github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-06-28 16:41:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

remove vermicelli.h and replace it with vermicelli.hpp

Browse Source
This commit is contained in:
Konstantinos Margaritis 2021-11-02 22:30:53 +02:00
parent 869d2bd53b
commit 210295a702
13 changed files with 7 additions and 210 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CMakeLists.txt
View File

@ -616,7 +616,7 @@ set (hs_exec_SRCS
src/nfa/tamarama_internal.h
src/nfa/truffle.cpp
src/nfa/truffle.h
src/nfa/vermicelli.h
src/nfa/vermicelli.hpp
src/nfa/vermicelli_run.h
src/nfa/vermicelli_simd.cpp
src/som/som.h

1
src/hwlm/hwlm.c
View File

@ -39,7 +39,6 @@
#include "nfa/accel.h"
#include "nfa/shufti.h"
#include "nfa/truffle.h"
#include "nfa/vermicelli.h"
#include "nfa/vermicelli.hpp"
#include <string.h>

1
src/nfa/accel.c
View File

@ -30,7 +30,6 @@
#include "accel.h"
#include "shufti.h"
#include "truffle.h"
#include "vermicelli.h"
#include "vermicelli.hpp"
#include "ue2common.h"

1
src/nfa/castle.c
View File

@ -40,7 +40,6 @@
#include "repeat.h"
#include "shufti.h"
#include "truffle.h"
#include "vermicelli.h"
#include "vermicelli.hpp"
#include "util/bitutils.h"
#include "util/multibit.h"

3
src/nfa/lbr.c
View File

@ -40,7 +40,6 @@
#include "repeat_internal.h"
#include "shufti.h"
#include "truffle.h"
#include "vermicelli.h"
#include "vermicelli.hpp"
#include "util/partial_store.h"
#include "util/unaligned.h"
@ -534,4 +533,4 @@ char lbrFwdScanTruf(const struct NFA *nfa, const u8 *buf,
#ifdef HAVE_SVE2
#include "lbr_sve.h"
#endif
#endif

2
src/nfa/limex_accel.c
View File

@ -40,7 +40,7 @@
#include "shufti.h"
#include "truffle.h"
#include "ue2common.h"
#include "vermicelli.h"
#include "vermicelli.hpp"
#include "util/arch.h"
#include "util/bitutils.h"
#include "util/simd_utils.h"

2
src/nfa/mpv.c
View File

@ -36,7 +36,7 @@
#include "shufti.h"
#include "truffle.h"
#include "ue2common.h"
#include "vermicelli.h"
#include "vermicelli.hpp"
#include "vermicelli_run.h"
#include "util/multibit.h"
#include "util/partial_store.h"

1
src/nfa/nfa_rev_api.h
View File

@ -35,7 +35,6 @@
#include "accel.h"
#include "nfa_internal.h"
#include "vermicelli.h"
#include "vermicelli.hpp"
#include "util/unaligned.h"

119
src/nfa/vermicelli.h
View File

@ -1,119 +0,0 @@
/*
* Copyright (c) 2015-2020, Intel Corporation
* Copyright (c) 2021, Arm Limited
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* * Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of Intel Corporation nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
/** \file
* \brief Vermicelli: single-byte and double-byte acceleration.
*/
#ifndef VERMICELLI_H
#define VERMICELLI_H
#include "util/bitutils.h"
#include "util/simd_utils.h"
#include "util/unaligned.h"
#if !defined(HAVE_AVX512)
#include "vermicelli_common.h"
#endif
#ifdef HAVE_SVE2
#include "vermicelli_sve.h"
#endif
static really_inline
const u8 *vermicelliDoubleMaskedExec(char c1, char c2, char m1, char m2,
const u8 *buf, const u8 *buf_end) {
DEBUG_PRINTF("double verm scan (\\x%02hhx&\\x%02hhx)(\\x%02hhx&\\x%02hhx) "
"over %zu bytes\n", c1, m1, c2, m2, (size_t)(buf_end - buf));
assert(buf < buf_end);
VERM_TYPE chars1 = VERM_SET_FN(c1);
VERM_TYPE chars2 = VERM_SET_FN(c2);
VERM_TYPE mask1 = VERM_SET_FN(m1);
VERM_TYPE mask2 = VERM_SET_FN(m2);
#ifdef HAVE_AVX512
if (buf_end - buf <= VERM_BOUNDARY) {
const u8 *ptr = dvermMiniMasked(chars1, chars2, mask1, mask2, buf,
buf_end);
if (ptr) {
return ptr;
}
/* check for partial match at end */
if ((buf_end[-1] & m1) == (u8)c1) {
DEBUG_PRINTF("partial!!!\n");
return buf_end - 1;
}
return buf_end;
}
#endif
assert((buf_end - buf) >= VERM_BOUNDARY);
uintptr_t min = (uintptr_t)buf % VERM_BOUNDARY;
if (min) {
// Input isn't aligned, so we need to run one iteration with an
// unaligned load, then skip buf forward to the next aligned address.
// There's some small overlap here, but we don't mind scanning it twice
// if we can do it quickly, do we?
const u8 *p = dvermPreconditionMasked(chars1, chars2, mask1, mask2, buf);
if (p) {
return p;
}
buf += VERM_BOUNDARY - min;
assert(buf < buf_end);
}
// Aligned loops from here on in
const u8 *ptr = dvermSearchAlignedMasked(chars1, chars2, mask1, mask2, c1,
c2, m1, m2, buf, buf_end);
if (ptr) {
return ptr;
}
// Tidy up the mess at the end
ptr = dvermPreconditionMasked(chars1, chars2, mask1, mask2,
buf_end - VERM_BOUNDARY);
if (ptr) {
return ptr;
}
/* check for partial match at end */
if ((buf_end[-1] & m1) == (u8)c1) {
DEBUG_PRINTF("partial!!!\n");
return buf_end - 1;
}
return buf_end;
}
#endif /* VERMICELLI_H */

79
src/nfa/vermicelli_common.h
View File

@ -1,79 +0,0 @@
/*
* Copyright (c) 2015-2020, Intel Corporation
* Copyright (c) 2021, Arm Limited
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* * Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of Intel Corporation nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
/** \file
* \brief Vermicelli: Implementation shared between architectures.
*
* (users should include vermicelli.h instead of this)
*/
#define VERM_BOUNDARY 16
#define VERM_TYPE m128
#define VERM_SET_FN set1_16x8
// returns NULL if not found
static really_inline
const u8 *dvermPreconditionMasked(m128 chars1, m128 chars2,
m128 mask1, m128 mask2, const u8 *buf) {
m128 data = loadu128(buf); // unaligned
m128 v1 = eq128(chars1, and128(data, mask1));
m128 v2 = eq128(chars2, and128(data, mask2));
u32 z = movemask128(and128(v1, rshiftbyte_m128(v2, 1)));
/* no fixup of the boundary required - the aligned run will pick it up */
if (unlikely(z)) {
u32 pos = ctz32(z);
return buf + pos;
}
return NULL;
}
static really_inline
const u8 *dvermSearchAlignedMasked(m128 chars1, m128 chars2,
m128 mask1, m128 mask2, u8 c1, u8 c2, u8 m1,
u8 m2, const u8 *buf, const u8 *buf_end) {
assert((size_t)buf % 16 == 0);
for (; buf + 16 < buf_end; buf += 16) {
m128 data = load128(buf);
m128 v1 = eq128(chars1, and128(data, mask1));
m128 v2 = eq128(chars2, and128(data, mask2));
u32 z = movemask128(and128(v1, rshiftbyte_m128(v2, 1)));
if ((buf[15] & m1) == c1 && (buf[16] & m2) == c2) {
z |= (1 << 15);
}
if (unlikely(z)) {
u32 pos = ctz32(z);
return buf + pos;
}
}
return NULL;
}

4
src/nfa/vermicelli_run.h
View File

@ -26,9 +26,11 @@
* POSSIBILITY OF SUCH DAMAGE.
*/
#include "vermicelli.h"
#include "vermicelli.hpp"
#define VERM_BOUNDARY 16
#define VERM_TYPE m128
static really_inline
const u8 *find_xverm_run(char c, char nocase, u32 repeat, UNUSED const u8 *buf,
const u8 *buf_start, const u8 *buf_end, char negate) {

1
unit/internal/rvermicelli.cpp
View File

@ -30,7 +30,6 @@
#include "config.h"
#include "gtest/gtest.h"
#include "nfa/vermicelli.h"
#include "nfa/vermicelli.hpp"
#define BOUND (~(VERM_BOUNDARY - 1))

1
unit/internal/vermicelli.cpp
View File

@ -30,7 +30,6 @@
#include "config.h"
#include "gtest/gtest.h"
#include "nfa/vermicelli.h"
#include "nfa/vermicelli.hpp"
TEST(Vermicelli, ExecNoMatch1) {