Skip to content

Commit b888a6a

Browse files
weijinrongyuanweijinrongyuan
authored
Update parse_with_padding.c
[optimize] Add arm sve intrinsic to replace x86 avx intrinsic for arm platform
1 parent 53b981b commit b888a6a

File tree

1 file changed

+134
-5
lines changed

1 file changed

+134
-5
lines changed

‎native/parse_with_padding.c

Lines changed: 134 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -587,6 +587,135 @@ static always_inline ssize_t handle_unicode(char** sp, char **dp) {
587587
return 0;
588588
}
589589

590+
#define get_string_block { \
591+
if (unlikely((opts & F_VALIDATE_STRING) != 0)) { \
592+
bsresult = svcmpeq_n_u8(pg, v, '\\'); \
593+
quoteresult = svcmpeq_n_u8(pg, v, '"'); \
594+
escresult = svcmpeq_n_u8(pg, v, '\x1f'); \
595+
} else { \
596+
bsresult = svcmpeq_n_u8(pg, v, '\\'); \
597+
quoteresult = svcmpeq_n_u8(pg, v, '"'); \
598+
escresult = svpfalse_b(); \
599+
}}
600+
601+
#define has_first_quote { \
602+
svbool_t res = svorr_b_z(svptrue_b8(), bsresult, escresult); \
603+
if (!svptest_any(svptrue_b8(), res)) { \
604+
temp = quoteresult; \
605+
} else { \
606+
temp = svbrkb_b_z(svptrue_b8(), res); \
607+
temp = svand_b_z(svptrue_b8(), temp, quoteresult); \
608+
} \
609+
}
610+
611+
#define has_first_backslash { \
612+
if (!svptest_any(svptrue_b8(), quoteresult)) { \
613+
temp = bsresult; \
614+
} else { \
615+
temp = svbrkb_b_z(svptrue_b8(), quoteresult); \
616+
temp = svand_b_z(svptrue_b8(), temp, bsresult); \
617+
} \
618+
}
619+
620+
#define has_first_unescaped { \
621+
if (!svptest_any(svptrue_b8(), quoteresult)) { \
622+
temp = escresult; \
623+
} else { \
624+
temp = svbrkb_b_z(svptrue_b8(), quoteresult); \
625+
temp = svand_b_z(svptrue_b8(), temp, escresult); \
626+
} \
627+
}
628+
629+
static always_inline long parse_string_inplace_sve(uint8_t** cur, bool* has_esc, uint64_t opts) {
630+
uint8_t* start = *cur;
631+
svbool_t bsresult;
632+
svbool_t quoteresult;
633+
svbool_t escresult;
634+
svbool_t temp;
635+
svbool_t pg = svptrue_b8();
636+
svuint8_t v;
637+
while (true) {
638+
v = svld1_u8(pg, (uint8_t*)(*cur));
639+
get_string_block;
640+
has_first_quote;
641+
if (svptest_any(svptrue_b8(), temp)) {
642+
*cur += svcntp_b8(svptrue_b8(), svbrkb_b_z(svptrue_b8(), quoteresult)) + 1; // skip the quote char
643+
*has_esc = false;
644+
return *cur - start - 1;
645+
}
646+
has_first_backslash;
647+
if (unlikely(svptest_any(svptrue_b8(), temp))) {
648+
break;
649+
}
650+
has_first_unescaped;
651+
if (unlikely((opts & F_VALIDATE_STRING) != 0 && svptest_any(svptrue_b8(), temp))) {
652+
*cur += svcntp_b8(svptrue_b8(), svbrkb_b_z(svptrue_b8(), escresult));
653+
return -SONIC_CONTROL_CHAR;
654+
}
655+
*cur += 32;
656+
}
657+
// deal with the escaped string
658+
*has_esc = true;
659+
has_first_backslash;
660+
*cur += svcntp_b8(svptrue_b8(), svbrkb_b_z(svptrue_b8(), bsresult));
661+
uint8_t* dst = *cur;
662+
uint8_t esc;
663+
escape:
664+
esc = *(*cur + 1);
665+
if (likely(esc) != 'u') {
666+
if (unlikely(ESCAPED_TAB[esc]) == 0) {
667+
return -SONIC_INVALID_ESCAPED;
668+
}
669+
*cur += 2;
670+
*dst++ = ESCAPED_TAB[esc];
671+
} else if (handle_unicode((char**)cur, (char**)&dst) != 0) {
672+
return -SONIC_INVALID_ESCAPED_UTF;
673+
}
674+
// check continuous escaped char
675+
if (**cur == '\\') {
676+
goto escape;
677+
}
678+
find_and_move:
679+
v = svld1_u8(pg, (uint8_t *)(*cur));
680+
get_string_block;
681+
has_first_quote;
682+
if (svptest_any(svptrue_b8(), temp)) {
683+
while (true) {
684+
repeat_8( {
685+
if (**cur != '"') {
686+
*dst++ = **cur;
687+
*cur += 1;
688+
} else {
689+
*cur += 1;
690+
return dst - start;
691+
}
692+
});
693+
}
694+
}
695+
has_first_unescaped;
696+
if (unlikely((opts & F_VALIDATE_STRING) != 0 && svptest_any(svptrue_b8(), temp))) {
697+
*cur += svcntp_b8(svptrue_b8(), svbrkb_b_z(svptrue_b8(), escresult));
698+
return -SONIC_CONTROL_CHAR;
699+
}
700+
has_first_backslash;
701+
if (svptest_any(svptrue_b8(), temp)) {
702+
while (true) {
703+
repeat_8( {
704+
if (**cur != '\\') {
705+
*dst++ = **cur;
706+
*cur += 1;
707+
} else {
708+
goto escape;
709+
}
710+
});
711+
}
712+
}
713+
svst1_u8(svptrue_b8(), dst, v);
714+
*cur += 32;
715+
dst += 32;
716+
goto find_and_move;
717+
}
718+
590719
// positive is length
591720
// negative is - error_code
592721
static always_inline long parse_string_inplace(uint8_t** cur, bool* has_esc, uint64_t opts) {
@@ -1112,7 +1241,7 @@ static always_inline error_code parse(GoParser* slf, reader* rdr, visitor* vis)
11121241
neg = false;
11131242
break;
11141243
case '"':
1115-
slen = parse_string_inplace(cur, &has_esc, slf->opt);
1244+
slen = parse_string_inplace_sve(cur, &has_esc, slf->opt);
11161245
if (slen < 0) {
11171246
err = (error_code)(-slen);
11181247
}
@@ -1149,7 +1278,7 @@ static always_inline error_code parse(GoParser* slf, reader* rdr, visitor* vis)
11491278
// parse key
11501279
pos = offset_from(*cur);
11511280

1152-
slen = parse_string_inplace(cur, &has_esc, slf->opt);
1281+
slen = parse_string_inplace_sve(cur, &has_esc, slf->opt);
11531282
if (slen < 0) {
11541283
err = (error_code)(-slen);
11551284
return err;
@@ -1205,7 +1334,7 @@ static always_inline error_code parse(GoParser* slf, reader* rdr, visitor* vis)
12051334
neg = false;
12061335
break;
12071336
case '"':
1208-
slen = parse_string_inplace(cur, &has_esc, slf->opt);
1337+
slen = parse_string_inplace_sve(cur, &has_esc, slf->opt);
12091338
if (slen < 0) {
12101339
err = (error_code)(-slen);
12111340
}
@@ -1293,7 +1422,7 @@ static always_inline error_code parse(GoParser* slf, reader* rdr, visitor* vis)
12931422
neg = false;
12941423
break;
12951424
case '"':
1296-
slen = parse_string_inplace(cur, &has_esc, slf->opt);
1425+
slen = parse_string_inplace_sve(cur, &has_esc, slf->opt);
12971426
if (slen < 0) {
12981427
err = (error_code)(-slen);
12991428
}
@@ -1379,4 +1508,4 @@ long parse_with_padding(void* p) {
13791508
}
13801509

13811510
return err;
1382-
}
1511+
}

0 commit comments

Comments
 (0)