walkero
/
pcre2
1
0
Fork 0
Code Issues Pull Requests Projects Releases 1 Wiki Activity

Improve star and upto performance in JIT when unicode is disabled.

This commit is contained in:
Zoltán Herczeg 2015-06-21 05:10:48 +00:00
parent bac822ba58
commit c1355dedc2
1 changed files with 94 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

123
src/pcre2_jit_compile.c
View File

@ -8086,7 +8086,8 @@ BOOL charpos_enabled;
PCRE2_UCHAR charpos_char; PCRE2_UCHAR charpos_char;
unsigned int charpos_othercasebit; unsigned int charpos_othercasebit;
PCRE2_SPTR end; PCRE2_SPTR end;
jump_list *nomatch = NULL; jump_list *no_match = NULL;
jump_list *no_char1_match = NULL;
struct sljit_jump *jump = NULL; struct sljit_jump *jump = NULL;
struct sljit_label *label; struct sljit_label *label;
int private_data_ptr = PRIVATE_DATA(cc); int private_data_ptr = PRIVATE_DATA(cc);
@ -8242,13 +8243,13 @@ switch(opcode)
if (opcode == OP_UPTO) if (opcode == OP_UPTO)
{ {
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1); OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
add_jump(compiler, &nomatch, JUMP(SLJIT_ZERO)); add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
} }
/* Search the last instance of charpos_char. */ /* Search the last instance of charpos_char. */
label = LABEL(); label = LABEL();
compile_char1_matchingpath(common, type, cc, &nomatch, FALSE); compile_char1_matchingpath(common, type, cc, &no_match, FALSE);
detect_partial_match(common, &nomatch); detect_partial_match(common, &no_match);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
if (charpos_othercasebit != 0) if (charpos_othercasebit != 0)
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, charpos_othercasebit); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, charpos_othercasebit);
@ -8263,8 +8264,22 @@ switch(opcode)
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
JUMPHERE(jump); JUMPHERE(jump);
} }
if (opcode == OP_UPTO)
{
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
else
JUMPTO(SLJIT_JUMP, label);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
} }
else #if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
else if (common->utf)
{ {
if (private_data_ptr == 0) if (private_data_ptr == 0)
allocate_stack(common, 2); allocate_stack(common, 2);
@ -8276,24 +8291,45 @@ switch(opcode)
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max); OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
label = LABEL(); label = LABEL();
compile_char1_matchingpath(common, type, cc, &nomatch, TRUE); compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
}
if (opcode == OP_STAR) if (opcode == OP_UPTO)
JUMPTO(SLJIT_JUMP, label); {
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
else
JUMPTO(SLJIT_JUMP, label);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
}
#endif
else else
{ {
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1); if (private_data_ptr == 0)
JUMPTO(SLJIT_NOT_ZERO, label); allocate_stack(common, 2);
}
set_jumps(nomatch, LABEL()); OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); if (opcode == OP_UPTO)
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
if (charpos_enabled) label = LABEL();
{ detect_partial_match(common, &no_match);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
if (opcode == OP_UPTO)
{
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
else
JUMPTO(SLJIT_JUMP, label);
set_jumps(no_char1_match, LABEL());
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
} }
} }
@ -8329,32 +8365,61 @@ switch(opcode)
break; break;
case OP_POSSTAR: case OP_POSSTAR:
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); #if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
label = LABEL();
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
JUMPTO(SLJIT_JUMP, label);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset);
break;
}
#endif
label = LABEL(); label = LABEL();
compile_char1_matchingpath(common, type, cc, &nomatch, TRUE); detect_partial_match(common, &no_match);
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
JUMPTO(SLJIT_JUMP, label); JUMPTO(SLJIT_JUMP, label);
set_jumps(nomatch, LABEL()); set_jumps(no_char1_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset); OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_match, LABEL());
break; break;
case OP_POSUPTO: case OP_POSUPTO:
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1, STR_PTR, 0); #if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1, STR_PTR, 0);
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
label = LABEL();
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1, STR_PTR, 0);
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1);
break;
}
#endif
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max); OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
label = LABEL(); label = LABEL();
compile_char1_matchingpath(common, type, cc, &nomatch, TRUE); detect_partial_match(common, &no_match);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1, STR_PTR, 0); compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1); OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label); JUMPTO(SLJIT_NOT_ZERO, label);
set_jumps(nomatch, LABEL()); OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1); set_jumps(no_char1_match, LABEL());
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_match, LABEL());
break; break;
case OP_POSQUERY: case OP_POSQUERY:
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
compile_char1_matchingpath(common, type, cc, &nomatch, TRUE); compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
set_jumps(nomatch, LABEL()); set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset); OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset);
break; break;