@matheusbsbs / matheus-bagliano / commits / 06a604e670

attr: avoid heavy work when we know the specified attr is not defined

If we have never seen attr 'X' in any .gitattributes file we have examined so far, we can be sure that 'X' is not defined. So no need to go over all the attr stack to look for attr 'X'. This is the purpose behind this new field maybe_real. This optimization breaks down if macros are involved because we can't know for sure what macro would expand to 'X' at attr parsing time. But if we go the pessimistic way and assume all macros are expanded, we hit the builtin "binary" macro. At least the "diff" attr defined in this macro will disable this optimization for git-grep. So we wait until any attr lines _may_ reference to a macro before we turn this off. In git.git, this reduces the number of fill_one() call for "git grep abcdefghi" from ~5348 to 2955. The optimization stops when it reads t/.gitattributes, which uses 'binary' macro. We could probably reduce it further by limiting the 'binary' reference to t/ and subdirs only in this case. "git grep" is actually a good example to justify this patch. The command checks "diff" attribute on every file. People usually don't define this attribute. But they pay the attr lookup penalty anyway without this patch, proportional to the number of attr lines they have in repo. Helped-by: Junio C Hamano <gitster@pobox.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Nguyễn Thái Ngọc Duy committed Dec 28, 2014 at 06:39 UTC 06a604e67051b9342c158432a49e42a5440f9280

1 file changed +28 -5

attr.c

+28 -5

@@ -33,9 +33,11 @@ struct git_attr {

33 unsigned h;

34 int attr_nr;

35 int maybe_macro;

36 + int maybe_real;

37 char name[FLEX_ARRAY];

38 };

39 static int attr_nr;

40 +static int cannot_trust_maybe_real;

42 static struct git_attr_check *check_all_attr;

43 static struct git_attr *(git_attr_hash[HASHSIZE]);

@@ -97,6 +99,7 @@ static struct git_attr *git_attr_internal(const char *name, int len)

99 a->next = git_attr_hash[pos];

100 a->attr_nr = attr_nr++;

101 a->maybe_macro = 0;

102 + a->maybe_real = 0;

103 git_attr_hash[pos] = a;

104

105 REALLOC_ARRAY(check_all_attr, attr_nr);

@@ -269,6 +272,10 @@ static struct match_attr *parse_attr_line(const char *line, const char *src,

272 /* Second pass to fill the attr_states */

273 for (cp = states, i = 0; *cp; i++) {

274 cp = parse_attr(src, lineno, cp, &(res->state[i]));

275 + if (!is_macro)

276 + res->state[i].attr->maybe_real = 1;

277 + if (res->state[i].attr->maybe_macro)

278 + cannot_trust_maybe_real = 1;

279 }

280

281 return res;

@@ -710,10 +717,13 @@ static int macroexpand_one(int nr, int rem)

717 }

718

719 /*

713 - * Collect all attributes for path into the array pointed to by

714 - * check_all_attr.

720 + * Collect attributes for path into the array pointed to by

721 + * check_all_attr. If num is non-zero, only attributes in check[] are

722 + * collected. Otherwise all attributes are collected.

723 */

716 -static void collect_all_attrs(const char *path)

724 +static void collect_some_attrs(const char *path, int num,

725 + struct git_attr_check *check)

726 +

727 {

728 struct attr_stack *stk;

729 int i, pathlen, rem, dirlen;

@@ -736,6 +746,19 @@ static void collect_all_attrs(const char *path)

746 prepare_attr_stack(path, dirlen);

747 for (i = 0; i < attr_nr; i++)

748 check_all_attr[i].value = ATTR__UNKNOWN;

749 + if (num && !cannot_trust_maybe_real) {

750 + rem = 0;

751 + for (i = 0; i < num; i++) {

752 + if (!check[i].attr->maybe_real) {

753 + struct git_attr_check *c;

754 + c = check_all_attr + check[i].attr->attr_nr;

755 + c->value = ATTR__UNSET;

756 + rem++;

757 + }

758 + }

759 + if (rem == num)

760 + return;

761 + }

762

763 rem = attr_nr;

764 for (stk = attr_stack; 0 < rem && stk; stk = stk->prev)

@@ -746,7 +769,7 @@ int git_check_attr(const char *path, int num, struct git_attr_check *check)

769 {

770 int i;

771

749 - collect_all_attrs(path);

772 + collect_some_attrs(path, num, check);

773

774 for (i = 0; i < num; i++) {

775 const char *value = check_all_attr[check[i].attr->attr_nr].value;

@@ -762,7 +785,7 @@ int git_all_attrs(const char *path, int *num, struct git_attr_check **check)

785 {

786 int i, count, j;

787

765 - collect_all_attrs(path);

788 + collect_some_attrs(path, 0, NULL);

789

790 /* Count the number of attributes that are set. */

791 count = 0;