1 /*
2 * Copyright 2009 Sun Microsystems, Inc. All rights reserved.
3 * Use is subject to license terms.
4 */
5 /*
6 * Copyright 2013 Saso Kiselkov. All rights reserved.
7 */
8
9 /*
10 * The basic framework for this code came from the reference
11 * implementation for MD5. That implementation is Copyright (C)
12 * 1991-2, RSA Data Security, Inc. Created 1991. All rights reserved.
13 *
14 * License to copy and use this software is granted provided that it
15 * is identified as the "RSA Data Security, Inc. MD5 Message-Digest
16 * Algorithm" in all material mentioning or referencing this software
17 * or this function.
18 *
19 * License is also granted to make and use derivative works provided
20 * that such works are identified as "derived from the RSA Data
21 * Security, Inc. MD5 Message-Digest Algorithm" in all material
22 * mentioning or referencing the derived work.
23 *
24 * RSA Data Security, Inc. makes no representations concerning either
25 * the merchantability of this software or the suitability of this
26 * software for any particular purpose. It is provided "as is"
27 * without express or implied warranty of any kind.
28 *
29 * These notices must be retained in any copies of any part of this
30 * documentation and/or software.
31 *
32 * NOTE: Cleaned-up and optimized, version of SHA2, based on the FIPS 180-2
33 * standard, available at
34 * http://csrc.nist.gov/publications/fips/fips180-2/fips180-2.pdf
35 * Not as fast as one would like -- further optimizations are encouraged
36 * and appreciated.
37 */
38
39 #include <sys/zfs_context.h>
40 #define _SHA2_IMPL
41 #include <sys/sha2.h>
42 #include <sha2/sha2_consts.h>
43
44 #define _RESTRICT_KYWD
45
46 #ifdef _ZFS_LITTLE_ENDIAN
47 #include <sys/byteorder.h>
48 #define HAVE_HTONL
49 #endif
50 #include <sys/isa_defs.h> /* for _ILP32 */
51
52 static void Encode(uint8_t *, uint32_t *, size_t);
53 static void Encode64(uint8_t *, uint64_t *, size_t);
54
55 /* userspace only supports the generic version */
56 #if defined(__amd64) && defined(_KERNEL)
57 #define SHA512Transform(ctx, in) SHA512TransformBlocks((ctx), (in), 1)
58 #define SHA256Transform(ctx, in) SHA256TransformBlocks((ctx), (in), 1)
59
60 void SHA512TransformBlocks(SHA2_CTX *ctx, const void *in, size_t num);
61 void SHA256TransformBlocks(SHA2_CTX *ctx, const void *in, size_t num);
62
63 #else
64 static void SHA256Transform(SHA2_CTX *, const uint8_t *);
65 static void SHA512Transform(SHA2_CTX *, const uint8_t *);
66 #endif /* __amd64 && _KERNEL */
67
68 static uint8_t PADDING[128] = { 0x80, /* all zeros */ };
69
70 /*
71 * The low-level checksum routines use a lot of stack space. On systems where
72 * small stacks are enforced (like 32-bit kernel builds), insert compiler memory
73 * barriers to reduce stack frame size. This can reduce the SHA512Transform()
74 * stack frame usage from 3k to <1k on ARM32, for example.
75 */
76 #if defined(_ILP32) || defined(__powerpc) /* small stack */
77 #define SMALL_STACK_MEMORY_BARRIER asm volatile("": : :"memory");
78 #else
79 #define SMALL_STACK_MEMORY_BARRIER
80 #endif
81
82 /* Ch and Maj are the basic SHA2 functions. */
83 #define Ch(b, c, d) (((b) & (c)) ^ ((~b) & (d)))
84 #define Maj(b, c, d) (((b) & (c)) ^ ((b) & (d)) ^ ((c) & (d)))
85
86 /* Rotates x right n bits. */
87 #define ROTR(x, n) \
88 (((x) >> (n)) | ((x) << ((sizeof (x) * NBBY)-(n))))
89
90 /* Shift x right n bits */
91 #define SHR(x, n) ((x) >> (n))
92
93 /* SHA256 Functions */
94 #define BIGSIGMA0_256(x) (ROTR((x), 2) ^ ROTR((x), 13) ^ ROTR((x), 22))
95 #define BIGSIGMA1_256(x) (ROTR((x), 6) ^ ROTR((x), 11) ^ ROTR((x), 25))
96 #define SIGMA0_256(x) (ROTR((x), 7) ^ ROTR((x), 18) ^ SHR((x), 3))
97 #define SIGMA1_256(x) (ROTR((x), 17) ^ ROTR((x), 19) ^ SHR((x), 10))
98
99 #define SHA256ROUND(a, b, c, d, e, f, g, h, i, w) \
100 T1 = h + BIGSIGMA1_256(e) + Ch(e, f, g) + SHA256_CONST(i) + w; \
101 d += T1; \
102 T2 = BIGSIGMA0_256(a) + Maj(a, b, c); \
103 h = T1 + T2
104
105 /* SHA384/512 Functions */
106 #define BIGSIGMA0(x) (ROTR((x), 28) ^ ROTR((x), 34) ^ ROTR((x), 39))
107 #define BIGSIGMA1(x) (ROTR((x), 14) ^ ROTR((x), 18) ^ ROTR((x), 41))
108 #define SIGMA0(x) (ROTR((x), 1) ^ ROTR((x), 8) ^ SHR((x), 7))
109 #define SIGMA1(x) (ROTR((x), 19) ^ ROTR((x), 61) ^ SHR((x), 6))
110 #define SHA512ROUND(a, b, c, d, e, f, g, h, i, w) \
111 T1 = h + BIGSIGMA1(e) + Ch(e, f, g) + SHA512_CONST(i) + w; \
112 d += T1; \
113 T2 = BIGSIGMA0(a) + Maj(a, b, c); \
114 h = T1 + T2; \
115 SMALL_STACK_MEMORY_BARRIER;
116
117 /*
118 * sparc optimization:
119 *
120 * on the sparc, we can load big endian 32-bit data easily. note that
121 * special care must be taken to ensure the address is 32-bit aligned.
122 * in the interest of speed, we don't check to make sure, since
123 * careful programming can guarantee this for us.
124 */
125
126 #if defined(_ZFS_BIG_ENDIAN)
127 #define LOAD_BIG_32(addr) (*(uint32_t *)(addr))
128 #define LOAD_BIG_64(addr) (*(uint64_t *)(addr))
129
130 #elif defined(HAVE_HTONL)
131 #define LOAD_BIG_32(addr) htonl(*((uint32_t *)(addr)))
132 #define LOAD_BIG_64(addr) htonll(*((uint64_t *)(addr)))
133
134 #else
135 /* little endian -- will work on big endian, but slowly */
136 #define LOAD_BIG_32(addr) \
137 (((addr)[0] << 24) | ((addr)[1] << 16) | ((addr)[2] << 8) | (addr)[3])
138 #define LOAD_BIG_64(addr) \
139 (((uint64_t)(addr)[0] << 56) | ((uint64_t)(addr)[1] << 48) | \
140 ((uint64_t)(addr)[2] << 40) | ((uint64_t)(addr)[3] << 32) | \
141 ((uint64_t)(addr)[4] << 24) | ((uint64_t)(addr)[5] << 16) | \
142 ((uint64_t)(addr)[6] << 8) | (uint64_t)(addr)[7])
143 #endif /* _BIG_ENDIAN */
144
145
146 #if !defined(__amd64) || !defined(_KERNEL)
147 /* SHA256 Transform */
148
149 static void
SHA256Transform(SHA2_CTX * ctx,const uint8_t * blk)150 SHA256Transform(SHA2_CTX *ctx, const uint8_t *blk)
151 {
152 uint32_t a = ctx->state.s32[0];
153 uint32_t b = ctx->state.s32[1];
154 uint32_t c = ctx->state.s32[2];
155 uint32_t d = ctx->state.s32[3];
156 uint32_t e = ctx->state.s32[4];
157 uint32_t f = ctx->state.s32[5];
158 uint32_t g = ctx->state.s32[6];
159 uint32_t h = ctx->state.s32[7];
160
161 uint32_t w0, w1, w2, w3, w4, w5, w6, w7;
162 uint32_t w8, w9, w10, w11, w12, w13, w14, w15;
163 uint32_t T1, T2;
164
165 #if defined(__sparc)
166 static const uint32_t sha256_consts[] = {
167 SHA256_CONST_0, SHA256_CONST_1, SHA256_CONST_2,
168 SHA256_CONST_3, SHA256_CONST_4, SHA256_CONST_5,
169 SHA256_CONST_6, SHA256_CONST_7, SHA256_CONST_8,
170 SHA256_CONST_9, SHA256_CONST_10, SHA256_CONST_11,
171 SHA256_CONST_12, SHA256_CONST_13, SHA256_CONST_14,
172 SHA256_CONST_15, SHA256_CONST_16, SHA256_CONST_17,
173 SHA256_CONST_18, SHA256_CONST_19, SHA256_CONST_20,
174 SHA256_CONST_21, SHA256_CONST_22, SHA256_CONST_23,
175 SHA256_CONST_24, SHA256_CONST_25, SHA256_CONST_26,
176 SHA256_CONST_27, SHA256_CONST_28, SHA256_CONST_29,
177 SHA256_CONST_30, SHA256_CONST_31, SHA256_CONST_32,
178 SHA256_CONST_33, SHA256_CONST_34, SHA256_CONST_35,
179 SHA256_CONST_36, SHA256_CONST_37, SHA256_CONST_38,
180 SHA256_CONST_39, SHA256_CONST_40, SHA256_CONST_41,
181 SHA256_CONST_42, SHA256_CONST_43, SHA256_CONST_44,
182 SHA256_CONST_45, SHA256_CONST_46, SHA256_CONST_47,
183 SHA256_CONST_48, SHA256_CONST_49, SHA256_CONST_50,
184 SHA256_CONST_51, SHA256_CONST_52, SHA256_CONST_53,
185 SHA256_CONST_54, SHA256_CONST_55, SHA256_CONST_56,
186 SHA256_CONST_57, SHA256_CONST_58, SHA256_CONST_59,
187 SHA256_CONST_60, SHA256_CONST_61, SHA256_CONST_62,
188 SHA256_CONST_63
189 };
190 #endif /* __sparc */
191
192 if ((uintptr_t)blk & 0x3) { /* not 4-byte aligned? */
193 bcopy(blk, ctx->buf_un.buf32, sizeof (ctx->buf_un.buf32));
194 blk = (uint8_t *)ctx->buf_un.buf32;
195 }
196
197 /* LINTED E_BAD_PTR_CAST_ALIGN */
198 w0 = LOAD_BIG_32(blk + 4 * 0);
199 SHA256ROUND(a, b, c, d, e, f, g, h, 0, w0);
200 /* LINTED E_BAD_PTR_CAST_ALIGN */
201 w1 = LOAD_BIG_32(blk + 4 * 1);
202 SHA256ROUND(h, a, b, c, d, e, f, g, 1, w1);
203 /* LINTED E_BAD_PTR_CAST_ALIGN */
204 w2 = LOAD_BIG_32(blk + 4 * 2);
205 SHA256ROUND(g, h, a, b, c, d, e, f, 2, w2);
206 /* LINTED E_BAD_PTR_CAST_ALIGN */
207 w3 = LOAD_BIG_32(blk + 4 * 3);
208 SHA256ROUND(f, g, h, a, b, c, d, e, 3, w3);
209 /* LINTED E_BAD_PTR_CAST_ALIGN */
210 w4 = LOAD_BIG_32(blk + 4 * 4);
211 SHA256ROUND(e, f, g, h, a, b, c, d, 4, w4);
212 /* LINTED E_BAD_PTR_CAST_ALIGN */
213 w5 = LOAD_BIG_32(blk + 4 * 5);
214 SHA256ROUND(d, e, f, g, h, a, b, c, 5, w5);
215 /* LINTED E_BAD_PTR_CAST_ALIGN */
216 w6 = LOAD_BIG_32(blk + 4 * 6);
217 SHA256ROUND(c, d, e, f, g, h, a, b, 6, w6);
218 /* LINTED E_BAD_PTR_CAST_ALIGN */
219 w7 = LOAD_BIG_32(blk + 4 * 7);
220 SHA256ROUND(b, c, d, e, f, g, h, a, 7, w7);
221 /* LINTED E_BAD_PTR_CAST_ALIGN */
222 w8 = LOAD_BIG_32(blk + 4 * 8);
223 SHA256ROUND(a, b, c, d, e, f, g, h, 8, w8);
224 /* LINTED E_BAD_PTR_CAST_ALIGN */
225 w9 = LOAD_BIG_32(blk + 4 * 9);
226 SHA256ROUND(h, a, b, c, d, e, f, g, 9, w9);
227 /* LINTED E_BAD_PTR_CAST_ALIGN */
228 w10 = LOAD_BIG_32(blk + 4 * 10);
229 SHA256ROUND(g, h, a, b, c, d, e, f, 10, w10);
230 /* LINTED E_BAD_PTR_CAST_ALIGN */
231 w11 = LOAD_BIG_32(blk + 4 * 11);
232 SHA256ROUND(f, g, h, a, b, c, d, e, 11, w11);
233 /* LINTED E_BAD_PTR_CAST_ALIGN */
234 w12 = LOAD_BIG_32(blk + 4 * 12);
235 SHA256ROUND(e, f, g, h, a, b, c, d, 12, w12);
236 /* LINTED E_BAD_PTR_CAST_ALIGN */
237 w13 = LOAD_BIG_32(blk + 4 * 13);
238 SHA256ROUND(d, e, f, g, h, a, b, c, 13, w13);
239 /* LINTED E_BAD_PTR_CAST_ALIGN */
240 w14 = LOAD_BIG_32(blk + 4 * 14);
241 SHA256ROUND(c, d, e, f, g, h, a, b, 14, w14);
242 /* LINTED E_BAD_PTR_CAST_ALIGN */
243 w15 = LOAD_BIG_32(blk + 4 * 15);
244 SHA256ROUND(b, c, d, e, f, g, h, a, 15, w15);
245
246 w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
247 SHA256ROUND(a, b, c, d, e, f, g, h, 16, w0);
248 w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
249 SHA256ROUND(h, a, b, c, d, e, f, g, 17, w1);
250 w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
251 SHA256ROUND(g, h, a, b, c, d, e, f, 18, w2);
252 w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
253 SHA256ROUND(f, g, h, a, b, c, d, e, 19, w3);
254 w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
255 SHA256ROUND(e, f, g, h, a, b, c, d, 20, w4);
256 w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
257 SHA256ROUND(d, e, f, g, h, a, b, c, 21, w5);
258 w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
259 SHA256ROUND(c, d, e, f, g, h, a, b, 22, w6);
260 w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
261 SHA256ROUND(b, c, d, e, f, g, h, a, 23, w7);
262 w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
263 SHA256ROUND(a, b, c, d, e, f, g, h, 24, w8);
264 w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
265 SHA256ROUND(h, a, b, c, d, e, f, g, 25, w9);
266 w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
267 SHA256ROUND(g, h, a, b, c, d, e, f, 26, w10);
268 w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
269 SHA256ROUND(f, g, h, a, b, c, d, e, 27, w11);
270 w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
271 SHA256ROUND(e, f, g, h, a, b, c, d, 28, w12);
272 w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
273 SHA256ROUND(d, e, f, g, h, a, b, c, 29, w13);
274 w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
275 SHA256ROUND(c, d, e, f, g, h, a, b, 30, w14);
276 w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
277 SHA256ROUND(b, c, d, e, f, g, h, a, 31, w15);
278
279 w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
280 SHA256ROUND(a, b, c, d, e, f, g, h, 32, w0);
281 w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
282 SHA256ROUND(h, a, b, c, d, e, f, g, 33, w1);
283 w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
284 SHA256ROUND(g, h, a, b, c, d, e, f, 34, w2);
285 w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
286 SHA256ROUND(f, g, h, a, b, c, d, e, 35, w3);
287 w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
288 SHA256ROUND(e, f, g, h, a, b, c, d, 36, w4);
289 w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
290 SHA256ROUND(d, e, f, g, h, a, b, c, 37, w5);
291 w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
292 SHA256ROUND(c, d, e, f, g, h, a, b, 38, w6);
293 w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
294 SHA256ROUND(b, c, d, e, f, g, h, a, 39, w7);
295 w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
296 SHA256ROUND(a, b, c, d, e, f, g, h, 40, w8);
297 w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
298 SHA256ROUND(h, a, b, c, d, e, f, g, 41, w9);
299 w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
300 SHA256ROUND(g, h, a, b, c, d, e, f, 42, w10);
301 w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
302 SHA256ROUND(f, g, h, a, b, c, d, e, 43, w11);
303 w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
304 SHA256ROUND(e, f, g, h, a, b, c, d, 44, w12);
305 w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
306 SHA256ROUND(d, e, f, g, h, a, b, c, 45, w13);
307 w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
308 SHA256ROUND(c, d, e, f, g, h, a, b, 46, w14);
309 w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
310 SHA256ROUND(b, c, d, e, f, g, h, a, 47, w15);
311
312 w0 = SIGMA1_256(w14) + w9 + SIGMA0_256(w1) + w0;
313 SHA256ROUND(a, b, c, d, e, f, g, h, 48, w0);
314 w1 = SIGMA1_256(w15) + w10 + SIGMA0_256(w2) + w1;
315 SHA256ROUND(h, a, b, c, d, e, f, g, 49, w1);
316 w2 = SIGMA1_256(w0) + w11 + SIGMA0_256(w3) + w2;
317 SHA256ROUND(g, h, a, b, c, d, e, f, 50, w2);
318 w3 = SIGMA1_256(w1) + w12 + SIGMA0_256(w4) + w3;
319 SHA256ROUND(f, g, h, a, b, c, d, e, 51, w3);
320 w4 = SIGMA1_256(w2) + w13 + SIGMA0_256(w5) + w4;
321 SHA256ROUND(e, f, g, h, a, b, c, d, 52, w4);
322 w5 = SIGMA1_256(w3) + w14 + SIGMA0_256(w6) + w5;
323 SHA256ROUND(d, e, f, g, h, a, b, c, 53, w5);
324 w6 = SIGMA1_256(w4) + w15 + SIGMA0_256(w7) + w6;
325 SHA256ROUND(c, d, e, f, g, h, a, b, 54, w6);
326 w7 = SIGMA1_256(w5) + w0 + SIGMA0_256(w8) + w7;
327 SHA256ROUND(b, c, d, e, f, g, h, a, 55, w7);
328 w8 = SIGMA1_256(w6) + w1 + SIGMA0_256(w9) + w8;
329 SHA256ROUND(a, b, c, d, e, f, g, h, 56, w8);
330 w9 = SIGMA1_256(w7) + w2 + SIGMA0_256(w10) + w9;
331 SHA256ROUND(h, a, b, c, d, e, f, g, 57, w9);
332 w10 = SIGMA1_256(w8) + w3 + SIGMA0_256(w11) + w10;
333 SHA256ROUND(g, h, a, b, c, d, e, f, 58, w10);
334 w11 = SIGMA1_256(w9) + w4 + SIGMA0_256(w12) + w11;
335 SHA256ROUND(f, g, h, a, b, c, d, e, 59, w11);
336 w12 = SIGMA1_256(w10) + w5 + SIGMA0_256(w13) + w12;
337 SHA256ROUND(e, f, g, h, a, b, c, d, 60, w12);
338 w13 = SIGMA1_256(w11) + w6 + SIGMA0_256(w14) + w13;
339 SHA256ROUND(d, e, f, g, h, a, b, c, 61, w13);
340 w14 = SIGMA1_256(w12) + w7 + SIGMA0_256(w15) + w14;
341 SHA256ROUND(c, d, e, f, g, h, a, b, 62, w14);
342 w15 = SIGMA1_256(w13) + w8 + SIGMA0_256(w0) + w15;
343 SHA256ROUND(b, c, d, e, f, g, h, a, 63, w15);
344
345 ctx->state.s32[0] += a;
346 ctx->state.s32[1] += b;
347 ctx->state.s32[2] += c;
348 ctx->state.s32[3] += d;
349 ctx->state.s32[4] += e;
350 ctx->state.s32[5] += f;
351 ctx->state.s32[6] += g;
352 ctx->state.s32[7] += h;
353 }
354
355
356 /* SHA384 and SHA512 Transform */
357
358 static void
SHA512Transform(SHA2_CTX * ctx,const uint8_t * blk)359 SHA512Transform(SHA2_CTX *ctx, const uint8_t *blk)
360 {
361
362 uint64_t a = ctx->state.s64[0];
363 uint64_t b = ctx->state.s64[1];
364 uint64_t c = ctx->state.s64[2];
365 uint64_t d = ctx->state.s64[3];
366 uint64_t e = ctx->state.s64[4];
367 uint64_t f = ctx->state.s64[5];
368 uint64_t g = ctx->state.s64[6];
369 uint64_t h = ctx->state.s64[7];
370
371 uint64_t w0, w1, w2, w3, w4, w5, w6, w7;
372 uint64_t w8, w9, w10, w11, w12, w13, w14, w15;
373 uint64_t T1, T2;
374
375 #if defined(__sparc)
376 static const uint64_t sha512_consts[] = {
377 SHA512_CONST_0, SHA512_CONST_1, SHA512_CONST_2,
378 SHA512_CONST_3, SHA512_CONST_4, SHA512_CONST_5,
379 SHA512_CONST_6, SHA512_CONST_7, SHA512_CONST_8,
380 SHA512_CONST_9, SHA512_CONST_10, SHA512_CONST_11,
381 SHA512_CONST_12, SHA512_CONST_13, SHA512_CONST_14,
382 SHA512_CONST_15, SHA512_CONST_16, SHA512_CONST_17,
383 SHA512_CONST_18, SHA512_CONST_19, SHA512_CONST_20,
384 SHA512_CONST_21, SHA512_CONST_22, SHA512_CONST_23,
385 SHA512_CONST_24, SHA512_CONST_25, SHA512_CONST_26,
386 SHA512_CONST_27, SHA512_CONST_28, SHA512_CONST_29,
387 SHA512_CONST_30, SHA512_CONST_31, SHA512_CONST_32,
388 SHA512_CONST_33, SHA512_CONST_34, SHA512_CONST_35,
389 SHA512_CONST_36, SHA512_CONST_37, SHA512_CONST_38,
390 SHA512_CONST_39, SHA512_CONST_40, SHA512_CONST_41,
391 SHA512_CONST_42, SHA512_CONST_43, SHA512_CONST_44,
392 SHA512_CONST_45, SHA512_CONST_46, SHA512_CONST_47,
393 SHA512_CONST_48, SHA512_CONST_49, SHA512_CONST_50,
394 SHA512_CONST_51, SHA512_CONST_52, SHA512_CONST_53,
395 SHA512_CONST_54, SHA512_CONST_55, SHA512_CONST_56,
396 SHA512_CONST_57, SHA512_CONST_58, SHA512_CONST_59,
397 SHA512_CONST_60, SHA512_CONST_61, SHA512_CONST_62,
398 SHA512_CONST_63, SHA512_CONST_64, SHA512_CONST_65,
399 SHA512_CONST_66, SHA512_CONST_67, SHA512_CONST_68,
400 SHA512_CONST_69, SHA512_CONST_70, SHA512_CONST_71,
401 SHA512_CONST_72, SHA512_CONST_73, SHA512_CONST_74,
402 SHA512_CONST_75, SHA512_CONST_76, SHA512_CONST_77,
403 SHA512_CONST_78, SHA512_CONST_79
404 };
405 #endif /* __sparc */
406
407
408 if ((uintptr_t)blk & 0x7) { /* not 8-byte aligned? */
409 bcopy(blk, ctx->buf_un.buf64, sizeof (ctx->buf_un.buf64));
410 blk = (uint8_t *)ctx->buf_un.buf64;
411 }
412
413 /* LINTED E_BAD_PTR_CAST_ALIGN */
414 w0 = LOAD_BIG_64(blk + 8 * 0);
415 SHA512ROUND(a, b, c, d, e, f, g, h, 0, w0);
416 /* LINTED E_BAD_PTR_CAST_ALIGN */
417 w1 = LOAD_BIG_64(blk + 8 * 1);
418 SHA512ROUND(h, a, b, c, d, e, f, g, 1, w1);
419 /* LINTED E_BAD_PTR_CAST_ALIGN */
420 w2 = LOAD_BIG_64(blk + 8 * 2);
421 SHA512ROUND(g, h, a, b, c, d, e, f, 2, w2);
422 /* LINTED E_BAD_PTR_CAST_ALIGN */
423 w3 = LOAD_BIG_64(blk + 8 * 3);
424 SHA512ROUND(f, g, h, a, b, c, d, e, 3, w3);
425 /* LINTED E_BAD_PTR_CAST_ALIGN */
426 w4 = LOAD_BIG_64(blk + 8 * 4);
427 SHA512ROUND(e, f, g, h, a, b, c, d, 4, w4);
428 /* LINTED E_BAD_PTR_CAST_ALIGN */
429 w5 = LOAD_BIG_64(blk + 8 * 5);
430 SHA512ROUND(d, e, f, g, h, a, b, c, 5, w5);
431 /* LINTED E_BAD_PTR_CAST_ALIGN */
432 w6 = LOAD_BIG_64(blk + 8 * 6);
433 SHA512ROUND(c, d, e, f, g, h, a, b, 6, w6);
434 /* LINTED E_BAD_PTR_CAST_ALIGN */
435 w7 = LOAD_BIG_64(blk + 8 * 7);
436 SHA512ROUND(b, c, d, e, f, g, h, a, 7, w7);
437 /* LINTED E_BAD_PTR_CAST_ALIGN */
438 w8 = LOAD_BIG_64(blk + 8 * 8);
439 SHA512ROUND(a, b, c, d, e, f, g, h, 8, w8);
440 /* LINTED E_BAD_PTR_CAST_ALIGN */
441 w9 = LOAD_BIG_64(blk + 8 * 9);
442 SHA512ROUND(h, a, b, c, d, e, f, g, 9, w9);
443 /* LINTED E_BAD_PTR_CAST_ALIGN */
444 w10 = LOAD_BIG_64(blk + 8 * 10);
445 SHA512ROUND(g, h, a, b, c, d, e, f, 10, w10);
446 /* LINTED E_BAD_PTR_CAST_ALIGN */
447 w11 = LOAD_BIG_64(blk + 8 * 11);
448 SHA512ROUND(f, g, h, a, b, c, d, e, 11, w11);
449 /* LINTED E_BAD_PTR_CAST_ALIGN */
450 w12 = LOAD_BIG_64(blk + 8 * 12);
451 SHA512ROUND(e, f, g, h, a, b, c, d, 12, w12);
452 /* LINTED E_BAD_PTR_CAST_ALIGN */
453 w13 = LOAD_BIG_64(blk + 8 * 13);
454 SHA512ROUND(d, e, f, g, h, a, b, c, 13, w13);
455 /* LINTED E_BAD_PTR_CAST_ALIGN */
456 w14 = LOAD_BIG_64(blk + 8 * 14);
457 SHA512ROUND(c, d, e, f, g, h, a, b, 14, w14);
458 /* LINTED E_BAD_PTR_CAST_ALIGN */
459 w15 = LOAD_BIG_64(blk + 8 * 15);
460 SHA512ROUND(b, c, d, e, f, g, h, a, 15, w15);
461
462 w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
463 SHA512ROUND(a, b, c, d, e, f, g, h, 16, w0);
464 w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
465 SHA512ROUND(h, a, b, c, d, e, f, g, 17, w1);
466 w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
467 SHA512ROUND(g, h, a, b, c, d, e, f, 18, w2);
468 w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
469 SHA512ROUND(f, g, h, a, b, c, d, e, 19, w3);
470 w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
471 SHA512ROUND(e, f, g, h, a, b, c, d, 20, w4);
472 w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
473 SHA512ROUND(d, e, f, g, h, a, b, c, 21, w5);
474 w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
475 SHA512ROUND(c, d, e, f, g, h, a, b, 22, w6);
476 w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
477 SHA512ROUND(b, c, d, e, f, g, h, a, 23, w7);
478 w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
479 SHA512ROUND(a, b, c, d, e, f, g, h, 24, w8);
480 w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
481 SHA512ROUND(h, a, b, c, d, e, f, g, 25, w9);
482 w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
483 SHA512ROUND(g, h, a, b, c, d, e, f, 26, w10);
484 w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
485 SHA512ROUND(f, g, h, a, b, c, d, e, 27, w11);
486 w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
487 SHA512ROUND(e, f, g, h, a, b, c, d, 28, w12);
488 w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
489 SHA512ROUND(d, e, f, g, h, a, b, c, 29, w13);
490 w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
491 SHA512ROUND(c, d, e, f, g, h, a, b, 30, w14);
492 w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
493 SHA512ROUND(b, c, d, e, f, g, h, a, 31, w15);
494
495 w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
496 SHA512ROUND(a, b, c, d, e, f, g, h, 32, w0);
497 w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
498 SHA512ROUND(h, a, b, c, d, e, f, g, 33, w1);
499 w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
500 SHA512ROUND(g, h, a, b, c, d, e, f, 34, w2);
501 w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
502 SHA512ROUND(f, g, h, a, b, c, d, e, 35, w3);
503 w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
504 SHA512ROUND(e, f, g, h, a, b, c, d, 36, w4);
505 w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
506 SHA512ROUND(d, e, f, g, h, a, b, c, 37, w5);
507 w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
508 SHA512ROUND(c, d, e, f, g, h, a, b, 38, w6);
509 w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
510 SHA512ROUND(b, c, d, e, f, g, h, a, 39, w7);
511 w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
512 SHA512ROUND(a, b, c, d, e, f, g, h, 40, w8);
513 w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
514 SHA512ROUND(h, a, b, c, d, e, f, g, 41, w9);
515 w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
516 SHA512ROUND(g, h, a, b, c, d, e, f, 42, w10);
517 w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
518 SHA512ROUND(f, g, h, a, b, c, d, e, 43, w11);
519 w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
520 SHA512ROUND(e, f, g, h, a, b, c, d, 44, w12);
521 w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
522 SHA512ROUND(d, e, f, g, h, a, b, c, 45, w13);
523 w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
524 SHA512ROUND(c, d, e, f, g, h, a, b, 46, w14);
525 w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
526 SHA512ROUND(b, c, d, e, f, g, h, a, 47, w15);
527
528 w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
529 SHA512ROUND(a, b, c, d, e, f, g, h, 48, w0);
530 w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
531 SHA512ROUND(h, a, b, c, d, e, f, g, 49, w1);
532 w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
533 SHA512ROUND(g, h, a, b, c, d, e, f, 50, w2);
534 w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
535 SHA512ROUND(f, g, h, a, b, c, d, e, 51, w3);
536 w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
537 SHA512ROUND(e, f, g, h, a, b, c, d, 52, w4);
538 w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
539 SHA512ROUND(d, e, f, g, h, a, b, c, 53, w5);
540 w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
541 SHA512ROUND(c, d, e, f, g, h, a, b, 54, w6);
542 w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
543 SHA512ROUND(b, c, d, e, f, g, h, a, 55, w7);
544 w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
545 SHA512ROUND(a, b, c, d, e, f, g, h, 56, w8);
546 w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
547 SHA512ROUND(h, a, b, c, d, e, f, g, 57, w9);
548 w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
549 SHA512ROUND(g, h, a, b, c, d, e, f, 58, w10);
550 w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
551 SHA512ROUND(f, g, h, a, b, c, d, e, 59, w11);
552 w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
553 SHA512ROUND(e, f, g, h, a, b, c, d, 60, w12);
554 w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
555 SHA512ROUND(d, e, f, g, h, a, b, c, 61, w13);
556 w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
557 SHA512ROUND(c, d, e, f, g, h, a, b, 62, w14);
558 w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
559 SHA512ROUND(b, c, d, e, f, g, h, a, 63, w15);
560
561 w0 = SIGMA1(w14) + w9 + SIGMA0(w1) + w0;
562 SHA512ROUND(a, b, c, d, e, f, g, h, 64, w0);
563 w1 = SIGMA1(w15) + w10 + SIGMA0(w2) + w1;
564 SHA512ROUND(h, a, b, c, d, e, f, g, 65, w1);
565 w2 = SIGMA1(w0) + w11 + SIGMA0(w3) + w2;
566 SHA512ROUND(g, h, a, b, c, d, e, f, 66, w2);
567 w3 = SIGMA1(w1) + w12 + SIGMA0(w4) + w3;
568 SHA512ROUND(f, g, h, a, b, c, d, e, 67, w3);
569 w4 = SIGMA1(w2) + w13 + SIGMA0(w5) + w4;
570 SHA512ROUND(e, f, g, h, a, b, c, d, 68, w4);
571 w5 = SIGMA1(w3) + w14 + SIGMA0(w6) + w5;
572 SHA512ROUND(d, e, f, g, h, a, b, c, 69, w5);
573 w6 = SIGMA1(w4) + w15 + SIGMA0(w7) + w6;
574 SHA512ROUND(c, d, e, f, g, h, a, b, 70, w6);
575 w7 = SIGMA1(w5) + w0 + SIGMA0(w8) + w7;
576 SHA512ROUND(b, c, d, e, f, g, h, a, 71, w7);
577 w8 = SIGMA1(w6) + w1 + SIGMA0(w9) + w8;
578 SHA512ROUND(a, b, c, d, e, f, g, h, 72, w8);
579 w9 = SIGMA1(w7) + w2 + SIGMA0(w10) + w9;
580 SHA512ROUND(h, a, b, c, d, e, f, g, 73, w9);
581 w10 = SIGMA1(w8) + w3 + SIGMA0(w11) + w10;
582 SHA512ROUND(g, h, a, b, c, d, e, f, 74, w10);
583 w11 = SIGMA1(w9) + w4 + SIGMA0(w12) + w11;
584 SHA512ROUND(f, g, h, a, b, c, d, e, 75, w11);
585 w12 = SIGMA1(w10) + w5 + SIGMA0(w13) + w12;
586 SHA512ROUND(e, f, g, h, a, b, c, d, 76, w12);
587 w13 = SIGMA1(w11) + w6 + SIGMA0(w14) + w13;
588 SHA512ROUND(d, e, f, g, h, a, b, c, 77, w13);
589 w14 = SIGMA1(w12) + w7 + SIGMA0(w15) + w14;
590 SHA512ROUND(c, d, e, f, g, h, a, b, 78, w14);
591 w15 = SIGMA1(w13) + w8 + SIGMA0(w0) + w15;
592 SHA512ROUND(b, c, d, e, f, g, h, a, 79, w15);
593
594 ctx->state.s64[0] += a;
595 ctx->state.s64[1] += b;
596 ctx->state.s64[2] += c;
597 ctx->state.s64[3] += d;
598 ctx->state.s64[4] += e;
599 ctx->state.s64[5] += f;
600 ctx->state.s64[6] += g;
601 ctx->state.s64[7] += h;
602
603 }
604 #endif /* !__amd64 || !_KERNEL */
605
606
607 /*
608 * Encode()
609 *
610 * purpose: to convert a list of numbers from little endian to big endian
611 * input: uint8_t * : place to store the converted big endian numbers
612 * uint32_t * : place to get numbers to convert from
613 * size_t : the length of the input in bytes
614 * output: void
615 */
616
617 static void
Encode(uint8_t * _RESTRICT_KYWD output,uint32_t * _RESTRICT_KYWD input,size_t len)618 Encode(uint8_t *_RESTRICT_KYWD output, uint32_t *_RESTRICT_KYWD input,
619 size_t len)
620 {
621 size_t i, j;
622
623 #if defined(__sparc)
624 if (IS_P2ALIGNED(output, sizeof (uint32_t))) {
625 for (i = 0, j = 0; j < len; i++, j += 4) {
626 /* LINTED E_BAD_PTR_CAST_ALIGN */
627 *((uint32_t *)(output + j)) = input[i];
628 }
629 } else {
630 #endif /* little endian -- will work on big endian, but slowly */
631 for (i = 0, j = 0; j < len; i++, j += 4) {
632 output[j] = (input[i] >> 24) & 0xff;
633 output[j + 1] = (input[i] >> 16) & 0xff;
634 output[j + 2] = (input[i] >> 8) & 0xff;
635 output[j + 3] = input[i] & 0xff;
636 }
637 #if defined(__sparc)
638 }
639 #endif
640 }
641
642 static void
Encode64(uint8_t * _RESTRICT_KYWD output,uint64_t * _RESTRICT_KYWD input,size_t len)643 Encode64(uint8_t *_RESTRICT_KYWD output, uint64_t *_RESTRICT_KYWD input,
644 size_t len)
645 {
646 size_t i, j;
647
648 #if defined(__sparc)
649 if (IS_P2ALIGNED(output, sizeof (uint64_t))) {
650 for (i = 0, j = 0; j < len; i++, j += 8) {
651 /* LINTED E_BAD_PTR_CAST_ALIGN */
652 *((uint64_t *)(output + j)) = input[i];
653 }
654 } else {
655 #endif /* little endian -- will work on big endian, but slowly */
656 for (i = 0, j = 0; j < len; i++, j += 8) {
657
658 output[j] = (input[i] >> 56) & 0xff;
659 output[j + 1] = (input[i] >> 48) & 0xff;
660 output[j + 2] = (input[i] >> 40) & 0xff;
661 output[j + 3] = (input[i] >> 32) & 0xff;
662 output[j + 4] = (input[i] >> 24) & 0xff;
663 output[j + 5] = (input[i] >> 16) & 0xff;
664 output[j + 6] = (input[i] >> 8) & 0xff;
665 output[j + 7] = input[i] & 0xff;
666 }
667 #if defined(__sparc)
668 }
669 #endif
670 }
671
672
673 void
SHA2Init(uint64_t mech,SHA2_CTX * ctx)674 SHA2Init(uint64_t mech, SHA2_CTX *ctx)
675 {
676
677 switch (mech) {
678 case SHA256_MECH_INFO_TYPE:
679 case SHA256_HMAC_MECH_INFO_TYPE:
680 case SHA256_HMAC_GEN_MECH_INFO_TYPE:
681 ctx->state.s32[0] = 0x6a09e667U;
682 ctx->state.s32[1] = 0xbb67ae85U;
683 ctx->state.s32[2] = 0x3c6ef372U;
684 ctx->state.s32[3] = 0xa54ff53aU;
685 ctx->state.s32[4] = 0x510e527fU;
686 ctx->state.s32[5] = 0x9b05688cU;
687 ctx->state.s32[6] = 0x1f83d9abU;
688 ctx->state.s32[7] = 0x5be0cd19U;
689 break;
690 case SHA384_MECH_INFO_TYPE:
691 case SHA384_HMAC_MECH_INFO_TYPE:
692 case SHA384_HMAC_GEN_MECH_INFO_TYPE:
693 ctx->state.s64[0] = 0xcbbb9d5dc1059ed8ULL;
694 ctx->state.s64[1] = 0x629a292a367cd507ULL;
695 ctx->state.s64[2] = 0x9159015a3070dd17ULL;
696 ctx->state.s64[3] = 0x152fecd8f70e5939ULL;
697 ctx->state.s64[4] = 0x67332667ffc00b31ULL;
698 ctx->state.s64[5] = 0x8eb44a8768581511ULL;
699 ctx->state.s64[6] = 0xdb0c2e0d64f98fa7ULL;
700 ctx->state.s64[7] = 0x47b5481dbefa4fa4ULL;
701 break;
702 case SHA512_MECH_INFO_TYPE:
703 case SHA512_HMAC_MECH_INFO_TYPE:
704 case SHA512_HMAC_GEN_MECH_INFO_TYPE:
705 ctx->state.s64[0] = 0x6a09e667f3bcc908ULL;
706 ctx->state.s64[1] = 0xbb67ae8584caa73bULL;
707 ctx->state.s64[2] = 0x3c6ef372fe94f82bULL;
708 ctx->state.s64[3] = 0xa54ff53a5f1d36f1ULL;
709 ctx->state.s64[4] = 0x510e527fade682d1ULL;
710 ctx->state.s64[5] = 0x9b05688c2b3e6c1fULL;
711 ctx->state.s64[6] = 0x1f83d9abfb41bd6bULL;
712 ctx->state.s64[7] = 0x5be0cd19137e2179ULL;
713 break;
714 case SHA512_224_MECH_INFO_TYPE:
715 ctx->state.s64[0] = 0x8C3D37C819544DA2ULL;
716 ctx->state.s64[1] = 0x73E1996689DCD4D6ULL;
717 ctx->state.s64[2] = 0x1DFAB7AE32FF9C82ULL;
718 ctx->state.s64[3] = 0x679DD514582F9FCFULL;
719 ctx->state.s64[4] = 0x0F6D2B697BD44DA8ULL;
720 ctx->state.s64[5] = 0x77E36F7304C48942ULL;
721 ctx->state.s64[6] = 0x3F9D85A86A1D36C8ULL;
722 ctx->state.s64[7] = 0x1112E6AD91D692A1ULL;
723 break;
724 case SHA512_256_MECH_INFO_TYPE:
725 ctx->state.s64[0] = 0x22312194FC2BF72CULL;
726 ctx->state.s64[1] = 0x9F555FA3C84C64C2ULL;
727 ctx->state.s64[2] = 0x2393B86B6F53B151ULL;
728 ctx->state.s64[3] = 0x963877195940EABDULL;
729 ctx->state.s64[4] = 0x96283EE2A88EFFE3ULL;
730 ctx->state.s64[5] = 0xBE5E1E2553863992ULL;
731 ctx->state.s64[6] = 0x2B0199FC2C85B8AAULL;
732 ctx->state.s64[7] = 0x0EB72DDC81C52CA2ULL;
733 break;
734 #ifdef _KERNEL
735 default:
736 cmn_err(CE_PANIC,
737 "sha2_init: failed to find a supported algorithm: 0x%x",
738 (uint32_t)mech);
739
740 #endif /* _KERNEL */
741 }
742
743 ctx->algotype = (uint32_t)mech;
744 ctx->count.c64[0] = ctx->count.c64[1] = 0;
745 }
746
747 #ifndef _KERNEL
748
749 // #pragma inline(SHA256Init, SHA384Init, SHA512Init)
750 void
SHA256Init(SHA256_CTX * ctx)751 SHA256Init(SHA256_CTX *ctx)
752 {
753 SHA2Init(SHA256, ctx);
754 }
755
756 void
SHA384Init(SHA384_CTX * ctx)757 SHA384Init(SHA384_CTX *ctx)
758 {
759 SHA2Init(SHA384, ctx);
760 }
761
762 void
SHA512Init(SHA512_CTX * ctx)763 SHA512Init(SHA512_CTX *ctx)
764 {
765 SHA2Init(SHA512, ctx);
766 }
767
768 #endif /* _KERNEL */
769
770 /*
771 * SHA2Update()
772 *
773 * purpose: continues an sha2 digest operation, using the message block
774 * to update the context.
775 * input: SHA2_CTX * : the context to update
776 * void * : the message block
777 * size_t : the length of the message block, in bytes
778 * output: void
779 */
780
781 void
SHA2Update(SHA2_CTX * ctx,const void * inptr,size_t input_len)782 SHA2Update(SHA2_CTX *ctx, const void *inptr, size_t input_len)
783 {
784 uint32_t i, buf_index, buf_len, buf_limit;
785 const uint8_t *input = inptr;
786 uint32_t algotype = ctx->algotype;
787
788 /* check for noop */
789 if (input_len == 0)
790 return;
791
792 if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
793 buf_limit = 64;
794
795 /* compute number of bytes mod 64 */
796 buf_index = (ctx->count.c32[1] >> 3) & 0x3F;
797
798 /* update number of bits */
799 if ((ctx->count.c32[1] += (input_len << 3)) < (input_len << 3))
800 ctx->count.c32[0]++;
801
802 ctx->count.c32[0] += (input_len >> 29);
803
804 } else {
805 buf_limit = 128;
806
807 /* compute number of bytes mod 128 */
808 buf_index = (ctx->count.c64[1] >> 3) & 0x7F;
809
810 /* update number of bits */
811 if ((ctx->count.c64[1] += (input_len << 3)) < (input_len << 3))
812 ctx->count.c64[0]++;
813
814 ctx->count.c64[0] += (input_len >> 29);
815 }
816
817 buf_len = buf_limit - buf_index;
818
819 /* transform as many times as possible */
820 i = 0;
821 if (input_len >= buf_len) {
822
823 /*
824 * general optimization:
825 *
826 * only do initial bcopy() and SHA2Transform() if
827 * buf_index != 0. if buf_index == 0, we're just
828 * wasting our time doing the bcopy() since there
829 * wasn't any data left over from a previous call to
830 * SHA2Update().
831 */
832 if (buf_index) {
833 bcopy(input, &ctx->buf_un.buf8[buf_index], buf_len);
834 if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE)
835 SHA256Transform(ctx, ctx->buf_un.buf8);
836 else
837 SHA512Transform(ctx, ctx->buf_un.buf8);
838
839 i = buf_len;
840 }
841
842 #if !defined(__amd64) || !defined(_KERNEL)
843 if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
844 for (; i + buf_limit - 1 < input_len; i += buf_limit) {
845 SHA256Transform(ctx, &input[i]);
846 }
847 } else {
848 for (; i + buf_limit - 1 < input_len; i += buf_limit) {
849 SHA512Transform(ctx, &input[i]);
850 }
851 }
852
853 #else
854 uint32_t block_count;
855 if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
856 block_count = (input_len - i) >> 6;
857 if (block_count > 0) {
858 SHA256TransformBlocks(ctx, &input[i],
859 block_count);
860 i += block_count << 6;
861 }
862 } else {
863 block_count = (input_len - i) >> 7;
864 if (block_count > 0) {
865 SHA512TransformBlocks(ctx, &input[i],
866 block_count);
867 i += block_count << 7;
868 }
869 }
870 #endif /* !__amd64 || !_KERNEL */
871
872 /*
873 * general optimization:
874 *
875 * if i and input_len are the same, return now instead
876 * of calling bcopy(), since the bcopy() in this case
877 * will be an expensive noop.
878 */
879
880 if (input_len == i)
881 return;
882
883 buf_index = 0;
884 }
885
886 /* buffer remaining input */
887 bcopy(&input[i], &ctx->buf_un.buf8[buf_index], input_len - i);
888 }
889
890
891 /*
892 * SHA2Final()
893 *
894 * purpose: ends an sha2 digest operation, finalizing the message digest and
895 * zeroing the context.
896 * input: uchar_t * : a buffer to store the digest
897 * : The function actually uses void* because many
898 * : callers pass things other than uchar_t here.
899 * SHA2_CTX * : the context to finalize, save, and zero
900 * output: void
901 */
902
903 void
SHA2Final(void * digest,SHA2_CTX * ctx)904 SHA2Final(void *digest, SHA2_CTX *ctx)
905 {
906 uint8_t bitcount_be[sizeof (ctx->count.c32)];
907 uint8_t bitcount_be64[sizeof (ctx->count.c64)];
908 uint32_t index;
909 uint32_t algotype = ctx->algotype;
910
911 if (algotype <= SHA256_HMAC_GEN_MECH_INFO_TYPE) {
912 index = (ctx->count.c32[1] >> 3) & 0x3f;
913 Encode(bitcount_be, ctx->count.c32, sizeof (bitcount_be));
914 SHA2Update(ctx, PADDING, ((index < 56) ? 56 : 120) - index);
915 SHA2Update(ctx, bitcount_be, sizeof (bitcount_be));
916 Encode(digest, ctx->state.s32, sizeof (ctx->state.s32));
917 } else {
918 index = (ctx->count.c64[1] >> 3) & 0x7f;
919 Encode64(bitcount_be64, ctx->count.c64,
920 sizeof (bitcount_be64));
921 SHA2Update(ctx, PADDING, ((index < 112) ? 112 : 240) - index);
922 SHA2Update(ctx, bitcount_be64, sizeof (bitcount_be64));
923 if (algotype <= SHA384_HMAC_GEN_MECH_INFO_TYPE) {
924 ctx->state.s64[6] = ctx->state.s64[7] = 0;
925 Encode64(digest, ctx->state.s64,
926 sizeof (uint64_t) * 6);
927 } else if (algotype == SHA512_224_MECH_INFO_TYPE) {
928 uint8_t last[sizeof (uint64_t)];
929 /*
930 * Since SHA-512/224 doesn't align well to 64-bit
931 * boundaries, we must do the encoding in three steps:
932 * 1) encode the three 64-bit words that fit neatly
933 * 2) encode the last 64-bit word to a temp buffer
934 * 3) chop out the lower 32-bits from the temp buffer
935 * and append them to the digest
936 */
937 Encode64(digest, ctx->state.s64, sizeof (uint64_t) * 3);
938 Encode64(last, &ctx->state.s64[3], sizeof (uint64_t));
939 bcopy(last, (uint8_t *)digest + 24, 4);
940 } else if (algotype == SHA512_256_MECH_INFO_TYPE) {
941 Encode64(digest, ctx->state.s64, sizeof (uint64_t) * 4);
942 } else {
943 Encode64(digest, ctx->state.s64,
944 sizeof (ctx->state.s64));
945 }
946 }
947
948 /* zeroize sensitive information */
949 bzero(ctx, sizeof (*ctx));
950 }
951
952 #ifdef _KERNEL
953 EXPORT_SYMBOL(SHA2Init);
954 EXPORT_SYMBOL(SHA2Update);
955 EXPORT_SYMBOL(SHA2Final);
956 #endif
957