--- /dev/null
+/*
+ * Copyright (c) 2015-2017, Intel Corporation
+ * Copyright (c) 2020-2021, VectorCamp PC
+ * 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 Shufti: character class acceleration.
+ *
+ */
+
+template <uint16_t S>
+static really_inline
+const SuperVector<S> blockSingleMask(SuperVector<S> mask_lo, SuperVector<S> mask_hi, SuperVector<S> chars) {
+ const SuperVector<S> low4bits = SuperVector<S>::dup_u8(0xf);
+
+ SuperVector<S> c_lo = chars & low4bits;
+ SuperVector<S> c_hi = chars.template vshr_8_imm<4>();
+ c_lo = mask_lo.template pshufb<false>(c_lo);
+ c_hi = mask_hi.template pshufb<false>(c_hi);
+
+ return (c_lo & c_hi) > (SuperVector<S>::Zeroes());
+}
+
+template <uint16_t S>
+static really_inline
+SuperVector<S> blockDoubleMask(SuperVector<S> mask1_lo, SuperVector<S> mask1_hi, SuperVector<S> mask2_lo, SuperVector<S> mask2_hi, SuperVector<S> chars) {
+
+ const SuperVector<S> low4bits = SuperVector<S>::dup_u8(0xf);
+ SuperVector<S> chars_lo = chars & low4bits;
+ chars_lo.print8("chars_lo");
+ SuperVector<S> chars_hi = chars.template vshr_64_imm<4>() & low4bits;
+ chars_hi.print8("chars_hi");
+ SuperVector<S> c1_lo = mask1_lo.template pshufb<true>(chars_lo);
+ c1_lo.print8("c1_lo");
+ SuperVector<S> c1_hi = mask1_hi.template pshufb<true>(chars_hi);
+ c1_hi.print8("c1_hi");
+ SuperVector<S> t1 = c1_lo | c1_hi;
+ t1.print8("t1");
+
+ SuperVector<S> c2_lo = mask2_lo.template pshufb<true>(chars_lo);
+ c2_lo.print8("c2_lo");
+ SuperVector<S> c2_hi = mask2_hi.template pshufb<true>(chars_hi);
+ c2_hi.print8("c2_hi");
+ SuperVector<S> t2 = c2_lo | c2_hi;
+ t2.print8("t2");
+ t2.template vshr_128_imm<1>().print8("t2.vshr_128(1)");
+ SuperVector<S> t = t1 | (t2.template vshr_128_imm<1>());
+ t.print8("t");
+
+ return !t.eq(SuperVector<S>::Ones());
+}
* Utilises the SSSE3 pshufb shuffle instruction
*/
+#include <algorithm>
+
#include "shufti.h"
#include "ue2common.h"
#include "util/arch.h"
#include "util/supervector/supervector.hpp"
#include "util/match.hpp"
-#include <asm/unistd.h>
-#include <linux/perf_event.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <sys/ioctl.h>
-#include <unistd.h>
-
-#include <inttypes.h>
-#include <sys/types.h>
-
template <uint16_t S>
static really_inline
-const SuperVector<S> blockSingleMask(SuperVector<S> mask_lo, SuperVector<S> mask_hi, SuperVector<S> chars) {
- const SuperVector<S> low4bits = SuperVector<S>::dup_u8(0xf);
-
- SuperVector<S> c_lo = chars & low4bits;
- SuperVector<S> c_hi = chars.template vshr_8_imm<4>();
- c_lo = mask_lo.template pshufb<false>(c_lo);
- c_hi = mask_hi.template pshufb<false>(c_hi);
-
- return (c_lo & c_hi) > (SuperVector<S>::Zeroes());
-}
-
+const SuperVector<S> blockSingleMask(SuperVector<S> mask_lo, SuperVector<S> mask_hi, SuperVector<S> chars);
template <uint16_t S>
static really_inline
-SuperVector<S> blockDoubleMask(SuperVector<S> mask1_lo, SuperVector<S> mask1_hi, SuperVector<S> mask2_lo, SuperVector<S> mask2_hi, SuperVector<S> chars) {
-
- const SuperVector<S> low4bits = SuperVector<S>::dup_u8(0xf);
- SuperVector<S> chars_lo = chars & low4bits;
- chars_lo.print8("chars_lo");
- SuperVector<S> chars_hi = chars.template vshr_64_imm<4>() & low4bits;
- chars_hi.print8("chars_hi");
- SuperVector<S> c1_lo = mask1_lo.template pshufb<true>(chars_lo);
- c1_lo.print8("c1_lo");
- SuperVector<S> c1_hi = mask1_hi.template pshufb<true>(chars_hi);
- c1_hi.print8("c1_hi");
- SuperVector<S> t1 = c1_lo | c1_hi;
- t1.print8("t1");
-
- SuperVector<S> c2_lo = mask2_lo.template pshufb<true>(chars_lo);
- c2_lo.print8("c2_lo");
- SuperVector<S> c2_hi = mask2_hi.template pshufb<true>(chars_hi);
- c2_hi.print8("c2_hi");
- SuperVector<S> t2 = c2_lo | c2_hi;
- t2.print8("t2");
- t2.template vshr_128_imm<1>().print8("t2.vshr_128(1)");
- SuperVector<S> t = t1 | (t2.template vshr_128_imm<1>());
- t.print8("t");
-
- return !t.eq(SuperVector<S>::Ones());
-}
+SuperVector<S> blockDoubleMask(SuperVector<S> mask1_lo, SuperVector<S> mask1_hi, SuperVector<S> mask2_lo, SuperVector<S> mask2_hi, SuperVector<S> chars);
+
+#if defined(ARCH_IA32) || defined(ARCH_X86_64)
+#include "x86/shufti.hpp"
+#elif defined(ARCH_ARM32) || defined(ARCH_AARCH64)
+#include "arm/shufti.hpp"
+#endif
template <uint16_t S>
static really_inline
d = ROUNDUP_PTR(d, S);
}
- while(d + S <= buf_end) {
+ while(d + S <= buf_end) {
__builtin_prefetch(d + 64);
DEBUG_PRINTF("d %p \n", d);
SuperVector<S> chars = SuperVector<S>::load(d);
rv = fwdBlock(wide_mask_lo, wide_mask_hi, chars, d);
if (rv) return rv;
- d += S;
+ d += S;
}
}
// finish off tail
if (d != buf_end) {
- SuperVector<S> chars = SuperVector<S>::loadu(d);
+ SuperVector<S> chars = SuperVector<S>::loadu_maskz(d, buf_end - d);
rv = fwdBlock(wide_mask_lo, wide_mask_hi, chars, d);
DEBUG_PRINTF("rv %p \n", rv);
- if (rv) return rv;
+ if (rv && rv < buf_end) return rv;
}
return buf_end;
SuperVector<S> chars = SuperVector<S>::loadu(buf);
rv = revBlock(wide_mask_lo, wide_mask_hi, chars, buf);
DEBUG_PRINTF("rv %p \n", rv);
- if (rv) return rv;
+ if (rv && rv < buf_end) return rv;
}
return buf - 1;
d = ROUNDUP_PTR(d, S);
}
- while(d + S <= buf_end) {
+ while(d + S <= buf_end) {
__builtin_prefetch(d + 64);
DEBUG_PRINTF("d %p \n", d);
SuperVector<S> chars = SuperVector<S>::load(d);
rv = fwdBlockDouble(wide_mask1_lo, wide_mask1_hi, wide_mask2_lo, wide_mask2_hi, chars, d);
if (rv) return rv;
- d += S;
+ d += S;
}
}
// finish off tail
if (d != buf_end) {
- SuperVector<S> chars = SuperVector<S>::loadu(buf_end - S);
- rv = fwdBlockDouble(wide_mask1_lo, wide_mask1_hi, wide_mask2_lo, wide_mask2_hi, chars, buf_end - S);
+ SuperVector<S> chars = SuperVector<S>::loadu(d);
+ rv = fwdBlockDouble(wide_mask1_lo, wide_mask1_hi, wide_mask2_lo, wide_mask2_hi, chars, d);
DEBUG_PRINTF("rv %p \n", rv);
- if (rv) return rv;
+ if (rv && rv < buf_end) return rv;
}
return buf_end;
--- /dev/null
+/*
+ * Copyright (c) 2015-2017, Intel Corporation
+ * Copyright (c) 2020-2021, VectorCamp PC
+ * 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 Shufti: character class acceleration.
+ *
+ */
+
+#ifndef SHUFTI_SIMD_X86_HPP
+#define SHUFTI_SIMD_X86_HPP
+
+#include "util/supervector/supervector.hpp"
+#include "util/match.hpp"
+
+template <uint16_t S>
+static really_inline
+const SuperVector<S> blockSingleMask(SuperVector<S> mask_lo, SuperVector<S> mask_hi, SuperVector<S> chars) {
+ const SuperVector<S> low4bits = SuperVector<S>::dup_u8(0xf);
+
+ SuperVector<S> c_lo = chars & low4bits;
+ SuperVector<S> c_hi = chars.template vshr_64_imm<4>() & low4bits;
+ c_lo = mask_lo.template pshufb(c_lo);
+ c_hi = mask_hi.template pshufb(c_hi);
+
+ SuperVector c = c_lo & c_hi;
+
+ return c.eq(SuperVector<S>::Zeroes());
+}
+
+template <uint16_t S>
+static really_inline
+SuperVector<S> blockDoubleMask(SuperVector<S> mask1_lo, SuperVector<S> mask1_hi, SuperVector<S> mask2_lo, SuperVector<S> mask2_hi, SuperVector<S> chars) {
+
+ const SuperVector<S> low4bits = SuperVector<S>::dup_u8(0xf);
+ SuperVector<S> chars_lo = chars & low4bits;
+ chars_lo.print8("chars_lo");
+ SuperVector<S> chars_hi = low4bits.opandnot(chars).template vshr_64_imm<4>();
+ chars_hi.print8("chars_hi");
+ SuperVector<S> c1_lo = mask1_lo.pshufb(chars_lo);
+ c1_lo.print8("c1_lo");
+ SuperVector<S> c1_hi = mask1_hi.pshufb(chars_hi);
+ c1_hi.print8("c1_hi");
+ SuperVector<S> c1 = c1_lo | c1_hi;
+ c1.print8("c1");
+
+ SuperVector<S> c2_lo = mask2_lo.pshufb(chars_lo);
+ c2_lo.print8("c2_lo");
+ SuperVector<S> c2_hi = mask2_hi.pshufb(chars_hi);
+ c2_hi.print8("c2_hi");
+ SuperVector<S> c2 = c2_lo | c2_hi;
+ c2.print8("c2");
+ c2.template vshr_128_imm<1>().print8("c2.vshr_128(1)");
+ SuperVector<S> c = c1 | (c2.template vshr_128_imm<1>());
+ c.print8("c");
+
+ return c.eq(SuperVector<S>::Ones());
+}
+
+#endif // SHUFTI_SIMD_X86_HPP
projects / thirdparty / vectorscan.git / commitdiff
