]>
Commit | Line | Data |
---|---|---|
7c673cae FG |
1 | /********************************************************************** |
2 | Copyright(c) 2011-2016 Intel Corporation All rights reserved. | |
3 | ||
4 | Redistribution and use in source and binary forms, with or without | |
1e59de90 | 5 | modification, are permitted provided that the following conditions |
7c673cae FG |
6 | are met: |
7 | * Redistributions of source code must retain the above copyright | |
8 | notice, this list of conditions and the following disclaimer. | |
9 | * Redistributions in binary form must reproduce the above copyright | |
10 | notice, this list of conditions and the following disclaimer in | |
11 | the documentation and/or other materials provided with the | |
12 | distribution. | |
13 | * Neither the name of Intel Corporation nor the names of its | |
14 | contributors may be used to endorse or promote products derived | |
15 | from this software without specific prior written permission. | |
16 | ||
17 | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS | |
18 | "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT | |
19 | LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR | |
20 | A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT | |
21 | OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, | |
22 | SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT | |
23 | LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, | |
24 | DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY | |
25 | THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT | |
26 | (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE | |
27 | OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. | |
28 | **********************************************************************/ | |
29 | ||
1e59de90 TL |
30 | #if defined(__clang__) |
31 | # pragma clang attribute push (__attribute__((target("avx2"))), apply_to=function) | |
32 | #elif defined(__ICC) | |
33 | # pragma intel optimization_parameter target_arch=AVX2 | |
34 | #elif defined(__ICL) | |
35 | # pragma [intel] optimization_parameter target_arch=AVX2 | |
36 | #elif (__GNUC__ >= 5) | |
37 | # pragma GCC target("avx2") | |
38 | #endif | |
39 | ||
7c673cae FG |
40 | #include "sha1_mb.h" |
41 | #include "memcpy_inline.h" | |
1e59de90 | 42 | #include "endian_helper.h" |
7c673cae FG |
43 | |
44 | #ifdef _MSC_VER | |
45 | # include <intrin.h> | |
46 | # define inline __inline | |
47 | #endif | |
48 | ||
49 | static inline void hash_init_digest(SHA1_WORD_T * digest); | |
1e59de90 | 50 | static inline uint32_t hash_pad(uint8_t padblock[SHA1_BLOCK_SIZE * 2], uint64_t total_len); |
7c673cae FG |
51 | static SHA1_HASH_CTX *sha1_ctx_mgr_resubmit(SHA1_HASH_CTX_MGR * mgr, SHA1_HASH_CTX * ctx); |
52 | ||
53 | void sha1_ctx_mgr_init_avx2(SHA1_HASH_CTX_MGR * mgr) | |
54 | { | |
55 | sha1_mb_mgr_init_avx2(&mgr->mgr); | |
56 | } | |
57 | ||
58 | SHA1_HASH_CTX *sha1_ctx_mgr_submit_avx2(SHA1_HASH_CTX_MGR * mgr, SHA1_HASH_CTX * ctx, | |
59 | const void *buffer, uint32_t len, HASH_CTX_FLAG flags) | |
60 | { | |
61 | if (flags & (~HASH_ENTIRE)) { | |
62 | // User should not pass anything other than FIRST, UPDATE, or LAST | |
63 | ctx->error = HASH_CTX_ERROR_INVALID_FLAGS; | |
64 | return ctx; | |
65 | } | |
66 | ||
67 | if (ctx->status & HASH_CTX_STS_PROCESSING) { | |
68 | // Cannot submit to a currently processing job. | |
69 | ctx->error = HASH_CTX_ERROR_ALREADY_PROCESSING; | |
70 | return ctx; | |
71 | } | |
72 | ||
73 | if ((ctx->status & HASH_CTX_STS_COMPLETE) && !(flags & HASH_FIRST)) { | |
74 | // Cannot update a finished job. | |
75 | ctx->error = HASH_CTX_ERROR_ALREADY_COMPLETED; | |
76 | return ctx; | |
77 | } | |
78 | ||
79 | if (flags & HASH_FIRST) { | |
80 | // Init digest | |
81 | hash_init_digest(ctx->job.result_digest); | |
82 | ||
83 | // Reset byte counter | |
84 | ctx->total_length = 0; | |
85 | ||
86 | // Clear extra blocks | |
87 | ctx->partial_block_buffer_length = 0; | |
88 | } | |
89 | // If we made it here, there were no errors during this call to submit | |
90 | ctx->error = HASH_CTX_ERROR_NONE; | |
91 | ||
92 | // Store buffer ptr info from user | |
93 | ctx->incoming_buffer = buffer; | |
94 | ctx->incoming_buffer_length = len; | |
95 | ||
96 | // Store the user's request flags and mark this ctx as currently being processed. | |
97 | ctx->status = (flags & HASH_LAST) ? | |
98 | (HASH_CTX_STS) (HASH_CTX_STS_PROCESSING | HASH_CTX_STS_LAST) : | |
99 | HASH_CTX_STS_PROCESSING; | |
100 | ||
101 | // Advance byte counter | |
102 | ctx->total_length += len; | |
103 | ||
104 | // If there is anything currently buffered in the extra blocks, append to it until it contains a whole block. | |
105 | // Or if the user's buffer contains less than a whole block, append as much as possible to the extra block. | |
106 | if ((ctx->partial_block_buffer_length) | (len < SHA1_BLOCK_SIZE)) { | |
107 | // Compute how many bytes to copy from user buffer into extra block | |
108 | uint32_t copy_len = SHA1_BLOCK_SIZE - ctx->partial_block_buffer_length; | |
109 | if (len < copy_len) | |
110 | copy_len = len; | |
111 | ||
112 | if (copy_len) { | |
113 | // Copy and update relevant pointers and counters | |
114 | memcpy_varlen(&ctx->partial_block_buffer | |
115 | [ctx->partial_block_buffer_length], buffer, copy_len); | |
116 | ||
117 | ctx->partial_block_buffer_length += copy_len; | |
118 | ctx->incoming_buffer = (const void *)((const char *)buffer + copy_len); | |
119 | ctx->incoming_buffer_length = len - copy_len; | |
120 | } | |
121 | // The extra block should never contain more than 1 block here | |
122 | assert(ctx->partial_block_buffer_length <= SHA1_BLOCK_SIZE); | |
123 | ||
124 | // If the extra block buffer contains exactly 1 block, it can be hashed. | |
125 | if (ctx->partial_block_buffer_length >= SHA1_BLOCK_SIZE) { | |
126 | ctx->partial_block_buffer_length = 0; | |
127 | ||
128 | ctx->job.buffer = ctx->partial_block_buffer; | |
129 | ctx->job.len = 1; | |
130 | ||
131 | ctx = (SHA1_HASH_CTX *) sha1_mb_mgr_submit_avx2(&mgr->mgr, &ctx->job); | |
132 | } | |
133 | } | |
134 | ||
135 | return sha1_ctx_mgr_resubmit(mgr, ctx); | |
136 | } | |
137 | ||
138 | SHA1_HASH_CTX *sha1_ctx_mgr_flush_avx2(SHA1_HASH_CTX_MGR * mgr) | |
139 | { | |
140 | SHA1_HASH_CTX *ctx; | |
141 | ||
142 | while (1) { | |
143 | ctx = (SHA1_HASH_CTX *) sha1_mb_mgr_flush_avx2(&mgr->mgr); | |
144 | ||
145 | // If flush returned 0, there are no more jobs in flight. | |
146 | if (!ctx) | |
147 | return NULL; | |
148 | ||
149 | // If flush returned a job, verify that it is safe to return to the user. | |
150 | // If it is not ready, resubmit the job to finish processing. | |
151 | ctx = sha1_ctx_mgr_resubmit(mgr, ctx); | |
152 | ||
153 | // If sha1_ctx_mgr_resubmit returned a job, it is ready to be returned. | |
154 | if (ctx) | |
155 | return ctx; | |
156 | ||
157 | // Otherwise, all jobs currently being managed by the SHA1_HASH_CTX_MGR still need processing. Loop. | |
158 | } | |
159 | } | |
160 | ||
161 | static SHA1_HASH_CTX *sha1_ctx_mgr_resubmit(SHA1_HASH_CTX_MGR * mgr, SHA1_HASH_CTX * ctx) | |
162 | { | |
163 | while (ctx) { | |
164 | if (ctx->status & HASH_CTX_STS_COMPLETE) { | |
165 | ctx->status = HASH_CTX_STS_COMPLETE; // Clear PROCESSING bit | |
166 | return ctx; | |
167 | } | |
168 | // If the extra blocks are empty, begin hashing what remains in the user's buffer. | |
169 | if (ctx->partial_block_buffer_length == 0 && ctx->incoming_buffer_length) { | |
170 | const void *buffer = ctx->incoming_buffer; | |
171 | uint32_t len = ctx->incoming_buffer_length; | |
172 | ||
173 | // Only entire blocks can be hashed. Copy remainder to extra blocks buffer. | |
174 | uint32_t copy_len = len & (SHA1_BLOCK_SIZE - 1); | |
175 | ||
176 | if (copy_len) { | |
177 | len -= copy_len; | |
178 | memcpy_fixedlen(ctx->partial_block_buffer, | |
179 | ((const char *)buffer + len), copy_len); | |
180 | ctx->partial_block_buffer_length = copy_len; | |
181 | } | |
182 | ||
183 | ctx->incoming_buffer_length = 0; | |
184 | ||
185 | // len should be a multiple of the block size now | |
186 | assert((len % SHA1_BLOCK_SIZE) == 0); | |
187 | ||
188 | // Set len to the number of blocks to be hashed in the user's buffer | |
189 | len >>= SHA1_LOG2_BLOCK_SIZE; | |
190 | ||
191 | if (len) { | |
192 | ctx->job.buffer = (uint8_t *) buffer; | |
193 | ctx->job.len = len; | |
194 | ctx = (SHA1_HASH_CTX *) sha1_mb_mgr_submit_avx2(&mgr->mgr, | |
195 | &ctx->job); | |
196 | continue; | |
197 | } | |
198 | } | |
199 | // If the extra blocks are not empty, then we are either on the last block(s) | |
200 | // or we need more user input before continuing. | |
201 | if (ctx->status & HASH_CTX_STS_LAST) { | |
202 | uint8_t *buf = ctx->partial_block_buffer; | |
203 | uint32_t n_extra_blocks = hash_pad(buf, ctx->total_length); | |
204 | ||
205 | ctx->status = | |
206 | (HASH_CTX_STS) (HASH_CTX_STS_PROCESSING | HASH_CTX_STS_COMPLETE); | |
207 | ctx->job.buffer = buf; | |
208 | ctx->job.len = (uint32_t) n_extra_blocks; | |
209 | ctx = (SHA1_HASH_CTX *) sha1_mb_mgr_submit_avx2(&mgr->mgr, &ctx->job); | |
210 | continue; | |
211 | } | |
212 | ||
213 | if (ctx) | |
214 | ctx->status = HASH_CTX_STS_IDLE; | |
215 | return ctx; | |
216 | } | |
217 | ||
218 | return NULL; | |
219 | } | |
220 | ||
221 | static inline void hash_init_digest(SHA1_WORD_T * digest) | |
222 | { | |
223 | static const SHA1_WORD_T hash_initial_digest[SHA1_DIGEST_NWORDS] = | |
224 | { SHA1_INITIAL_DIGEST }; | |
225 | memcpy_fixedlen(digest, hash_initial_digest, sizeof(hash_initial_digest)); | |
226 | } | |
227 | ||
1e59de90 | 228 | static inline uint32_t hash_pad(uint8_t padblock[SHA1_BLOCK_SIZE * 2], uint64_t total_len) |
7c673cae | 229 | { |
1e59de90 | 230 | uint32_t i = (uint32_t) (total_len & (SHA1_BLOCK_SIZE - 1)); |
7c673cae FG |
231 | |
232 | memclr_fixedlen(&padblock[i], SHA1_BLOCK_SIZE); | |
233 | padblock[i] = 0x80; | |
234 | ||
235 | // Move i to the end of either 1st or 2nd extra block depending on length | |
236 | i += ((SHA1_BLOCK_SIZE - 1) & (0 - (total_len + SHA1_PADLENGTHFIELD_SIZE + 1))) + 1 + | |
237 | SHA1_PADLENGTHFIELD_SIZE; | |
238 | ||
239 | #if SHA1_PADLENGTHFIELD_SIZE == 16 | |
240 | *((uint64_t *) & padblock[i - 16]) = 0; | |
241 | #endif | |
242 | ||
1e59de90 | 243 | *((uint64_t *) & padblock[i - 8]) = to_be64((uint64_t) total_len << 3); |
7c673cae FG |
244 | |
245 | return i >> SHA1_LOG2_BLOCK_SIZE; // Number of extra blocks to hash | |
246 | } | |
247 | ||
248 | struct slver { | |
249 | uint16_t snum; | |
250 | uint8_t ver; | |
251 | uint8_t core; | |
252 | }; | |
253 | struct slver sha1_ctx_mgr_init_avx2_slver_04020145; | |
254 | struct slver sha1_ctx_mgr_init_avx2_slver = { 0x0145, 0x02, 0x04 }; | |
255 | ||
256 | struct slver sha1_ctx_mgr_submit_avx2_slver_04020146; | |
257 | struct slver sha1_ctx_mgr_submit_avx2_slver = { 0x0146, 0x02, 0x04 }; | |
258 | ||
259 | struct slver sha1_ctx_mgr_flush_avx2_slver_04020147; | |
260 | struct slver sha1_ctx_mgr_flush_avx2_slver = { 0x0147, 0x02, 0x04 }; | |
1e59de90 TL |
261 | |
262 | #if defined(__clang__) | |
263 | # pragma clang attribute pop | |
264 | #endif |