void *in, size_t num);
72 
73 #else
74 static void SHA256Transform(SHA2_CTX *, const uint8_t *);
75 static void SHA512Transform(SHA2_CTX *, const uint8_t *);
76 #endif	/* __amd64 */
77 
78 static uint8_t PADDING[128] = { 0x80, /* all zeros */ };
79 
80 /* Ch and Maj are the basic SHA2 functions. */
81 #define	Ch(b, c, d)	(((b) & (c)) ^ ((~b) & (d)))
82 #define	Maj(b, c, d)	(((b) & (c)) ^ ((b) & (d)) ^ ((c) & (d)))
83 
84 /* Rotates x right n bits. */
85 #define	ROTR(x, n)	\
86 	(((x) >> (n)) | ((x) << ((sizeof (x) * NBBY)-(n))))
87 
88 /* Shift x right n bits */
89 #define	SHR(x, n)	((x) >> (n))
90 
91 /* SHA256 Functions */
92 #define	BIGSIGMA0_256(x)	(ROTR((x), 2) ^ ROTR((x), 13) ^ ROTR((x), 22))
93 #define	BIGSIGMA1_256(x)	(ROTR((x), 6) ^ ROTR((x), 11) ^ ROTR((x), 25))
94 #define	SIGMA0_256(x)		(ROTR((x), 7) ^ ROTR((x), 18) ^ SHR((x), 3))
95 #define	SIGMA1_256(x)		(ROTR((x), 17) ^ ROTR((x), 19) ^ SHR((x), 10))
96 
97 #define	SHA256ROUND(a, b, c, d, e, f, g, h, i, w)			\
98 	T1 = h + BIGSIGMA1_256(e) + Ch(e, f, g) + SHA256_CONST(i) + w;	\
99 	d += T1;							\
100 	T2 = BIGSIGMA0_256(a) + Maj(a, b, c);				\
101 	h = T1 + T2
102 
103 /* SHA384/512 Functions */
104 #define	BIGSIGMA0(x)	(ROTR((x), 28) ^ ROTR((x), 34) ^ ROTR((x), 39))
105 #define	BIGSIGMA1(x)	(ROTR((x), 14) ^ ROTR((x), 18) ^ ROTR((x), 41))
106 #define	SIGMA0(x)	(ROTR((x), 1) ^ ROTR((x), 8) ^ SHR((x), 7))
107 #define	SIGMA1(x)	(ROTR((x), 19) ^ ROTR((x), 61) ^ SHR((x), 6))
108 #define	SHA512ROUND(a, b, c, d, e, f, g, h, i, w)			\
109 	T1 = h + BIGSIGMA1(e) + Ch(e, f, g) + SHA512_CONST(i) + w;	\
110 	d += T1;							\
111 	T2 = BIGSIGMA0(a) + Maj(a, b, c);				\
112 	h = T1 + T2
113 
114 /*
115  * sparc optimization:
116  *
117  * on the sparc, we can load big endian 32-bit data easily.  note that
118  * special care must be taken to ensure the address is 32-bit aligned.
119  * in the interest of speed, we don't check to make sure, since
120  * careful programming can guarantee this for us.
121  */
122 
123 #if	defined(_BIG_ENDIAN)
124 
125 #define	LOAD_BIG_32(addr)	(*(uint32_t *)(addr))
126 
127 #else	/* little endian -- will work on big endian, but slowly */
128 
129 #define	LOAD_BIG_32(addr)	\
130 	(((addr)[0] << 24) | ((addr)[1] << 16) | ((addr)[2] << 8) | (addr)[3])
131 #endif
132 
133 
134 #if	defined(_BIG_ENDIAN)
135 
136 #define	LOAD_BIG_64(addr)	(*(uint64_t *)(addr))
137 
138 #else	/* little endian -- will work on big endian, but slowly */
139 
140 #define	LOAD_BIG_64(addr)	\
141 	(((uint64_t)(addr)[0] << 56) | ((uint64_t)(addr)[1] << 48) |	\
142 	    ((uint64_t)(addr)[2] << 40) | ((uint64_t)(addr)[3] << 32) |	\
143 	    ((uint64_t)(addr)[4] << 24) | ((uint64_t)(addr)[5] << 16) |	\
144 	    ((uint64_t)(addr)[6] << 8) | (uint64_t)(addr)[7])
145 #endif
146 
147 
148 #if	!defined(__amd64)
149 /* SHA256 Transform */
150 
151 static void
152 SHA256Transform(SHA2_CTX *ctx, const uint8_t *blk)
153 {
154 	uint32_t a = ctx->state.s32[0];
155 	uint32_t b = ctx->state.s32[1];
156 	uint32_t c = ctx->state.s32[2];
157 	uint32_t d = ctx->state.s32[3];
158 	uint32_t e = ctx->state.s32[4];
159 	uint32_t f = ctx->state.s32[5];
160 	uint32_t g = ctx->state.s32[6];
161 	uint32_t h = ctx->state.s32[7];
162 
163 	uint32_t w0, w1, w2, w3, w4, w5, w6, w7;
164 	uint32_t w8, w9, w10, w11, w12, w13, w14, w15;
165 	uint32_t T1, T2;
166 
167 #if	defined(__sparc)
168 	static const uint32_t sha256_consts[] = {
169 		SHA256_CONST_0, SHA256_CONST_1, SHA256_CONST_2,
170 		SHA256_CONST_3, SHA256_CONST_4, SHA256_CONST_5,
171 		SHA256_CONST_6, SHA256_CONST_7, SHA256_CONST_8,
172 		SHA256_CONST_9, SHA256_CONST_10, SHA256_CONST_11,
173 		SHA256_CONST_12, SHA256_CONST_13, SHA256_CONST_14,
174 		SHA256_CONST_15, SHA256_CONST_16, SHA256_CONST_17,
175 		SHA256_CONST_18, SHA256_CONST_19, SHA256_CONST_20,
176 		SHA256_CONST_21, SHA256_CONST_22, SHA256_CONST_23,
177 		SHA256_CONST_24, SHA256_CONST_25, SHA256_CONST_26,
178 		SHA256_CONST_27, SHA256_CONST_28, SHA256_CONST_29,
179 		SHA256_CONST_30, SHA256_CONST_31, SHA256_CONST_32,
180 		SHA256_CONST_33, SHA256_CONST_34, SHA256_CONST_35,
181 		SHA256_CONST_36, SHA256_CONST_37, SHA256_CONST_38,
182 		SHA256_CONST_39, SHA256_CONST_40, SHA256_CONST_41,
183 		SHA256_CONST_42, SHA256_CONST_43, SHA256_CONST_44,
184 		SHA256_CONST_45, SHA256_CONST_46, SHA256_CONST_47,
185 		SHA256_CONST_48, SHA256_CONST_49, SHA256_CONST_50,
186 		SHA256_CONST_51, SHA256_CONST_52, SHA256_CONST_53,
187 		SHA256_CONST_54, SHA256_CONST_55, SHA256_CONST_56,
188 		SHA256_CONST_57, SHA256_CONST_58, SHA256_CONST_59,
189 		SHA256_CONST_60, SHA256_CONST_61, SHA256_CONST_62,
190 		SHA256_CONST_63
191 	};
192 #endif	/* __sparc */
193 
194 	if ((uintptr_t)blk & 0x3) {		/* not 4-byte aligned? */
195 		bcopy(blk, ctx->buf_un.buf32,  sizeof (ctx->buf_un.buf32));
196 		blk = (uint8_t *)ctx->buf_un.buf32;
197 	}
198 
199 	/* LINTED E_BAD_PTR_CAST_ALIGN */
200 	w0 =  LOAD_BIG_32(blk + 4 * 0);
201 	SHA256ROUND(a, b, c, d, e, f, g, h, 0, w0);
202 	/* LINTED E_BAD_PTR_CAST_ALIGN */
203 	w1 =  LOAD_BIG_32(blk + 4 * 1);
204 	SHA256ROUND(h, a, b, c, d, e, f, g, 1, w1);
205 	/* LINTED E_BAD_PTR_CAST_ALIGN */
206 	w2 =  LOAD_BIG_32(blk + 4 * 2);
207 	SHA256ROUND(g, h, a, b, c, d, e, f, 2, w2);
208 	/* LINTED E_BAD_PTR_CAST_ALIGN */
209 	w3 =  LOAD_BIG_32(blk + 4 * 3);
210 	SHA256ROUND(f, g, h, a, b, c, d, e, 3, w3);
211 	/* LINTED E_BAD_PTR_CAST_ALIGN */
212 	w4 =  LOAD_BIG_32(blk + 4 * 4);
213 	SHA256ROUND(e, f, g, h, a, b, c, d, 4, w4);
214 	/* LINTED E_BAD_PTR_CAST_ALIGN */
215 	w5 =  LOAD_BIG_32(blk + 4 * 5);
216 	SHA256ROUND(d, e, f, g, h, a, b, c, 5, w5);
217 	/* LINTED E_BAD_PTR_CAST_ALIGN */
218 	w6 =  LOAD_BIG_32(blk + 4 * 6);
219 	SHA256ROUND(c, d, e, f, g, h, a, b, 6, w6);
220 	/* LINTED E_BAD_PTR_CAST_ALIGN */
221 	w7 =  LOAD_BIG_32(blk + 4 * 7);
222 	SHA256ROUND(b, c, d, e, f, g, h, a, 7, w7);
223 	/* LINTED E_BAD_PTR_CAST_ALIGN */
224 	w8 =  LOAD_BIG_32(blk + 4 * 8);
225 	SHA256ROUND(a, b, c, d, e, f, g, h, 8, w8);
226 	/* LINTED E_BAD_PTR_CAST_ALIGN */
227 	w9 =  LOAD_BIG_32(blk + 4 * 9);
228 	SHA256ROUND(h, a, b, c, d, e, f, g, 9, w9);
229 	/* LINTED E_BAD_PTR_CAST_ALIGN */
230 	w10 =  LOAD_BIG_32(blk + 4 * 10);
231 	SHA256ROUND(g, h, a, b, c, d, e, f, 10, w10);
232 	/* LINTED E_BAD_PTR_CAST_ALIGN */
233 	w11 =  LOAD_BIG_32(blk + 4 * 11);
234 	SHA256ROUND(f, g, h, a, b, c, d, e, 11, w11);
235 	/* LINTED E_BAD_PTR_CAST_ALIGN */
236 	w12 =  LOAD_BIG_32(blk + 4 * 12);
237 	SHA256ROUND(e, f, g, h, a, b, c, d, 12, w12);
238 	/* LINTED E_BAD_PTR_CAST_ALIGN */
239 	w13 =  LOAD_BIG_32(blk + 4 * 13);
240 	SHA256ROUND(d, e, f, g, h, a, b, c, 13, w13);
241 	/* LINTED E_BAD_PTR_CAST_ALIGN */
242 	w14 =  LOAD_BIG_32(blk + 4 * 14);
243 	SHA256ROUND(c, d, e, f, g, h, a, b, 14, w14);
244 	/* LINTED E_BAD_PTR_CAST_ALIGN */
245 	w15 =  LOAD_BIG_32(blk + 4 * 15);
246 	SHA256ROUND(b, c, d, e, f, g, h, a, 15, w15);
247 
248 	w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
249 	SHA256ROUND(a, b, c, d, e, f, g, h, 16, w0);
250 	w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
251 	SHA256ROUND(h, a, b, c, d, e, f, g, 17, w1);
252 	w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
253 	SHA256ROUND(g, h, a, b, c, d, e, f, 18, w2);
254 	w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
255 	SHA256ROUND(f, g, h, a, b, c, d, e, 19, w3);
256 	w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
257 	SHA256ROUND(e, f, g, h, a, b, c, d, 20, w4);
258 	w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
259 	SHA256ROUND(d, e, f, g, h, a, b, c, 21, w5);
260 	w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
261 	SHA256ROUND(c, d, e, f, g, h, a, b, 22, w6);
262 	w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
263 	SHA256ROUND(b, c, d, e, f, g, h, a, 23, w7);
264 	w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
265 	SHA256ROUND(a, b, c, d, e, f, g, h, 24, w8);
266 	w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
267 	SHA256ROUND(h, a, b, c, d, e, f, g, 25, w9);
268 	w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
269 	SHA256ROUND(g, h, a, b, c, d, e, f, 26, w10);
270 	w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
271 	SHA256ROUND(f, g, h, a, b, c, d, e, 27, w11);
272 	w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
273 	SHA256ROUND(e, f, g, h, a, b, c, d, 28, w12);
274 	w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
275 	SHA256ROUND(d, e, f, g, h, a, b, c, 29, w13);
276 	w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
277 	SHA256ROUND(c, d, e, f, g, h, a, b, 30, w14);
278 	w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
279 	SHA256ROUND(b, c, d, e, f, g, h, a, 31, w15);
280 
281 	w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
282 	SHA256ROUND(a, b, c, d, e, f, g, h, 32, w0);
283 	w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
284 	SHA256ROUND(h, a, b, c, d, e, f, g, 33, w1);
285 	w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
286 	SHA256ROUND(g, h, a, b, c, d, e, f, 34, w2);
287 	w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
288 	SHA256ROUND(f, g, h, a, b, c, d, e, 35, w3);
289 	w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
290 	SHA256ROUND(e, f, g, h, a, b, c, d, 36, w4);
291 	w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
292 	SHA256ROUND(d, e, f, g, h, a, b, c, 37, w5);
293 	w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
294 	SHA256ROUND(c, d, e, f, g, h, a, b, 38, w6);
295 	w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
296 	SHA256ROUND(b, c, d, e, f, g, h, a, 39, w7);
297 	w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
298 	SHA256ROUND(a, b, c, d, e, f, g, h, 40, w8);
299 	w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
300 	SHA256ROUND(h, a, b, c, d, e, f, g, 41, w9);
301 	w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
302 	SHA256ROUND(g, h, a, b, c, d, e, f, 42, w10);
303 	w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
304 	SHA256ROUND(f, g, h, a, b, c, d, e, 43, w11);
305 	w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
306 	SHA256ROUND(e, f, g, h, a, b, c, d, 44, w12);
307 	w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
308 	SHA256ROUND(d, e, f, g, h, a, b, c, 45, w13);
309 	w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
310 	SHA256ROUND(c, d, e, f, g, h, a, b, 46, w14);
311 	w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
312 	SHA256ROUND(b, c, d, e, f, g, h, a, 47, w15);
313 
314 	w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
315 	SHA256ROUND(a, b, c, d, e, f, g, h, 48, w0);
316 	w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
317 	SHA256ROUND(h, a, b, c, d, e, f, g, 49, w1);
318 	w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
319 	SHA256ROUND(g, h, a, b, c, d, e, f, 50, w2);
320 	w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
321 	SHA256ROUND(f, g, h, a, b, c, d, e, 51, w3);
322 	w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
323 	SHA256ROUND(e, f, g, h, a, b, c, d, 52, w4);
324 	w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
325 	SHA256ROUND(d, e, f, g, h, a, b, c, 53, w5);
326 	w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
327 	SHA256ROUND(c, d, e, f, g, h, a, b, 54, w6);
328 	w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
329 	SHA256ROUND(b, c, d, e, f, g, h, a, 55, w7);
330 	w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
331 	SHA256ROUND(a, b, c, d, e, f, g, h, 56, w8);
332 	w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
333 	SHA256ROUND(h, a, b, c, d, e, f, g, 57, w9);
334 	w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
335 	SHA256ROUND(g, h, a, b, c, d, e, f, 58, w10);
336 	w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
337 	SHA256ROUND(f, g, h, a, b, c, d, e, 59, w11);
338 	w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
339 	SHA256ROUND(e, f, g, h, a, b, c, d, 60, w12);
340 	w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
341 	SHA256ROUND(d, e, f, g, h, a, b, c, 61, w13);
342 	w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
343 	SHA256ROUND(c, d, e, f, g, h, a, b, 62, w14);
344 	w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
345 	SHA256ROUND(b, c, d, e, f, g, h, a, 63, w15);
346 
347 	ctx->state.s32[0] += a;
348 	ctx->state.s32[1] += b;
349 	ctx->state.s32[2] += c;
350 	ctx->state.s32[3] += d;
351 	ctx->state.s32[4] += e;
352 	ctx->state.s32[5] += f;
353 	ctx->state.s32[6] += g;
354 	ctx->state.s32[7] += h;
355 }
356 
357 
358 /* SHA384 and SHA512 Transform */
359 
360 static void
361 SHA512Transform(SHA2_CTX *ctx, const uint8_t *blk)
362 {
363 
364 	uint64_t a = ctx->state.s64[0];
365 	uint64_t b = ctx->state.s64[1];
366 	uint64_t c = ctx->state.s64[2];
367 	uint64_t d = ctx->state.s64[3];
368 	uint64_t e = ctx->state.s64[4];
369 	uint64_t f = ctx->state.s64[5];
370 	uint64_t g = ctx->state.s64[6];
371 	uint64_t h = ctx->state.s64[7];
372 
373 	uint64_t w0, w1, w2, w3, w4, w5, w6, w7;
374 	uint64_t w8, w9, w10, w11, w12, w13, w14, w15;
375 	uint64_t T1, T2;
376 
377 #if	defined(__sparc)
378 	static const uint64_t sha512_consts[] = {
379 		SHA512_CONST_0, SHA512_CONST_1, SHA512_CONST_2,
380 		SHA512_CONST_3, SHA512_CONST_4, SHA512_CONST_5,
381 		SHA512_CONST_6, SHA512_CONST_7, SHA512_CONST_8,
382 		SHA512_CONST_9, SHA512_CONST_10, SHA512_CONST_11,
383 		SHA512_CONST_12, SHA512_CONST_13, SHA512_CONST_14,
384 		SHA512_CONST_15, SHA512_CONST_16, SHA512_CONST_17,
385 		SHA512_CONST_18, SHA512_CONST_19, SHA512_CONST_20,
386 		SHA512_CONST_21, SHA512_CONST_22, SHA512_CONST_23,
387 		SHA512_CONST_24, SHA512_CONST_25, SHA512_CONST_26,
388 		SHA512_CONST_27, SHA512_CONST_28, SHA512_CONST_29,
389 		SHA512_CONST_30, SHA512_CONST_31, SHA512_CONST_32,
390 		SHA512_CONST_33, SHA512_CONST_34, SHA512_CONST_35,
391 		SHA512_CONST_36, SHA512_CONST_37, SHA512_CONST_38,
392 		SHA512_CONST_39, SHA512_CONST_40, SHA512_CONST_41,
393 		SHA512_CONST_42, SHA512_CONST_43, SHA512_CONST_44,
394 		SHA512_CONST_45, SHA512_CONST_46, SHA512_CONST_47,
395 		SHA512_CONST_48, SHA512_CONST_49, SHA512_CONST_50,
396 		SHA512_CONST_51, SHA512_CONST_52, SHA512_CONST_53,
397 		SHA512_CONST_54, SHA512_CONST_55, SHA512_CONST_56,
398 		SHA512_CONST_57, SHA512_CONST_58, SHA512_CONST_59,
399 		SHA512_CONST_60, SHA512_CONST_61, SHA512_CONST_62,
400 		SHA512_CONST_63, SHA512_CONST_64, SHA512_CONST_65,
401 		SHA512_CONST_66, SHA512_CONST_67, SHA512_CONST_68,
402 		SHA512_CONST_69, SHA512_CONST_70, SHA512_CONST_71,
403 		SHA512_CONST_72, SHA512_CONST_73, SHA512_CONST_74,
404 		SHA512_CONST_75, SHA512_CONST_76, SHA512_CONST_77,
405 		SHA512_CONST_78, SHA512_CONST_79
406 	};
407 #endif	/* __sparc */
408 
409 
410 	if ((uintptr_t)blk & 0x7) {		/* not 8-byte aligned? */
411 		bcopy(blk, ctx->buf_un.buf64,  sizeof (ctx->buf_un.buf64));
412 		blk = (uint8_t *)ctx->buf_un.buf64;
413 	}
414 
415 	/* LINTED E_BAD_PTR_CAST_ALIGN */
416 	w0 =  LOAD_BIG_64(blk + 8 * 0);
417 	SHA512ROUND(a, b, c, d, e, f, g, h, 0, w0);
418 	/* LINTED E_BAD_PTR_CAST_ALIGN */
419 	w1 =  LOAD_BIG_64(blk + 8 * 1);
420 	SHA512ROUND(h, a, b, c, d, e, f, g, 1, w1);
421 	/* LINTED E_BAD_PTR_CAST_ALIGN */
422 	w2 =  LOAD_BIG_64(blk + 8 * 2);
423 	SHA512ROUND(g, h, a, b, c, d, e, f, 2, w2);
424 	/* LINTED E_BAD_PTR_CAST_ALIGN */
425 	w3 =  LOAD_BIG_64(blk + 8 * 3);
426 	SHA512ROUND(f, g, h, a, b, c, d, e, 3, w3);
427 	/* LINTED E_BAD_PTR_CAST_ALIGN */
428 	w4 =  LOAD_BIG_64(blk + 8 * 4);
429 	SHA512ROUND(e, f, g, h, a, b, c, d, 4, w4);
430 	/* LINTED E_BAD_PTR_CAST_ALIGN */
431 	w5 =  LOAD_BIG_64(blk + 8 * 5);
432 	SHA512ROUND(d, e, f, g, h, a, b, c, 5, w5);
433 	/* LINTED E_BAD_PTR_CAST_ALIGN */
434 	w6 =  LOAD_BIG_64(blk + 8 * 6);
435 	SHA512ROUND(c, d, e, f, g, h, a, b, 6, w6);
436 	/* LINTED E_BAD_PTR_CAST_ALIGN */
437 	w7 =  LOAD_BIG_64(blk + 8 * 7);
438 	SHA512ROUND(b, c, d, e, f, g, h, a, 7, w7);
439 	/* LINTED E_BAD_PTR_CAST_ALIGN */
440 	w8 =  LOAD_BIG_64(blk + 8 * 8);
441 	SHA512ROUND(a, b, c, d, e, f, g, h, 8, w8);
442 	/* LINTED E_BAD_PTR_CAST_ALIGN */
443 	w9 =  LOAD_BIG_64(blk + 8 * 9);
444 	SHA512ROUND(h, a, b, c, d, e, f, g, 9, w9);
445 	/* LINTED E_BAD_PTR_CAST_ALIGN */
446 	w10 =  LOAD_BIG_64(blk + 8 * 10);
447 	SHA512ROUND(g, h, a, b, c, d, e, f, 10, w10);
448 	/* LINTED E_BAD_PTR_CAST_ALIGN */
449 	w11 =  LOAD_BIG_64(blk + 8 * 11);
450 	SHA512ROUND(f, g, h, a, b, c, d, e, 11, w11);
451 	/* LINTED E_BAD_PTR_CAST_ALIGN */
452 	w12 =  LOAD_BIG_64(blk + 8 * 12);
453 	SHA512ROUND(e, f, g, h, a, b, c, d, 12, w12);
454 	/* LINTED E_BAD_PTR_CAST_ALIGN */
455 	w13 =  LOAD_BIG_64(blk + 8 * 13);
456 	SHA512ROUND(d, e, f, g, h, a, b, c, 13, w13);
457 	/* LINTED E_BAD_PTR_CAST_ALIGN */
458 	w14 =  LOAD_BIG_64(blk + 8 * 14);
459 	SHA512ROUND(c, d, e, f, g, h, a, b, 14, w14);
460 	/* LINTED E_BAD_PTR_CAST_ALIGN */
461 	w15 =  LOAD_BIG_64(blk + 8 * 15);
462 	SHA512ROUND(b, c, d, e, f, g, h, a, 15, w15);
463 
464 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
465 	SHA512ROUND(a, b, c, d, e, f, g, h, 16, w0);
466 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
467 	SHA512ROUND(h, a, b, c, d, e, f, g, 17, w1);
468 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
469 	SHA512ROUND(g, h, a, b, c, d, e, f, 18, w2);
470 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
471 	SHA512ROUND(f, g, h, a, b, c, d, e, 19, w3);
472 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
473 	SHA512ROUND(e, f, g, h, a, b, c, d, 20, w4);
474 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
475 	SHA512ROUND(d, e, f, g, h, a, b, c, 21, w5);
476 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
477 	SHA512ROUND(c, d, e, f, g, h, a, b, 22, w6);
478 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
479 	SHA512ROUND(b, c, d, e, f, g, h, a, 23, w7);
480 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
481 	SHA512ROUND(a, b, c, d, e, f, g, h, 24, w8);
482 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
483 	SHA512ROUND(h, a, b, c, d, e, f, g, 25, w9);
484 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
485 	SHA512ROUND(g, h, a, b, c, d, e, f, 26, w10);
486 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
487 	SHA512ROUND(f, g, h, a, b, c, d, e, 27, w11);
488 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
489 	SHA512ROUND(e, f, g, h, a, b, c, d, 28, w12);
490 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
491 	SHA512ROUND(d, e, f, g, h, a, b, c, 29, w13);
492 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
493 	SHA512ROUND(c, d, e, f, g, h, a, b, 30, w14);
494 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
495 	SHA512ROUND(b, c, d, e, f, g, h, a, 31, w15);
496 
497 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
498 	SHA512ROUND(a, b, c, d, e, f, g, h, 32, w0);
499 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
500 	SHA512ROUND(h, a, b, c, d, e, f, g, 33, w1);
501 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
502 	SHA512ROUND(g, h, a, b, c, d, e, f, 34, w2);
503 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
504 	SHA512ROUND(f, g, h, a, b, c, d, e, 35, w3);
505 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
506 	SHA512ROUND(e, f, g, h, a, b, c, d, 36, w4);
507 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
508 	SHA512ROUND(d, e, f, g, h, a, b, c, 37, w5);
509 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
510 	SHA512ROUND(c, d, e, f, g, h, a, b, 38, w6);
511 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
512 	SHA512ROUND(b, c, d, e, f, g, h, a, 39, w7);
513 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
514 	SHA512ROUND(a, b, c, d, e, f, g, h, 40, w8);
515 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
516 	SHA512ROUND(h, a, b, c, d, e, f, g, 41, w9);
517 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
518 	SHA512ROUND(g, h, a, b, c, d, e, f, 42, w10);
519 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
520 	SHA512ROUND(f, g, h, a, b, c, d, e, 43, w11);
521 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
522 	SHA512ROUND(e, f, g, h, a, b, c, d, 44, w12);
523 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
524 	SHA512ROUND(d, e, f, g, h, a, b, c, 45, w13);
525 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
526 	SHA512ROUND(c, d, e, f, g, h, a, b, 46, w14);
527 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
528 	SHA512ROUND(b, c, d, e, f, g, h, a, 47, w15);
529 
530 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
531 	SHA512ROUND(a, b, c, d, e, f, g, h, 48, w0);
532 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
533 	SHA512ROUND(h, a, b, c, d, e, f, g, 49, w1);
534 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
535 	SHA512ROUND(g, h, a, b, c, d, e, f, 50, w2);
536 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
537 	SHA512ROUND(f, g, h, a, b, c, d, e, 51, w3);
538 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
539 	SHA512ROUND(e, f, g, h, a, b, c, d, 52, w4);
540 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
541 	SHA512ROUND(d, e, f, g, h, a, b, c, 53, w5);
542 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
543 	SHA512ROUND(c, d, e, f, g, h, a, b, 54, w6);
544 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
545 	SHA512ROUND(b, c, d, e, f, g, h, a, 55, w7);
546 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
547 	SHA512ROUND(a, b, c, d, e, f, g, h, 56, w8);
548 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
549 	SHA512ROUND(h, a, b, c, d, e, f, g, 57, w9);
550 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
551 	SHA512ROUND(g, h, a, b, c, d, e, f, 58, w10);
552 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
553 	SHA512ROUND(f, g, h, a, b, c, d, e, 59, w11);
554 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
555 	SHA512ROUND(e, f, g, h, a, b, c, d, 60, w12);
556 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
557 	SHA512ROUND(d, e, f, g, h, a, b, c, 61, w13);
558 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
559 	SHA512ROUND(c, d, e, f, g, h, a, b, 62, w14);
560 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
561 	SHA512ROUND(b, c, d, e, f, g, h, a, 63, w15);
562 
563 	w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
564 	SHA512ROUND(a, b, c, d, e, f, g, h, 64, w0);
565 	w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
566 	SHA512ROUND(h, a, b, c, d, e, f, g, 65, w1);
567 	w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
568 	SHA512ROUND(g, h, a, b, c, d, e, f, 66, w2);
569 	w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
570 	SHA512ROUND(f, g, h, a, b, c, d, e, 67, w3);
571 	w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
572 	SHA512ROUND(e, f, g, h, a, b, c, d, 68, w4);
573 	w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
574 	SHA512ROUND(d, e, f, g, h, a, b, c, 69, w5);
575 	w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
576 	SHA512ROUND(c, d, e, f, g, h, a, b, 70, w6);
577 	w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
578 	SHA512ROUND(b, c, d, e, f, g, h, a, 71, w7);
579 	w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
580 	SHA512ROUND(a, b, c, d, e, f, g, h, 72, w8);
581 	w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
582 	SHA512ROUND(h, a, b, c, d, e, f, g, 73, w9);
583 	w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
584 	SHA512ROUND(g, h, a, b, c, d, e, f, 74, w10);
585 	w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
586 	SHA512ROUND(f, g, h, a, b, c, d, e, 75, w11);
587 	w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
588 	SHA512ROUND(e, f, g, h, a, b, c, d, 76, w12);
589 	w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
590 	SHA512ROUND(d, e, f, g, h, a, b, c, 77, w13);
591 	w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
592 	SHA512ROUND(c, d, e, f, g, h, a, b, 78, w14);
593 	w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
594 	SHA512ROUND(b, c, d, e, f, g, h, a, 79, w15);
595 
596 	ctx->state.s64[0] += a;
597 	ctx->state.s64[1] += b;
598 	ctx->state.s64[2] += c;
599 	ctx->state.s64[3] += d;
600 	ctx->state.s64[4] += e;
601 	ctx->state.s64[5] += f;
602 	ctx->state.s64[6] += g;
603 	ctx->state.s64[7] += h;
604 
605 }
606 #endif	/* !__amd64 */
607 
608 
609 /*
610  * Encode()
611  *
612  * purpose: to convert a list of numbers from little endian to big endian
613  *   input: uint8_t *	: place to store the converted big endian numbers
614  *	    uint32_t *	: place to get numbers to convert from
615  *          size_t	: the length of the input in bytes
616  *  output: void
617  */
618 
619 static void
620 Encode(uint8_t *_RESTRICT_KYWD output, uint32_t *_RESTRICT_KYWD input,
621     size_t len)
622 {
623 	size_t		i, j;
624 
625 #if	defined(__sparc)
626 	if (IS_P2ALIGNED(output, sizeof (uint32_t))) {
627 		for (i = 0, j = 0; j < len; i++, j += 4) {
628 			/* LINTED: pointer alignment */
629 			*((uint32_t *)(output + j)) = input[i];
630 		}
631 	} else {
632 #endif	/* little endian -- will work on big endian, but slowly */
633 		for (i = 0, j = 0; j < len; i++, j += 4) {
634 			output[j]	= (input[i] >> 24) & 0xff;
635 			output[j + 1]	= (input[i] >> 16) & 0xff;
636 			output[j + 2]	= (input[i] >>  8) & 0xff;
637 			output[j + 3]	= input[i] & 0xff;
638 		}
639 #if	defined(__sparc)
640 	}
641 #endif
642 }
643 
644 static void
645 Encode64(uint8_t *_RESTRICT_KYWD output, uint64_t *_RESTRICT_KYWD input,
646     size_t len)
647 {
648 	size_t		i, j;
649 
650 #if	defined(__sparc)
651 	if (IS_P2ALIGNED(output, sizeof (uint64_t))) {
652 		for (i = 0, j = 0; j < len; i++, j += 8) {
653 			/* LINTED: pointer alignment */
654 			*((uint64_t *)(output + j)) = input[i];
655 		}
656 	} else {
657 #endif	/* little endian -- will work on big endian, but slowly */
658 		for (i = 0, j = 0; j < len; i++, j += 8) {
659 
660 			output[j]	= (input[i] >> 56) & 0xff;
661 			output[j + 1]	= (input[i] >> 48) & 0xff;
662 			output[j + 2]	= (input[i] >> 40) & 0xff;
663 			output[j + 3]	= (input[i] >> 32) & 0xff;
664 			output[j + 4]	= (input[i] >> 24) & 0xff;
665 			output[j + 5]	= (input[i] >> 16) & 0xff;
666 			output[j + 6]	= (input[i] >>  8) & 0xff;
667 			output[j + 7]	= input[i] & 0xff;
668 		}
669 #if	defined(__sparc)
670 	}
671 #endif
672 }
673 
674 
675 void
676 SHA2Init(uint64_t mech, SHA2_CTX *ctx)
677 {
678 
679 	switch (mech) {
680 	case SHA256_MECH_INFO_TYPE:
681 	case SHA256_HMAC_MECH_INFO_TYPE:
682 	case SHA256_HMAC_GEN_MECH_INFO_TYPE:
683 		ctx->state.s32[0] = 0x6a09e667U;
684 		ctx->state.s32[1] = 0xbb67ae85U;
685 		ctx->state.s32[2] = 0x3c6ef372U;
686 		ctx->state.s32[3] = 0xa54ff53aU;
687 		ctx->state.s32[4] = 0x510e527fU;
688 		ctx->state.s32[5] = 0x9b05688cU;
689 		ctx->state.s32[6] = 0x1f83d9abU;
690 		ctx->state.s32[7] = 0x5be0cd19U;
691 		break;
692 	case SHA384_MECH_INFO_TYPE:
693 	case SHA384_HMAC_MECH_INFO_TYPE:
694 	case SHA384_HMAC_GEN_MECH_INFO_TYPE:
695 		ctx->state.s64[0] = 0xcbbb9d5dc1059ed8ULL;
696 		ctx->state.s64[1] = 0x629a292a367cd507ULL;
697 		ctx->state.s64[2] = 0x9159015a3070dd17ULL;
698 		ctx->state.s64[3] = 0x152fecd8f70e5939ULL;
699 		ctx->state.s64[4] = 0x67332667ffc00b31ULL;
700 		ctx->state.s64[5] = 0x8eb44a8768581511ULL;
701 		ctx->state.s64[6] = 0xdb0c2e0d64f98fa7ULL;
702 		ctx->state.s64[7] = 0x47b5481dbefa4fa4ULL;
703 		break;
704 	case SHA512_MECH_INFO_TYPE:
705 	case SHA512_HMAC_MECH_INFO_TYPE:
706 	case SHA512_HMAC_GEN_MECH_INFO_TYPE:
707 		ctx->state.s64[0] = 0x6a09e667f3bcc908ULL;
708 		ctx->state.s64[1] = 0xbb67ae8584caa73bULL;
709 		ctx->state.s64[2] = 0x3c6ef372fe94f82bULL;
710 		ctx->state.s64[3] = 0xa54ff53a5f1d36f1ULL;
711 		ctx->state.s64[4] = 0x510e527fade682d1ULL;
712 		ctx->state.s64[5] = 0x9b05688c2b3e6c1fULL;
713 		ctx->state.s64[6] = 0x1f83d9abfb41bd6bULL;
714 		ctx->state.s64[7] = 0x5be0cd19137e2179ULL;
715 		break;
716 #ifdef _KERNEL
717 	default:
718 		cmn_err(CE_PANIC, "sha2_init: "
719 		    "failed to find a supported algorithm: 0x%x",
720 		    (uint32_t)mech);
721 
722 #endif /* _KERNEL */
723 	}
724 
725 	ctx->algotype = mech;
726 	ctx->count.c64[0] = ctx->count.c64[1] = 0;
727 }
728 
729 #ifndef _KERNEL
730 
731 #pragma inline(SHA256Init, SHA384Init, SHA512Init)
732 void
733 SHA256Init(SHA256_CTX *ctx)
734 {
735 	SHA2Init(SHA256, ctx);
736 }
737 
738 void
739 SHA384Init(SHA384_CTX *ctx)
740 {
741 	SHA2Init(SHA384, ctx);
742 }
743 
744 void
745 SHA512Init(SHA512_CTX *ctx)
746 {
747 	SHA2Init(SHA512, ctx);
748 }
749 
750 #endif /* _KERNEL */
751 
752 /*
753  * SHA2Update()
754  *
755  * purpose: continues an sha2 digest operation, using the message block
756  *          to update the context.
757  *   input: SHA2_CTX *	: the context to update
758  *          void *	: the message block
759  *          size_t      : the length of the message block, in bytes
760  *  output: void
761  */
762 
763 void
764 SHA2Update(SHA2_CTX *ctx, const void *inptr, size_t input_len)
765 {
766 	uint32_t	i, buf_index, buf_len, buf_limit;
767 	const uint8_t	*input = inptr;
768 	uint32_t	algotype = ctx->algotype;
769 #if defined(__amd64)
770 	uint32_t	block_count;
771 #endif	/* !__amd64 */
772 
773 
774 	/* check for noop */
775 	if (input_len == 0)
776 		return;
777 
778 	if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
779 		buf_limit = 64;
780 
781 		/* compute number of bytes mod 64 */
782 		buf_index = (ctx->count.c32[1] >> 3) & 0x3F;
783 
784 		/* update number of bits */
785 		if ((ctx->count.c32[1] += (input_len << 3)) < (input_len << 3))
786 			ctx->count.c32[0]++;
787 
788 		ctx->count.c32[0] += (input_len >> 29);
789 
790 	} else {
791 		buf_limit = 128;
792 
793 		/* compute number of bytes mod 128 */
794 		buf_index = (ctx->count.c64[1] >> 3) & 0x7F;
795 
796 		/* update number of bits */
797 		if ((ctx->count.c64[1] += (input_len << 3)) < (input_len << 3))
798 			ctx->count.c64[0]++;
799 
800 		ctx->count.c64[0] += (input_len >> 29);
801 	}
802 
803 	buf_len = buf_limit - buf_index;
804 
805 	/* transform as many times as possible */
806 	i = 0;
807 	if (input_len >= buf_len) {
808 
809 		/*
810 		 * general optimization:
811 		 *
812 		 * only do initial bcopy() and SHA2Transform() if
813 		 * buf_index != 0.  if buf_index == 0, we're just
814 		 * wasting our time doing the bcopy() since there
815 		 * wasn't any data left over from a previous call to
816 		 * SHA2Update().
817 		 */
818 		if (buf_index) {
819 			bcopy(input, &ctx->buf_un.buf8[buf_index], buf_len);
820 			if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE)
821 				SHA256Transform(ctx, ctx->buf_un.buf8);
822 			else
823 				SHA512Transform(ctx, ctx->buf_un.buf8);
824 
825 			i = buf_len;
826 		}
827 
828 #if !defined(__amd64)
829 		if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
830 			for (; i + buf_limit - 1 < input_len; i += buf_limit) {
831 				SHA256Transform(ctx, &input[i]);
832 			}
833 		} else {
834 			for (; i + buf_limit - 1 < input_len; i += buf_limit) {
835 				SHA512Transform(ctx, &input[i]);
836 			}
837 		}
838 
839 #else
840 		if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
841 			block_count = (input_len - i) >> 6;
842 			if (block_count > 0) {
843 				SHA256TransformBlocks(ctx, &input[i],
844 				    block_count);
845 				i += block_count << 6;
846 			}
847 		} else {
848 			block_count = (input_len - i) >> 7;
849 			if (block_count > 0) {
850 				SHA512TransformBlocks(ctx, &input[i],
851 				    block_count);
852 				i += block_count << 7;
853 			}
854 		}
855 #endif	/* !__amd64 */
856 
857 		/*
858 		 * general optimization:
859 		 *
860 		 * if i and input_len are the same, return now instead
861 		 * of calling bcopy(), since the bcopy() in this case
862 		 * will be an expensive noop.
863 		 */
864 
865 		if (input_len == i)
866 			return;
867 
868 		buf_index = 0;
869 	}
870 
871 	/* buffer remaining input */
872 	bcopy(&input[i], &ctx->buf_un.buf8[buf_index], input_len - i);
873 }
874 
875 
876 /*
877  * SHA2Final()
878  *
879  * purpose: ends an sha2 digest operation, finalizing the message digest and
880  *          zeroing the context.
881  *   input: uchar_t *	: a buffer to store the digest
882  *			: The function actually uses void* because many
883  *			: callers pass things other than uchar_t here.
884  *          SHA2_CTX *  : the context to finalize, save, and zero
885  *  output: void
886  */
887 
888 void
889 SHA2Final(void *digest, SHA2_CTX *ctx)
890 {
891 	uint8_t		bitcount_be[sizeof (ctx->count.c32)];
892 	uint8_t		bitcount_be64[sizeof (ctx->count.c64)];
893 	uint32_t	index;
894 	uint32_t	algotype = ctx->algotype;
895 
896 	if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
897 		index  = (ctx->count.c32[1] >> 3) & 0x3f;
898 		Encode(bitcount_be, ctx->count.c32, sizeof (bitcount_be));
899 		SHA2Update(ctx, PADDING, ((index < 56) ? 56 : 120) - index);
900 		SHA2Update(ctx, bitcount_be, sizeof (bitcount_be));
901 		Encode(digest, ctx->state.s32, sizeof (ctx->state.s32));
902 
903 	} else {
904 		index  = (ctx->count.c64[1] >> 3) & 0x7f;
905 		Encode64(bitcount_be64, ctx->count.c64,
906 		    sizeof (bitcount_be64));
907 		SHA2Update(ctx, PADDING, ((index < 112) ? 112 : 240) - index);
908 		SHA2Update(ctx, bitcount_be64, sizeof (bitcount_be64));
909 		if (algotype <= SHA384_HMAC_GEN_MECH_INFO_TYPE) {
910 			ctx->state.s64[6] = ctx->state.s64[7] = 0;
911 			Encode64(digest, ctx->state.s64,
912 			    sizeof (uint64_t) * 6);
913 		} else
914 			Encode64(digest, ctx->state.s64,
915 			    sizeof (ctx->state.s64));
916 	}
917 
918 	/* zeroize sensitive information */
919 	bzero(ctx, sizeof (*ctx));
920 }
921