Justin Viiret
9eb349a343
rose: expose smwr builder, tidy up engine build
2016-08-10 14:59:10 +10:00
Justin Viiret
a427a2843b
rose_build_anchored: clean up remapping
...
Note that there are no EOD reports in the anchored matcher raw_dfas.
2016-08-10 14:59:10 +10:00
Justin Viiret
210246af01
rose_build_anchored: remove unused forward decls
2016-08-10 14:59:10 +10:00
Justin Viiret
8754cbbd24
rose: use program offset, not final_id, in atable
...
This removes the need to look up the program offset in a table when
handling an anchored literal match.
2016-08-10 14:59:10 +10:00
Justin Viiret
4dbbc4eaa5
rose: add RECORD_ANCHORED instruction to program
...
Moves recordAnchoredLiteralMatch from an unconditional call in the
anchored callback to being driven by a program instruction.
2016-08-10 14:59:10 +10:00
Alex Coyte
981b59fd05
minor eager prefixes improvements
...
- count eager prefixes as always run engine when comparing with smwr
- only check if a prefix is vacuous after adding back literal fragments
2016-08-10 14:59:10 +10:00
Xiang Wang
e95a251935
UE-2991: avoid copying one unnecessary subqueue item
2016-08-10 14:59:06 +10:00
Justin Viiret
aed2e721f4
rose: tidy up CHECK_{BYTE,MASK} dump code
2016-08-10 14:57:48 +10:00
Xu, Chi
4d7469392d
rose: add CHECK_BYTE/CHECK_MASK instructions
...
These instructions are specialisations of the "lookaround" code for
performance.
2016-08-10 14:57:48 +10:00
Justin Viiret
1bab10698f
rose_build_convert: improve history selection
...
Fixes assertion failures introduced by last commit.
2016-08-10 14:57:42 +10:00
Justin Viiret
3e96cd48ef
rose: sanity check CHECK_BOUNDS instruction
2016-08-10 14:57:36 +10:00
Justin Viiret
8d31607556
rose: only use anch history when there are bounds
2016-08-10 14:57:36 +10:00
Alex Coyte
f55e968692
tamarama dump: use correct base offset
2016-08-10 14:57:20 +10:00
Justin Viiret
6ec93a54c4
rose: only dump small-block table in block mode
2016-08-10 14:57:20 +10:00
Justin Viiret
39c6a0c7bf
rose: check literal bounds when building SB table
...
Literals that cannot lead to a report in the first ROSE_SMALL_BLOCK_LEN
bytes may be dropped from the small block table.
2016-08-10 14:56:12 +10:00
Justin Viiret
9346a9090e
fdr: remove groups from struct FDR_Runtime_Args
2016-08-10 14:55:52 +10:00
Justin Viiret
42f23c2c91
teddy: no need to write control out at the end
2016-08-10 14:55:51 +10:00
Justin Viiret
b6a77b7329
teddy: remove extra control ptr
2016-08-10 14:55:51 +10:00
Justin Viiret
85f049edb2
fdr: remove extra control ptr
2016-08-10 14:55:51 +10:00
Justin Viiret
49bb3b5c82
simd_utils: setbit/clearbit by loading 1-bit mask
2016-08-10 14:52:56 +10:00
Justin Viiret
790683b641
rose: don't always dedupe small-block lit variants
2016-08-10 14:52:56 +10:00
Alex Coyte
c58424cab3
reduce character classes for hwlm accel
2016-08-10 14:52:56 +10:00
Alex Coyte
69933edf68
truffle hwlm accel
2016-08-10 14:52:56 +10:00
Alex Coyte
691b08d170
use NGHolder::foo in favour of NFAGraph::foo
2016-08-10 14:52:56 +10:00
Alex Coyte
2471b770a8
we no longer store the history len
2016-08-10 14:52:56 +10:00
Alex Coyte
3a1429a621
group_weak_end is no longer used
2016-08-10 14:52:56 +10:00
Matthew Barr
22b451b59b
Ensure that m256 is 32-aligned on non-avx2 builds
2016-08-10 14:52:56 +10:00
Justin Viiret
d497a1259a
rose: use normal callback for pure-literal cases
2016-08-10 14:52:56 +10:00
Justin Viiret
8435f918d1
limex: use cyclics mask for repeat compress
2016-08-10 14:52:56 +10:00
Justin Viiret
ae5e347778
limex: only compress active repeats
2016-08-10 14:52:56 +10:00
Justin Viiret
aa54352a3a
limex: switch off stale repeats during compress
2016-08-10 14:52:56 +10:00
Justin Viiret
19e79be87d
limex: compress repeats before state
2016-08-10 14:52:56 +10:00
Justin Viiret
7cc5346c11
assign groups: turn pair<pair> into a tuple
2016-08-10 14:49:24 +10:00
Justin Viiret
bfaa0acaea
rose: preserve lit properties when building masks
...
This fixes a bug with commit 6a6b0e5, which did not preserve the
requires_explode and requires_benefits properties when a new literal was
generated to add an HWLM and/cmp mask.
Also extends the requires_explode handling to allow masked literals.
2016-08-10 14:48:29 +10:00
Boris Nagaev
b73bd9b6e1
new test: deserializers fail with garbage input
2016-08-10 14:08:59 +10:00
Boris Nagaev
66c2a53d33
hs_serialized_database_info: check whole input
...
Most lines of hs_serialized_database_info are not needed,
as the information is provided by db_decode_header.
Moreover, db_decode_header changes its first argument,
so it points to end of header after db_decode_header.
2016-08-10 13:27:44 +10:00
Boris Nagaev
373a624bad
simplegrep: open file in binary mode ("rb")
...
Otherwise it hangs on binary files (platform MinGW).
2016-08-10 13:20:26 +10:00
Justin Viiret
e9cfbae68f
workaround for freebsd/clang/libc++ build issues
...
Rather than relying on set's constructor from {}, explicitly construct
the set.
2016-07-08 11:07:51 +10:00
Matthew Barr
e3d416a6ea
Apply some consistency to the names we give shifts
2016-07-08 11:07:50 +10:00
Matthew Barr
c76ff285e7
remove unnecessary function proto
2016-07-08 11:07:50 +10:00
Matthew Barr
9c915cc936
remove only use of cmpmsk8 and unused cmpmsk16
2016-07-08 11:07:50 +10:00
Matthew Barr
0722b5db5b
Remove GCC-style compound statements
...
These do not appear to give us benefits over inlining on recent compilers.
2016-07-08 11:07:50 +10:00
Matthew Barr
1b3e795fc9
teddy: we only need the upper lane
...
Just use an extract, no need to shuffle first.
2016-07-08 11:07:50 +10:00
Matthew Barr
adf820bbba
simd: simplify the set-all-ones util funcs
...
Modern compilers (gcc, icc) get this right, with the benefit of
removing our last use of inline asm in this file.
2016-07-08 11:07:50 +10:00
Matthew Barr
4d6934fc77
Move limex specific shuffle utils and ssse3 funcs
2016-07-08 11:07:50 +10:00
Justin Viiret
9f98f4c7b2
nfa: standardise callback start, end naming
2016-07-08 11:02:05 +10:00
Justin Viiret
cf9e40ae1c
nfa: unify NfaCallback and SomNfaCallback
...
Use just one callback type, with both start and end offsets.
2016-07-08 11:01:56 +10:00
Xiang Wang
9087d59be5
tamarama: add container engine for exclusive nfas
...
Add the new Tamarama engine that acts as a container for infix/suffix
engines that can be proven to run exclusively of one another.
This reduces stream state for pattern sets with many exclusive engines.
2016-07-08 11:01:34 +10:00
Justin Viiret
013dbd3b3c
rose: re-inline literal handling program exec
2016-07-08 11:01:34 +10:00
Justin Viiret
76d96809f8
rose: move roseRunProgram into its own unit
...
The roseRunProgram function had gotten very large for the number of
sites it was being inlined into, with negative effects on performance in
large cases. This change moves it into its own translation unit.
2016-07-08 11:01:34 +10:00