aolo2
/
cparse
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

SIMD fast path for skipping long multi-line comments. ~190MB/sec

master
A.Olokhtonov 2 years ago
parent
c72eecac85
commit
ff1826e43e
2 changed files with 149 additions and 158 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 20
      input.c
  2. 287
      main.c

20
input.c
Unescape View File

@ -8,6 +8,8 @@ multi
line line
comment */ comment */
/* short com */
123 123
0707 0707
0xfAb 0xfAb
@ -71,10 +73,22 @@ u"\1\12\123"
U"\xffaab" U"\xffaab"
L"\uffFF\xff" L"\uffFF\xff"
[ ] ( ) { } . -> [] ( ) { } . ->
++ -- & * + - ~ ! ++ -- & * + - ~ !
/ % << >> < > <= >= / %
<<
>>
<
>
<=
>=
? : ; ... ? : ; ...
= *= /= %= += -= <<= = *= /= %= += -= <<=
, # ## , # ##
<: :> <% %> %: %:%: <:
:>
<%
%> %: %:%:
<stdio.h>
<stdio".h>

287
main.c
Unescape View File

@ -10,8 +10,9 @@
#include <immintrin.h> #include <immintrin.h>
#define MAX(a, b) ((a) > (b) ? (a) : (b))
enum token_kind { enum token_kind {
TOKEN_KEYWORD, TOKEN_KEYWORD,
TOKEN_IDENTIFIER, TOKEN_IDENTIFIER,
@ -58,41 +59,29 @@ advance(struct str *s, int by)
static int static int
nondigit(struct str s) nondigit(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') || (c == '_'));
return(('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') || (c == '_'));
}
return(0);
} }
static int static int
digit(struct str s) digit(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return('0' <= c && c <= '9');
return('0' <= c && c <= '9');
}
return(0);
} }
static int static int
nonzero_digit(struct str s) nonzero_digit(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return('1' <= c && c <= '9');
return('1' <= c && c <= '9');
}
return(0);
} }
static int static int
octal_digit(struct str s) octal_digit(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return('0' <= c && c <= '7');
return('0' <= c && c <= '7');
}
return(0);
} }
static int static int
@ -111,31 +100,22 @@ hexadecimal_prefix(struct str s)
static int static int
hexadecimal_digit(struct str s) hexadecimal_digit(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(digit(s) || ('A' <= c && c <= 'F') || ('a' <= c && c <= 'f'));
return(digit(s) || ('A' <= c && c <= 'F') || ('a' <= c && c <= 'f'));
}
return(0);
} }
static int static int
unsigned_suffix(struct str s) unsigned_suffix(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c == 'u' || c == 'U');
return(c == 'u' || c == 'U');
}
return(0);
} }
static int static int
long_suffix(struct str s) long_suffix(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c == 'l' || c == 'L');
return(c == 'l' || c == 'L');
}
return(0);
} }
static int static int
@ -154,21 +134,15 @@ long_long_suffix(struct str s)
static int static int
sign(struct str s) sign(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c == '+' || c == '-');
return(c == '+' || c == '-');
}
return(0);
} }
static int static int
floating_suffix(struct str s) floating_suffix(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c == 'f' || c == 'l' || c == 'F' || c == 'L');
return(c == 'f' || c == 'l' || c == 'F' || c == 'L');
}
return(0);
} }
static int static int
@ -291,11 +265,7 @@ hex_quad(struct str s)
static int static int
universal_character_name(struct str s) universal_character_name(struct str s)
{ {
if (s.size < 2) { if (s.size >= 2 && s.text[0] == '\\') {
return(0);
}
if (s.text[0] == '\\') {
if (s.text[1] == 'u') { if (s.text[1] == 'u') {
advance(&s, 2); advance(&s, 2);
int hq = hex_quad(s); int hq = hex_quad(s);
@ -382,20 +352,18 @@ octal_constant(struct str s)
{ {
int start = s.size; int start = s.size;
if (s.size) { if (s.text[0] == '0') {
if (s.text[0] == '0') { advance(&s, 1);
advance(&s, 1);
for (;;) {
for (;;) { if (octal_digit(s)) {
if (octal_digit(s)) { advance(&s, 1);
advance(&s, 1); } else {
} else { break;
break;
}
} }
return(start - s.size);
} }
return(start - s.size);
} }
return(0); return(0);
@ -471,16 +439,14 @@ fractional_constant(struct str s)
s.text += ds1; s.text += ds1;
s.size -= ds1; s.size -= ds1;
if (s.size) { if (s.text[0] == '.') {
if (s.text[0] == '.') { advance(&s, 1);
advance(&s, 1);
int ds2 = digit_sequence(s);
int ds2 = digit_sequence(s); advance(&s, ds2);
advance(&s, ds2);
if (ds1 > 0 || ds2 > 0) {
if (ds1 > 0 || ds2 > 0) { return(ds1 + ds2 + 1);
return(ds1 + ds2 + 1);
}
} }
} }
@ -492,18 +458,16 @@ exponent_part(struct str s)
{ {
int start = s.size; int start = s.size;
if (s.size) { if (s.text[0] == 'e' || s.text[0] == 'E') {
if (s.text[0] == 'e' || s.text[0] == 'E') { advance(&s, 1);
advance(&s, 1);
int sgn = sign(s);
int sgn = sign(s); advance(&s, sgn);
advance(&s, sgn);
int ds = digit_sequence(s);
int ds = digit_sequence(s); if (ds) {
if (ds) { advance(&s, ds);
advance(&s, ds); return(start - s.size);
return(start - s.size);
}
} }
} }
@ -564,7 +528,7 @@ hexadecimal_fractional_constant(struct str s)
advance(&s, hds1); advance(&s, hds1);
if (s.size && s.text[0] == '.') { if (s.text[0] == '.') {
advance(&s, 1); advance(&s, 1);
int hds2 = hexadecimal_digit_sequence(s); int hds2 = hexadecimal_digit_sequence(s);
@ -582,7 +546,7 @@ binary_exponent_part(struct str s)
{ {
int start = s.size; int start = s.size;
if (s.size && (s.text[0] == 'p' || s.text[0] == 'P')) { if (s.text[0] == 'p' || s.text[0] == 'P') {
advance(&s, 1); advance(&s, 1);
int sgn = sign(s); int sgn = sign(s);
@ -673,34 +637,23 @@ c_char(struct str s)
return(sym); return(sym);
} }
if (s.size) { char c = s.text[0];
char c = s.text[0];
return(c != '\'' && c != '\\' && c != '\n');
}
return(0); return(c != '\'' && c != '\\' && c != '\n');
} }
static int static int
h_char(struct str s) h_char(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c != '\n' && c != '>');
return(c != '\n' && c != '>');
}
return(0);
} }
static int static int
q_char(struct str s) q_char(struct str s)
{ {
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c != '\n' && c != '\"');
return(c != '\n' && c != '\"');
}
return(0);
} }
static int static int
@ -732,7 +685,7 @@ character_constant(struct str s)
int start = s.size; int start = s.size;
int ok = 0; int ok = 0;
if (s.size && s.text[0] == '\'') { if (s.text[0] == '\'') {
advance(&s, 1); advance(&s, 1);
ok = 1; ok = 1;
} else if (s.size >= 2 && s.text[0] == 'L' && s.text[1] == '\'') { } else if (s.size >= 2 && s.text[0] == 'L' && s.text[1] == '\'') {
@ -750,7 +703,7 @@ character_constant(struct str s)
int ccs = c_char_sequence(s); int ccs = c_char_sequence(s);
if (ccs) { if (ccs) {
advance(&s, ccs); advance(&s, ccs);
if (s.size && s.text[0] == '\'') { if (s.text[0] == '\'') {
advance(&s, 1); advance(&s, 1);
return(start - s.size); return(start - s.size);
} }
@ -777,8 +730,9 @@ static int
whitespace(struct str s) whitespace(struct str s)
{ {
int start = s.size; int start = s.size;
//int spaces = 0x20090a0d;
while (s.size && (s.text[0] == ' ' || s.text[0] == '\t' || s.text[0] == '\n' || s.text[0] == '\r')) { while (s.text[0] == ' ' || s.text[0] == '\t' || s.text[0] == '\n' || s.text[0] == '\r') {
advance(&s, 1); advance(&s, 1);
} }
@ -794,12 +748,8 @@ s_char(struct str s)
return(sym); return(sym);
} }
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c != '\"' && c != '\\' && c != '\n');
return(c != '\"' && c != '\\' && c != '\n');
}
return(0);
} }
static int static int
@ -832,14 +782,8 @@ encoding_prefix(struct str s)
return(2); return(2);
} }
if (s.size) { char c = s.text[0];
char c = s.text[0]; return(c == 'u' || c == 'U' || c == 'L');
if (c == 'u' || c == 'U' || c == 'L') {
return(1);
}
}
return(0);
} }
static int static int
@ -850,13 +794,13 @@ string_literal(struct str s)
int ep = encoding_prefix(s); int ep = encoding_prefix(s);
advance(&s, ep); advance(&s, ep);
if (s.size && s.text[0] == '\"') { if (s.text[0] == '\"') {
advance(&s, 1); advance(&s, 1);
int scs = s_char_sequence(s); int scs = s_char_sequence(s);
advance(&s, scs); advance(&s, scs);
if (s.size && s.text[0] == '\"') { if (s.text[0] == '\"') {
advance(&s, 1); advance(&s, 1);
return(start - s.size); return(start - s.size);
} }
@ -908,18 +852,16 @@ punctuator(struct str s)
} }
} }
if (s.size) { char c = s.text[0];
char c = s.text[0]; if (c == '[' || c == ']' || c == '(' || c == ')' ||
if (c == '[' || c == ']' || c == '(' || c == ')' || c == '{' || c == '}' || c == '.' || c == '&' ||
c == '{' || c == '}' || c == '.' || c == '&' || c == '*' || c == '+' || c == '-' || c == '~' ||
c == '*' || c == '+' || c == '-' || c == '~' || c == '!' || c == '/' || c == '%' || c == '<' ||
c == '!' || c == '/' || c == '%' || c == '<' || c == '>' || c == '^' || c == '|' || c == '?' ||
c == '>' || c == '^' || c == '|' || c == '?' || c == ':' || c == ';' || c == '=' || c == ',' ||
c == ':' || c == ';' || c == '=' || c == ',' || c == '#')
c == '#') {
{ return(1);
return(1);
}
} }
return(0); return(0);
@ -974,27 +916,27 @@ q_char_sequence(struct str s)
static int static int
header_name(struct str s) header_name(struct str s)
{ {
if (s.size && s.text[0] == '<') { if (s.text[0] == '<') {
advance(&s, 1); advance(&s, 1);
int hcs = h_char_sequence(s); int hcs = h_char_sequence(s);
if (hcs) { if (hcs) {
advance(&s, hcs); advance(&s, hcs);
if (s.size && s.text[0] == '>') { if (s.text[0] == '>') {
return(hcs + 2); return(hcs + 2);
} }
} }
} }
if (s.size && s.text[0] == '\"') { if (s.text[0] == '\"') {
advance(&s, 1); advance(&s, 1);
int qcs = q_char_sequence(s); int qcs = q_char_sequence(s);
if (qcs) { if (qcs) {
advance(&s, qcs); advance(&s, qcs);
if (s.size && s.text[0] == '\"') { if (s.text[0] == '\"') {
return(qcs + 2); return(qcs + 2);
} }
} }
@ -1028,6 +970,28 @@ comment(struct str s)
if (s.text[0] == '/' && s.text[1] == '*') { if (s.text[0] == '/' && s.text[1] == '*') {
/* multi-line comment */ /* multi-line comment */
advance(&s, 2);
__m128i mask = _mm_setr_epi8('*', '/', '*', '/', '*', '/', '*', '/', '*', '/', '*', '/', '*', '/', '*', '/');
__m128i mask_sus = _mm_setr_epi8(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, '*');
while (s.size > 16) {
__m128i chunk = _mm_loadu_si128((__m128i *)(s.text));
__m128i v1 = _mm_cmpeq_epi16(chunk, mask);
__m128i v2 = _mm_cmpeq_epi16(_mm_bslli_si128(chunk, 1), mask);
__m128i v3 = _mm_cmpeq_epi8(chunk, mask_sus);
__m128i v12 = _mm_or_si128(v1, v2);
__m128i v123 = _mm_or_si128(v12, v3);
if (!_mm_testz_si128(v123, v123)) {
break;
}
advance(&s, 16);
}
while (s.size) { while (s.size) {
if (s.size >= 2 && s.text[0] == '*' && s.text[1] == '/') { if (s.size >= 2 && s.text[0] == '*' && s.text[1] == '/') {
advance(&s, 2); advance(&s, 2);
@ -1057,36 +1021,42 @@ lex(char *text, int size)
if ((sym = comment(s))) { if ((sym = comment(s))) {
//printf("Comment: "); //printf("Comment: ");
} else if ((sym = constant(s))) { //printf("%.*s\n", sym, s.text);
//printf("Constant: "); } else {
int sym_constant = constant(s);
int sym_punctuator = punctuator(s);
int sym_string = string_literal(s);
int sym_header = header_name(s);
int sym_identifier = identifier(s);
sym = MAX(sym_constant, MAX(sym_punctuator, MAX(sym_string, MAX(sym_header, sym_identifier))));
#if 0 #if 0
} else if ((sym == pp_number(s))) { if (sym == sym_constant) {
printf("PP number: "); printf("Constant: ");
} else if (sym == sym_punctuator) {
printf("Punctuator: ");
} else if (sym == sym_string) {
printf("String: ");
} else if (sym == sym_header) {
printf("Header: ");
} else if (sym == sym_identifier) {
printf("Identifier: ");
}
#endif #endif
} else if ((sym = punctuator(s))) {
//printf("Punctuator: ");
} else if ((sym = string_literal(s))) {
//printf("String literal: ");
} else if ((sym = header_name(s))) {
//printf("Header name: ");
} else if ((sym = identifier(s))) {
//printf("Identifier: ");
} }
//printf("%.*s\n", sym, s.text);
if (sym) { if (sym) {
advance(&s, sym); advance(&s, sym);
} else if (s.size) { } else if (s.size == 1 && s.text[0] == '\0') {
break;
} else {
fprintf(stderr, "Error!\n"); fprintf(stderr, "Error!\n");
break; break;
} }
} }
return(NULL); return(NULL);
} }
@ -1169,6 +1139,13 @@ main(int argc, char **argv)
int size = (int) sb.st_size; int size = (int) sb.st_size;
char *data = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0); char *data = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
if (size && data[size - 1] != '\n') {
fprintf(stderr, "No terminating new line. Fuck you!\n");
return(1);
}
data[size - 1] = '\0';
if (data == MAP_FAILED) { if (data == MAP_FAILED) {
perror("mmap"); perror("mmap");
return(1); return(1);

Write Preview
Loading…
Cancel
Save