Add core.bigFileThreshold config variable.
[git/git-bigfiles.git] / fsck.c
blob511b82cba9977c14d6dcba5efbd6d38591d3aacc
1 #include "cache.h"
2 #include "object.h"
3 #include "blob.h"
4 #include "tree.h"
5 #include "tree-walk.h"
6 #include "commit.h"
7 #include "tag.h"
8 #include "fsck.h"
10 static int fsck_walk_tree(struct tree *tree, fsck_walk_func walk, void *data)
12 struct tree_desc desc;
13 struct name_entry entry;
14 int res = 0;
16 if (parse_tree(tree))
17 return -1;
19 init_tree_desc(&desc, tree->buffer, tree->size);
20 while (tree_entry(&desc, &entry)) {
21 int result;
23 if (S_ISGITLINK(entry.mode))
24 continue;
25 if (S_ISDIR(entry.mode))
26 result = walk(&lookup_tree(entry.sha1)->object, OBJ_TREE, data);
27 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode))
28 result = walk(&lookup_blob(entry.sha1)->object, OBJ_BLOB, data);
29 else {
30 result = error("in tree %s: entry %s has bad mode %.6o\n",
31 sha1_to_hex(tree->object.sha1), entry.path, entry.mode);
33 if (result < 0)
34 return result;
35 if (!res)
36 res = result;
38 return res;
41 static int fsck_walk_commit(struct commit *commit, fsck_walk_func walk, void *data)
43 struct commit_list *parents;
44 int res;
45 int result;
47 if (parse_commit(commit))
48 return -1;
50 result = walk((struct object *)commit->tree, OBJ_TREE, data);
51 if (result < 0)
52 return result;
53 res = result;
55 parents = commit->parents;
56 while (parents) {
57 result = walk((struct object *)parents->item, OBJ_COMMIT, data);
58 if (result < 0)
59 return result;
60 if (!res)
61 res = result;
62 parents = parents->next;
64 return res;
67 static int fsck_walk_tag(struct tag *tag, fsck_walk_func walk, void *data)
69 if (parse_tag(tag))
70 return -1;
71 return walk(tag->tagged, OBJ_ANY, data);
74 int fsck_walk(struct object *obj, fsck_walk_func walk, void *data)
76 if (!obj)
77 return -1;
78 switch (obj->type) {
79 case OBJ_BLOB:
80 return 0;
81 case OBJ_TREE:
82 return fsck_walk_tree((struct tree *)obj, walk, data);
83 case OBJ_COMMIT:
84 return fsck_walk_commit((struct commit *)obj, walk, data);
85 case OBJ_TAG:
86 return fsck_walk_tag((struct tag *)obj, walk, data);
87 default:
88 error("Unknown object type for %s", sha1_to_hex(obj->sha1));
89 return -1;
94 * The entries in a tree are ordered in the _path_ order,
95 * which means that a directory entry is ordered by adding
96 * a slash to the end of it.
98 * So a directory called "a" is ordered _after_ a file
99 * called "a.c", because "a/" sorts after "a.c".
101 #define TREE_UNORDERED (-1)
102 #define TREE_HAS_DUPS (-2)
104 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
106 int len1 = strlen(name1);
107 int len2 = strlen(name2);
108 int len = len1 < len2 ? len1 : len2;
109 unsigned char c1, c2;
110 int cmp;
112 cmp = memcmp(name1, name2, len);
113 if (cmp < 0)
114 return 0;
115 if (cmp > 0)
116 return TREE_UNORDERED;
119 * Ok, the first <len> characters are the same.
120 * Now we need to order the next one, but turn
121 * a '\0' into a '/' for a directory entry.
123 c1 = name1[len];
124 c2 = name2[len];
125 if (!c1 && !c2)
127 * git-write-tree used to write out a nonsense tree that has
128 * entries with the same name, one blob and one tree. Make
129 * sure we do not have duplicate entries.
131 return TREE_HAS_DUPS;
132 if (!c1 && S_ISDIR(mode1))
133 c1 = '/';
134 if (!c2 && S_ISDIR(mode2))
135 c2 = '/';
136 return c1 < c2 ? 0 : TREE_UNORDERED;
139 static int fsck_tree(struct tree *item, int strict, fsck_error error_func)
141 int retval;
142 int has_full_path = 0;
143 int has_empty_name = 0;
144 int has_zero_pad = 0;
145 int has_bad_modes = 0;
146 int has_dup_entries = 0;
147 int not_properly_sorted = 0;
148 struct tree_desc desc;
149 unsigned o_mode;
150 const char *o_name;
152 init_tree_desc(&desc, item->buffer, item->size);
154 o_mode = 0;
155 o_name = NULL;
157 while (desc.size) {
158 unsigned mode;
159 const char *name;
161 tree_entry_extract(&desc, &name, &mode);
163 if (strchr(name, '/'))
164 has_full_path = 1;
165 if (!*name)
166 has_empty_name = 1;
167 has_zero_pad |= *(char *)desc.buffer == '0';
168 update_tree_entry(&desc);
170 switch (mode) {
172 * Standard modes..
174 case S_IFREG | 0755:
175 case S_IFREG | 0644:
176 case S_IFLNK:
177 case S_IFDIR:
178 case S_IFGITLINK:
179 break;
181 * This is nonstandard, but we had a few of these
182 * early on when we honored the full set of mode
183 * bits..
185 case S_IFREG | 0664:
186 if (!strict)
187 break;
188 default:
189 has_bad_modes = 1;
192 if (o_name) {
193 switch (verify_ordered(o_mode, o_name, mode, name)) {
194 case TREE_UNORDERED:
195 not_properly_sorted = 1;
196 break;
197 case TREE_HAS_DUPS:
198 has_dup_entries = 1;
199 break;
200 default:
201 break;
205 o_mode = mode;
206 o_name = name;
209 retval = 0;
210 if (has_full_path)
211 retval += error_func(&item->object, FSCK_WARN, "contains full pathnames");
212 if (has_empty_name)
213 retval += error_func(&item->object, FSCK_WARN, "contains empty pathname");
214 if (has_zero_pad)
215 retval += error_func(&item->object, FSCK_WARN, "contains zero-padded file modes");
216 if (has_bad_modes)
217 retval += error_func(&item->object, FSCK_WARN, "contains bad file modes");
218 if (has_dup_entries)
219 retval += error_func(&item->object, FSCK_ERROR, "contains duplicate file entries");
220 if (not_properly_sorted)
221 retval += error_func(&item->object, FSCK_ERROR, "not properly sorted");
222 return retval;
225 static int fsck_commit(struct commit *commit, fsck_error error_func)
227 char *buffer = commit->buffer;
228 unsigned char tree_sha1[20], sha1[20];
229 struct commit_graft *graft;
230 int parents = 0;
232 if (!commit->date)
233 return error_func(&commit->object, FSCK_ERROR, "invalid author/committer line");
235 if (memcmp(buffer, "tree ", 5))
236 return error_func(&commit->object, FSCK_ERROR, "invalid format - expected 'tree' line");
237 if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
238 return error_func(&commit->object, FSCK_ERROR, "invalid 'tree' line format - bad sha1");
239 buffer += 46;
240 while (!memcmp(buffer, "parent ", 7)) {
241 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
242 return error_func(&commit->object, FSCK_ERROR, "invalid 'parent' line format - bad sha1");
243 buffer += 48;
244 parents++;
246 graft = lookup_commit_graft(commit->object.sha1);
247 if (graft) {
248 struct commit_list *p = commit->parents;
249 parents = 0;
250 while (p) {
251 p = p->next;
252 parents++;
254 if (graft->nr_parent == -1 && !parents)
255 ; /* shallow commit */
256 else if (graft->nr_parent != parents)
257 return error_func(&commit->object, FSCK_ERROR, "graft objects missing");
258 } else {
259 struct commit_list *p = commit->parents;
260 while (p && parents) {
261 p = p->next;
262 parents--;
264 if (p || parents)
265 return error_func(&commit->object, FSCK_ERROR, "parent objects missing");
267 if (memcmp(buffer, "author ", 7))
268 return error_func(&commit->object, FSCK_ERROR, "invalid format - expected 'author' line");
269 if (!commit->tree)
270 return error_func(&commit->object, FSCK_ERROR, "could not load commit's tree %s", sha1_to_hex(tree_sha1));
272 return 0;
275 static int fsck_tag(struct tag *tag, fsck_error error_func)
277 struct object *tagged = tag->tagged;
279 if (!tagged)
280 return error_func(&tag->object, FSCK_ERROR, "could not load tagged object");
281 return 0;
284 int fsck_object(struct object *obj, int strict, fsck_error error_func)
286 if (!obj)
287 return error_func(obj, FSCK_ERROR, "no valid object to fsck");
289 if (obj->type == OBJ_BLOB)
290 return 0;
291 if (obj->type == OBJ_TREE)
292 return fsck_tree((struct tree *) obj, strict, error_func);
293 if (obj->type == OBJ_COMMIT)
294 return fsck_commit((struct commit *) obj, error_func);
295 if (obj->type == OBJ_TAG)
296 return fsck_tag((struct tag *) obj, error_func);
298 return error_func(obj, FSCK_ERROR, "unknown type '%d' (internal fsck error)",
299 obj->type);
302 int fsck_error_function(struct object *obj, int type, const char *fmt, ...)
304 va_list ap;
305 int len;
306 struct strbuf sb = STRBUF_INIT;
308 strbuf_addf(&sb, "object %s:", obj->sha1?sha1_to_hex(obj->sha1):"(null)");
310 va_start(ap, fmt);
311 len = vsnprintf(sb.buf + sb.len, strbuf_avail(&sb), fmt, ap);
312 va_end(ap);
314 if (len < 0)
315 len = 0;
316 if (len >= strbuf_avail(&sb)) {
317 strbuf_grow(&sb, len + 2);
318 va_start(ap, fmt);
319 len = vsnprintf(sb.buf + sb.len, strbuf_avail(&sb), fmt, ap);
320 va_end(ap);
321 if (len >= strbuf_avail(&sb))
322 die("this should not happen, your snprintf is broken");
325 error("%s", sb.buf);
326 strbuf_release(&sb);
327 return 1;