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

Add SIMD support for fast forward newline in JIT.

This commit is contained in:
Zoltán Herczeg 2020-06-02 10:54:37 +00:00
parent e44976f929
commit 0652de5597
1 changed files with 102 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

137
src/pcre2_jit_compile.c
View File

@ -6168,9 +6168,9 @@ static SLJIT_INLINE void fast_forward_newline(compiler_common *common)
{ {
DEFINE_COMPILER; DEFINE_COMPILER;
struct sljit_label *loop; struct sljit_label *loop;
struct sljit_jump *lastchar; struct sljit_jump *lastchar = NULL;
struct sljit_jump *firstchar; struct sljit_jump *firstchar;
struct sljit_jump *quit; struct sljit_jump *quit = NULL;
struct sljit_jump *foundcr = NULL; struct sljit_jump *foundcr = NULL;
struct sljit_jump *notfoundnl; struct sljit_jump *notfoundnl;
jump_list *newline = NULL; jump_list *newline = NULL;
@ -6183,39 +6183,69 @@ if (common->match_end_ptr != 0)
if (common->nltype == NLTYPE_FIXED && common->newline > 255) if (common->nltype == NLTYPE_FIXED && common->newline > 255)
{ {
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); if (JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD && common->mode == PCRE2_JIT_COMPLETE)
if (HAS_VIRTUAL_REGISTERS)
{ {
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); if (HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); {
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
}
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_NOT_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
fast_forward_char_pair_simd(common, 1, common->newline & 0xff, common->newline & 0xff, 0, (common->newline >> 8) & 0xff, (common->newline >> 8) & 0xff);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
} }
else else
{ {
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str)); lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin)); if (HAS_VIRTUAL_REGISTERS)
} {
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0); OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
}
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2));
OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_GREATER_EQUAL); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_GREATER_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 #if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif #endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
loop = LABEL(); loop = LABEL();
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop);
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop);
JUMPHERE(quit);
JUMPHERE(lastchar);
}
JUMPHERE(quit);
JUMPHERE(firstchar); JUMPHERE(firstchar);
JUMPHERE(lastchar);
if (common->match_end_ptr != 0) if (common->match_end_ptr != 0)
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
@ -6232,22 +6262,57 @@ else
/* Example: match /^/ to \r\n from offset 1. */ /* Example: match /^/ to \r\n from offset 1. */
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0); firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
move_back(common, NULL, FALSE);
if (common->nltype == NLTYPE_ANY)
move_back(common, NULL, FALSE);
else
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
loop = LABEL(); loop = LABEL();
common->ff_newline_shortcut = loop; common->ff_newline_shortcut = loop;
read_char(common, common->nlmin, common->nlmax, NULL, READ_CHAR_NEWLINE); if (JIT_HAS_FAST_FORWARD_CHAR_SIMD && (common->nltype == NLTYPE_FIXED || common->nltype == NLTYPE_ANYCRLF))
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); {
if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) if (common->nltype == NLTYPE_ANYCRLF)
foundcr = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); {
check_newlinechar(common, common->nltype, &newline, FALSE); fast_forward_char_simd(common, CHAR_CR, CHAR_LF, 0);
set_jumps(newline, loop); if (common->mode != PCRE2_JIT_COMPLETE)
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
quit = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
}
else
{
fast_forward_char_simd(common, common->newline, common->newline, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (common->mode != PCRE2_JIT_COMPLETE)
{
OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_PTR, 0, STR_END, 0);
CMOV(SLJIT_GREATER, STR_PTR, STR_END, 0);
}
}
}
else
{
read_char(common, common->nlmin, common->nlmax, NULL, READ_CHAR_NEWLINE);
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
foundcr = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
check_newlinechar(common, common->nltype, &newline, FALSE);
set_jumps(newline, loop);
}
if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
{ {
quit = JUMP(SLJIT_JUMP); if (quit == NULL)
JUMPHERE(foundcr); {
quit = JUMP(SLJIT_JUMP);
JUMPHERE(foundcr);
}
notfoundnl = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); notfoundnl = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL);
@ -6259,7 +6324,9 @@ if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
JUMPHERE(notfoundnl); JUMPHERE(notfoundnl);
JUMPHERE(quit); JUMPHERE(quit);
} }
JUMPHERE(lastchar);
if (lastchar)
JUMPHERE(lastchar);
JUMPHERE(firstchar); JUMPHERE(firstchar);
if (common->match_end_ptr != 0) if (common->match_end_ptr != 0)