@@ -21,13 +21,13 @@ Changes with version 0.11 22 Feb 2013
2121
2222 * Bugfix: variables, structures, functions changed their names from
2323 *GOST3411* to fully-qualified *GOST34112012*
24-
24+
2525 * Bugfix: API: GOST34112012Init() now operates on allocated memory.
2626 Previously, memory allocation for context has been done with
2727 GOST34112012Init() call - this guarantee proper alignment for context's
2828 members. As for now, alignment is done with ALIGN keyword in object
2929 declaration and GOST34112012Init() does not allocate any byte of memory.
30-
30+
3131 * Bugfix: API: GOST34112012Final() now returns digest in memory pointed by
3232 second argument. Context member "hexdigest" not used anymore thus
3333 abandoned. It is up to calling application to present digest in
@@ -39,18 +39,18 @@ Changes with version 0.10 29 Jan 2013
3939 substitution by Pi[] while generating multiplication lookup table.
4040 Lookup table arranged in a such way totally eliminates the need to do
4141 whole S-cycle in LPS.
42-
42+
4343 * Feature: use SSE4.1 instructions set in LPS if available.
4444
4545 * Feature: multiplication lookup table is now statically precomputed in
4646 header file and is in constant memory.
47-
47+
4848 * Bugfix: rewrite GOST3411Update() function to implement streaming mode
4949 when processing large amount of data.
50-
50+
5151 * Bugfix: rename round2() and round3() to stage2() and stage3()
5252 respectively.
53-
53+
5454 * Bugfix: in auto configuration probes: delete temporary files.
5555
5656Changes with version 0.09 22 Jan 2013
@@ -60,10 +60,10 @@ Changes with version 0.09 22 Jan 2013
6060 quadruples. This avoid copying of temporary key and internal state
6161 between rounds in g(), thus reduce instructions count per plaintext
6262 block.
63-
63+
6464 * Feature: with code optimization above, performance on 32-bit systems
6565 greatly improved and is now about as fast as with 64-bit systems.
66-
66+
6767 * Feature: detect MMX and SSE2 capabilities on compile time and use faster
6868 code when possible. From this version onward there are at least two
6969 versions of implementation: portable (not using any extensions like MMX
@@ -76,7 +76,7 @@ Changes with version 0.09 22 Jan 2013
7676
7777 * Bugfix: integer overflow incorrectly handled in addition by modulo 2.
7878 This could lead to incorrect results on some files.
79-
79+
8080Changes with version 0.08 19 Jan 2013
8181
8282 * Feature: boost performance on 32-bit systems by using 64-bit MMX XOR
@@ -94,31 +94,31 @@ Changes with version 0.07 17 Jan 2013
9494
9595 * Bugfix: key generation function K replaced with XLPS macros to avoid
9696 inlining issues when max inline instruction limit reached.
97-
97+
9898 * Bugfix: encryption function E inlined to compression g to avoid
9999 compile-time inlining issues when max inline instruction limit reached.
100-
100+
101101 * Bugfix: use unsigned integers where possible.
102102
103103 * Bugfix: fix typo in update() when comparing bufsize.
104104
105105 * Bugfix: memory allocator rewritten to use posix_memalign().
106106
107107Changes with version 0.06 09 Jan 2013
108-
108+
109109 * Feature: performance optimization in LPS: hint compiler to use eight
110110 registers instead of union buffer.
111-
111+
112112 * Feature: performance optimization in permutation P: rotate matrix by
113113 cyclic right shifting bits in source matrix columns to rows in
114114 destination matrix.
115-
115+
116116 * Feature: as result of previous two optimizations: replace LPS macros
117117 with improved macros XLPS.
118118
119119 * Bugfix: improve constant table C presentation to avoid runtime
120120 conversion in key generator K.
121-
121+
122122 * Bugfix: unroll key generation loop in E.
123123
124124 * Bugfix: rename union member names according to their natural meaning.
0 commit comments