vectorscan/src/nfa/multivermicelli_avx2.h
2016-03-01 11:21:39 +11:00

284 lines
8.3 KiB
C

/*
* Copyright (c) 2015, Intel Corporation
*
* 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.
*/
#include "util/bitutils.h"
#include "util/simd_utils.h"
#include "util/unaligned.h"
#include "multiaccel_common.h"
static really_inline
const u8 *JOIN(MATCH_ALGO, vermUnalignNocase)(m256 chars,
const u8 *buf,
const u8 run_len
#ifdef MULTIACCEL_DOUBLE
, const u8 run_len2
#endif
) {
m256 casemask = set32x8(CASE_CLEAR);
const u8 *ptr;
m256 data = loadu256(buf);
u32 z = movemask256(eq256(chars, and256(casemask, data)));
ptr = (*JOIN4(MATCH_ALGO, match_funcs, _, 64)[run_len])
(buf, z
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
return NULL;
}
static really_inline
const u8 *JOIN(MATCH_ALGO, vermUnalign)(m256 chars,
const u8 *buf,
const u8 run_len
#ifdef MULTIACCEL_DOUBLE
, const u8 run_len2
#endif
) {
const u8 *ptr;
m256 data = loadu256(buf);
u32 z = movemask256(eq256(chars, data));
ptr = (*JOIN4(MATCH_ALGO, match_funcs, _, 64)[run_len])
(buf, z
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
return NULL;
}
/*
* 32-byte pipeline
*/
static really_inline
const u8 *JOIN(MATCH_ALGO, vermPipeline)(m256 chars,
const u8 *buf,
const u8 *buf_end,
const u8 run_len
#ifdef MULTIACCEL_DOUBLE
, const u8 run_len2
#endif
) {
const u8* ptr, *last_buf;
u32 last_res;
// pipeline prologue: scan first 32 bytes
m256 data = load256(buf);
u32 z = movemask256(eq256(chars, data));
last_res = z;
last_buf = buf;
buf += 32;
// now, start the pipeline!
assert((size_t)buf % 32 == 0);
for (; buf + 31 < buf_end; buf += 32) {
// scan more data
data = load256(buf);
z = movemask256(eq256(chars, data));
// do a comparison on previous result
ptr = (*JOIN4(MATCH_ALGO, match_funcs, _, 64)[run_len])
(last_buf, last_res
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
last_buf = buf;
last_res = z;
}
assert(buf <= buf_end && buf >= buf_end - 32);
// epilogue: compare final results
ptr = (*JOIN4(MATCH_ALGO, match_funcs, _, 64)[run_len])
(last_buf, last_res
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
return NULL;
}
/*
* 32-byte caseless pipeline
*/
static really_inline
const u8 *JOIN(MATCH_ALGO, vermPipelineNocase)(m256 chars,
const u8 *buf,
const u8 *buf_end,
const u8 run_len
#ifdef MULTIACCEL_DOUBLE
, const u8 run_len2
#endif
) {
m256 casemask = set32x8(CASE_CLEAR);
const u8* ptr, *last_buf;
u32 last_res;
// pipeline prologue: scan first 32 bytes
m256 data = load256(buf);
u32 z = movemask256(eq256(chars, and256(casemask, data)));
last_res = z;
last_buf = buf;
buf += 32;
// now, start the pipeline!
assert((size_t)buf % 32 == 0);
for (; buf + 31 < buf_end; buf += 32) {
// scan more data
data = load256(buf);
z = movemask256(eq256(chars, and256(casemask, data)));
// do a comparison on previous result
ptr = (*JOIN4(MATCH_ALGO, match_funcs, _, 64)[run_len])
(last_buf, last_res
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
last_buf = buf;
last_res = z;
}
assert(buf <= buf_end && buf >= buf_end - 32);
// epilogue: compare final results
ptr = (*JOIN4(MATCH_ALGO, match_funcs, _, 64)[run_len])
(last_buf, last_res
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
return NULL;
}
const u8 *JOIN(MATCH_ALGO, vermicelliExec)(char c, char nocase,
const u8 *buf,
const u8 *buf_end,
const u8 run_len
#ifdef MULTIACCEL_DOUBLE
, const u8 run_len2
#endif
) {
DEBUG_PRINTF("verm scan %s\\x%02hhx over %zu bytes\n",
nocase ? "nocase " : "", c, (size_t)(buf_end - buf));
assert(buf < buf_end);
const u8 *ptr;
// Handle small scans.
if (buf_end - buf < 32) {
for (; buf < buf_end; buf++) {
char cur = (char)*buf;
if (nocase) {
cur &= CASE_CLEAR;
}
if (cur == c) {
break;
}
}
return buf;
}
m256 chars = set32x8(c); /* nocase already uppercase */
uintptr_t min = (uintptr_t)buf % 32;
if (min) {
ptr = nocase ? JOIN(MATCH_ALGO, vermUnalignNocase)(chars,
buf, run_len
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
) : JOIN(MATCH_ALGO, vermUnalign)(chars,
buf, run_len
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
buf += 32 - min;
}
if (buf_end - buf >= 32){
ptr = nocase ? JOIN(MATCH_ALGO, vermPipelineNocase)(chars,
buf, buf_end, run_len
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
) : JOIN(MATCH_ALGO, vermPipeline)(chars,
buf, buf_end, run_len
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
if (unlikely(ptr)) {
return ptr;
}
}
// final unaligned scan
ptr = nocase ? JOIN(MATCH_ALGO, vermUnalignNocase)(chars,
buf_end - 32, run_len
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
) : JOIN(MATCH_ALGO, vermUnalign)(chars,
buf_end - 32, run_len
#ifdef MULTIACCEL_DOUBLE
, run_len2
#endif
);
// run our pipeline
return ptr ? ptr : buf_end;
}