@matheusbsbs / matheus-bagliano / commits / 8d2dfc49b1

Fork 0

process_{tree,blob}: show objects without buffering

Here's a less trivial thing, and slightly more dubious one. I was looking at that "struct object_array objects", and wondering why we do that. I have honestly totally forgotten. Why not just call the "show()" function as we encounter the objects? Rather than add the objects to the object_array, and then at the very end going through the array and doing a 'show' on all, just do things more incrementally. Now, there are possible downsides to this: - the "buffer using object_array" _can_ in theory result in at least better I-cache usage (two tight loops rather than one more spread out one). I don't think this is a real issue, but in theory.. - this _does_ change the order of the objects printed. Instead of doing a "process_tree(revs, commit->tree, &objects, NULL, "");" in the loop over the commits (which puts all the root trees _first_ in the object list, this patch just adds them to the list of pending objects, and then we'll traverse them in that order (and thus show each root tree object together with the objects we discover under it) I _think_ the new ordering actually makes more sense, but the object ordering is actually a subtle thing when it comes to packing efficiency, so any change in order is going to have implications for packing. Good or bad, I dunno. - There may be some reason why we did it that odd way with the object array, that I have simply forgotten. Anyway, now that we don't buffer up the objects before showing them that may actually result in lower memory usage during that whole traverse_commit_list() phase. This is seriously not very deeply tested. It makes sense to me, it seems to pass all the tests, it looks ok, but... Does anybody remember why we did that "object_array" thing? It used to be an "object_list" a long long time ago, but got changed into the array due to better memory usage patterns (those linked lists of obejcts are horrible from a memory allocation standpoint). But I wonder why we didn't do this back then. Maybe there's a reason for it. Or maybe there _used_ to be a reason, and no longer is. Signed-off-by: Junio C Hamano <gitster@pobox.com>

Linus Torvalds committed Apr 10, 2009 at 17:27 UTC 8d2dfc49b199c7da6faefd7993630f24bd37fee0

7 files changed +48 -41

builtin-pack-objects.c

+10 -6

@@ -1856,13 +1856,17 @@ static void show_commit(struct commit *commit)

1856 commit->object.flags |= OBJECT_ADDED;

1857 }

1858

1859 -static void show_object(struct object_array_entry *p)

1859 +static void show_object(struct object *obj, const char *name)

1860 {

1861 - add_preferred_base_object(p->name);

1862 - add_object_entry(p->item->sha1, p->item->type, p->name, 0);

1863 - p->item->flags |= OBJECT_ADDED;

1864 - free((char *)p->name);

1865 - p->name = NULL;

1861 + add_preferred_base_object(name);

1862 + add_object_entry(obj->sha1, obj->type, name, 0);

1863 + obj->flags |= OBJECT_ADDED;

1864 +

1865 + /*

1866 + * We will have generated the hash from the name,

1867 + * but not saved a pointer to it - we can free it

1868 + */

1869 + free((char *)name);

1870 }

1871

1872 static void show_edge(struct commit *commit)

builtin-rev-list.c

+10 -10

@@ -169,27 +169,27 @@ static void finish_commit(struct commit *commit)

169 commit->buffer = NULL;

170 }

171

172 -static void finish_object(struct object_array_entry *p)

172 +static void finish_object(struct object *obj, const char *name)

173 {

174 - if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))

175 - die("missing blob object '%s'", sha1_to_hex(p->item->sha1));

174 + if (obj->type == OBJ_BLOB && !has_sha1_file(obj->sha1))

175 + die("missing blob object '%s'", sha1_to_hex(obj->sha1));

176 }

177

178 -static void show_object(struct object_array_entry *p)

178 +static void show_object(struct object *obj, const char *name)

179 {

180 /* An object with name "foo\n0000000..." can be used to

181 * confuse downstream "git pack-objects" very badly.

182 */

183 - const char *ep = strchr(p->name, '\n');

183 + const char *ep = strchr(name, '\n');

184

185 - finish_object(p);

185 + finish_object(obj, name);

186 if (ep) {

187 - printf("%s %.*s\n", sha1_to_hex(p->item->sha1),

188 - (int) (ep - p->name),

189 - p->name);

187 + printf("%s %.*s\n", sha1_to_hex(obj->sha1),

188 + (int) (ep - name),

189 + name);

190 }

191 else

192 - printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);

192 + printf("%s %s\n", sha1_to_hex(obj->sha1), name);

193 }

194

195 static void show_edge(struct commit *commit)

list-objects.c

+18 -17

@@ -10,7 +10,7 @@

11 static void process_blob(struct rev_info *revs,

12 struct blob *blob,

13 - struct object_array *p,

13 + show_object_fn show,

14 struct name_path *path,

15 const char *name)

16 {

@@ -23,7 +23,7 @@ static void process_blob(struct rev_info *revs,

23 if (obj->flags & (UNINTERESTING | SEEN))

24 return;

25 obj->flags |= SEEN;

26 - add_object(obj, p, path, name);

26 + show(obj, path_name(path, name));

27 }

29 /*

@@ -50,7 +50,7 @@ static void process_blob(struct rev_info *revs,

50 */

51 static void process_gitlink(struct rev_info *revs,

52 const unsigned char *sha1,

53 - struct object_array *p,

53 + show_object_fn show,

54 struct name_path *path,

55 const char *name)

56 {

@@ -59,7 +59,7 @@ static void process_gitlink(struct rev_info *revs,

60 static void process_tree(struct rev_info *revs,

61 struct tree *tree,

62 - struct object_array *p,

62 + show_object_fn show,

63 struct name_path *path,

64 const char *name)

65 {

@@ -77,7 +77,7 @@ static void process_tree(struct rev_info *revs,

77 if (parse_tree(tree) < 0)

78 die("bad tree object %s", sha1_to_hex(obj->sha1));

79 obj->flags |= SEEN;

80 - add_object(obj, p, path, name);

80 + show(obj, path_name(path, name));

81 me.up = path;

82 me.elem = name;

83 me.elem_len = strlen(name);

@@ -88,14 +88,14 @@ static void process_tree(struct rev_info *revs,

88 if (S_ISDIR(entry.mode))

89 process_tree(revs,

90 lookup_tree(entry.sha1),

91 - p, &me, entry.path);

91 + show, &me, entry.path);

92 else if (S_ISGITLINK(entry.mode))

93 process_gitlink(revs, entry.sha1,

94 - p, &me, entry.path);

94 + show, &me, entry.path);

95 else

96 process_blob(revs,

97 lookup_blob(entry.sha1),

98 - p, &me, entry.path);

98 + show, &me, entry.path);

99 }

100 free(tree->buffer);

101 tree->buffer = NULL;

@@ -134,16 +134,20 @@ void mark_edges_uninteresting(struct commit_list *list,

134 }

135 }

136

137 +static void add_pending_tree(struct rev_info *revs, struct tree *tree)

138 +{

139 + add_pending_object(revs, &tree->object, "");

140 +}

141 +

142 void traverse_commit_list(struct rev_info *revs,

143 void (*show_commit)(struct commit *),

139 - void (*show_object)(struct object_array_entry *))

144 + void (*show_object)(struct object *, const char *))

145 {

146 int i;

147 struct commit *commit;

143 - struct object_array objects = { 0, 0, NULL };

148

149 while ((commit = get_revision(revs)) != NULL) {

146 - process_tree(revs, commit->tree, &objects, NULL, "");

150 + add_pending_tree(revs, commit->tree);

151 show_commit(commit);

152 }

153 for (i = 0; i < revs->pending.nr; i++) {

@@ -154,25 +158,22 @@ void traverse_commit_list(struct rev_info *revs,

158 continue;

159 if (obj->type == OBJ_TAG) {

160 obj->flags |= SEEN;

157 - add_object_array(obj, name, &objects);

161 + show_object(obj, name);

162 continue;

163 }

164 if (obj->type == OBJ_TREE) {

161 - process_tree(revs, (struct tree *)obj, &objects,

165 + process_tree(revs, (struct tree *)obj, show_object,

166 NULL, name);

167 continue;

168 }

169 if (obj->type == OBJ_BLOB) {

166 - process_blob(revs, (struct blob *)obj, &objects,

170 + process_blob(revs, (struct blob *)obj, show_object,

171 NULL, name);

172 continue;

173 }

174 die("unknown pending object %s (%s)",

175 sha1_to_hex(obj->sha1), name);

176 }

173 - for (i = 0; i < objects.nr; i++)

174 - show_object(&objects.objects[i]);

175 - free(objects.objects);

177 if (revs->pending.nr) {

178 free(revs->pending.objects);

179 revs->pending.nr = 0;

list-objects.h

+1 -1

@@ -2,7 +2,7 @@

2 #define LIST_OBJECTS_H

4 typedef void (*show_commit_fn)(struct commit *);

5 -typedef void (*show_object_fn)(struct object_array_entry *);

5 +typedef void (*show_object_fn)(struct object *, const char *);

6 typedef void (*show_edge_fn)(struct commit *);

8 void traverse_commit_list(struct rev_info *revs, show_commit_fn, show_object_fn);

revision.c

+1 -1

@@ -14,7 +14,7 @@

15 volatile show_early_output_fn_t show_early_output;

17 -static char *path_name(struct name_path *path, const char *name)

17 +char *path_name(struct name_path *path, const char *name)

18 {

19 struct name_path *p;

20 char *n, *m;

revision.h

@@ -141,6 +141,8 @@ struct name_path {

141 const char *elem;

142 };

143

144 +char *path_name(struct name_path *path, const char *name);

145 +

146 extern void add_object(struct object *obj,

147 struct object_array *p,

148 struct name_path *path,

upload-pack.c

+6 -6

@@ -78,20 +78,20 @@ static void show_commit(struct commit *commit)

78 commit->buffer = NULL;

79 }

81 -static void show_object(struct object_array_entry *p)

81 +static void show_object(struct object *obj, const char *name)

82 {

83 /* An object with name "foo\n0000000..." can be used to

84 * confuse downstream git-pack-objects very badly.

85 */

86 - const char *ep = strchr(p->name, '\n');

86 + const char *ep = strchr(name, '\n');

87 if (ep) {

88 - fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(p->item->sha1),

89 - (int) (ep - p->name),

90 - p->name);

88 + fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(obj->sha1),

89 + (int) (ep - name),

90 + name);

91 }

92 else

93 fprintf(pack_pipe, "%s %s\n",

94 - sha1_to_hex(p->item->sha1), p->name);

94 + sha1_to_hex(obj->sha1), name);

95 }

97 static void show_edge(struct commit *commit)