AtlatestRepositorysigil-crypto
sigil-crypto / tree / vendor / monocypherCHANGELOG.md
1
4.0.32
-----3
2026/06/155
- Fixed timing leak vulnerability in EdDSA/Ed25519 signatures.6
- Various minor documentation fixes.7
- Various minor build system fixes.8
- Various minor compiler warning fixes.11
4.0.212
-----13
2023/08/2415
- Fixed multiple-lanes Argon2.16
- Improved Poly1305 performance.17
- Improved Argon2 performance.18
- Makefiles no longer override standard environment variables.21
4.0.122
-----23
2023/03/0625
- Fixed Ed25519 secret key size in function prototype.26
- Fixed soname (should have been changed in 4.0.0)27
- Added convenience sub-targets to makefile.28
- Briefly specified wire format of Elligator and incremental AEAD.31
4.0.032
-----33
2023/02/2035
- Fixed unsafe signature API.36
- Simpler, more flexible low-level signature API.37
- Fully specified, consensus-friendly signatures.38
- Added Argon2d and Argon2id, support multiple lanes.39
- Added safe and fast streaming AEAD.40
- Added HKDF-SHA-512 and documented BLAKE2b KDF.41
- More consistent and memorable function names.42
- POSIX makefile.45
3.1.346
-----47
2022/04/2549
- Fixed many typos in the documentation.50
- Fixed buffer overflow in speed benchmarks.51
- Fixed some MSVC warnings.52
- Fixed a minor violation of the Elligator2 reverse map specs.53
- Added `change-prefix.sh` to help changing the `crypto_` prefix.54
- Added the `MONOCYPHER_CPP_NAMESPACE` preprocessor definition to55
support namespaces for C++.56
- Deprecated `crypto_key_exchange()`57
- Use GitHub actions to automate the regular test suite.60
3.1.261
-----62
2020/12/2764
- Addressed issues from Cure53's audit:65
- MON-01-001: Clarified which CSPRNG to use on Darwin.66
- MON-01-002: Won't fix (nonce handling is a core design decision).67
- MON-01-004: Compared with Kleshni's implementation.68
- MON-01-005: Split a dedicated "advanced" folder in the manual.69
- Quality assurance for 2^255-19 arithmetic (elliptic curves):70
- Documented carry propagation.71
- Enforced slightly safer invariants.72
- Improved the speed of EdDSA signature generation.73
- Made the vectors.h header more compact and easier to modify.74
- TIS-CI integration.75
- Added speed benchmark for ed25519-donna.76
- Documented lengths limits of `crypto_ietf_chacha20()`79
3.1.180
-----81
2020/06/1583
- Various documentation fixes.84
- Fixed various compiler warnings.85
- Fixed some integer overflows (16-bit platforms only).88
3.1.089
-----90
2020/04/0392
- Added Elligator 2 mappings (hash to curve, curve to hash).93
- Added OPRF support (with scalar inversion).94
- Added Edwards25519 -> Curve25519 conversions.97
3.0.098
-----99
2020/01/19101
- Deprecated the incremental AEAD interface.102
- Deprecated the incremental Chacha20, added a direct interface.103
- Added IETF Chacha20 (96-bit nonce), as described in RFC 8439.104
- Moved deprecated interfaces to a separate `src/deprecated` folder.105
- Removed the `ED25519_SHA512` preprocessor flag.106
- `crypto_x25519()` and `crypto_key_exchange()` now return `void`.107
- Added a custom hash interface to EdDSA. Several instances of EdDSA108
can share the same binary.109
- Added optional support for HMAC SHA-512.110
- Moved SHA-512 operations to `src/optional/monocypher-ed25519.(h|c)`.111
- Optional support for Ed25519 no longer requires a preprocessor flag.112
Add `src/optional/monocypher-ed25519.(h|c)` to your project instead.115
2.0.6116
-----117
2019/10/21119
- Added the `BLAKE2_NO_UNROLLING` preprocessor definition. Activating120
it makes the binary about 5KB smaller and speeds up processing times121
on many embedded processors.122
- Reduced the stack usage of signature verification by about123
40%. Signature verification now fits in smaller machines.124
- Fixed many implicit casts warnings.125
- Fixed the manual here and there.126
- Lots of small nitpicks.129
2.0.5130
-----131
2018/08/23133
- Faster EdDSA signatures and verification. Like, 4 times as fast.136
2.0.4137
-----138
2018/06/24140
- Corrected a critical vulnerability found by Mike Pechkin in EdDSA,141
where crypto_check() was accepting invalid signatures. The current142
fix removes a buggy optimisation, effectively halving the performance143
of EdDSA.144
- The test suite no longer tries to allocate zero bytes (some platforms145
fail such an allocation).147
2.0.3148
-----149
2018/06/16151
- Corrected undefined behaviour in BLAKE2b.152
- Improved the test suite (faster, better coverage).154
2.0.2155
-----156
2018/04/23158
- Corrected a couple failures to wipe secret buffers.159
- Corrected a bug that prevented compilation in Ed25519 mode.160
- Adjusted the number of test vectors in the test suite.161
- Improved tests for incremental interfaces.162
- Replaced the GNU all permissive licence by a public domain dedication163
(Creative Commons CC-0). The BSD licence remains as a fallback.165
2.0.1166
-----167
2018/03/07169
- Followed a systematic pattern for the loading code of symmetric170
crypto. It is now easier to review.171
- Tweaked Poly1305 code to make it easier to prove correct.173
2.0.0174
-----175
2018/02/14177
- Changed the authenticated encryption format. It now conforms to178
RFC 7539, with one exception: it uses XChacha20 initialisation instead179
of the IETF version of Chacha20. This new format conforms to180
libsodium's `crypto_aead_xchacha20poly1305_ietf_encrypt`.181
- Removed `crypto_lock_encrypt()` and `crypto_lock_auth()`.182
- Renamed `crypto_lock_aead_auth()` to `crypto_lock_auth_ad()`.183
- Renamed `crypto_unlock_aead_auth()` to `crypto_unlock_auth_ad()`.184
- Added `crypto_lock_auth_message()` and `crypto_unlock_auth_message()`.185
- Renamed `crypto_aead_lock` to `crypto_lock_aead`.186
- Renamed `crypto_aead_unlock` to `crypto_unlock_aead`.188
The format change facilitates optimisation by aligning data to block189
boundaries. The API changes increase consistency.191
1.1.0192
-----193
2018/02/06195
- Rewrote the manual into proper man pages.196
- Added incremental interfaces for authenticated encryption and197
signatures.198
- Replaced `crypto_memcmp()` by 3 fixed size buffer comparisons (16, 32,199
and 64 bytes), to make sure the generated code remains constant time.200
- A couple breaking API changes, easily fixed by renaming the affected201
functions.203
1.0.1204
-----205
2017/07/23207
- Optimised the loading and unloading code of the symmetric crypto208
(BLAKE2b, SHA-512, Chacha20, and Poly1305).209
- Fused self-contained tests together for easier analysis with Frama-C210
and the TIS interpreter.212
1.0213
---214
2017/07/18216
- Renamed `crypto_chacha20_Xinit` to `crypto_chacha20_x_init`, for217
consistency reasons (snake case everywhere).218
- Fixed signed integer overflow detected by UBSan.219
- Doubled the speed of EdDSA by performing the scalar product in220
Montgomery space.222
0.8223
---224
2017/07/06226
- Added about a hundred lines of code to improve performance of public227
key cryptography. Diffie-Hellman is now 20% faster than before.228
The effects are less pronounced for EdDSA.229
- Added random self-consistency tests.230
- Added a speed benchmark against libsodium.232
0.7233
---234
2017/06/07236
- Slightly changed the authenticated encryption API. Functions are237
now all in "detached" mode. The reason is better support for238
authenticated encryption _without_ additional data.239
- Rewrote BLAKE2b from spec so it can use the same licence as240
everything else.241
- Added random tests that compare Monocypher with libsodium and242
ed25519-donna.243
- Added explicit support for Frama-C analysis (this doesn't affect the244
source code).246
0.6247
---248
2017/03/17250
- Fixed incorrect Poly1305 output on empty messages. (Found by Mike251
Pechkin.)253
0.5254
---255
2017/03/10257
- Fixed many undefined behaviours in Curve25519 that occur whenever258
we perform a left shift on a signed negative integer. It doesn't259
affect the generated code, but you never know. (Found with Frama-C260
by André Maroneze.)262
Fun fact: TweetNaCl and ref10 have the same bug. Libsodium has263
corrected the issue, though.265
For those who don't comprehend the magnitude of this madness, the266
expression `-1 << 3` is undefined in C. This is explained in267
section 6.5.7(§4) of the C11 standard.269
0.4270
---271
2017/03/09273
- Fixed critical bug causing Argon2i to fail whenever it uses more274
than 512 blocks. It was reading uninitialised memory and the275
results were incorrect. (Found by Mike Pechkin.)276
- Fixed an undefined behaviour in Curve25519 (`fe_tobytes()`). It was277
accessing uninitialised memory before throwing it away. It didn't278
affect the compiled code nor the results, but you never know.279
(Found with [Frama-C](http://frama-c.com) by André Maroneze.)281
0.3282
---283
2017/02/27285
- Got the invariants of Poly1305 right and put them in the comments.286
There was no bug, but that was lucky (turned out the IETF test287
vectors were designed to trigger the bugs I was afraid of).288
- Simplified Poly1305 finalisation (replaced conditional subtraction289
by a carry propagation).290
- Made a few cosmetic changes here and there.292
0.2293
---294
????/??/??296
- Public interface significantly reworked. Removed redundant, hard to297
mess up constructions.298
- Added AEAD.299
- Sped up Curve25519 by a factor of more than 6 (switched to ref10300
arithmetic).301
- Added various test vectors and completed the consistency tests.303
0.1304
---305
2016/??/??