Skip to content

Commit d180a8a

Browse files
nodejs-github-botaduh95
nodejs-github-bot
authored and
aduh95
committed
deps: update simdutf to 5.6.3
PR-URL: #55973 Reviewed-By: Luigi Pinca <[email protected]> Reviewed-By: Rafael Gonzaga <[email protected]>
1 parent 1fb30d6 commit d180a8a

File tree

2 files changed

+59
-46
lines changed

2 files changed

+59
-46
lines changed

‎deps/simdutf/simdutf.cpp

+51-38
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
/* auto-generated on 2024-11-14 14:52:31 -0500. Do not edit! */
1+
/* auto-generated on 2024-11-21 10:33:28 -0500. Do not edit! */
22
/* begin file src/simdutf.cpp */
33
#include "simdutf.h"
44
// We include base64_tables once.
@@ -23495,7 +23495,7 @@ size_t encode_base64(char *dst, const char *src, size_t srclen,
2349523495
}
2349623496

2349723497
template <bool base64_url>
23498-
static inline uint64_t to_base64_mask(block64 *b, bool *error) {
23498+
static inline uint64_t to_base64_mask(block64 *b, uint64_t *error) {
2349923499
__m512i input = b->chunks[0];
2350023500
const __m512i ascii_space_tbl = _mm512_set_epi8(
2350123501
0, 0, 13, 12, 0, 10, 9, 0, 0, 0, 0, 0, 0, 0, 0, 32, 0, 0, 13, 12, 0, 10,
@@ -23538,7 +23538,7 @@ static inline uint64_t to_base64_mask(block64 *b, bool *error) {
2353823538
if (mask) {
2353923539
const __mmask64 spaces = _mm512_cmpeq_epi8_mask(
2354023540
_mm512_shuffle_epi8(ascii_space_tbl, input), input);
23541-
*error |= (mask != spaces);
23541+
*error = (mask ^ spaces);
2354223542
}
2354323543
b->chunks[0] = translated;
2354423544

@@ -23646,16 +23646,13 @@ compress_decode_base64(char *dst, const chartype *src, size_t srclen,
2364623646
block64 b;
2364723647
load_block(&b, src);
2364823648
src += 64;
23649-
bool error = false;
23649+
uint64_t error = 0;
2365023650
uint64_t badcharmask = to_base64_mask<base64_url>(&b, &error);
2365123651
if (error) {
2365223652
src -= 64;
23653-
while (src < srcend && scalar::base64::is_eight_byte(*src) &&
23654-
to_base64[uint8_t(*src)] <= 64) {
23655-
src++;
23656-
}
23657-
return {error_code::INVALID_BASE64_CHARACTER, size_t(src - srcinit),
23658-
size_t(dst - dstinit)};
23653+
size_t error_offset = _tzcnt_u64(error);
23654+
return {error_code::INVALID_BASE64_CHARACTER,
23655+
size_t(src - srcinit + error_offset), size_t(dst - dstinit)};
2365923656
}
2366023657
if (badcharmask != 0) {
2366123658
// optimization opportunity: check for simple masks like those made of
@@ -28240,7 +28237,7 @@ struct block64 {
2824028237
};
2824128238

2824228239
template <bool base64_url>
28243-
static inline uint32_t to_base64_mask(__m256i *src, bool *error) {
28240+
static inline uint32_t to_base64_mask(__m256i *src, uint32_t *error) {
2824428241
const __m256i ascii_space_tbl =
2824528242
_mm256_setr_epi8(0x20, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x9, 0xa,
2824628243
0x0, 0xc, 0xd, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x0, 0x0,
@@ -28324,17 +28321,19 @@ static inline uint32_t to_base64_mask(__m256i *src, bool *error) {
2832428321
if (mask) {
2832528322
__m256i ascii_space =
2832628323
_mm256_cmpeq_epi8(_mm256_shuffle_epi8(ascii_space_tbl, *src), *src);
28327-
*error |= (mask != _mm256_movemask_epi8(ascii_space));
28324+
*error = (mask ^ _mm256_movemask_epi8(ascii_space));
2832828325
}
2832928326
*src = out;
2833028327
return (uint32_t)mask;
2833128328
}
2833228329

2833328330
template <bool base64_url>
28334-
static inline uint64_t to_base64_mask(block64 *b, bool *error) {
28335-
*error = 0;
28336-
uint64_t m0 = to_base64_mask<base64_url>(&b->chunks[0], error);
28337-
uint64_t m1 = to_base64_mask<base64_url>(&b->chunks[1], error);
28331+
static inline uint64_t to_base64_mask(block64 *b, uint64_t *error) {
28332+
uint32_t err0 = 0;
28333+
uint32_t err1 = 0;
28334+
uint64_t m0 = to_base64_mask<base64_url>(&b->chunks[0], &err0);
28335+
uint64_t m1 = to_base64_mask<base64_url>(&b->chunks[1], &err1);
28336+
*error = err0 | ((uint64_t)err1 << 32);
2833828337
return m0 | (m1 << 32);
2833928338
}
2834028339

@@ -28466,16 +28465,13 @@ compress_decode_base64(char *dst, const chartype *src, size_t srclen,
2846628465
block64 b;
2846728466
load_block(&b, src);
2846828467
src += 64;
28469-
bool error = false;
28468+
uint64_t error = 0;
2847028469
uint64_t badcharmask = to_base64_mask<base64_url>(&b, &error);
2847128470
if (error) {
2847228471
src -= 64;
28473-
while (src < srcend && scalar::base64::is_eight_byte(*src) &&
28474-
to_base64[uint8_t(*src)] <= 64) {
28475-
src++;
28476-
}
28477-
return {error_code::INVALID_BASE64_CHARACTER, size_t(src - srcinit),
28478-
size_t(dst - dstinit)};
28472+
size_t error_offset = _tzcnt_u64(error);
28473+
return {error_code::INVALID_BASE64_CHARACTER,
28474+
size_t(src - srcinit + error_offset), size_t(dst - dstinit)};
2847928475
}
2848028476
if (badcharmask != 0) {
2848128477
// optimization opportunity: check for simple masks like those made of
@@ -37992,7 +37988,7 @@ struct block64 {
3799237988
};
3799337989

3799437990
template <bool base64_url>
37995-
static inline uint16_t to_base64_mask(__m128i *src, bool *error) {
37991+
static inline uint16_t to_base64_mask(__m128i *src, uint32_t *error) {
3799637992
const __m128i ascii_space_tbl =
3799737993
_mm_setr_epi8(0x20, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x9, 0xa, 0x0,
3799837994
0xc, 0xd, 0x0, 0x0);
@@ -38059,22 +38055,42 @@ static inline uint16_t to_base64_mask(__m128i *src, bool *error) {
3805938055
if (mask) {
3806038056
__m128i ascii_space =
3806138057
_mm_cmpeq_epi8(_mm_shuffle_epi8(ascii_space_tbl, *src), *src);
38062-
*error |= (mask != _mm_movemask_epi8(ascii_space));
38058+
*error = (mask ^ _mm_movemask_epi8(ascii_space));
3806338059
}
3806438060
*src = out;
3806538061
return (uint16_t)mask;
3806638062
}
3806738063

3806838064
template <bool base64_url>
38069-
static inline uint64_t to_base64_mask(block64 *b, bool *error) {
38070-
*error = 0;
38071-
uint64_t m0 = to_base64_mask<base64_url>(&b->chunks[0], error);
38072-
uint64_t m1 = to_base64_mask<base64_url>(&b->chunks[1], error);
38073-
uint64_t m2 = to_base64_mask<base64_url>(&b->chunks[2], error);
38074-
uint64_t m3 = to_base64_mask<base64_url>(&b->chunks[3], error);
38065+
static inline uint64_t to_base64_mask(block64 *b, uint64_t *error) {
38066+
uint32_t err0 = 0;
38067+
uint32_t err1 = 0;
38068+
uint32_t err2 = 0;
38069+
uint32_t err3 = 0;
38070+
uint64_t m0 = to_base64_mask<base64_url>(&b->chunks[0], &err0);
38071+
uint64_t m1 = to_base64_mask<base64_url>(&b->chunks[1], &err1);
38072+
uint64_t m2 = to_base64_mask<base64_url>(&b->chunks[2], &err2);
38073+
uint64_t m3 = to_base64_mask<base64_url>(&b->chunks[3], &err3);
38074+
*error = (err0) | ((uint64_t)err1 << 16) | ((uint64_t)err2 << 32) |
38075+
((uint64_t)err3 << 48);
3807538076
return m0 | (m1 << 16) | (m2 << 32) | (m3 << 48);
3807638077
}
3807738078

38079+
#if defined(_MSC_VER) && !defined(__clang__)
38080+
static inline size_t simdutf_tzcnt_u64(uint64_t num) {
38081+
unsigned long ret;
38082+
if (num == 0) {
38083+
return 64;
38084+
}
38085+
_BitScanForward64(&ret, num);
38086+
return ret;
38087+
}
38088+
#else // GCC or Clang
38089+
static inline size_t simdutf_tzcnt_u64(uint64_t num) {
38090+
return num ? __builtin_ctzll(num) : 64;
38091+
}
38092+
#endif
38093+
3807838094
static inline void copy_block(block64 *b, char *output) {
3807938095
_mm_storeu_si128(reinterpret_cast<__m128i *>(output), b->chunks[0]);
3808038096
_mm_storeu_si128(reinterpret_cast<__m128i *>(output + 16), b->chunks[1]);
@@ -38222,16 +38238,13 @@ compress_decode_base64(char *dst, const chartype *src, size_t srclen,
3822238238
block64 b;
3822338239
load_block(&b, src);
3822438240
src += 64;
38225-
bool error = false;
38241+
uint64_t error = 0;
3822638242
uint64_t badcharmask = to_base64_mask<base64_url>(&b, &error);
3822738243
if (error) {
3822838244
src -= 64;
38229-
while (src < srcend && scalar::base64::is_eight_byte(*src) &&
38230-
to_base64[uint8_t(*src)] <= 64) {
38231-
src++;
38232-
}
38233-
return {error_code::INVALID_BASE64_CHARACTER, size_t(src - srcinit),
38234-
size_t(dst - dstinit)};
38245+
size_t error_offset = simdutf_tzcnt_u64(error);
38246+
return {error_code::INVALID_BASE64_CHARACTER,
38247+
size_t(src - srcinit + error_offset), size_t(dst - dstinit)};
3823538248
}
3823638249
if (badcharmask != 0) {
3823738250
// optimization opportunity: check for simple masks like those made of

‎deps/simdutf/simdutf.h

+8-8
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
/* auto-generated on 2024-11-14 14:52:31 -0500. Do not edit! */
1+
/* auto-generated on 2024-11-21 10:33:28 -0500. Do not edit! */
22
/* begin file include/simdutf.h */
33
#ifndef SIMDUTF_H
44
#define SIMDUTF_H
@@ -155,11 +155,11 @@
155155
// RISC-V 64-bit
156156
#define SIMDUTF_IS_RISCV64 1
157157

158-
#if __clang_major__ >= 19
159-
// Does the compiler support target regions for RISC-V
160-
#define SIMDUTF_HAS_RVV_TARGET_REGION 1
161-
#endif
162-
158+
// #if __riscv_v_intrinsic >= 1000000
159+
// #define SIMDUTF_HAS_RVV_INTRINSICS 1
160+
// #define SIMDUTF_HAS_RVV_TARGET_REGION 1
161+
// #elif ...
162+
// Check for special compiler versions that implement pre v1.0 intrinsics
163163
#if __riscv_v_intrinsic >= 11000
164164
#define SIMDUTF_HAS_RVV_INTRINSICS 1
165165
#endif
@@ -670,7 +670,7 @@ SIMDUTF_DISABLE_UNDESIRED_WARNINGS
670670
#define SIMDUTF_SIMDUTF_VERSION_H
671671

672672
/** The version of simdutf being used (major.minor.revision) */
673-
#define SIMDUTF_VERSION "5.6.2"
673+
#define SIMDUTF_VERSION "5.6.3"
674674

675675
namespace simdutf {
676676
enum {
@@ -685,7 +685,7 @@ enum {
685685
/**
686686
* The revision (major.minor.REVISION) of simdutf being used.
687687
*/
688-
SIMDUTF_VERSION_REVISION = 2
688+
SIMDUTF_VERSION_REVISION = 3
689689
};
690690
} // namespace simdutf
691691

0 commit comments

Comments
 (0)