4 #include "git-compat-util.h"
8 #include "string-list.h"
14 * Basic data structures for the directory cache
17 #define CACHE_SIGNATURE 0x44495243 /* "DIRC" */
19 uint32_t hdr_signature
;
24 #define INDEX_FORMAT_LB 2
25 #define INDEX_FORMAT_UB 4
28 struct hashmap_entry ent
;
29 struct stat_data ce_stat_data
;
31 unsigned int ce_flags
;
32 unsigned int mem_pool_allocated
;
33 unsigned int ce_namelen
;
34 unsigned int index
; /* for link extension */
36 char name
[FLEX_ARRAY
]; /* more */
39 #define CE_STAGEMASK (0x3000)
40 #define CE_EXTENDED (0x4000)
41 #define CE_VALID (0x8000)
42 #define CE_STAGESHIFT 12
45 * Range 0xFFFF0FFF in ce_flags is divided into
46 * two parts: in-memory flags and on-disk ones.
47 * Flags in CE_EXTENDED_FLAGS will get saved on-disk
48 * if you want to save a new flag, add it in
51 * In-memory only flags
53 #define CE_UPDATE (1 << 16)
54 #define CE_REMOVE (1 << 17)
55 #define CE_UPTODATE (1 << 18)
56 #define CE_ADDED (1 << 19)
58 #define CE_HASHED (1 << 20)
59 #define CE_FSMONITOR_VALID (1 << 21)
60 #define CE_WT_REMOVE (1 << 22) /* remove in work directory */
61 #define CE_CONFLICTED (1 << 23)
63 #define CE_UNPACKED (1 << 24)
64 #define CE_NEW_SKIP_WORKTREE (1 << 25)
66 /* used to temporarily mark paths matched by pathspecs */
67 #define CE_MATCHED (1 << 26)
69 #define CE_UPDATE_IN_BASE (1 << 27)
70 #define CE_STRIP_NAME (1 << 28)
73 * Extended on-disk flags
75 #define CE_INTENT_TO_ADD (1 << 29)
76 #define CE_SKIP_WORKTREE (1 << 30)
77 /* CE_EXTENDED2 is for future extension */
78 #define CE_EXTENDED2 (1U << 31)
80 #define CE_EXTENDED_FLAGS (CE_INTENT_TO_ADD | CE_SKIP_WORKTREE)
83 * Safeguard to avoid saving wrong flags:
84 * - CE_EXTENDED2 won't get saved until its semantic is known
85 * - Bits in 0x0000FFFF have been saved in ce_flags already
86 * - Bits in 0x003F0000 are currently in-memory flags
88 #if CE_EXTENDED_FLAGS & 0x803FFFFF
89 #error "CE_EXTENDED_FLAGS out of range"
92 /* Forward structure decls */
97 * Copy the sha1 and stat state of a cache entry from one to
98 * another. But we never change the name, or the hash state!
100 static inline void copy_cache_entry(struct cache_entry
*dst
,
101 const struct cache_entry
*src
)
103 unsigned int state
= dst
->ce_flags
& CE_HASHED
;
104 int mem_pool_allocated
= dst
->mem_pool_allocated
;
106 /* Don't copy hash chain and name */
107 memcpy(&dst
->ce_stat_data
, &src
->ce_stat_data
,
108 offsetof(struct cache_entry
, name
) -
109 offsetof(struct cache_entry
, ce_stat_data
));
111 /* Restore the hash state */
112 dst
->ce_flags
= (dst
->ce_flags
& ~CE_HASHED
) | state
;
114 /* Restore the mem_pool_allocated flag */
115 dst
->mem_pool_allocated
= mem_pool_allocated
;
118 static inline unsigned create_ce_flags(unsigned stage
)
120 return (stage
<< CE_STAGESHIFT
);
123 #define ce_namelen(ce) ((ce)->ce_namelen)
124 #define ce_size(ce) cache_entry_size(ce_namelen(ce))
125 #define ce_stage(ce) ((CE_STAGEMASK & (ce)->ce_flags) >> CE_STAGESHIFT)
126 #define ce_uptodate(ce) ((ce)->ce_flags & CE_UPTODATE)
127 #define ce_skip_worktree(ce) ((ce)->ce_flags & CE_SKIP_WORKTREE)
128 #define ce_mark_uptodate(ce) ((ce)->ce_flags |= CE_UPTODATE)
129 #define ce_intent_to_add(ce) ((ce)->ce_flags & CE_INTENT_TO_ADD)
131 static inline unsigned int ce_mode_from_stat(const struct cache_entry
*ce
,
134 extern int trust_executable_bit
, has_symlinks
;
135 if (!has_symlinks
&& S_ISREG(mode
) &&
136 ce
&& S_ISLNK(ce
->ce_mode
))
138 if (!trust_executable_bit
&& S_ISREG(mode
)) {
139 if (ce
&& S_ISREG(ce
->ce_mode
))
141 return create_ce_mode(0666);
143 return create_ce_mode(mode
);
145 static inline int ce_to_dtype(const struct cache_entry
*ce
)
147 unsigned ce_mode
= ntohl(ce
->ce_mode
);
148 if (S_ISREG(ce_mode
))
150 else if (S_ISDIR(ce_mode
) || S_ISGITLINK(ce_mode
))
152 else if (S_ISLNK(ce_mode
))
158 static inline int ce_path_match(struct index_state
*istate
,
159 const struct cache_entry
*ce
,
160 const struct pathspec
*pathspec
,
163 return match_pathspec(istate
, pathspec
, ce
->name
, ce_namelen(ce
), 0, seen
,
164 S_ISDIR(ce
->ce_mode
) || S_ISGITLINK(ce
->ce_mode
));
167 #define cache_entry_size(len) (offsetof(struct cache_entry,name) + (len) + 1)
169 #define SOMETHING_CHANGED (1 << 0) /* unclassified changes go here */
170 #define CE_ENTRY_CHANGED (1 << 1)
171 #define CE_ENTRY_REMOVED (1 << 2)
172 #define CE_ENTRY_ADDED (1 << 3)
173 #define RESOLVE_UNDO_CHANGED (1 << 4)
174 #define CACHE_TREE_CHANGED (1 << 5)
175 #define SPLIT_INDEX_ORDERED (1 << 6)
176 #define UNTRACKED_CHANGED (1 << 7)
177 #define FSMONITOR_CHANGED (1 << 8)
180 struct untracked_cache
;
184 enum sparse_index_mode
{
186 * There are no sparse directories in the index at all.
188 * Repositories that don't use cone-mode sparse-checkout will
189 * always have their indexes in this mode.
194 * The index has already been collapsed to sparse directories
195 * whereever possible.
200 * The sparse directories that exist are outside the
201 * sparse-checkout boundary, but it is possible that some file
202 * entries could collapse to sparse directory entries.
204 INDEX_PARTIALLY_SPARSE
,
208 struct cache_entry
**cache
;
209 unsigned int version
;
210 unsigned int cache_nr
, cache_alloc
, cache_changed
;
211 struct string_list
*resolve_undo
;
212 struct cache_tree
*cache_tree
;
213 struct split_index
*split_index
;
214 struct cache_time timestamp
;
215 unsigned name_hash_initialized
: 1,
219 updated_skipworktree
: 1,
220 fsmonitor_has_run_once
: 1;
221 enum sparse_index_mode sparse_index
;
222 struct hashmap name_hash
;
223 struct hashmap dir_hash
;
224 struct object_id oid
;
225 struct untracked_cache
*untracked
;
226 char *fsmonitor_last_update
;
227 struct ewah_bitmap
*fsmonitor_dirty
;
228 struct mem_pool
*ce_mem_pool
;
229 struct progress
*progress
;
230 struct repository
*repo
;
231 struct pattern_list
*sparse_checkout_patterns
;
235 * A "struct index_state istate" must be initialized with
236 * INDEX_STATE_INIT or the corresponding index_state_init().
238 * If the variable won't be used again, use release_index() to free()
239 * its resources. If it needs to be used again use discard_index(),
240 * which does the same thing, but will use use index_state_init() at
241 * the end. The discard_index() will use its own "istate->repo" as the
242 * "r" argument to index_state_init() in that case.
244 #define INDEX_STATE_INIT(r) { \
247 void index_state_init(struct index_state
*istate
, struct repository
*r
);
248 void release_index(struct index_state
*istate
);
251 int test_lazy_init_name_hash(struct index_state
*istate
, int try_threaded
);
252 void add_name_hash(struct index_state
*istate
, struct cache_entry
*ce
);
253 void remove_name_hash(struct index_state
*istate
, struct cache_entry
*ce
);
254 void free_name_hash(struct index_state
*istate
);
256 /* Cache entry creation and cleanup */
259 * Create cache_entry intended for use in the specified index. Caller
260 * is responsible for discarding the cache_entry with
261 * `discard_cache_entry`.
263 struct cache_entry
*make_cache_entry(struct index_state
*istate
,
265 const struct object_id
*oid
,
268 unsigned int refresh_options
);
270 struct cache_entry
*make_empty_cache_entry(struct index_state
*istate
,
274 * Create a cache_entry that is not intended to be added to an index. If
275 * `ce_mem_pool` is not NULL, the entry is allocated within the given memory
276 * pool. Caller is responsible for discarding "loose" entries with
277 * `discard_cache_entry()` and the memory pool with
278 * `mem_pool_discard(ce_mem_pool, should_validate_cache_entries())`.
280 struct cache_entry
*make_transient_cache_entry(unsigned int mode
,
281 const struct object_id
*oid
,
284 struct mem_pool
*ce_mem_pool
);
286 struct cache_entry
*make_empty_transient_cache_entry(size_t len
,
287 struct mem_pool
*ce_mem_pool
);
290 * Discard cache entry.
292 void discard_cache_entry(struct cache_entry
*ce
);
295 * Check configuration if we should perform extra validation on cache
298 int should_validate_cache_entries(void);
301 * Duplicate a cache_entry. Allocate memory for the new entry from a
302 * memory_pool. Takes into account cache_entry fields that are meant
303 * for managing the underlying memory allocation of the cache_entry.
305 struct cache_entry
*dup_cache_entry(const struct cache_entry
*ce
, struct index_state
*istate
);
308 * Validate the cache entries in the index. This is an internal
309 * consistency check that the cache_entry structs are allocated from
310 * the expected memory pool.
312 void validate_cache_entries(const struct index_state
*istate
);
315 * Bulk prefetch all missing cache entries that are not GITLINKs and that match
316 * the given predicate. This function should only be called if
317 * repo_has_promisor_remote() returns true.
319 typedef int (*must_prefetch_predicate
)(const struct cache_entry
*);
320 void prefetch_cache_entries(const struct index_state
*istate
,
321 must_prefetch_predicate must_prefetch
);
323 #ifdef USE_THE_INDEX_VARIABLE
324 extern struct index_state the_index
;
327 #define INIT_DB_QUIET 0x0001
328 #define INIT_DB_EXIST_OK 0x0002
330 int init_db(const char *git_dir
, const char *real_git_dir
,
331 const char *template_dir
, int hash_algo
,
332 const char *initial_branch
, unsigned int flags
);
333 void initialize_repository_version(int hash_algo
, int reinit
);
335 /* Initialize and use the cache information */
337 void preload_index(struct index_state
*index
,
338 const struct pathspec
*pathspec
,
339 unsigned int refresh_flags
);
340 int do_read_index(struct index_state
*istate
, const char *path
,
341 int must_exist
); /* for testting only! */
342 int read_index_from(struct index_state
*, const char *path
,
344 int is_index_unborn(struct index_state
*);
346 void ensure_full_index(struct index_state
*istate
);
348 /* For use with `write_locked_index()`. */
349 #define COMMIT_LOCK (1 << 0)
350 #define SKIP_IF_UNCHANGED (1 << 1)
353 * Write the index while holding an already-taken lock. Close the lock,
354 * and if `COMMIT_LOCK` is given, commit it.
356 * Unless a split index is in use, write the index into the lockfile.
358 * With a split index, write the shared index to a temporary file,
359 * adjust its permissions and rename it into place, then write the
360 * split index to the lockfile. If the temporary file for the shared
361 * index cannot be created, fall back to the behavior described in
362 * the previous paragraph.
364 * With `COMMIT_LOCK`, the lock is always committed or rolled back.
365 * Without it, the lock is closed, but neither committed nor rolled
368 * If `SKIP_IF_UNCHANGED` is given and the index is unchanged, nothing
369 * is written (and the lock is rolled back if `COMMIT_LOCK` is given).
371 int write_locked_index(struct index_state
*, struct lock_file
*lock
, unsigned flags
);
373 void discard_index(struct index_state
*);
374 void move_index_extensions(struct index_state
*dst
, struct index_state
*src
);
375 int unmerged_index(const struct index_state
*);
378 * Returns 1 if istate differs from tree, 0 otherwise. If tree is NULL,
379 * compares istate to HEAD. If tree is NULL and on an unborn branch,
380 * returns 1 if there are entries in istate, 0 otherwise. If an strbuf is
381 * provided, the space-separated list of files that differ will be appended
384 int repo_index_has_changes(struct repository
*repo
,
388 int verify_path(const char *path
, unsigned mode
);
389 int strcmp_offset(const char *s1
, const char *s2
, size_t *first_change
);
390 int index_dir_exists(struct index_state
*istate
, const char *name
, int namelen
);
391 void adjust_dirname_case(struct index_state
*istate
, char *name
);
392 struct cache_entry
*index_file_exists(struct index_state
*istate
, const char *name
, int namelen
, int igncase
);
395 * Searches for an entry defined by name and namelen in the given index.
396 * If the return value is positive (including 0) it is the position of an
397 * exact match. If the return value is negative, the negated value minus 1
398 * is the position where the entry would be inserted.
399 * Example: The current index consists of these files and its stages:
403 * index_name_pos(&index, "a", 1) -> -1
404 * index_name_pos(&index, "b", 1) -> 0
405 * index_name_pos(&index, "c", 1) -> -2
406 * index_name_pos(&index, "d", 1) -> 1
407 * index_name_pos(&index, "e", 1) -> -3
408 * index_name_pos(&index, "f", 1) -> -3
409 * index_name_pos(&index, "g", 1) -> -5
411 int index_name_pos(struct index_state
*, const char *name
, int namelen
);
414 * Like index_name_pos, returns the position of an entry of the given name in
415 * the index if one exists, otherwise returns a negative value where the negated
416 * value minus 1 is the position where the index entry would be inserted. Unlike
417 * index_name_pos, however, a sparse index is not expanded to find an entry
418 * inside a sparse directory.
420 int index_name_pos_sparse(struct index_state
*, const char *name
, int namelen
);
423 * Determines whether an entry with the given name exists within the
424 * given index. The return value is 1 if an exact match is found, otherwise
425 * it is 0. Note that, unlike index_name_pos, this function does not expand
426 * the index if it is sparse. If an item exists within the full index but it
427 * is contained within a sparse directory (and not in the sparse index), 0 is
430 int index_entry_exists(struct index_state
*, const char *name
, int namelen
);
433 * Some functions return the negative complement of an insert position when a
434 * precise match was not found but a position was found where the entry would
435 * need to be inserted. This helper protects that logic from any integer
438 static inline int index_pos_to_insert_pos(uintmax_t pos
)
441 die("overflow: -1 - %"PRIuMAX
, pos
);
442 return -1 - (int)pos
;
445 #define ADD_CACHE_OK_TO_ADD 1 /* Ok to add */
446 #define ADD_CACHE_OK_TO_REPLACE 2 /* Ok to replace file/directory */
447 #define ADD_CACHE_SKIP_DFCHECK 4 /* Ok to skip DF conflict checks */
448 #define ADD_CACHE_JUST_APPEND 8 /* Append only */
449 #define ADD_CACHE_NEW_ONLY 16 /* Do not replace existing ones */
450 #define ADD_CACHE_KEEP_CACHE_TREE 32 /* Do not invalidate cache-tree */
451 #define ADD_CACHE_RENORMALIZE 64 /* Pass along HASH_RENORMALIZE */
452 int add_index_entry(struct index_state
*, struct cache_entry
*ce
, int option
);
453 void rename_index_entry_at(struct index_state
*, int pos
, const char *new_name
);
455 /* Remove entry, return true if there are more entries to go. */
456 int remove_index_entry_at(struct index_state
*, int pos
);
458 void remove_marked_cache_entries(struct index_state
*istate
, int invalidate
);
459 int remove_file_from_index(struct index_state
*, const char *path
);
460 #define ADD_CACHE_VERBOSE 1
461 #define ADD_CACHE_PRETEND 2
462 #define ADD_CACHE_IGNORE_ERRORS 4
463 #define ADD_CACHE_IGNORE_REMOVAL 8
464 #define ADD_CACHE_INTENT 16
466 * These two are used to add the contents of the file at path
467 * to the index, marking the working tree up-to-date by storing
468 * the cached stat info in the resulting cache entry. A caller
469 * that has already run lstat(2) on the path can call
470 * add_to_index(), and all others can call add_file_to_index();
471 * the latter will do necessary lstat(2) internally before
472 * calling the former.
474 int add_to_index(struct index_state
*, const char *path
, struct stat
*, int flags
);
475 int add_file_to_index(struct index_state
*, const char *path
, int flags
);
477 int chmod_index_entry(struct index_state
*, struct cache_entry
*ce
, char flip
);
478 int ce_same_name(const struct cache_entry
*a
, const struct cache_entry
*b
);
479 void set_object_name_for_intent_to_add_entry(struct cache_entry
*ce
);
480 int index_name_is_other(struct index_state
*, const char *, int);
481 void *read_blob_data_from_index(struct index_state
*, const char *, unsigned long *);
483 /* do stat comparison even if CE_VALID is true */
484 #define CE_MATCH_IGNORE_VALID 01
485 /* do not check the contents but report dirty on racily-clean entries */
486 #define CE_MATCH_RACY_IS_DIRTY 02
487 /* do stat comparison even if CE_SKIP_WORKTREE is true */
488 #define CE_MATCH_IGNORE_SKIP_WORKTREE 04
489 /* ignore non-existent files during stat update */
490 #define CE_MATCH_IGNORE_MISSING 0x08
491 /* enable stat refresh */
492 #define CE_MATCH_REFRESH 0x10
493 /* don't refresh_fsmonitor state or do stat comparison even if CE_FSMONITOR_VALID is true */
494 #define CE_MATCH_IGNORE_FSMONITOR 0X20
495 int is_racy_timestamp(const struct index_state
*istate
,
496 const struct cache_entry
*ce
);
497 int has_racy_timestamp(struct index_state
*istate
);
498 int ie_match_stat(struct index_state
*, const struct cache_entry
*, struct stat
*, unsigned int);
499 int ie_modified(struct index_state
*, const struct cache_entry
*, struct stat
*, unsigned int);
502 * Record to sd the data from st that we use to check whether a file
503 * might have changed.
505 void fill_stat_data(struct stat_data
*sd
, struct stat
*st
);
508 * Return 0 if st is consistent with a file not having been changed
509 * since sd was filled. If there are differences, return a
510 * combination of MTIME_CHANGED, CTIME_CHANGED, OWNER_CHANGED,
511 * INODE_CHANGED, and DATA_CHANGED.
513 int match_stat_data(const struct stat_data
*sd
, struct stat
*st
);
514 int match_stat_data_racy(const struct index_state
*istate
,
515 const struct stat_data
*sd
, struct stat
*st
);
517 void fill_stat_cache_info(struct index_state
*istate
, struct cache_entry
*ce
, struct stat
*st
);
519 #define REFRESH_REALLY (1 << 0) /* ignore_valid */
520 #define REFRESH_UNMERGED (1 << 1) /* allow unmerged */
521 #define REFRESH_QUIET (1 << 2) /* be quiet about it */
522 #define REFRESH_IGNORE_MISSING (1 << 3) /* ignore non-existent */
523 #define REFRESH_IGNORE_SUBMODULES (1 << 4) /* ignore submodules */
524 #define REFRESH_IN_PORCELAIN (1 << 5) /* user friendly output, not "needs update" */
525 #define REFRESH_PROGRESS (1 << 6) /* show progress bar if stderr is tty */
526 #define REFRESH_IGNORE_SKIP_WORKTREE (1 << 7) /* ignore skip_worktree entries */
527 int refresh_index(struct index_state
*, unsigned int flags
, const struct pathspec
*pathspec
, char *seen
, const char *header_msg
);
529 * Refresh the index and write it to disk.
531 * 'refresh_flags' is passed directly to 'refresh_index()', while
532 * 'COMMIT_LOCK | write_flags' is passed to 'write_locked_index()', so
533 * the lockfile is always either committed or rolled back.
535 * If 'gentle' is passed, errors locking the index are ignored.
537 * Return 1 if refreshing the index returns an error, -1 if writing
538 * the index to disk fails, 0 on success.
540 * Note that if refreshing the index returns an error, we still write
541 * out the index (unless locking fails).
543 int repo_refresh_and_write_index(struct repository
*, unsigned int refresh_flags
, unsigned int write_flags
, int gentle
, const struct pathspec
*, char *seen
, const char *header_msg
);
545 struct cache_entry
*refresh_cache_entry(struct index_state
*, struct cache_entry
*, unsigned int);
547 void set_alternate_index_output(const char *);
549 extern int verify_index_checksum
;
550 extern int verify_ce_order
;
552 #define MTIME_CHANGED 0x0001
553 #define CTIME_CHANGED 0x0002
554 #define OWNER_CHANGED 0x0004
555 #define MODE_CHANGED 0x0008
556 #define INODE_CHANGED 0x0010
557 #define DATA_CHANGED 0x0020
558 #define TYPE_CHANGED 0x0040
560 int cmp_cache_name_compare(const void *a_
, const void *b_
);
564 * return 0 if success, 1 - if addition of a file failed and
565 * ADD_FILES_IGNORE_ERRORS was specified in flags
567 int add_files_to_cache(const char *prefix
, const struct pathspec
*pathspec
, int flags
);
570 extern int diff_auto_refresh_index
;
573 void overlay_tree_on_index(struct index_state
*istate
,
574 const char *tree_name
, const char *prefix
);
578 int try_merge_command(struct repository
*r
,
579 const char *strategy
, size_t xopts_nr
,
580 const char **xopts
, struct commit_list
*common
,
581 const char *head_arg
, struct commit_list
*remotes
);
582 int checkout_fast_forward(struct repository
*r
,
583 const struct object_id
*from
,
584 const struct object_id
*to
,
585 int overwrite_ignore
);
588 int sane_execvp(const char *file
, char *const argv
[]);
591 * A struct to encapsulate the concept of whether a file has changed
592 * since we last checked it. This uses criteria similar to those used
595 struct stat_validity
{
596 struct stat_data
*sd
;
599 void stat_validity_clear(struct stat_validity
*sv
);
602 * Returns 1 if the path is a regular file (or a symlink to a regular
603 * file) and matches the saved stat_validity, 0 otherwise. A missing
604 * or inaccessible file is considered a match if the struct was just
605 * initialized, or if the previous update found an inaccessible file.
607 int stat_validity_check(struct stat_validity
*sv
, const char *path
);
610 * Update the stat_validity from a file opened at descriptor fd. If
611 * the file is missing, inaccessible, or not a regular file, then
612 * future calls to stat_validity_check will match iff one of those
613 * conditions continues to be true.
615 void stat_validity_update(struct stat_validity
*sv
, int fd
);