Debian Codegraph @

NEWS

raw
1 libxcrypt NEWS -- history of user-visible changes.
2
3 Please send bug reports, questions and suggestions to
4 <https://github.com/besser82/libxcrypt/issues>.
5
6 Version 4.5.1
7 * Do not include undefined symbols in version-script. (issue #181, #213).
8 * Fix build with clang-20+ on macOS (issue #216).
9
10 Version 4.5.0
11 * Implement the sm3crypt ($sm3$) hashing algorithm (issue #188).
12 * Implement the sm3-yescrypt ($sm3y$) hashing algorithm (issue #206).
13 * Fix the implementation of the crypt(3) functions and the crypt_gensalt(3)
14 functions to not overwrite the output buffer too early. (issue #209).
15 * Fix the strcpy_or_abort() function to call abort() in -DNDEBUG builds.
16 * Add some more testcases.
17 * Several fixes for issues found by Coverity.
18
19 Version 4.4.38
20 * Fix several "-Wunterminated-string-initialization", which are seen by
21 upcoming GCC 15.x (issue #194).
22 * Fix "-Wmaybe-uninitialized" in crypt.c, which is seen by GCC 13.3.0.
23 * Skip test/explicit-bzero if compiling with ASAN.
24 * Drop hard requirement for the pkg-config binary (issue #198).
25
26 Version 4.4.37
27 * Several fixes to the manpages (issue #185).
28 * Add binary compatibility for x86_64 GNU/Hurd (issue #189).
29 * Only test the needed makecontext signature during configure (issue #178).
30 * Fix -Werror=strict-overflow in lib/crypt-bcrypt.c, which is seen
31 by GCC 4.8.5 (issue #197).
32
33 Version 4.4.36
34 * Fix left over bits failing with Perl v5.38.0 (issue #173).
35
36 Version 4.4.35
37 * Fix build with Perl v5.38.0 (issue #170).
38 * Fix build with MinGW-w(32|64).
39
40 Version 4.4.34
41 * Update build-aux/m4/ax_valgrind_check.m4 to v23.
42 * Optimize some cast operation for performance in
43 lib/alg-yescrypt-platform.c.
44 * Add SHA-2 Maj() optimization proposed by Wei Dai in lib/alg-sha512.c.
45 * Explicitly clean the stack and context state after computation in
46 lib/alg-gost3411-2012-hmac.c, lib/alg-hmac-sha1.c, and lib/alg-sha256.c
47 (issue #168).
48
49 Version 4.4.33
50 * Fix -Werror=sign-conversion in lib/alg-yescrypt-platform.c.
51 With commit 894aee75433b4dc8d9724b126da6e79fa5f6814b we introduced some
52 changes to huge page handling, that show this error when building with
53 GCC v12.2.1, and thus need a small fix.
54
55 Version 4.4.32
56 * Improvements to huge page handling in lib/alg-yescrypt-platform.c.
57 When explicitly using huge pages, request the 2 MiB page size.
58 This should fix the issue where on a system configured to use 1 GiB
59 huge pages we'd fail on munmap() as we're only rounding the size up
60 to a multiple of 2 MiB. With the fix, we wouldn't use huge pages on
61 such a system. Unfortunately, now we also wouldn't use huge pages on
62 Linux kernels too old to have MAP_HUGE_2MB (issue #152).
63
64 Version 4.4.31
65 * Fix -Werror=conversion in lib/alg-yescrypt-opt.c
66 (issues #161 and #162).
67 * Add some SHA-2 Maj() optimization in lib/alg-sha256.c.
68 * Fix issues found by Covscan in test/getrandom-fallback.c.
69 * Fix -Werror=strict-overflow in lib/crypt-des.c, which is seen
70 by GCC 12.x (issues #155 and #163).
71
72 Version 4.4.30
73 * configure: Restore ucontext api functionality check.
74 In c3f01c72b303cbbb0cc8983120677edee2f3fa4b the use of the ucontext api
75 in the main program was removed, and with it the configure check for it.
76 However, the ucontext api is still used in the "explicit_bzero" test and
77 thus this test still needs to be in place.
78 See also: https://bugs.gentoo.org/838172
79 * configure: Restore the functionality of the '--disable-symvers' switch.
80 Without this fix the build was simply broken, if symbol versioning was
81 disabled for any reason, e.g. whether the compiler nor the linker
82 supporting it, or if disabled on purpose by the user (issue #142).
83 * Fix variable name in crypt(3) for a datamember of 'struct crypt_data'
84 (issue #153).
85
86 Version 4.4.29
87 * Add glibc-on-loongarch-lp64 (Loongson LA464 / LA664) entry to
88 libcrypt.minver. This was added in GNU libc 2.36.
89
90 Version 4.4.28
91 * Add glibc-on-or1k (OpenRISC 1000) entry to libcrypt.minver.
92 This was added in GNU libc 2.35.
93
94 Version 4.4.27
95 * Limit the maximum amount of rbytes to 64 bytes (512 bits) for
96 yescrypt, gost-yescrypt, and scrypt. Also reflect this limit
97 in the documentation (issue #145).
98
99 Version 4.4.26
100 * Fix compilation on systems with GCC >= 10, that do not support
101 declarations with __attribute__((symver)).
102
103 Version 4.4.25
104 * Add support for Python 3.11 in the configure script.
105 * Stricter checking of invalid salt characters (issue #135).
106 Hashed passphrases are always entirely printable ASCII, and do
107 not contain any whitespace or the characters ':', ';', '*', '!',
108 or '\'. (These characters are used as delimiters and special
109 markers in the passwd(5) and shadow(5) files.)
110
111 Version 4.4.24
112 * Add hash group for Debian in lib/hashes.conf.
113 Debian has switched to use the yescrypt hashing algorithm as
114 the default for new user passwords, so we should add a group
115 for this distribution.
116 * Overhaul the badsalt test.
117 Test patterns are now mostly generated rather than manually coded
118 into a big table. Not reading past the end of the “setting” part
119 of the string is tested more thoroughly (this would have caught the
120 sunmd5 $$ bug if it had been available at the time).
121 Test logs are tidier.
122 * Add ‘test-programs’ utility target to Makefile.
123 It is sometimes useful to compile all the test programs but not run
124 them. Add a Makefile target that does this.
125 * Fix incorrect bcrypt-related ifdeffage in test/badsalt.c.
126 The four variants of bcrypt are independently configurable, but the
127 badsalt tests for them were all being toggled by INCLUDE_bcrypt,
128 which is only the macro for the $2b$ variant.
129 * Fix bigcrypt-related test cases in test/badsalt.c.
130 The test spec was only correct when both or neither of bigcrypt and
131 descrypt were enabled.
132 * Detect ASan in configure and disable incompatible tests.
133 ASan’s “interceptors” for crypt and crypt_r have a semantic conflict
134 with libxcrypt, requiring a few tests to be disabled for builds with
135 -fsanitize-address. See commentary in test/crypt-badargs.c for an
136 explanation of the conflict, and the commentary in
137 build-aux/m4/zw_detect_asan.m4 for why a configure test is required.
138 * Fix several issues found by Covscan in the testsuite. These include:
139 - CWE-170: String not null terminated (STRING_NULL)
140 - CWE-188: Reliance on integer endianness (INCOMPATIBLE_CAST)
141 - CWE-190: Unintentional integer overflow (OVERFLOW_BEFORE_WIDEN)
142 - CWE-569: Wrong sizeof argument (SIZEOF_MISMATCH)
143 - CWE-573: Missing varargs init or cleanup (VARARGS)
144 - CWE-687: Argument cannot be negative (NEGATIVE_RETURNS)
145
146 Version 4.4.23
147 * Fix output calculation for gensalt_yescrypt_rn().
148 * Fix -Werror=conversion in lib/crypt-des-obsolete.c,
149 test/des-obsolete.c, and test/des-obsolete_r.c.
150
151 Version 4.4.22
152 * The crypt_checksalt() function has been fixed to correctly return
153 with 'CRYPT_SALT_INVALID', in case the setting, that is passed
154 to be checked, represents an empty passphrase or an uncomputed
155 setting for descrypt without any salt characters.
156
157 Version 4.4.21
158 * The crypt_checksalt() function will now return the value
159 'CRYPT_SALT_METHOD_LEGACY' in case the setting, that is passed
160 to be checked, uses a hashing method, which is considered to be
161 too weak for use with new passphrases.
162
163 Version 4.4.20
164 * Fix build when the CFLAGS variable, that is passed into the
165 configure script, has a leading whitespace character in it
166 (issue #123).
167
168 Version 4.4.19
169 * Improve fallback implementation of explicit_bzero.
170 * Add glibc-on-CSKY, ARC, and RISCV-32 entries to libcrypt.minver.
171 These were added in GNU libc 2.29, 2.32, and 2.33 respectively
172 (issue #122).
173 * Do not build xcrypt.h if we’re not going to install it.
174 * Do not apply --enable-obsolete-api-enosys mode to fcrypt.
175 * Compilation fix for NetBSD. NetBSD’s <unistd.h> declares encrypt
176 and setkey to return int, contrary to POSIX (which says they return
177 void). Rename those declarations out of the way with macros.
178 * Compilation fixes for building with GCC 11.
179 Basically fixes for explicit type-casting.
180 * Force update of existing symlinks during installation (issue #120).
181
182 Version 4.4.18
183 * Fix compilation errors on (Free)BSD (issue #110).
184 * Fix conversion error in lib/alg-gost3411-core.c, which is seen by
185 some sensitive compilers.
186 * Convert build scripts to Perl.
187 The minimum version of Perl required is 5.14.
188
189 Version 4.4.17
190 * Fix compilation error in 'alignas (type)' with older versions
191 of glibc and/or gcc (issue #107).
192 * Salt string compatibility with generic implementations (issue #105).
193 All other existing implementations of the md5crypt, sha256crypt,
194 and sha512crypt hashing methods allow any ASCII character to be
195 present in the salt string.
196 We changed our implementation of these hash methods to be compliant
197 with the general behaviour, except for the colon ':' character,
198 which serves a special purpose in the Unix shadow file, and the
199 newline '\n' character, since all parameters of the user data must
200 be on the same line within the Unix shadow file.
201 * Fix for GCC 10.2 on s390x.
202
203 Version 4.4.16
204 * Add support for the e2k architecture.
205
206 Version 4.4.15
207 * The compatibility symbols crypt_gensalt_r, xcrypt, xcrypt_r,
208 xcrypt_gensalt, and xcrypt_gensalt_r are deprecated further.
209 (These are alternative names for crypt_gensalt_rn, crypt, crypt_r,
210 crypt_gensalt, and crypt_gensalt_rn, respectively; there is no
211 difference in behavior.)
212
213 In an --enable-xcrypt-compat-files configuration, newly compiled
214 programs can still use functions with these names if they include
215 xcrypt.h, but a program that attempts to use one of these functions
216 *without* including xcrypt.h will fail to link. In particular, this
217 means AC_CHECK_FUNCS([xcrypt]) will not detect xcrypt, because the
218 test program generated by AC_CHECK_FUNCS does not include xcrypt.h.
219
220 In a --disable-xcrypt-compat-files --enable-obsolete-api
221 configuration, xcrypt.h is not installed and newly compiled programs
222 cannot use functions with these names, but existing binaries that
223 expect to find these symbols in a shared libcrypt.so.1 will still
224 work.
225
226 In a --disable-obsolete-api configuration, these symbols are not
227 defined at all; moreover, --disable-obsolete-api now implies
228 --disable-xcrypt-compat-files (which means xcrypt.h isn't installed).
229
230 Version 4.4.14
231 * Renamed bootstrap script: 'bootstrap' -> 'autogen.sh'.
232 If building from a Git checkout instead of a tarball release,
233 use './autogen.sh' to create the configure script.
234
235 Version 4.4.13
236 * libxcrypt now builds (including working tests) on Mac OSX.
237 * Speed up ka-sunmd5 by skipping most of the test phrases.
238 ka-sunmd5 is slower than the entire rest of the testsuite put
239 together, because the sunmd5 hash is Just That Slow and we have to do
240 extra tests for it to ensure bug-compatibility. No easy optimizations
241 are possible and the difficult optimizations are not worth the
242 engineering effort for this obsolete hash. Instead, skip most of the
243 test phrases. See comments in test/ka-table-gen.py!h_sunmd5 for
244 further explanation.
245
246 Version 4.4.12
247 * Another fix for GCC v10.x, which occurs on s390 architectures only.
248
249 Version 4.4.11
250 * Fixes for GCC v10.x (issue #95).
251 * Change how the known-answer tests are parallelized.
252
253 Version 4.4.10
254 * Fix ordering of hash methods in crypt-hashes.h when generated
255 using gawk < 4.1.0.
256
257 Version 4.4.9
258 * Fix false positive finding from CovScan.
259
260 Version 4.4.8
261 * Add binary compatibility for GNU/Hurd and GNU/kFreeBSD.
262
263 Version 4.4.7
264 * LICENSING: Add missing files and update license information.
265 * Fix -Wformat in test/crypt-kat.c.
266
267 Version 4.4.6
268 * Make unalignment test really unaligned.
269
270 Version 4.4.5
271 * Fix alignment problem for GOST 34.11 (Streebog) in gost-yestcrypt.
272 Some architectures in some circumstances do not allow unaligned
273 memory access (such as ARM, MIPS, SPARC) triggering SIGBUS. This
274 patch very crudely fixes this issue.
275 Being unfixed this would trigger SIGBUS when password buffer is
276 unaligned. Crash and fix are tested on UltraSparc T5 on GCC Compile
277 farm.
278
279 Version 4.4.4
280 * The crypt_* functions will now all fail and set errno to ERANGE if
281 their 'phrase' argument is longer than CRYPT_MAX_PASSPHRASE_SIZE
282 characters (this is currently 512). Formerly, longer passphrases
283 would either be silently accepted, silently truncated, or the
284 library would crash, depending on the hashing method.
285 * The NT hashing method no longer truncates passphrases at 128
286 characters; Windows does not do this. (The Windows login dialog
287 _limits_ interactively entered passphrases to 127 characters.
288 Passphrases set via the low-level API can be longer.)
289
290 Version 4.4.3
291 * Fix the value of SUNMD5_MAX_ROUNDS.
292 * Add generated C++-guards to <xcrypt.h>.
293 * Add --enable-obsolete-api-enosys configure option.
294 If enabled, this option replaces the obsolete APIs (fcrypt,
295 encrypt{,_r}, and setkey{,_r}) with stubs that set errno to
296 ENOSYS and return without performing any real operations.
297 This allows one to disable DEScrypt support while preserving
298 POSIX compliance.
299 For security reasons, the encrypt{,r} functions will also
300 overwrite their data-block argument with random bits.
301 The fcrypt function will always produce a failure token
302 (*0 or *1), unless the library was also configured with
303 --disable-failure-tokens, in which case it will always
304 return NULL.
305
306 Version 4.4.2
307 * Add test-alg-yescrypt for improved coverage.
308 * Add x32 specific inline asm.
309
310 Version 4.4.1
311 * Change the output of the gensalt function for the NT hashing method
312 ($3$) to output just its prefix ($3$) instead of a fake salt.
313 The output buffer for the returned string must be at least 4 bytes
314 long.
315 * Decrase the minimum required amount of random bytes for the NT
316 hashing method to 1 byte, which is actually not used.
317 * The default linker flags now generate binaries with full read-only
318 relocations (including full read-only re-mapping of the GOT), if
319 supported by the linker and the system's native binary format.
320
321 Version 4.4.0
322 * Implement the crypt_preferred_method function.
323 This function can be used as a convenience function to get the prefix
324 of the preferred hash method.
325
326 Version 4.3.4
327 * --enable-hashes now supports 'fedora' as a group of hashing methods.
328 * Add aliases for xcrypt{,_r} and xcrypt_gensalt{,_r}.
329 They were added for code compatibility with libxcrypt v3.1.1
330 and earlier.
331 * Install the <xcrypt.h> header file, declaring the previously named
332 aliases, and a symlink from libxcrypt.so to libcrypt.so, if a shared
333 library is build. For static libraries a corresponding symlink for
334 the archive file will be installed.
335 The installation of the compatibility files can be disabled by
336 passing the '--disable-xcrypt-compat-files' flag to the configure
337 script.
338 * Replace the prototype for the crypt_gensalt_r function with a
339 declaration through a macro, so new compiled applications link
340 against the identical crypt_gensalt_rn function directly.
341
342 Version 4.3.3
343 * Add an alias for crypt_gensalt_r.
344 The function was available in older versions (v3.1.1 and earlier)
345 of libxcrypt. It has the same semantics and the same prototype as
346 the crypt_gensalt_rn function.
347
348 Version 4.3.2
349 * Fix the gensalt function for the NT hashing method ($3$) to
350 properly terminate its output string. The output buffer for the
351 returned string must be at least 30 bytes long.
352 * Remove the gensalt function for the bcrypt x variant ($2x$).
353 This prefix was never intended for use when hashing new passphrases.
354 The only use case originally intended was to manually edit '$2a$' to
355 '$2x$' in passphrase hashes to knowingly take the risk yet enable
356 users to continue to log in when upgrading systems with buggy bcrypt
357 implementations to fixed versions.
358
359 Version 4.3.1
360 * Reduce the number of methods that can be the default for new hashes.
361 We don’t want to accidentally encourage use of gost-yescrypt,
362 scrypt, or the bug-compatibility bcrypt variants by people who do
363 not have a specific need for them.
364 * sha256crypt ($5$) is not considered to be strong anymore. Thus
365 it was dropped from the STRONG and DEFAULT sets.
366 * The four variants of bcrypt ($2b$, $2a$, $2y$, $2x$) are now
367 independently selectable at configure time. The $2x$ variant has
368 been dropped from the STRONG set, because it has a severe bug
369 causing it to be easy to find collisions for some passwords. The
370 $2x$ and $2y$ variants have been dropped from the FreeBSD, NetBSD,
371 OpenBSD, and Solaris compatibility sets, because those operating
372 systems never supported those variants.
373
374 Version 4.3.0
375 * Implement the gost-yescrypt ($gy$) hashing algorithm.
376 * Remove all of the nonnull annotations, which are a questionable
377 optimization that have caused problems in the past and may cause
378 future problems on some systems.
379 * Rename all hash methods to match naming in John the Ripper.
380 * Make the configure switches --enable-obsolete-api and
381 --enable-hashes process their arguments case-insensitively; for
382 instance, "--enable-hashes=OpenBSD" and "--enable-hashes=openbsd"
383 are now understood as synonymous.
384 * Fix gensalt for bigcrypt if descrypt hash method is not selected
385 at compile time.
386 * Fix the bigcrypt hash method to reject too short settings and/or
387 descrypt hashes with a truncated phrase. Hashes with a setting
388 shorter than 13 characters and phrases shorter than 8 characters
389 are valid bigcrypt hashes, although they are identical with the
390 output generated by descrypt under the same conditions.
391 This only applies when the descrypt hash method is not selected
392 at compile time.
393 * Implement crypt_checksalt, which can be used by portable users of
394 libxcrypt to check whether the desired hash method is supported.
395 * Make a default prefix available whenever at least one strong hash
396 is enabled.
397 * Fix the definition of 'CRYPT_GENSALT_IMPLEMENTS_DEFAULT_PREFIX'
398 to reflect whether the default prefix is available or not.
399
400 Version 4.2.3
401 * Add bootstrap script. If building from a Git checkout instead of a
402 tarball release, use './bootstrap' to create the configure script.
403 * Use sha512 implementation from Colin Percival. Thus we now have a
404 sha512 implementation under the BSD license.
405 * Use md5 implementation from Alexander Peslyak. Thus we now have a
406 md5 implementation in the public domain.
407 * 'make dist' now generates bzip2 and xz compressed tarballs along
408 with the default gzip compressed tarball.
409 * The group 'altlinux' has been renamed to 'alt' in --enable-hashes,
410 as the ALTLINUX distribution has been rebranded to ALT some time ago.
411 * New tagged versions are automatically submitted for Coverity Scan:
412 https://scan.coverity.com/projects/besser82-libxcrypt
413
414 Version 4.2.2
415 * Convert existing manpages to BSD mdoc format.
416
417 Version 4.2.1
418 * Fix tests for yescrypt and scrypt, when failure-tokens have been
419 disabled.
420
421 Version 4.2.0
422 * Implement yescrypt ($y$) and scrypt ($7$) hashing algorithms.
423 * For scrypt the implemented gensalt function ensures every new hash
424 is computed using at least 32 MiBytes of RAM.
425 * yescrypt is the new default method used for new hashes.
426 * --enable-hashes now supports additional groups of hashing methods:
427 'altlinux', 'owl', and 'suse', which select the methods historically
428 supported on those operating systems.
429 * Added 'XCRYPT_VERSION_*' macros.
430
431 Version 4.1.2
432 * Add optional 'check-valgrind' target to the Makefile.
433 * Replace crypt-sha{256,512}.c with an implementation in the Public
434 Domain.
435 * Add alias man-pages for other crypt functions.
436 * Add configure option --disable-failure-tokens, which causes crypt
437 and crypt_r to return NULL on failure, as crypt_rn and crypt_ra do,
438 instead of a special "failure token". Using this option improves
439 compatibility with programs written on the assumption that, like
440 most C library functions, crypt and crypt_r will return NULL on
441 failure; but it breaks compatibility with programs that assume these
442 functions never return NULL. We're not sure which type of program
443 is more common. Please let us know if you encounter either.
444 * Improved handling of out-of-range cost parameters in gensalt.
445 The behavior is now:
446 - for hashes with a fixed cost parameter (DES/trad, DES/big, NTHASH,
447 MD5/bsd), crypt_gensalt only accepts 0 as the rounds argument.
448 - for hashes with a linear cost parameter (DES/bsdi, MD5/sun, SHA1,
449 SHA256, SHA512), crypt_gensalt accepts 0 or any value in the range
450 [1, ULONG_MAX] and clips it to the actual valid range for the hash
451 function, if necessary. In the case of DES/bsdi, even numbers
452 become odd, as well.
453 - for hashes with an exponential cost parameter (bcrypt),
454 crypt_gensalt only accepts 0 or a value in the actual valid range.
455 - the documented valid range for SHA1 is now [4, 4294967295] instead
456 of [1, 4294967295].
457 - all of this is tested.
458
459 Version 4.1.1
460 * --enable-hashes now supports additional groups of hashing methods:
461 'freebsd', 'netbsd', 'openbsd', 'osx', and 'solaris', which select
462 the hashes historically supported on those operating systems.
463 * Predictable behavior when arguments to crypt() are NULL or invalid
464 (issue #15).
465 * Hash formats $5, $6, and $md5 once again allow an explicit rounds
466 parameter specifying the default number of rounds (issue #16).
467 * The library no longer uses swapcontext(), for ease of debugging and
468 better compatibility with hardening mechanisms like Intel CET
469 (issue #18).
470 * Can now be built with versions of GNU ld older than 2.26.2.
471 * crypt_gensalt_ra no longer leaks memory on failure.
472
473 Version 4.1.0
474 * Fix spelling of SUSE.
475 * Lower the minimum required automake version to 1.14.
476 * Fix build with USE_SWAPCONTEXT turned off.
477 * Extend --enable-weak-hashes configure option to accept optional
478 "glibc" parameter. When specified, it enables only those of
479 weak hashes that are supported by historic versions of the
480 GNU libc.
481 * Fix the leak of obtained random bytes.
482 * Check expected output strings for deterministic methods.
483 * Fix memory leak in crypt_sha1crypt_rn.
484 * Fix read of random bytes out of bounds in gensalt_sha1crypt_rn.
485 * Make it possible to disable individual hashes at configure time.
486 The default is --enable-hashes=all. --enable-hashes=strong is the
487 equivalent of the old --disable-weak-hashes. You could even do
488 --enable-hashes=bcrypt,des to get a binary-compatible libcrypt.so.1
489 that still supports almost nothing other than bcrypt.
490 * Make salt validation pickier.
491 * Replace crypt-sunmd5.c with BSD-licensed cleanroom reimplementation.
492 * Make crypt_gensalt for $sha1 deterministic.
493 * Fix incorrect output-size computation in crypt_sha1crypt_rn.
494 * Add docs for SHA1, MD5/Sun, NTHASH.
495 * Introduce CRYPT_GENSALT_IMPLEMENTS_* feature test macros.
496 * Install libcrypt.pc symlink along with libxcrypt.pc.
497 * Extend --enable-obsolete-api configure option.
498 Make vendor specific parts of compatibility ABI that are enabled by
499 --enable-obsolete-api option configurable.
500 This allows vendors to enable only those parts of compatibility ABI
501 that are relevant to them.
502 * Extend overall test coverage.
503
504 Version 4.0.1
505 * Fixes for GCC v8.x
506 * Add symbol version for riscv64
507 * Fixed an uninitialized value in test-crypt-badsalt
508
509 Version 4.0.0
510 * Full binary backward compatibility with glibc libcrypt; all programs
511 compiled with glibc libcrypt, including vendor-patched versions that
512 include the Openwall extensions, should work with this libcrypt
513 (however, programs compiled against this libcrypt will NOT work with
514 glibc libcrypt).
515 * struct crypt_data is now only 32kB (from 128kB), and divided into a
516 public-API section and a properly opaque internal section.
517 * New feature: supplying a null pointer as the "prefix" argument to
518 any of the crypt_gencrypt functions will cause it to select the best
519 available hash function (in this release, bcrypt in mode 'a').
520 CAUTION: it must be a null pointer, not an empty string. If you
521 supply an empty string, that selects DES, which is the *worst*
522 available hash function.
523 * New feature: supplying a null pointer as the "rbytes" argument to
524 any of the crypt_gencrypt functions will cause it to acquire random
525 bytes from the operating system.
526 * The legacy functions bigcrypt, fcrypt, encrypt, encrypt_r,
527 setkey, and setkey_r are no longer available for use by new programs.
528 All of these (except fcrypt, which was just another name for crypt)
529 force the use of DES, which is no longer safe for any application.
530 * New configure option --disable-obsolete-api removes the above
531 functions from the library. Since this breaks compatibility with
532 glibc's libcrypt, when this option is used the shared library will
533 be libcrypt.so.2 instead of libcrypt.so.1, and all of the
534 compatibility symbol versions for the crypt* functions will be
535 omitted. This option is the default on all operating systems where
536 there is no GNU C Library to be compatible with.
537 * New configure option --disable-weak-hashes removes all support for
538 DES and MD5 hashes from the library. This option implies
539 --disable-obsolete-api, and will prevent DES and MD5 password hashes
540 from being *verified* -- accounts with such hashes are effectively
541 locked. It is intended for use in high-security new installations.
542 * Sensitive intermediate data is now thoroughly scrubbed from the
543 stack and CPU registers before the crypt functions return to their
544 callers.
545 * UFC-crypt has been replaced with FreeSec; this enables the reduction in
546 size of crypt_data, and adds full support for BSD extended DES hashes.
547 * Extensive code cleanup and portability work. The static library
548 should now be buildable with any C99 compiler, although some
549 features may not be available (notably acquiring random bytes from
550 the operating system and scrubbing the stack). The shared library
551 does still require some GNU extensions for symbol versioning.
552 * The configure options --enable-Wno-cast-align and --enable-bootstrap
553 are no longer necessary and have been removed.
554 * If building from a Git checkout instead of a tarball release, use
555 'autoreconf -i' to create the configure script; autogen.sh has been
556 removed.
557 * More thoroughly tested.
558
559 Version 3.1.1
560 * Add '--enable-Wno-cast-align' to silence 'cast increases required alignment'
561 * Whitespace clean-up
562
563 Version 3.1.0
564 * Update upstream-contact
565 * Add './configure --enable-bootstrap' to skip some tests on initial build
566 * Add bootstrap-script for Autotools
567 * Add LICENSE for bcrypt
568 * Update crypt_blowfish to v1.3
569 * Add '-Wextra' to CFLAGS
570 * Fix warnings generated by gcc 5.1.1
571 * Update Autotools
572
573 Version 3.0.4
574 * Fix warnings generated by gcc 4.6
575
576 Version 3.0.3
577 * Fix memory leak
578 * Fix compiler warnings
579
580 Version 3.0.2
581 * Fix generating salts for MD5
582
583 Version 3.0.1
584 * Fix build failures
585
586 Version 3.0
587 * Add sha256 and sha512 hashes
588 * Move all hashes into plugins
589
590 Version 2.4
591 * Sync with crypt_blowfish 1.0
592
593 Version 2.3
594 * Fix problems with gcc > 4.0
595
596 Version 2.2
597 * Fix realloc call in md5-crypt
598
599 Version 2.1
600 * Enable x86.S for i386 again
601
602 Version 2.0
603 * Rename all crypt* functions to xcrypt* to avoid clash with normal
604 libcrypt from glibc
605
606 Version 1.4
607 * Fix compiling with glibc > 2.3.1
608 * Sync with crypt_blowfish 0.4.5
609
610 Version 1.3
611 * Sync with current glibc CVS (fix for not correct initialication of
612 internal data structs)
613
614 Version 1.2
615 * Fix building with glibc 2.3.x
616 * Add support for HPPA
617
618 Version 1.1
619 * merge md5.h with xcrypt.h
620 * Add SHA1 functions and crypt
621
622 Version 1.0
623 * First release