File: CHANGELOG.md

package info (click to toggle)
hyperscan 4.4.1-1
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 9,676 kB
  • ctags: 17,632
  • sloc: cpp: 126,339; ansic: 34,412; python: 330; sh: 45; makefile: 9
file content (151 lines) | stat: -rw-r--r-- 7,992 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
# Hyperscan Change Log

This is a list of notable changes to Hyperscan, in reverse chronological order.

## [4.4.1] 2017-02-28
- Bugfixes to fix issues where stale data was being referenced in scratch
  memory. In particular this may have resulted in hs_close_stream()
  referencing data from other previously scanned streams. This may result in
  incorrect matches being been reported.

## [4.4.0] 2017-01-20
- Introduce the "fat runtime" build. This will build several variants of the
  Hyperscan scanning engine specialised for different processor feature sets,
  and use the appropriate one for the host at runtime. This uses the "ifunc"
  indirect function attribute provided by GCC and is currently available on
  Linux only, where it is the default for release builds.
- New API function: add the `hs_valid_platform()` function. This function tests
  whether the host provides the SSSE3 instruction set required by Hyperscan.
- Introduce a new standard benchmarking tool, "hsbench". This provides an easy
  way to measure Hyperscan's performance for a particular set of patterns and
  corpus of data to be scanned.
- Introduce a 64-bit GPR LimEx NFA model, which uses 64-bit GPRs on 64-bit
  hosts and SSE registers on 32-bit hosts.
- Introduce a new DFA model ("McSheng") which is a hybrid of the existing
  McClellan and Sheng models. This improves scanning performance for some
  cases.
- Introduce lookaround specialisations to improve scanning performance.
- Improve the handling of long literals by moving confirmation to the Rose
  interpreter and simplifying the hash table used to track them in streaming
  mode.
- Improve compile time optimisation for removing redundant paths from
  expression graphs.
- Build: improve support for building with MSVC toolchain.
- Reduce the size of small write DFAs used for small scans in block mode.
- Introduce a custom graph type (`ue2_graph`) used in place of the Boost Graph
  Library's `adjacency_list` type. Improves compile time performance and type
  safety.
- Improve scanning performance of the McClellan DFA.
- Bugfix for a very unusual SOM case where the incorrect start offset was
  reported for a match.
- Bugfix for issue #37, removing execute permissions from some source files.
- Bugfix for issue #41, handle Windows line endings in pattern files.

## [4.3.2] 2016-11-15
- Bugfix for issue #39. This small change is a workaround for an issue in
  Boost 1.62. The fix has been submitted to Boost for inclusion in a future
  release.

## [4.3.1] 2016-08-29
- Bugfix for issue #30. In recent versions of Clang, a write to a variable was
  being elided, resulting in corrupted stream state after calling
  `hs_reset_stream()`.

## [4.3.0] 2016-08-24
- Introduce a new analysis pass ("Violet") used for decomposition of patterns
  into literals and smaller engines.
- Introduce a new container engine ("Tamarama") for infix and suffix engines
  that can be proven to run exclusively of one another. This reduces stream
  state for pattern sets with many such engines.
- Introduce a new shuffle-based DFA engine ("Sheng"). This improves scanning
  performance for pattern sets where small engines are generated.
- Improve the analysis used to extract extra mask information from short
  literals.
- Reduced compile time spent in equivalence class analysis.
- Build: frame pointers are now only omitted for 32-bit release builds.
- Build: Workaround for C++ issues reported on FreeBSD/libc++ platforms.
  (github issue #27)
- Simplify the LimEx NFA with a unified "variable shift" model, which reduces
  the number of different NFA code paths to one per model size.
- Allow some anchored prefixes that may squash the literal to which they are
  attached to run eagerly. This improves scanning performance for some
  patterns.
- Simplify and improve EOD ("end of data") matching, using the interpreter for
  all operations.
- Elide unnecessary instructions in the Rose interpreter at compile time.
- Reduce the number of inlined instantiations of the Rose interpreter in order
  to reduce instruction cache pressure.
- Small improvements to literal matcher acceleration.
- Parser: ignore `\E` metacharacters that are not preceded by `\Q`. This
  conforms to PCRE's behaviour, rather than returning a compile error.
- Check for misaligned memory when allocating an error structure in Hyperscan's
  compile path and return an appropriate error if detected.

## [4.2.0] 2016-05-31
- Introduce an interpreter for many complex actions to replace the use of
  internal reports within the core of Hyperscan (the "Rose" engine). This
  improves scanning performance and reduces database size for many pattern
  sets.
- Many enhancements to the acceleration framework used by NFA and DFA engines,
  including more flexible multibyte implementations and more AVX2 support. This
  improves scanning performance for many pattern sets.
- Improved prefiltering support for complex patterns containing very large
  bounded repeats (`R{M,N}` with large `N`).
- Improve scanning performance of pattern sets with a very large number of
  EOD-anchored patterns.
- Improve scanning performance of large pattern sets that use the
  `HS_FLAG_SINGLEMATCH` flag.
- Improve scanning performance of pattern sets that contain a single literal by
  improving the "Noodle" literal matcher.
- Small reductions in total stream state for many pattern sets.
- Improve runtime detection of AVX2 support.
- Disable -Werror for release builds, in order to behave better for packagers
  and users with different compiler combinations than those that we test.
- Improve support for building on Windows with MSVC 2015 (github issue #14).
  Support for Hyperscan on Windows is still experimental.
- Small updates to fix warnings identified by Coverity.
- Remove Python codegen for the "FDR" and "Teddy" literal matchers. These are
  now implemented directly in C code.
- Remove the specialist "Sidecar" engine in favour of using our more general
  repeat engines.
- New API function: add the `hs_expression_ext_info()` function. This is a
  variant of `hs_expression_info()` that can accept patterns with extended
  parameters.
- New API error value: add the `HS_SCRATCH_IN_USE` error, which is returned
  when Hyperscan detects that a scratch region is already in use on entry to an
  API function.

## [4.1.0] 2015-12-18
- Update version of PCRE used by testing tools as a syntax and semantic
  reference to PCRE 8.38.
- Small updates to fix warnings identified by Coverity.
- Clean up and unify exception handling behaviour across GPR and SIMD NFA
  models.
- Fix bug in handling of bounded repeat triggers with large gaps between them
  for sparse repeat model.
- Correctly reject POSIX collating elements (`[.ch.]`, `[=ch=]`) in the parser.
  These are not supported by Hyperscan.
- Add support for quoted sequences (`\Q...\E`) inside character classes.
- Simplify FDR literal matcher runtime by removing some static specialization.
- Fix handling of the POSIX `[:graph:]`, `[:print:]` and `[:punct:]` character
  classes to match the behaviour of PCRE 8.38 in both standard operation and
  with the UCP flag set. (Note: some bugs were fixed in this area in PCRE
  8.38.) Previously Hyperscan's behaviour was the same as versions of PCRE
  before 8.34.
- Improve performance when compiling pattern sets that include a large number
  of similar bounded repeat constructs. (github issue #9)

## [4.0.1] 2015-10-30
- Minor cleanups to test code.
- CMake and other build system improvements.
- API update: allow `hs_reset_stream()` and `hs_reset_and_copy_stream()` to be
  supplied with a NULL scratch pointer if no matches are required. This is in
  line with the behaviour of `hs_close_stream()`.
- Disallow bounded repeats with a very large minimum repeat but no maximum,
  i.e. {
    N,
} for very large N.
- Reduce compile memory usage in literal set explansion for some large cases.

## [4.0.0] 2015-10-20
- Original release of Hyperscan as open-source software.