/src/wolfssl-sp-math/wolfcrypt/src/poly1305.c
Line | Count | Source (jump to first uncovered line) |
1 | | /* poly1305.c |
2 | | * |
3 | | * Copyright (C) 2006-2025 wolfSSL Inc. |
4 | | * |
5 | | * This file is part of wolfSSL. |
6 | | * |
7 | | * wolfSSL is free software; you can redistribute it and/or modify |
8 | | * it under the terms of the GNU General Public License as published by |
9 | | * the Free Software Foundation; either version 3 of the License, or |
10 | | * (at your option) any later version. |
11 | | * |
12 | | * wolfSSL is distributed in the hope that it will be useful, |
13 | | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
14 | | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
15 | | * GNU General Public License for more details. |
16 | | * |
17 | | * You should have received a copy of the GNU General Public License |
18 | | * along with this program; if not, write to the Free Software |
19 | | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1335, USA |
20 | | */ |
21 | | /* |
22 | | |
23 | | DESCRIPTION |
24 | | This library contains implementation for the Poly1305 authenticator. |
25 | | |
26 | | Based off the public domain implementations by Andrew Moon |
27 | | and Daniel J. Bernstein |
28 | | |
29 | | */ |
30 | | |
31 | | |
32 | | /* |
33 | | * WOLFSSL_W64_WRAPPER Uses wrappers around word64 types for a system that does |
34 | | * not have word64 available. As expected it reduces |
35 | | * performance. Benchmarks collected July 2024 show |
36 | | * 303.004 MiB/s with and 1874.194 MiB/s without. |
37 | | */ |
38 | | |
39 | | #include <wolfssl/wolfcrypt/libwolfssl_sources.h> |
40 | | |
41 | | #ifdef HAVE_POLY1305 |
42 | | #include <wolfssl/wolfcrypt/poly1305.h> |
43 | | #include <wolfssl/wolfcrypt/cpuid.h> |
44 | | #ifdef NO_INLINE |
45 | | #include <wolfssl/wolfcrypt/misc.h> |
46 | | #else |
47 | | #define WOLFSSL_MISC_INCLUDED |
48 | | #include <wolfcrypt/src/misc.c> |
49 | | #endif |
50 | | #ifdef CHACHA_AEAD_TEST |
51 | | #include <stdio.h> |
52 | | #endif |
53 | | |
54 | | #ifdef _MSC_VER |
55 | | /* 4127 warning constant while(1) */ |
56 | | #pragma warning(disable: 4127) |
57 | | #endif |
58 | | |
59 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
60 | | #include <emmintrin.h> |
61 | | #include <immintrin.h> |
62 | | |
63 | | #if defined(__GNUC__) && ((__GNUC__ < 4) || \ |
64 | | (__GNUC__ == 4 && __GNUC_MINOR__ <= 8)) |
65 | | #undef NO_AVX2_SUPPORT |
66 | | #define NO_AVX2_SUPPORT |
67 | | #endif |
68 | | #if defined(__clang__) && ((__clang_major__ < 3) || \ |
69 | | (__clang_major__ == 3 && __clang_minor__ <= 5)) |
70 | | #define NO_AVX2_SUPPORT |
71 | | #elif defined(__clang__) && defined(NO_AVX2_SUPPORT) |
72 | | #undef NO_AVX2_SUPPORT |
73 | | #endif |
74 | | #if defined(_MSC_VER) && (_MSC_VER <= 1900) |
75 | | #undef NO_AVX2_SUPPORT |
76 | | #define NO_AVX2_SUPPORT |
77 | | #endif |
78 | | |
79 | | #define HAVE_INTEL_AVX1 |
80 | | #ifndef NO_AVX2_SUPPORT |
81 | | #define HAVE_INTEL_AVX2 |
82 | | #endif |
83 | | #endif |
84 | | |
85 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
86 | | static word32 intel_flags = 0; |
87 | | static word32 cpu_flags_set = 0; |
88 | | #endif |
89 | | |
90 | | #if defined(USE_INTEL_POLY1305_SPEEDUP) || defined(POLY130564) |
91 | | #if defined(_MSC_VER) |
92 | | #define POLY1305_NOINLINE __declspec(noinline) |
93 | | #elif defined(__GNUC__) |
94 | | #define POLY1305_NOINLINE __attribute__((noinline)) |
95 | | #else |
96 | | #define POLY1305_NOINLINE |
97 | | #endif |
98 | | |
99 | | #if defined(_MSC_VER) |
100 | | #include <intrin.h> |
101 | | |
102 | | typedef struct word128 { |
103 | | word64 lo; |
104 | | word64 hi; |
105 | | } word128; |
106 | | |
107 | | #define MUL(out, x, y) out.lo = _umul128((x), (y), &out.hi) |
108 | | #define ADD(out, in) { word64 t = out.lo; out.lo += in.lo; \ |
109 | | out.hi += (out.lo < t) + in.hi; } |
110 | | #define ADDLO(out, in) { word64 t = out.lo; out.lo += in; \ |
111 | | out.hi += (out.lo < t); } |
112 | | #define SHR(in, shift) (__shiftright128(in.lo, in.hi, (shift))) |
113 | | #define LO(in) (in.lo) |
114 | | |
115 | | #elif defined(__GNUC__) |
116 | | #if defined(__SIZEOF_INT128__) |
117 | | PEDANTIC_EXTENSION typedef unsigned __int128 word128; |
118 | | #else |
119 | | typedef unsigned word128 __attribute__((mode(TI))); |
120 | | #endif |
121 | | |
122 | 508k | #define MUL(out, x, y) out = ((word128)(x) * (y)) |
123 | 339k | #define ADD(out, in) (out) += (in) |
124 | 113k | #define ADDLO(out, in) (out) += (in) |
125 | 169k | #define SHR(in, shift) (word64)((in) >> (shift)) |
126 | 169k | #define LO(in) (word64)(in) |
127 | | #endif |
128 | | #endif |
129 | | |
130 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
131 | | #ifdef __cplusplus |
132 | | extern "C" { |
133 | | #endif |
134 | | |
135 | | #ifdef HAVE_INTEL_AVX1 |
136 | | /* Process one block (16 bytes) of data. |
137 | | * |
138 | | * ctx Poly1305 context. |
139 | | * m One block of message data. |
140 | | */ |
141 | | extern void poly1305_block_avx(Poly1305* ctx, const unsigned char *m); |
142 | | /* Process multiple blocks (n * 16 bytes) of data. |
143 | | * |
144 | | * ctx Poly1305 context. |
145 | | * m Blocks of message data. |
146 | | * bytes The number of bytes to process. |
147 | | */ |
148 | | extern void poly1305_blocks_avx(Poly1305* ctx, const unsigned char* m, |
149 | | size_t bytes); |
150 | | /* Set the key to use when processing data. |
151 | | * Initialize the context. |
152 | | * |
153 | | * ctx Poly1305 context. |
154 | | * key The key data (16 bytes). |
155 | | */ |
156 | | extern void poly1305_setkey_avx(Poly1305* ctx, const byte* key); |
157 | | /* Calculate the final result - authentication data. |
158 | | * Zeros out the private data in the context. |
159 | | * |
160 | | * ctx Poly1305 context. |
161 | | * mac Buffer to hold 16 bytes. |
162 | | */ |
163 | | extern void poly1305_final_avx(Poly1305* ctx, byte* mac); |
164 | | #endif |
165 | | |
166 | | #ifdef HAVE_INTEL_AVX2 |
167 | | /* Process multiple blocks (n * 16 bytes) of data. |
168 | | * |
169 | | * ctx Poly1305 context. |
170 | | * m Blocks of message data. |
171 | | * bytes The number of bytes to process. |
172 | | */ |
173 | | extern void poly1305_blocks_avx2(Poly1305* ctx, const unsigned char* m, |
174 | | size_t bytes); |
175 | | /* Calculate R^1, R^2, R^3 and R^4 and store them in the context. |
176 | | * |
177 | | * ctx Poly1305 context. |
178 | | */ |
179 | | extern void poly1305_calc_powers_avx2(Poly1305* ctx); |
180 | | /* Set the key to use when processing data. |
181 | | * Initialize the context. |
182 | | * Calls AVX set key function as final function calls AVX code. |
183 | | * |
184 | | * ctx Poly1305 context. |
185 | | * key The key data (16 bytes). |
186 | | */ |
187 | | extern void poly1305_setkey_avx2(Poly1305* ctx, const byte* key); |
188 | | /* Calculate the final result - authentication data. |
189 | | * Zeros out the private data in the context. |
190 | | * Calls AVX final function to quickly process last blocks. |
191 | | * |
192 | | * ctx Poly1305 context. |
193 | | * mac Buffer to hold 16 bytes - authentication data. |
194 | | */ |
195 | | extern void poly1305_final_avx2(Poly1305* ctx, byte* mac); |
196 | | #endif |
197 | | |
198 | | #ifdef __cplusplus |
199 | | } /* extern "C" */ |
200 | | #endif |
201 | | |
202 | | #elif defined(POLY130564) |
203 | | #if !defined(WOLFSSL_ARMASM) && !defined(WOLFSSL_RISCV_ASM) |
204 | | static word64 U8TO64(const byte* p) |
205 | 127k | { |
206 | 127k | return |
207 | 127k | (((word64)(p[0] & 0xff) ) | |
208 | 127k | ((word64)(p[1] & 0xff) << 8) | |
209 | 127k | ((word64)(p[2] & 0xff) << 16) | |
210 | 127k | ((word64)(p[3] & 0xff) << 24) | |
211 | 127k | ((word64)(p[4] & 0xff) << 32) | |
212 | 127k | ((word64)(p[5] & 0xff) << 40) | |
213 | 127k | ((word64)(p[6] & 0xff) << 48) | |
214 | 127k | ((word64)(p[7] & 0xff) << 56)); |
215 | 127k | } |
216 | | |
217 | 7.30k | static void U64TO8(byte* p, word64 v) { |
218 | 7.30k | p[0] = (byte)v; |
219 | 7.30k | p[1] = (byte)(v >> 8); |
220 | 7.30k | p[2] = (byte)(v >> 16); |
221 | 7.30k | p[3] = (byte)(v >> 24); |
222 | 7.30k | p[4] = (byte)(v >> 32); |
223 | 7.30k | p[5] = (byte)(v >> 40); |
224 | 7.30k | p[6] = (byte)(v >> 48); |
225 | 7.30k | p[7] = (byte)(v >> 56); |
226 | 7.30k | } |
227 | | #endif/* !WOLFSSL_ARMASM && !WOLFSSL_RISCV_ASM */ |
228 | | /* if not 64 bit then use 32 bit */ |
229 | | #elif !defined(WOLFSSL_ARMASM) |
230 | | |
231 | | static word32 U8TO32(const byte *p) |
232 | | { |
233 | | return |
234 | | (((word32)(p[0] & 0xff) ) | |
235 | | ((word32)(p[1] & 0xff) << 8) | |
236 | | ((word32)(p[2] & 0xff) << 16) | |
237 | | ((word32)(p[3] & 0xff) << 24)); |
238 | | } |
239 | | |
240 | | static void U32TO8(byte *p, word32 v) { |
241 | | p[0] = (byte)((v ) & 0xff); |
242 | | p[1] = (byte)((v >> 8) & 0xff); |
243 | | p[2] = (byte)((v >> 16) & 0xff); |
244 | | p[3] = (byte)((v >> 24) & 0xff); |
245 | | } |
246 | | #endif |
247 | | |
248 | | /* convert 32-bit unsigned to little endian 64 bit type as byte array */ |
249 | | static WC_INLINE void u32tole64(const word32 inLe32, byte outLe64[8]) |
250 | 7.27k | { |
251 | | #ifndef WOLFSSL_X86_64_BUILD |
252 | | outLe64[0] = (byte)(inLe32 & 0x000000FF); |
253 | | outLe64[1] = (byte)((inLe32 & 0x0000FF00) >> 8); |
254 | | outLe64[2] = (byte)((inLe32 & 0x00FF0000) >> 16); |
255 | | outLe64[3] = (byte)((inLe32 & 0xFF000000) >> 24); |
256 | | outLe64[4] = 0; |
257 | | outLe64[5] = 0; |
258 | | outLe64[6] = 0; |
259 | | outLe64[7] = 0; |
260 | | #else |
261 | 7.27k | *(word64*)outLe64 = inLe32; |
262 | 7.27k | #endif |
263 | 7.27k | } |
264 | | |
265 | | |
266 | | #if !defined(WOLFSSL_ARMASM) && !defined(WOLFSSL_RISCV_ASM) |
267 | | /* |
268 | | This local function operates on a message with a given number of bytes |
269 | | with a given ctx pointer to a Poly1305 structure. |
270 | | */ |
271 | | static int poly1305_blocks(Poly1305* ctx, const unsigned char *m, |
272 | | size_t bytes) |
273 | 11.5k | { |
274 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
275 | | /* AVX2 is handled in wc_Poly1305Update. */ |
276 | | SAVE_VECTOR_REGISTERS(return _svr_ret;); |
277 | | poly1305_blocks_avx(ctx, m, bytes); |
278 | | RESTORE_VECTOR_REGISTERS(); |
279 | | return 0; |
280 | | #elif defined(POLY130564) |
281 | 11.5k | const word64 hibit = (ctx->finished) ? 0 : ((word64)1 << 40); /* 1 << 128 */ |
282 | 11.5k | word64 r0,r1,r2; |
283 | 11.5k | word64 s1,s2; |
284 | 11.5k | word64 h0,h1,h2; |
285 | 11.5k | word64 c; |
286 | 11.5k | word128 d0,d1,d2,d; |
287 | | |
288 | 11.5k | r0 = ctx->r[0]; |
289 | 11.5k | r1 = ctx->r[1]; |
290 | 11.5k | r2 = ctx->r[2]; |
291 | | |
292 | 11.5k | h0 = ctx->h[0]; |
293 | 11.5k | h1 = ctx->h[1]; |
294 | 11.5k | h2 = ctx->h[2]; |
295 | | |
296 | 11.5k | s1 = r1 * (5 << 2); |
297 | 11.5k | s2 = r2 * (5 << 2); |
298 | | |
299 | 68.1k | while (bytes >= POLY1305_BLOCK_SIZE) { |
300 | 56.5k | word64 t0,t1; |
301 | | |
302 | | /* h += m[i] */ |
303 | 56.5k | t0 = U8TO64(&m[0]); |
304 | 56.5k | t1 = U8TO64(&m[8]); |
305 | | |
306 | 56.5k | h0 += (( t0 ) & 0xfffffffffff); |
307 | 56.5k | h1 += (((t0 >> 44) | (t1 << 20)) & 0xfffffffffff); |
308 | 56.5k | h2 += (((t1 >> 24) ) & 0x3ffffffffff) | hibit; |
309 | | |
310 | | /* h *= r */ |
311 | 56.5k | MUL(d0, h0, r0); MUL(d, h1, s2); ADD(d0, d); MUL(d, h2, s1); ADD(d0, d); |
312 | 56.5k | MUL(d1, h0, r1); MUL(d, h1, r0); ADD(d1, d); MUL(d, h2, s2); ADD(d1, d); |
313 | 56.5k | MUL(d2, h0, r2); MUL(d, h1, r1); ADD(d2, d); MUL(d, h2, r0); ADD(d2, d); |
314 | | |
315 | | /* (partial) h %= p */ |
316 | 56.5k | c = SHR(d0, 44); h0 = LO(d0) & 0xfffffffffff; |
317 | 56.5k | ADDLO(d1, c); c = SHR(d1, 44); h1 = LO(d1) & 0xfffffffffff; |
318 | 56.5k | ADDLO(d2, c); c = SHR(d2, 42); h2 = LO(d2) & 0x3ffffffffff; |
319 | 56.5k | h0 += c * 5; c = (h0 >> 44); h0 = h0 & 0xfffffffffff; |
320 | 56.5k | h1 += c; |
321 | | |
322 | 56.5k | m += POLY1305_BLOCK_SIZE; |
323 | 56.5k | bytes -= POLY1305_BLOCK_SIZE; |
324 | 56.5k | } |
325 | | |
326 | 11.5k | ctx->h[0] = h0; |
327 | 11.5k | ctx->h[1] = h1; |
328 | 11.5k | ctx->h[2] = h2; |
329 | | |
330 | 11.5k | return 0; |
331 | | |
332 | | #else /* if not 64 bit then use 32 bit */ |
333 | | const word32 hibit = (ctx->finished) ? 0 : ((word32)1 << 24); /* 1 << 128 */ |
334 | | word32 r0,r1,r2,r3,r4; |
335 | | word32 s1,s2,s3,s4; |
336 | | word32 h0,h1,h2,h3,h4; |
337 | | word32 c; |
338 | | #ifdef WOLFSSL_W64_WRAPPER |
339 | | #ifdef WOLFSSL_SMALL_STACK |
340 | | w64wrapper* d; |
341 | | |
342 | | d = (w64wrapper*)XMALLOC(5 * sizeof(w64wrapper), NULL, |
343 | | DYNAMIC_TYPE_TMP_BUFFER); |
344 | | if (d == NULL) { |
345 | | return MEMORY_E; |
346 | | } |
347 | | #else |
348 | | w64wrapper d[5]; |
349 | | #endif |
350 | | #else |
351 | | word64 d0,d1,d2,d3,d4; |
352 | | #endif |
353 | | |
354 | | |
355 | | r0 = ctx->r[0]; |
356 | | r1 = ctx->r[1]; |
357 | | r2 = ctx->r[2]; |
358 | | r3 = ctx->r[3]; |
359 | | r4 = ctx->r[4]; |
360 | | |
361 | | s1 = r1 * 5; |
362 | | s2 = r2 * 5; |
363 | | s3 = r3 * 5; |
364 | | s4 = r4 * 5; |
365 | | |
366 | | h0 = ctx->h[0]; |
367 | | h1 = ctx->h[1]; |
368 | | h2 = ctx->h[2]; |
369 | | h3 = ctx->h[3]; |
370 | | h4 = ctx->h[4]; |
371 | | |
372 | | while (bytes >= POLY1305_BLOCK_SIZE) { |
373 | | /* h += m[i] */ |
374 | | h0 += (U8TO32(m+ 0) ) & 0x3ffffff; |
375 | | h1 += (U8TO32(m+ 3) >> 2) & 0x3ffffff; |
376 | | h2 += (U8TO32(m+ 6) >> 4) & 0x3ffffff; |
377 | | h3 += (U8TO32(m+ 9) >> 6) & 0x3ffffff; |
378 | | h4 += (U8TO32(m+12) >> 8) | hibit; |
379 | | |
380 | | /* h *= r */ |
381 | | #ifdef WOLFSSL_W64_WRAPPER |
382 | | { |
383 | | w64wrapper tmp; |
384 | | |
385 | | d[0] = w64Mul(h0, r0); tmp = w64Mul(h1, s4); |
386 | | d[0] = w64Add(d[0], tmp, NULL); tmp = w64Mul(h2, s3); |
387 | | d[0] = w64Add(d[0], tmp, NULL); tmp = w64Mul(h3, s2); |
388 | | d[0] = w64Add(d[0], tmp, NULL); tmp = w64Mul(h4, s1); |
389 | | d[0] = w64Add(d[0], tmp, NULL); |
390 | | |
391 | | d[1] = w64Mul(h0, r1); tmp = w64Mul(h1, r0); |
392 | | d[1] = w64Add(d[1], tmp, NULL); tmp = w64Mul(h2, s4); |
393 | | d[1] = w64Add(d[1], tmp, NULL); tmp = w64Mul(h3, s3); |
394 | | d[1] = w64Add(d[1], tmp, NULL); tmp = w64Mul(h4, s2); |
395 | | d[1] = w64Add(d[1], tmp, NULL); |
396 | | |
397 | | d[2] = w64Mul(h0, r2); tmp = w64Mul(h1, r1); |
398 | | d[2] = w64Add(d[2], tmp, NULL); tmp = w64Mul(h2, r0); |
399 | | d[2] = w64Add(d[2], tmp, NULL); tmp = w64Mul(h3, s4); |
400 | | d[2] = w64Add(d[2], tmp, NULL); tmp = w64Mul(h4, s3); |
401 | | d[2] = w64Add(d[2], tmp, NULL); |
402 | | |
403 | | d[3] = w64Mul(h0, r3); tmp = w64Mul(h1, r2); |
404 | | d[3] = w64Add(d[3], tmp, NULL); tmp = w64Mul(h2, r1); |
405 | | d[3] = w64Add(d[3], tmp, NULL); tmp = w64Mul(h3, r0); |
406 | | d[3] = w64Add(d[3], tmp, NULL); tmp = w64Mul(h4, s4); |
407 | | d[3] = w64Add(d[3], tmp, NULL); |
408 | | |
409 | | d[4] = w64Mul(h0, r4); tmp = w64Mul(h1, r3); |
410 | | d[4] = w64Add(d[4], tmp, NULL); tmp = w64Mul(h2, r2); |
411 | | d[4] = w64Add(d[4], tmp, NULL); tmp = w64Mul(h3, r1); |
412 | | d[4] = w64Add(d[4], tmp, NULL); tmp = w64Mul(h4, r0); |
413 | | d[4] = w64Add(d[4], tmp, NULL); |
414 | | } |
415 | | #else |
416 | | d0 = ((word64)h0 * r0) + ((word64)h1 * s4) + ((word64)h2 * s3) + |
417 | | ((word64)h3 * s2) + ((word64)h4 * s1); |
418 | | d1 = ((word64)h0 * r1) + ((word64)h1 * r0) + ((word64)h2 * s4) + |
419 | | ((word64)h3 * s3) + ((word64)h4 * s2); |
420 | | d2 = ((word64)h0 * r2) + ((word64)h1 * r1) + ((word64)h2 * r0) + |
421 | | ((word64)h3 * s4) + ((word64)h4 * s3); |
422 | | d3 = ((word64)h0 * r3) + ((word64)h1 * r2) + ((word64)h2 * r1) + |
423 | | ((word64)h3 * r0) + ((word64)h4 * s4); |
424 | | d4 = ((word64)h0 * r4) + ((word64)h1 * r3) + ((word64)h2 * r2) + |
425 | | ((word64)h3 * r1) + ((word64)h4 * r0); |
426 | | #endif |
427 | | |
428 | | /* (partial) h %= p */ |
429 | | #ifdef WOLFSSL_W64_WRAPPER |
430 | | c = w64GetLow32(w64ShiftRight(d[0], 26)); |
431 | | h0 = w64GetLow32(d[0]) & 0x3ffffff; |
432 | | d[1] = w64Add32(d[1], c, NULL); |
433 | | c = w64GetLow32(w64ShiftRight(d[1], 26)); |
434 | | h1 = w64GetLow32(d[1]) & 0x3ffffff; |
435 | | d[2] = w64Add32(d[2], c, NULL); |
436 | | c = w64GetLow32(w64ShiftRight(d[2], 26)); |
437 | | h2 = w64GetLow32(d[2]) & 0x3ffffff; |
438 | | d[3] = w64Add32(d[3], c, NULL); |
439 | | c = w64GetLow32(w64ShiftRight(d[3], 26)); |
440 | | h3 = w64GetLow32(d[3]) & 0x3ffffff; |
441 | | d[4] = w64Add32(d[4], c, NULL); |
442 | | c = w64GetLow32(w64ShiftRight(d[4], 26)); |
443 | | h4 = w64GetLow32(d[4]) & 0x3ffffff; |
444 | | #else |
445 | | c = (word32)(d0 >> 26); h0 = (word32)d0 & 0x3ffffff; |
446 | | d1 += c; c = (word32)(d1 >> 26); h1 = (word32)d1 & 0x3ffffff; |
447 | | d2 += c; c = (word32)(d2 >> 26); h2 = (word32)d2 & 0x3ffffff; |
448 | | d3 += c; c = (word32)(d3 >> 26); h3 = (word32)d3 & 0x3ffffff; |
449 | | d4 += c; c = (word32)(d4 >> 26); h4 = (word32)d4 & 0x3ffffff; |
450 | | #endif |
451 | | h0 += c * 5; c = (h0 >> 26); h0 = h0 & 0x3ffffff; |
452 | | h1 += c; |
453 | | |
454 | | m += POLY1305_BLOCK_SIZE; |
455 | | bytes -= POLY1305_BLOCK_SIZE; |
456 | | } |
457 | | |
458 | | ctx->h[0] = h0; |
459 | | ctx->h[1] = h1; |
460 | | ctx->h[2] = h2; |
461 | | ctx->h[3] = h3; |
462 | | ctx->h[4] = h4; |
463 | | |
464 | | #if defined(WOLFSSL_W64_WRAPPER) && defined(WOLFSSL_SMALL_STACK) |
465 | | XFREE(d, NULL, DYNAMIC_TYPE_TMP_BUFFER); |
466 | | #endif |
467 | | |
468 | | return 0; |
469 | | |
470 | | #endif /* end of 64 bit cpu blocks or 32 bit cpu */ |
471 | 11.5k | } |
472 | | |
473 | | /* |
474 | | This local function is used for the last call when a message with a given |
475 | | number of bytes is less than the block size. |
476 | | */ |
477 | | static int poly1305_block(Poly1305* ctx, const unsigned char *m) |
478 | 7.08k | { |
479 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
480 | | /* No call to poly1305_block when AVX2, AVX2 does 4 blocks at a time. */ |
481 | | SAVE_VECTOR_REGISTERS(return _svr_ret;); |
482 | | poly1305_block_avx(ctx, m); |
483 | | RESTORE_VECTOR_REGISTERS(); |
484 | | return 0; |
485 | | #else |
486 | 7.08k | return poly1305_blocks(ctx, m, POLY1305_BLOCK_SIZE); |
487 | 7.08k | #endif |
488 | 7.08k | } |
489 | | |
490 | | int wc_Poly1305SetKey(Poly1305* ctx, const byte* key, word32 keySz) |
491 | 3.65k | { |
492 | 3.65k | #if defined(POLY130564) && !defined(USE_INTEL_POLY1305_SPEEDUP) |
493 | 3.65k | word64 t0,t1; |
494 | 3.65k | #endif |
495 | | |
496 | 3.65k | if (key == NULL) |
497 | 0 | return BAD_FUNC_ARG; |
498 | | |
499 | | #ifdef CHACHA_AEAD_TEST |
500 | | word32 k; |
501 | | printf("Poly key used:\n"); |
502 | | for (k = 0; k < keySz; k++) { |
503 | | printf("%02x", key[k]); |
504 | | if ((k+1) % 8 == 0) |
505 | | printf("\n"); |
506 | | } |
507 | | printf("\n"); |
508 | | #endif |
509 | | |
510 | 3.65k | if (keySz != 32 || ctx == NULL) |
511 | 0 | return BAD_FUNC_ARG; |
512 | | |
513 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
514 | | if (!cpu_flags_set) { |
515 | | intel_flags = cpuid_get_flags(); |
516 | | cpu_flags_set = 1; |
517 | | } |
518 | | SAVE_VECTOR_REGISTERS(return _svr_ret;); |
519 | | #ifdef HAVE_INTEL_AVX2 |
520 | | if (IS_INTEL_AVX2(intel_flags)) |
521 | | poly1305_setkey_avx2(ctx, key); |
522 | | else |
523 | | #endif |
524 | | poly1305_setkey_avx(ctx, key); |
525 | | RESTORE_VECTOR_REGISTERS(); |
526 | | ctx->started = 0; |
527 | | #elif defined(POLY130564) |
528 | | |
529 | | /* r &= 0xffffffc0ffffffc0ffffffc0fffffff */ |
530 | 3.65k | t0 = U8TO64(key + 0); |
531 | 3.65k | t1 = U8TO64(key + 8); |
532 | | |
533 | 3.65k | ctx->r[0] = ( t0 ) & 0xffc0fffffff; |
534 | 3.65k | ctx->r[1] = ((t0 >> 44) | (t1 << 20)) & 0xfffffc0ffff; |
535 | 3.65k | ctx->r[2] = ((t1 >> 24) ) & 0x00ffffffc0f; |
536 | | |
537 | | /* h (accumulator) = 0 */ |
538 | 3.65k | ctx->h[0] = 0; |
539 | 3.65k | ctx->h[1] = 0; |
540 | 3.65k | ctx->h[2] = 0; |
541 | | |
542 | | /* save pad for later */ |
543 | 3.65k | ctx->pad[0] = U8TO64(key + 16); |
544 | 3.65k | ctx->pad[1] = U8TO64(key + 24); |
545 | | |
546 | 3.65k | ctx->leftover = 0; |
547 | 3.65k | ctx->finished = 0; |
548 | | |
549 | | #else /* if not 64 bit then use 32 bit */ |
550 | | |
551 | | /* r &= 0xffffffc0ffffffc0ffffffc0fffffff */ |
552 | | ctx->r[0] = (U8TO32(key + 0) ) & 0x3ffffff; |
553 | | ctx->r[1] = (U8TO32(key + 3) >> 2) & 0x3ffff03; |
554 | | ctx->r[2] = (U8TO32(key + 6) >> 4) & 0x3ffc0ff; |
555 | | ctx->r[3] = (U8TO32(key + 9) >> 6) & 0x3f03fff; |
556 | | ctx->r[4] = (U8TO32(key + 12) >> 8) & 0x00fffff; |
557 | | |
558 | | /* h = 0 */ |
559 | | ctx->h[0] = 0; |
560 | | ctx->h[1] = 0; |
561 | | ctx->h[2] = 0; |
562 | | ctx->h[3] = 0; |
563 | | ctx->h[4] = 0; |
564 | | |
565 | | /* save pad for later */ |
566 | | ctx->pad[0] = U8TO32(key + 16); |
567 | | ctx->pad[1] = U8TO32(key + 20); |
568 | | ctx->pad[2] = U8TO32(key + 24); |
569 | | ctx->pad[3] = U8TO32(key + 28); |
570 | | |
571 | | ctx->leftover = 0; |
572 | | ctx->finished = 0; |
573 | | |
574 | | #endif |
575 | | |
576 | 3.65k | return 0; |
577 | 3.65k | } |
578 | | |
579 | | int wc_Poly1305Final(Poly1305* ctx, byte* mac) |
580 | 3.65k | { |
581 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
582 | | #elif defined(POLY130564) |
583 | | |
584 | 3.65k | word64 h0,h1,h2,c; |
585 | 3.65k | word64 g0,g1,g2; |
586 | 3.65k | word64 t0,t1; |
587 | | |
588 | | #else |
589 | | |
590 | | word32 h0,h1,h2,h3,h4,c; |
591 | | word32 g0,g1,g2,g3,g4; |
592 | | #ifdef WOLFSSL_W64_WRAPPER |
593 | | w64wrapper f; |
594 | | #else |
595 | | word64 f; |
596 | | #endif |
597 | | word32 mask; |
598 | | |
599 | | #endif |
600 | | |
601 | 3.65k | if (ctx == NULL || mac == NULL) |
602 | 0 | return BAD_FUNC_ARG; |
603 | | |
604 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
605 | | SAVE_VECTOR_REGISTERS(return _svr_ret;); |
606 | | #ifdef HAVE_INTEL_AVX2 |
607 | | if (IS_INTEL_AVX2(intel_flags)) |
608 | | poly1305_final_avx2(ctx, mac); |
609 | | else |
610 | | #endif |
611 | | poly1305_final_avx(ctx, mac); |
612 | | RESTORE_VECTOR_REGISTERS(); |
613 | | #elif defined(POLY130564) |
614 | | |
615 | | /* process the remaining block */ |
616 | 3.65k | if (ctx->leftover) { |
617 | 11 | size_t i = ctx->leftover; |
618 | 11 | ctx->buffer[i] = 1; |
619 | 67 | for (i = i + 1; i < POLY1305_BLOCK_SIZE; i++) |
620 | 56 | ctx->buffer[i] = 0; |
621 | 11 | ctx->finished = 1; |
622 | 11 | poly1305_block(ctx, ctx->buffer); |
623 | 11 | } |
624 | | |
625 | | /* fully carry h */ |
626 | 3.65k | h0 = ctx->h[0]; |
627 | 3.65k | h1 = ctx->h[1]; |
628 | 3.65k | h2 = ctx->h[2]; |
629 | | |
630 | 3.65k | c = (h1 >> 44); h1 &= 0xfffffffffff; |
631 | 3.65k | h2 += c; c = (h2 >> 42); h2 &= 0x3ffffffffff; |
632 | 3.65k | h0 += c * 5; c = (h0 >> 44); h0 &= 0xfffffffffff; |
633 | 3.65k | h1 += c; c = (h1 >> 44); h1 &= 0xfffffffffff; |
634 | 3.65k | h2 += c; c = (h2 >> 42); h2 &= 0x3ffffffffff; |
635 | 3.65k | h0 += c * 5; c = (h0 >> 44); h0 &= 0xfffffffffff; |
636 | 3.65k | h1 += c; |
637 | | |
638 | | /* compute h + -p */ |
639 | 3.65k | g0 = h0 + 5; c = (g0 >> 44); g0 &= 0xfffffffffff; |
640 | 3.65k | g1 = h1 + c; c = (g1 >> 44); g1 &= 0xfffffffffff; |
641 | 3.65k | g2 = h2 + c - ((word64)1 << 42); |
642 | | |
643 | | /* select h if h < p, or h + -p if h >= p */ |
644 | 3.65k | c = (g2 >> ((sizeof(word64) * 8) - 1)) - 1; |
645 | 3.65k | g0 &= c; |
646 | 3.65k | g1 &= c; |
647 | 3.65k | g2 &= c; |
648 | 3.65k | c = ~c; |
649 | 3.65k | h0 = (h0 & c) | g0; |
650 | 3.65k | h1 = (h1 & c) | g1; |
651 | 3.65k | h2 = (h2 & c) | g2; |
652 | | |
653 | | /* h = (h + pad) */ |
654 | 3.65k | t0 = ctx->pad[0]; |
655 | 3.65k | t1 = ctx->pad[1]; |
656 | | |
657 | 3.65k | h0 += (( t0 ) & 0xfffffffffff) ; |
658 | 3.65k | c = (h0 >> 44); h0 &= 0xfffffffffff; |
659 | 3.65k | h1 += (((t0 >> 44) | (t1 << 20)) & 0xfffffffffff) + c; |
660 | 3.65k | c = (h1 >> 44); h1 &= 0xfffffffffff; |
661 | 3.65k | h2 += (((t1 >> 24) ) & 0x3ffffffffff) + c; |
662 | 3.65k | h2 &= 0x3ffffffffff; |
663 | | |
664 | | /* mac = h % (2^128) */ |
665 | 3.65k | h0 = ((h0 ) | (h1 << 44)); |
666 | 3.65k | h1 = ((h1 >> 20) | (h2 << 24)); |
667 | | |
668 | 3.65k | U64TO8(mac + 0, h0); |
669 | 3.65k | U64TO8(mac + 8, h1); |
670 | | |
671 | | /* zero out the state */ |
672 | 3.65k | ctx->h[0] = 0; |
673 | 3.65k | ctx->h[1] = 0; |
674 | 3.65k | ctx->h[2] = 0; |
675 | 3.65k | ctx->r[0] = 0; |
676 | 3.65k | ctx->r[1] = 0; |
677 | 3.65k | ctx->r[2] = 0; |
678 | 3.65k | ctx->pad[0] = 0; |
679 | 3.65k | ctx->pad[1] = 0; |
680 | | |
681 | | #else /* if not 64 bit then use 32 bit */ |
682 | | |
683 | | /* process the remaining block */ |
684 | | if (ctx->leftover) { |
685 | | size_t i = ctx->leftover; |
686 | | ctx->buffer[i++] = 1; |
687 | | for (; i < POLY1305_BLOCK_SIZE; i++) |
688 | | ctx->buffer[i] = 0; |
689 | | ctx->finished = 1; |
690 | | poly1305_block(ctx, ctx->buffer); |
691 | | } |
692 | | |
693 | | /* fully carry h */ |
694 | | h0 = ctx->h[0]; |
695 | | h1 = ctx->h[1]; |
696 | | h2 = ctx->h[2]; |
697 | | h3 = ctx->h[3]; |
698 | | h4 = ctx->h[4]; |
699 | | |
700 | | c = h1 >> 26; h1 = h1 & 0x3ffffff; |
701 | | h2 += c; c = h2 >> 26; h2 = h2 & 0x3ffffff; |
702 | | h3 += c; c = h3 >> 26; h3 = h3 & 0x3ffffff; |
703 | | h4 += c; c = h4 >> 26; h4 = h4 & 0x3ffffff; |
704 | | h0 += c * 5; c = h0 >> 26; h0 = h0 & 0x3ffffff; |
705 | | h1 += c; |
706 | | |
707 | | /* compute h + -p */ |
708 | | g0 = h0 + 5; c = g0 >> 26; g0 &= 0x3ffffff; |
709 | | g1 = h1 + c; c = g1 >> 26; g1 &= 0x3ffffff; |
710 | | g2 = h2 + c; c = g2 >> 26; g2 &= 0x3ffffff; |
711 | | g3 = h3 + c; c = g3 >> 26; g3 &= 0x3ffffff; |
712 | | g4 = h4 + c - ((word32)1 << 26); |
713 | | |
714 | | /* select h if h < p, or h + -p if h >= p */ |
715 | | mask = ((word32)g4 >> ((sizeof(word32) * 8) - 1)) - 1; |
716 | | g0 &= mask; |
717 | | g1 &= mask; |
718 | | g2 &= mask; |
719 | | g3 &= mask; |
720 | | g4 &= mask; |
721 | | mask = ~mask; |
722 | | h0 = (h0 & mask) | g0; |
723 | | h1 = (h1 & mask) | g1; |
724 | | h2 = (h2 & mask) | g2; |
725 | | h3 = (h3 & mask) | g3; |
726 | | h4 = (h4 & mask) | g4; |
727 | | |
728 | | /* h = h % (2^128) */ |
729 | | h0 = ((h0 ) | (h1 << 26)) & 0xffffffff; |
730 | | h1 = ((h1 >> 6) | (h2 << 20)) & 0xffffffff; |
731 | | h2 = ((h2 >> 12) | (h3 << 14)) & 0xffffffff; |
732 | | h3 = ((h3 >> 18) | (h4 << 8)) & 0xffffffff; |
733 | | |
734 | | /* mac = (h + pad) % (2^128) */ |
735 | | #ifdef WOLFSSL_W64_WRAPPER |
736 | | f = w64From32(0, h0); |
737 | | f = w64Add32(f, ctx->pad[0], NULL); |
738 | | h0 = w64GetLow32(f); |
739 | | |
740 | | f = w64ShiftRight(f, 32); |
741 | | f = w64Add32(f, h1, NULL); |
742 | | f = w64Add32(f, ctx->pad[1], NULL); |
743 | | h1 = w64GetLow32(f); |
744 | | |
745 | | f = w64ShiftRight(f, 32); |
746 | | f = w64Add32(f, h2, NULL); |
747 | | f = w64Add32(f, ctx->pad[2], NULL); |
748 | | h2 = w64GetLow32(f); |
749 | | |
750 | | f = w64ShiftRight(f, 32); |
751 | | f = w64Add32(f, h3, NULL); |
752 | | f = w64Add32(f, ctx->pad[3], NULL); |
753 | | h3 = w64GetLow32(f); |
754 | | #else |
755 | | f = (word64)h0 + ctx->pad[0] ; h0 = (word32)f; |
756 | | f = (word64)h1 + ctx->pad[1] + (f >> 32); h1 = (word32)f; |
757 | | f = (word64)h2 + ctx->pad[2] + (f >> 32); h2 = (word32)f; |
758 | | f = (word64)h3 + ctx->pad[3] + (f >> 32); h3 = (word32)f; |
759 | | #endif |
760 | | |
761 | | U32TO8(mac + 0, h0); |
762 | | U32TO8(mac + 4, h1); |
763 | | U32TO8(mac + 8, h2); |
764 | | U32TO8(mac + 12, h3); |
765 | | |
766 | | /* zero out the state */ |
767 | | ctx->h[0] = 0; |
768 | | ctx->h[1] = 0; |
769 | | ctx->h[2] = 0; |
770 | | ctx->h[3] = 0; |
771 | | ctx->h[4] = 0; |
772 | | ctx->r[0] = 0; |
773 | | ctx->r[1] = 0; |
774 | | ctx->r[2] = 0; |
775 | | ctx->r[3] = 0; |
776 | | ctx->r[4] = 0; |
777 | | ctx->pad[0] = 0; |
778 | | ctx->pad[1] = 0; |
779 | | ctx->pad[2] = 0; |
780 | | ctx->pad[3] = 0; |
781 | | |
782 | | #endif |
783 | | |
784 | 3.65k | return 0; |
785 | 3.65k | } |
786 | | #endif /* !WOLFSSL_ARMASM && !WOLFSSL_RISCV_ASM */ |
787 | | |
788 | | |
789 | | int wc_Poly1305Update(Poly1305* ctx, const byte* m, word32 bytes) |
790 | 18.0k | { |
791 | 18.0k | size_t i; |
792 | | |
793 | 18.0k | if (ctx == NULL || (m == NULL && bytes > 0)) |
794 | 0 | return BAD_FUNC_ARG; |
795 | | |
796 | 18.0k | if (bytes == 0) { |
797 | | /* valid, but do nothing */ |
798 | 7 | return 0; |
799 | 7 | } |
800 | | #ifdef CHACHA_AEAD_TEST |
801 | | word32 k; |
802 | | printf("Raw input to poly:\n"); |
803 | | for (k = 0; k < bytes; k++) { |
804 | | printf("%02x", m[k]); |
805 | | if ((k+1) % 16 == 0) |
806 | | printf("\n"); |
807 | | } |
808 | | printf("\n"); |
809 | | #endif |
810 | | |
811 | | #if defined(WOLFSSL_ARMASM) && !defined(WOLFSSL_ARMASM_THUMB2) && \ |
812 | | !defined(WOLFSSL_ARMASM_NO_NEON) |
813 | | /* handle leftover */ |
814 | | if (ctx->leftover) { |
815 | | size_t want = sizeof(ctx->buffer) - ctx->leftover; |
816 | | if (want > bytes) |
817 | | want = bytes; |
818 | | |
819 | | for (i = 0; i < want; i++) |
820 | | ctx->buffer[ctx->leftover + i] = m[i]; |
821 | | bytes -= (word32)want; |
822 | | m += want; |
823 | | ctx->leftover += want; |
824 | | if (ctx->leftover < sizeof(ctx->buffer)) { |
825 | | return 0; |
826 | | } |
827 | | |
828 | | poly1305_blocks(ctx, ctx->buffer, sizeof(ctx->buffer)); |
829 | | ctx->leftover = 0; |
830 | | } |
831 | | |
832 | | /* process full blocks */ |
833 | | if (bytes >= sizeof(ctx->buffer)) { |
834 | | size_t want = bytes & ~((size_t)POLY1305_BLOCK_SIZE - 1); |
835 | | |
836 | | poly1305_blocks(ctx, m, want); |
837 | | m += want; |
838 | | bytes -= (word32)want; |
839 | | } |
840 | | |
841 | | /* store leftover */ |
842 | | if (bytes) { |
843 | | for (i = 0; i < bytes; i++) |
844 | | ctx->buffer[ctx->leftover + i] = m[i]; |
845 | | ctx->leftover += bytes; |
846 | | } |
847 | | #else |
848 | | #ifdef USE_INTEL_POLY1305_SPEEDUP |
849 | | #ifdef HAVE_INTEL_AVX2 |
850 | | if (IS_INTEL_AVX2(intel_flags)) { |
851 | | SAVE_VECTOR_REGISTERS(return _svr_ret;); |
852 | | |
853 | | /* handle leftover */ |
854 | | if (ctx->leftover) { |
855 | | size_t want = sizeof(ctx->buffer) - ctx->leftover; |
856 | | if (want > bytes) |
857 | | want = bytes; |
858 | | |
859 | | for (i = 0; i < want; i++) |
860 | | ctx->buffer[ctx->leftover + i] = m[i]; |
861 | | bytes -= (word32)want; |
862 | | m += want; |
863 | | ctx->leftover += want; |
864 | | if (ctx->leftover < sizeof(ctx->buffer)) { |
865 | | RESTORE_VECTOR_REGISTERS(); |
866 | | return 0; |
867 | | } |
868 | | |
869 | | if (!ctx->started) { |
870 | | poly1305_calc_powers_avx2(ctx); |
871 | | ctx->started = 1; |
872 | | } |
873 | | poly1305_blocks_avx2(ctx, ctx->buffer, sizeof(ctx->buffer)); |
874 | | ctx->leftover = 0; |
875 | | } |
876 | | |
877 | | /* process full blocks */ |
878 | | if (bytes >= sizeof(ctx->buffer)) { |
879 | | size_t want = bytes & ~(sizeof(ctx->buffer) - 1); |
880 | | |
881 | | if (!ctx->started) { |
882 | | poly1305_calc_powers_avx2(ctx); |
883 | | ctx->started = 1; |
884 | | } |
885 | | poly1305_blocks_avx2(ctx, m, want); |
886 | | m += want; |
887 | | bytes -= (word32)want; |
888 | | } |
889 | | |
890 | | /* store leftover */ |
891 | | if (bytes) { |
892 | | for (i = 0; i < bytes; i++) |
893 | | ctx->buffer[ctx->leftover + i] = m[i]; |
894 | | ctx->leftover += bytes; |
895 | | } |
896 | | RESTORE_VECTOR_REGISTERS(); |
897 | | } |
898 | | else |
899 | | #endif |
900 | | #endif |
901 | 18.0k | { |
902 | | /* handle leftover */ |
903 | 18.0k | if (ctx->leftover) { |
904 | 7.08k | size_t want = (POLY1305_BLOCK_SIZE - ctx->leftover); |
905 | 7.08k | if (want > bytes) |
906 | 12 | want = bytes; |
907 | 54.0k | for (i = 0; i < want; i++) |
908 | 46.9k | ctx->buffer[ctx->leftover + i] = m[i]; |
909 | 7.08k | bytes -= (word32)want; |
910 | 7.08k | m += want; |
911 | 7.08k | ctx->leftover += want; |
912 | 7.08k | if (ctx->leftover < POLY1305_BLOCK_SIZE) |
913 | 12 | return 0; |
914 | 7.07k | poly1305_block(ctx, ctx->buffer); |
915 | 7.07k | ctx->leftover = 0; |
916 | 7.07k | } |
917 | | |
918 | | /* process full blocks */ |
919 | 17.9k | if (bytes >= POLY1305_BLOCK_SIZE) { |
920 | 4.48k | size_t want = ((size_t)bytes & ~((size_t)POLY1305_BLOCK_SIZE - 1)); |
921 | 4.48k | #if !defined(WOLFSSL_ARMASM) && !defined(WOLFSSL_RISCV_ASM) |
922 | 4.48k | int ret; |
923 | 4.48k | ret = poly1305_blocks(ctx, m, want); |
924 | 4.48k | if (ret != 0) |
925 | 0 | return ret; |
926 | | #else |
927 | | poly1305_blocks(ctx, m, want); |
928 | | #endif |
929 | 4.48k | m += want; |
930 | 4.48k | bytes -= (word32)want; |
931 | 4.48k | } |
932 | | |
933 | | /* store leftover */ |
934 | 17.9k | if (bytes) { |
935 | 73.3k | for (i = 0; i < bytes; i++) |
936 | 66.2k | ctx->buffer[ctx->leftover + i] = m[i]; |
937 | 7.08k | ctx->leftover += bytes; |
938 | 7.08k | } |
939 | 17.9k | } |
940 | 0 | #endif |
941 | | |
942 | 0 | return 0; |
943 | 17.9k | } |
944 | | |
945 | | /* Takes a Poly1305 struct that has a key loaded and pads the provided length |
946 | | ctx : Initialized Poly1305 struct to use |
947 | | lenToPad : Current number of bytes updated that needs padding to 16 |
948 | | */ |
949 | | int wc_Poly1305_Pad(Poly1305* ctx, word32 lenToPad) |
950 | 7.27k | { |
951 | 7.27k | int ret = 0; |
952 | 7.27k | word32 paddingLen; |
953 | 7.27k | byte padding[WC_POLY1305_PAD_SZ - 1]; |
954 | | |
955 | 7.27k | if (ctx == NULL) { |
956 | 0 | return BAD_FUNC_ARG; |
957 | 0 | } |
958 | 7.27k | if (lenToPad == 0) { |
959 | 7 | return 0; /* nothing needs to be done */ |
960 | 7 | } |
961 | | |
962 | 7.26k | XMEMSET(padding, 0, sizeof(padding)); |
963 | | |
964 | | /* Pad length to 16 bytes */ |
965 | 7.26k | paddingLen = (-(int)lenToPad) & (WC_POLY1305_PAD_SZ - 1); |
966 | 7.26k | if ((paddingLen > 0) && (paddingLen < WC_POLY1305_PAD_SZ)) { |
967 | 7.04k | ret = wc_Poly1305Update(ctx, padding, paddingLen); |
968 | 7.04k | } |
969 | 7.26k | return ret; |
970 | 7.27k | } |
971 | | |
972 | | /* Takes a Poly1305 struct that has a key loaded and adds the AEAD length |
973 | | encoding in 64-bit little endian |
974 | | aadSz : Size of the additional authentication data |
975 | | dataSz : Size of the plaintext or ciphertext |
976 | | */ |
977 | | int wc_Poly1305_EncodeSizes(Poly1305* ctx, word32 aadSz, word32 dataSz) |
978 | 3.63k | { |
979 | 3.63k | int ret; |
980 | 3.63k | byte little64[16]; /* sizeof(word64) * 2 */ |
981 | | |
982 | 3.63k | if (ctx == NULL) { |
983 | 0 | return BAD_FUNC_ARG; |
984 | 0 | } |
985 | | |
986 | 3.63k | XMEMSET(little64, 0, sizeof(little64)); |
987 | | |
988 | | /* size of additional data and input data as little endian 64 bit types */ |
989 | 3.63k | u32tole64(aadSz, little64); |
990 | 3.63k | u32tole64(dataSz, little64 + 8); |
991 | 3.63k | ret = wc_Poly1305Update(ctx, little64, sizeof(little64)); |
992 | | |
993 | 3.63k | return ret; |
994 | 3.63k | } |
995 | | |
996 | | #ifdef WORD64_AVAILABLE |
997 | | int wc_Poly1305_EncodeSizes64(Poly1305* ctx, word64 aadSz, word64 dataSz) |
998 | 0 | { |
999 | 0 | int ret; |
1000 | 0 | word64 little64[2]; |
1001 | |
|
1002 | 0 | if (ctx == NULL) { |
1003 | 0 | return BAD_FUNC_ARG; |
1004 | 0 | } |
1005 | | |
1006 | | #ifdef BIG_ENDIAN_ORDER |
1007 | | little64[0] = ByteReverseWord64(aadSz); |
1008 | | little64[1] = ByteReverseWord64(dataSz); |
1009 | | #else |
1010 | 0 | little64[0] = aadSz; |
1011 | 0 | little64[1] = dataSz; |
1012 | 0 | #endif |
1013 | |
|
1014 | 0 | ret = wc_Poly1305Update(ctx, (byte *)little64, sizeof(little64)); |
1015 | |
|
1016 | 0 | return ret; |
1017 | 0 | } |
1018 | | #endif |
1019 | | |
1020 | | /* Takes in an initialized Poly1305 struct that has a key loaded and creates |
1021 | | a MAC (tag) using recent TLS AEAD padding scheme. |
1022 | | ctx : Initialized Poly1305 struct to use |
1023 | | additional : Additional data to use |
1024 | | addSz : Size of additional buffer |
1025 | | input : Input buffer to create tag from |
1026 | | sz : Size of input buffer |
1027 | | tag : Buffer to hold created tag |
1028 | | tagSz : Size of input tag buffer (must be at least |
1029 | | WC_POLY1305_MAC_SZ(16)) |
1030 | | */ |
1031 | | int wc_Poly1305_MAC(Poly1305* ctx, const byte* additional, word32 addSz, |
1032 | | const byte* input, word32 sz, byte* tag, word32 tagSz) |
1033 | 3.63k | { |
1034 | 3.63k | int ret; |
1035 | | |
1036 | | /* sanity check on arguments */ |
1037 | 3.63k | if (ctx == NULL || input == NULL || tag == NULL || |
1038 | 3.63k | tagSz < WC_POLY1305_MAC_SZ) { |
1039 | 0 | return BAD_FUNC_ARG; |
1040 | 0 | } |
1041 | | |
1042 | | /* additional allowed to be 0 */ |
1043 | 3.63k | if (addSz > 0) { |
1044 | 3.63k | if (additional == NULL) |
1045 | 0 | return BAD_FUNC_ARG; |
1046 | | |
1047 | | /* additional data plus padding */ |
1048 | 3.63k | if ((ret = wc_Poly1305Update(ctx, additional, addSz)) != 0) { |
1049 | 0 | return ret; |
1050 | 0 | } |
1051 | | /* pad additional data */ |
1052 | 3.63k | if ((ret = wc_Poly1305_Pad(ctx, addSz)) != 0) { |
1053 | 0 | return ret; |
1054 | 0 | } |
1055 | 3.63k | } |
1056 | | |
1057 | | /* input plus padding */ |
1058 | 3.63k | if ((ret = wc_Poly1305Update(ctx, input, sz)) != 0) { |
1059 | 0 | return ret; |
1060 | 0 | } |
1061 | | /* pad input data */ |
1062 | 3.63k | if ((ret = wc_Poly1305_Pad(ctx, sz)) != 0) { |
1063 | 0 | return ret; |
1064 | 0 | } |
1065 | | |
1066 | | /* encode size of AAD and input data as little endian 64 bit types */ |
1067 | 3.63k | if ((ret = wc_Poly1305_EncodeSizes(ctx, addSz, sz)) != 0) { |
1068 | 0 | return ret; |
1069 | 0 | } |
1070 | | |
1071 | | /* Finalize the auth tag */ |
1072 | 3.63k | ret = wc_Poly1305Final(ctx, tag); |
1073 | | |
1074 | 3.63k | return ret; |
1075 | | |
1076 | 3.63k | } |
1077 | | #endif /* HAVE_POLY1305 */ |