Blob


1 /*
2 * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
3 *
4 * Permission to use, copy, modify, and distribute this software for any
5 * purpose with or without fee is hereby granted, provided that the above
6 * copyright notice and this permission notice appear in all copies.
7 *
8 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15 */
17 #include "got_compat.h"
19 #include <sys/queue.h>
20 #include <sys/stat.h>
22 #include <errno.h>
23 #include <dirent.h>
24 #include <fcntl.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <limits.h>
29 #include <unistd.h>
31 #include "got_error.h"
32 #include "got_object.h"
33 #include "got_path.h"
35 #include "got_lib_hash.h"
36 #include "got_lib_fileindex.h"
37 #include "got_lib_worktree.h"
39 /* got_fileindex_entry flags */
40 #define GOT_FILEIDX_F_PATH_LEN 0x00000fff
41 #define GOT_FILEIDX_F_STAGE 0x0000f000
42 #define GOT_FILEIDX_F_STAGE_SHIFT 12
43 #define GOT_FILEIDX_F_NOT_FLUSHED 0x00010000
44 #define GOT_FILEIDX_F_NO_BLOB 0x00020000
45 #define GOT_FILEIDX_F_NO_COMMIT 0x00040000
46 #define GOT_FILEIDX_F_NO_FILE_ON_DISK 0x00080000
47 #define GOT_FILEIDX_F_REMOVE_ON_FLUSH 0x00100000
48 #define GOT_FILEIDX_F_SKIPPED 0x00200000
50 struct got_fileindex {
51 struct got_fileindex_tree entries;
52 int nentries; /* Does not include entries marked for removal. */
53 #define GOT_FILEIDX_MAX_ENTRIES INT_MAX
54 };
56 mode_t
57 got_fileindex_entry_perms_get(struct got_fileindex_entry *ie)
58 {
59 return ((ie->mode & GOT_FILEIDX_MODE_PERMS) >>
60 GOT_FILEIDX_MODE_PERMS_SHIFT);
61 }
63 static void
64 fileindex_entry_perms_set(struct got_fileindex_entry *ie, mode_t mode)
65 {
66 ie->mode &= ~GOT_FILEIDX_MODE_PERMS;
67 ie->mode |= ((mode << GOT_FILEIDX_MODE_PERMS_SHIFT) &
68 GOT_FILEIDX_MODE_PERMS);
69 }
71 mode_t
72 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
73 {
74 mode_t perms = got_fileindex_entry_perms_get(ie);
75 int type = got_fileindex_entry_filetype_get(ie);
76 uint32_t ftype;
78 if (type == GOT_FILEIDX_MODE_REGULAR_FILE ||
79 type == GOT_FILEIDX_MODE_BAD_SYMLINK)
80 ftype = S_IFREG;
81 else
82 ftype = S_IFLNK;
84 return (ftype | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
85 }
87 const struct got_error *
88 got_fileindex_entry_update(struct got_fileindex_entry *ie,
89 int wt_fd, const char *ondisk_path, uint8_t *blob_sha1,
90 uint8_t *commit_sha1, int update_timestamps)
91 {
92 struct stat sb;
94 if (fstatat(wt_fd, ondisk_path, &sb, AT_SYMLINK_NOFOLLOW) != 0) {
95 if (!((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) &&
96 errno == ENOENT))
97 return got_error_from_errno2("fstatat", ondisk_path);
98 sb.st_mode = GOT_DEFAULT_FILE_MODE;
99 } else {
100 if (sb.st_mode & S_IFDIR)
101 return got_error_set_errno(EISDIR, ondisk_path);
102 ie->flags &= ~GOT_FILEIDX_F_NO_FILE_ON_DISK;
105 if ((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0) {
106 if (update_timestamps) {
107 ie->ctime_sec = sb.st_ctim.tv_sec;
108 ie->ctime_nsec = sb.st_ctim.tv_nsec;
109 ie->mtime_sec = sb.st_mtim.tv_sec;
110 ie->mtime_nsec = sb.st_mtim.tv_nsec;
112 ie->uid = sb.st_uid;
113 ie->gid = sb.st_gid;
114 ie->size = (sb.st_size & 0xffffffff);
115 if (S_ISLNK(sb.st_mode)) {
116 got_fileindex_entry_filetype_set(ie,
117 GOT_FILEIDX_MODE_SYMLINK);
118 fileindex_entry_perms_set(ie, 0);
119 } else {
120 got_fileindex_entry_filetype_set(ie,
121 GOT_FILEIDX_MODE_REGULAR_FILE);
122 fileindex_entry_perms_set(ie,
123 sb.st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
127 if (blob_sha1) {
128 memmove(ie->blob_sha1, blob_sha1, SHA1_DIGEST_LENGTH);
129 ie->flags &= ~GOT_FILEIDX_F_NO_BLOB;
130 } else
131 ie->flags |= GOT_FILEIDX_F_NO_BLOB;
133 if (commit_sha1) {
134 memcpy(ie->commit_sha1, commit_sha1, SHA1_DIGEST_LENGTH);
135 ie->flags &= ~GOT_FILEIDX_F_NO_COMMIT;
136 } else
137 ie->flags |= GOT_FILEIDX_F_NO_COMMIT;
139 return NULL;
142 void
143 got_fileindex_entry_mark_deleted_from_disk(struct got_fileindex_entry *ie)
145 ie->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
148 void
149 got_fileindex_entry_mark_skipped(struct got_fileindex_entry *ie)
151 ie->flags |= GOT_FILEIDX_F_SKIPPED;
154 const struct got_error *
155 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
156 const char *relpath)
158 size_t len;
160 *ie = calloc(1, sizeof(**ie));
161 if (*ie == NULL)
162 return got_error_from_errno("calloc");
164 (*ie)->path = strdup(relpath);
165 if ((*ie)->path == NULL) {
166 const struct got_error *err = got_error_from_errno("strdup");
167 free(*ie);
168 *ie = NULL;
169 return err;
172 len = strlen(relpath);
173 if (len > GOT_FILEIDX_F_PATH_LEN)
174 len = GOT_FILEIDX_F_PATH_LEN;
175 (*ie)->flags |= len;
177 return NULL;
180 void
181 got_fileindex_entry_free(struct got_fileindex_entry *ie)
183 free(ie->path);
184 free(ie);
187 size_t
188 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
190 return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
193 uint32_t
194 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
196 return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
199 void
200 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
202 ie->flags &= ~GOT_FILEIDX_F_STAGE;
203 ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
204 GOT_FILEIDX_F_STAGE);
207 int
208 got_fileindex_entry_filetype_get(struct got_fileindex_entry *ie)
210 return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
213 void
214 got_fileindex_entry_filetype_set(struct got_fileindex_entry *ie, int type)
216 ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_ONDISK;
217 ie->mode |= (type & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
220 void
221 got_fileindex_entry_staged_filetype_set(struct got_fileindex_entry *ie,
222 int type)
224 ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_STAGED;
225 ie->mode |= ((type << GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT) &
226 GOT_FILEIDX_MODE_FILE_TYPE_STAGED);
229 int
230 got_fileindex_entry_staged_filetype_get(struct got_fileindex_entry *ie)
232 return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_STAGED) >>
233 GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT;
236 int
237 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
239 return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
242 int
243 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
245 return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
248 int
249 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
251 return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
254 int
255 got_fileindex_entry_was_skipped(struct got_fileindex_entry *ie)
257 return (ie->flags & GOT_FILEIDX_F_SKIPPED) != 0;
260 static const struct got_error *
261 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
263 if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
264 return got_error(GOT_ERR_NO_SPACE);
266 if (RB_INSERT(got_fileindex_tree, &fileindex->entries, ie) != NULL)
267 return got_error_path(ie->path, GOT_ERR_FILEIDX_DUP_ENTRY);
269 fileindex->nentries++;
270 return NULL;
273 const struct got_error *
274 got_fileindex_entry_add(struct got_fileindex *fileindex,
275 struct got_fileindex_entry *ie)
277 /* Flag this entry until it gets written out to disk. */
278 ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
280 return add_entry(fileindex, ie);
283 void
284 got_fileindex_entry_remove(struct got_fileindex *fileindex,
285 struct got_fileindex_entry *ie)
287 /*
288 * Removing an entry from the RB tree immediately breaks
289 * in-progress iterations over file index entries.
290 * So flag this entry for removal and remove it once the index
291 * is written out to disk. Meanwhile, pretend this entry no longer
292 * exists if we get queried for it again before then.
293 */
294 ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
295 fileindex->nentries--;
298 struct got_fileindex_entry *
299 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
300 size_t path_len)
302 struct got_fileindex_entry *ie;
303 struct got_fileindex_entry key;
304 memset(&key, 0, sizeof(key));
305 key.path = (char *)path;
306 key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
307 ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
308 if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
309 return NULL;
310 return ie;
313 const struct got_error *
314 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
315 got_fileindex_cb cb, void *cb_arg)
317 const struct got_error *err;
318 struct got_fileindex_entry *ie, *tmp;
320 RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
321 if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
322 continue;
323 err = (*cb)(cb_arg, ie);
324 if (err)
325 return err;
327 return NULL;
330 struct got_fileindex *
331 got_fileindex_alloc(void)
333 struct got_fileindex *fileindex;
335 fileindex = calloc(1, sizeof(*fileindex));
336 if (fileindex == NULL)
337 return NULL;
339 RB_INIT(&fileindex->entries);
340 return fileindex;
343 void
344 got_fileindex_free(struct got_fileindex *fileindex)
346 struct got_fileindex_entry *ie;
348 while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
349 RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
350 got_fileindex_entry_free(ie);
352 free(fileindex);
355 static const struct got_error *
356 write_fileindex_val64(struct got_hash *ctx, uint64_t val, FILE *outfile)
358 size_t n;
360 val = htobe64(val);
361 got_hash_update(ctx, &val, sizeof(val));
362 n = fwrite(&val, 1, sizeof(val), outfile);
363 if (n != sizeof(val))
364 return got_ferror(outfile, GOT_ERR_IO);
365 return NULL;
368 static const struct got_error *
369 write_fileindex_val32(struct got_hash *ctx, uint32_t val, FILE *outfile)
371 size_t n;
373 val = htobe32(val);
374 got_hash_update(ctx, &val, sizeof(val));
375 n = fwrite(&val, 1, sizeof(val), outfile);
376 if (n != sizeof(val))
377 return got_ferror(outfile, GOT_ERR_IO);
378 return NULL;
381 static const struct got_error *
382 write_fileindex_val16(struct got_hash *ctx, uint16_t val, FILE *outfile)
384 size_t n;
386 val = htobe16(val);
387 got_hash_update(ctx, &val, sizeof(val));
388 n = fwrite(&val, 1, sizeof(val), outfile);
389 if (n != sizeof(val))
390 return got_ferror(outfile, GOT_ERR_IO);
391 return NULL;
394 static const struct got_error *
395 write_fileindex_path(struct got_hash *ctx, const char *path, FILE *outfile)
397 size_t n, len, pad = 0;
398 static const uint8_t zero[8] = { 0 };
400 len = strlen(path);
401 while ((len + pad) % 8 != 0)
402 pad++;
403 if (pad == 0)
404 pad = 8; /* NUL-terminate */
406 got_hash_update(ctx, path, len);
407 n = fwrite(path, 1, len, outfile);
408 if (n != len)
409 return got_ferror(outfile, GOT_ERR_IO);
410 got_hash_update(ctx, zero, pad);
411 n = fwrite(zero, 1, pad, outfile);
412 if (n != pad)
413 return got_ferror(outfile, GOT_ERR_IO);
414 return NULL;
417 static const struct got_error *
418 write_fileindex_entry(struct got_hash *ctx, struct got_fileindex_entry *ie,
419 FILE *outfile)
421 const struct got_error *err;
422 size_t n;
423 uint32_t stage;
425 err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
426 if (err)
427 return err;
428 err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
429 if (err)
430 return err;
431 err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
432 if (err)
433 return err;
434 err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
435 if (err)
436 return err;
438 err = write_fileindex_val32(ctx, ie->uid, outfile);
439 if (err)
440 return err;
441 err = write_fileindex_val32(ctx, ie->gid, outfile);
442 if (err)
443 return err;
444 err = write_fileindex_val32(ctx, ie->size, outfile);
445 if (err)
446 return err;
448 err = write_fileindex_val16(ctx, ie->mode, outfile);
449 if (err)
450 return err;
452 got_hash_update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
453 n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
454 if (n != SHA1_DIGEST_LENGTH)
455 return got_ferror(outfile, GOT_ERR_IO);
457 got_hash_update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
458 n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
459 if (n != SHA1_DIGEST_LENGTH)
460 return got_ferror(outfile, GOT_ERR_IO);
462 err = write_fileindex_val32(ctx, ie->flags, outfile);
463 if (err)
464 return err;
466 err = write_fileindex_path(ctx, ie->path, outfile);
467 if (err)
468 return err;
470 stage = got_fileindex_entry_stage_get(ie);
471 if (stage == GOT_FILEIDX_STAGE_MODIFY ||
472 stage == GOT_FILEIDX_STAGE_ADD) {
473 got_hash_update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
474 n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
475 outfile);
476 if (n != SHA1_DIGEST_LENGTH)
477 return got_ferror(outfile, GOT_ERR_IO);
480 return NULL;
483 const struct got_error *
484 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
486 const struct got_error *err = NULL;
487 struct got_fileindex_hdr hdr;
488 struct got_hash ctx;
489 uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
490 size_t n;
491 struct got_fileindex_entry *ie, *tmp;
493 got_hash_init(&ctx, GOT_HASH_SHA1);
495 hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
496 hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
497 hdr.nentries = htobe32(fileindex->nentries);
499 got_hash_update(&ctx, &hdr.signature, sizeof(hdr.signature));
500 got_hash_update(&ctx, &hdr.version, sizeof(hdr.version));
501 got_hash_update(&ctx, &hdr.nentries, sizeof(hdr.nentries));
502 n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
503 if (n != sizeof(hdr.signature))
504 return got_ferror(outfile, GOT_ERR_IO);
505 n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
506 if (n != sizeof(hdr.version))
507 return got_ferror(outfile, GOT_ERR_IO);
508 n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
509 if (n != sizeof(hdr.nentries))
510 return got_ferror(outfile, GOT_ERR_IO);
512 RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
513 ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
514 ie->flags &= ~GOT_FILEIDX_F_SKIPPED;
515 if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
516 RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
517 got_fileindex_entry_free(ie);
518 continue;
520 err = write_fileindex_entry(&ctx, ie, outfile);
521 if (err)
522 return err;
525 got_hash_final(&ctx, hash);
526 n = fwrite(hash, 1, SHA1_DIGEST_LENGTH, outfile);
527 if (n != SHA1_DIGEST_LENGTH)
528 return got_ferror(outfile, GOT_ERR_IO);
530 if (fflush(outfile) != 0)
531 return got_error_from_errno("fflush");
533 return NULL;
536 static const struct got_error *
537 read_fileindex_val64(uint64_t *val, struct got_hash *ctx, FILE *infile)
539 size_t n;
541 n = fread(val, 1, sizeof(*val), infile);
542 if (n != sizeof(*val))
543 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
544 got_hash_update(ctx, val, sizeof(*val));
545 *val = be64toh(*val);
546 return NULL;
549 static const struct got_error *
550 read_fileindex_val32(uint32_t *val, struct got_hash *ctx, FILE *infile)
552 size_t n;
554 n = fread(val, 1, sizeof(*val), infile);
555 if (n != sizeof(*val))
556 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
557 got_hash_update(ctx, val, sizeof(*val));
558 *val = be32toh(*val);
559 return NULL;
562 static const struct got_error *
563 read_fileindex_val16(uint16_t *val, struct got_hash *ctx, FILE *infile)
565 size_t n;
567 n = fread(val, 1, sizeof(*val), infile);
568 if (n != sizeof(*val))
569 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
570 got_hash_update(ctx, val, sizeof(*val));
571 *val = be16toh(*val);
572 return NULL;
575 static const struct got_error *
576 read_fileindex_path(char **path, struct got_hash *ctx, FILE *infile)
578 const size_t chunk_size = 8;
579 char p[PATH_MAX];
580 size_t n, len = 0;
582 do {
583 if (len + chunk_size > sizeof(p))
584 return got_error(GOT_ERR_FILEIDX_BAD);
586 n = fread(&p[len], 1, chunk_size, infile);
587 if (n != chunk_size)
588 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
590 got_hash_update(ctx, &p[len], chunk_size);
591 len += chunk_size;
592 } while (memchr(&p[len - chunk_size], '\0', chunk_size) == NULL);
594 *path = strdup(p);
595 if (*path == NULL)
596 return got_error_from_errno("strdup");
597 return NULL;
600 static const struct got_error *
601 read_fileindex_entry(struct got_fileindex_entry **iep, struct got_hash *ctx,
602 FILE *infile, uint32_t version)
604 const struct got_error *err;
605 struct got_fileindex_entry *ie;
606 size_t n;
608 *iep = NULL;
610 ie = calloc(1, sizeof(*ie));
611 if (ie == NULL)
612 return got_error_from_errno("calloc");
614 err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
615 if (err)
616 goto done;
617 err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
618 if (err)
619 goto done;
620 err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
621 if (err)
622 goto done;
623 err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
624 if (err)
625 goto done;
627 err = read_fileindex_val32(&ie->uid, ctx, infile);
628 if (err)
629 goto done;
630 err = read_fileindex_val32(&ie->gid, ctx, infile);
631 if (err)
632 goto done;
633 err = read_fileindex_val32(&ie->size, ctx, infile);
634 if (err)
635 goto done;
637 err = read_fileindex_val16(&ie->mode, ctx, infile);
638 if (err)
639 goto done;
641 n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
642 if (n != SHA1_DIGEST_LENGTH) {
643 err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
644 goto done;
646 got_hash_update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
648 n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
649 if (n != SHA1_DIGEST_LENGTH) {
650 err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
651 goto done;
653 got_hash_update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
655 err = read_fileindex_val32(&ie->flags, ctx, infile);
656 if (err)
657 goto done;
659 err = read_fileindex_path(&ie->path, ctx, infile);
660 if (err)
661 goto done;
663 if (version >= 2) {
664 uint32_t stage = got_fileindex_entry_stage_get(ie);
665 if (stage == GOT_FILEIDX_STAGE_MODIFY ||
666 stage == GOT_FILEIDX_STAGE_ADD) {
667 n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
668 infile);
669 if (n != SHA1_DIGEST_LENGTH) {
670 err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
671 goto done;
673 got_hash_update(ctx, ie->staged_blob_sha1,
674 SHA1_DIGEST_LENGTH);
676 } else {
677 /* GOT_FILE_INDEX_VERSION 1 does not support staging. */
678 ie->flags &= ~GOT_FILEIDX_F_STAGE;
681 done:
682 if (err)
683 got_fileindex_entry_free(ie);
684 else
685 *iep = ie;
686 return err;
689 const struct got_error *
690 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
692 const struct got_error *err = NULL;
693 struct got_fileindex_hdr hdr;
694 struct got_hash ctx;
695 struct got_fileindex_entry *ie;
696 uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
697 uint8_t sha1[SHA1_DIGEST_LENGTH];
698 size_t n;
699 int i;
701 got_hash_init(&ctx, GOT_HASH_SHA1);
703 n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
704 if (n != sizeof(hdr.signature)) {
705 if (n == 0) /* EOF */
706 return NULL;
707 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
709 n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
710 if (n != sizeof(hdr.version)) {
711 if (n == 0) /* EOF */
712 return NULL;
713 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
715 n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
716 if (n != sizeof(hdr.nentries)) {
717 if (n == 0) /* EOF */
718 return NULL;
719 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
722 got_hash_update(&ctx, &hdr.signature, sizeof(hdr.signature));
723 got_hash_update(&ctx, &hdr.version, sizeof(hdr.version));
724 got_hash_update(&ctx, &hdr.nentries, sizeof(hdr.nentries));
726 hdr.signature = be32toh(hdr.signature);
727 hdr.version = be32toh(hdr.version);
728 hdr.nentries = be32toh(hdr.nentries);
730 if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
731 return got_error(GOT_ERR_FILEIDX_SIG);
732 if (hdr.version > GOT_FILE_INDEX_VERSION)
733 return got_error(GOT_ERR_FILEIDX_VER);
735 for (i = 0; i < hdr.nentries; i++) {
736 err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
737 if (err)
738 return err;
739 err = add_entry(fileindex, ie);
740 if (err) {
741 got_fileindex_entry_free(ie);
742 return err;
746 n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
747 if (n != sizeof(sha1_expected))
748 return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
749 got_hash_final(&ctx, sha1);
750 if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
751 return got_error(GOT_ERR_FILEIDX_CSUM);
753 return NULL;
756 static struct got_fileindex_entry *
757 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
759 struct got_fileindex_entry *next;
761 next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
763 /* Skip entries which were added or removed by diff callbacks. */
764 while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
765 GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
766 next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
768 return next;
771 static const struct got_error *
772 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
773 struct got_tree_object *tree, const char *, const char *,
774 struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
776 static const struct got_error *
777 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
778 struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
779 const char *path, const char *entry_name, struct got_repository *repo,
780 struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
782 const struct got_error *err = NULL;
783 struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
785 if (!got_object_tree_entry_is_submodule(te) &&
786 S_ISDIR(got_tree_entry_get_mode(te))) {
787 char *subpath;
788 struct got_tree_object *subtree;
790 if (asprintf(&subpath, "%s%s%s", path,
791 path[0] == '\0' ? "" : "/",
792 got_tree_entry_get_name(te)) == -1)
793 return got_error_from_errno("asprintf");
795 err = got_object_open_as_tree(&subtree, repo,
796 got_tree_entry_get_id(te));
797 if (err) {
798 free(subpath);
799 return err;
802 err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
803 entry_name, repo, cb, cb_arg);
804 free(subpath);
805 got_object_tree_close(subtree);
806 if (err)
807 return err;
810 (*tidx)++;
811 *next = got_object_tree_get_entry(tree, *tidx);
812 return NULL;
815 static const struct got_error *
816 diff_fileindex_tree(struct got_fileindex *fileindex,
817 struct got_fileindex_entry **ie, struct got_tree_object *tree,
818 const char *path, const char *entry_name, struct got_repository *repo,
819 struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
821 const struct got_error *err = NULL;
822 struct got_tree_entry *te = NULL;
823 size_t path_len = strlen(path);
824 struct got_fileindex_entry *next;
825 int tidx = 0;
827 te = got_object_tree_get_entry(tree, tidx);
828 while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
829 if (te && *ie) {
830 char *te_path;
831 const char *te_name = got_tree_entry_get_name(te);
832 int cmp;
833 if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
834 err = got_error_from_errno("asprintf");
835 break;
837 cmp = got_path_cmp((*ie)->path, te_path,
838 got_fileindex_entry_path_len(*ie), strlen(te_path));
839 free(te_path);
840 if (cmp == 0) {
841 if (got_path_is_child((*ie)->path, path,
842 path_len) &&
843 !got_object_tree_entry_is_submodule(te) &&
844 (entry_name == NULL ||
845 strcmp(te_name, entry_name) == 0)) {
846 err = cb->diff_old_new(cb_arg, *ie, te,
847 path);
848 if (err || entry_name)
849 break;
851 *ie = walk_fileindex(fileindex, *ie);
852 err = walk_tree(&te, fileindex, ie, tree, &tidx,
853 path, entry_name, repo, cb, cb_arg);
854 } else if (cmp < 0) {
855 next = walk_fileindex(fileindex, *ie);
856 if (got_path_is_child((*ie)->path, path,
857 path_len) && entry_name == NULL) {
858 err = cb->diff_old(cb_arg, *ie, path);
859 if (err || entry_name)
860 break;
862 *ie = next;
863 } else {
864 if ((entry_name == NULL ||
865 strcmp(te_name, entry_name) == 0)) {
866 err = cb->diff_new(cb_arg, te, path);
867 if (err || entry_name)
868 break;
870 err = walk_tree(&te, fileindex, ie, tree, &tidx,
871 path, entry_name, repo, cb, cb_arg);
873 if (err)
874 break;
875 } else if (*ie) {
876 next = walk_fileindex(fileindex, *ie);
877 if (got_path_is_child((*ie)->path, path, path_len) &&
878 (entry_name == NULL ||
879 (te && strcmp(got_tree_entry_get_name(te),
880 entry_name) == 0))) {
881 err = cb->diff_old(cb_arg, *ie, path);
882 if (err || entry_name)
883 break;
885 *ie = next;
886 } else if (te) {
887 if (!got_object_tree_entry_is_submodule(te) &&
888 (entry_name == NULL ||
889 strcmp(got_tree_entry_get_name(te), entry_name)
890 == 0)) {
891 err = cb->diff_new(cb_arg, te, path);
892 if (err || entry_name)
893 break;
895 err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
896 entry_name, repo, cb, cb_arg);
897 if (err)
898 break;
902 return err;
905 const struct got_error *
906 got_fileindex_diff_tree(struct got_fileindex *fileindex,
907 struct got_tree_object *tree, const char *path, const char *entry_name,
908 struct got_repository *repo,
909 struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
911 struct got_fileindex_entry *ie;
912 ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
913 while (ie && !got_path_is_child(ie->path, path, strlen(path)))
914 ie = walk_fileindex(fileindex, ie);
915 return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
916 cb, cb_arg);
919 static const struct got_error *
920 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
921 struct got_pathlist_head *, int, const char *, const char *,
922 struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
924 static const struct got_error *
925 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
927 const struct got_error *err = NULL;
928 struct got_pathlist_entry *new = NULL;
929 struct dirent *dep = NULL;
930 struct dirent *de = NULL;
932 for (;;) {
933 de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
934 if (de == NULL) {
935 err = got_error_from_errno("malloc");
936 break;
939 if (readdir_r(dir, de, &dep) != 0) {
940 err = got_error_from_errno("readdir_r");
941 free(de);
942 break;
944 if (dep == NULL) {
945 free(de);
946 break;
949 if (strcmp(de->d_name, ".") == 0 ||
950 strcmp(de->d_name, "..") == 0 ||
951 (path[0] == '\0' &&
952 strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0) ||
953 (path[0] == '\0' &&
954 strcmp(de->d_name, GOT_WORKTREE_CVG_DIR) == 0)) {
955 free(de);
956 continue;
959 err = got_pathlist_insert(&new, dirlist, de->d_name, de);
960 if (err) {
961 free(de);
962 break;
964 if (new == NULL) {
965 err = got_error(GOT_ERR_DIR_DUP_ENTRY);
966 free(de);
967 break;
971 return err;
974 static int
975 have_tracked_file_in_dir(struct got_fileindex *fileindex, const char *path)
977 struct got_fileindex_entry *ie;
978 size_t path_len = strlen(path);
979 int cmp;
981 ie = RB_ROOT(&fileindex->entries);
982 while (ie) {
983 if (got_path_is_child(ie->path, path, path_len))
984 return 1;
985 cmp = got_path_cmp(path, ie->path, path_len,
986 got_fileindex_entry_path_len(ie));
987 if (cmp < 0)
988 ie = RB_LEFT(ie, entry);
989 else if (cmp > 0)
990 ie = RB_RIGHT(ie, entry);
991 else
992 break;
995 return 0;
998 static const struct got_error *
999 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
1000 struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
1001 const char *path, const char *rootpath, struct got_repository *repo,
1002 int ignore, struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1004 const struct got_error *err = NULL;
1005 struct dirent *de = dle->data;
1006 DIR *subdir = NULL;
1007 int subdirfd = -1;
1009 *next = NULL;
1011 /* Must traverse ignored directories if they contain tracked files. */
1012 if (de->d_type == DT_DIR && ignore &&
1013 have_tracked_file_in_dir(fileindex, path))
1014 ignore = 0;
1016 if (de->d_type == DT_DIR && !ignore) {
1017 char *subpath;
1018 char *subdirpath;
1019 struct got_pathlist_head subdirlist;
1021 TAILQ_INIT(&subdirlist);
1023 if (asprintf(&subpath, "%s%s%s", path,
1024 path[0] == '\0' ? "" : "/", de->d_name) == -1)
1025 return got_error_from_errno("asprintf");
1027 if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
1028 free(subpath);
1029 return got_error_from_errno("asprintf");
1032 subdirfd = openat(fd, de->d_name,
1033 O_RDONLY | O_NOFOLLOW | O_DIRECTORY | O_CLOEXEC);
1034 if (subdirfd == -1) {
1035 if (errno == EACCES) {
1036 *next = TAILQ_NEXT(dle, entry);
1037 return NULL;
1039 err = got_error_from_errno2("openat", subdirpath);
1040 free(subpath);
1041 free(subdirpath);
1042 return err;
1045 subdir = fdopendir(subdirfd);
1046 if (subdir == NULL)
1047 return got_error_from_errno2("fdopendir", path);
1048 subdirfd = -1;
1049 err = read_dirlist(&subdirlist, subdir, subdirpath);
1050 if (err) {
1051 free(subpath);
1052 free(subdirpath);
1053 closedir(subdir);
1054 return err;
1056 err = diff_fileindex_dir(fileindex, ie, &subdirlist,
1057 dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
1058 if (subdir && closedir(subdir) == -1 && err == NULL)
1059 err = got_error_from_errno2("closedir", subdirpath);
1060 free(subpath);
1061 free(subdirpath);
1062 got_pathlist_free(&subdirlist, GOT_PATHLIST_FREE_DATA);
1063 if (err)
1064 return err;
1067 *next = TAILQ_NEXT(dle, entry);
1068 return NULL;
1071 static const struct got_error *
1072 dirent_type_fixup(struct dirent *de, const char *rootpath, const char *path)
1074 const struct got_error *err;
1075 char *dir_path;
1076 int type;
1078 if (de->d_type != DT_UNKNOWN)
1079 return NULL;
1081 /* DT_UNKNOWN occurs on NFS mounts without "readdir plus" RPC. */
1082 if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
1083 return got_error_from_errno("asprintf");
1084 err = got_path_dirent_type(&type, dir_path, de);
1085 free(dir_path);
1086 if (err)
1087 return err;
1089 de->d_type = type;
1090 return NULL;
1093 static const struct got_error *
1094 diff_fileindex_dir(struct got_fileindex *fileindex,
1095 struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1096 int dirfd, const char *rootpath, const char *path,
1097 struct got_repository *repo,
1098 struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1100 const struct got_error *err = NULL;
1101 struct dirent *de = NULL;
1102 size_t path_len = strlen(path);
1103 struct got_pathlist_entry *dle;
1104 int ignore;
1106 if (cb->diff_traverse) {
1107 err = cb->diff_traverse(cb_arg, path, dirfd);
1108 if (err)
1109 return err;
1112 dle = TAILQ_FIRST(dirlist);
1113 while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1114 if (dle && *ie) {
1115 char *de_path;
1116 int cmp;
1117 de = dle->data;
1118 err = dirent_type_fixup(de, rootpath, path);
1119 if (err)
1120 break;
1121 if (asprintf(&de_path, "%s/%s", path,
1122 de->d_name) == -1) {
1123 err = got_error_from_errno("asprintf");
1124 break;
1126 cmp = got_path_cmp((*ie)->path, de_path,
1127 got_fileindex_entry_path_len(*ie),
1128 strlen(path) + 1 + strlen(de->d_name));
1129 free(de_path);
1130 if (cmp == 0) {
1131 err = cb->diff_old_new(cb_arg, *ie, de, path,
1132 dirfd);
1133 if (err)
1134 break;
1135 *ie = walk_fileindex(fileindex, *ie);
1136 err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1137 path, rootpath, repo, 0, cb, cb_arg);
1138 } else if (cmp < 0 ) {
1139 err = cb->diff_old(cb_arg, *ie, path);
1140 if (err)
1141 break;
1142 *ie = walk_fileindex(fileindex, *ie);
1143 } else {
1144 err = cb->diff_new(&ignore, cb_arg, de, path,
1145 dirfd);
1146 if (err)
1147 break;
1148 err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1149 path, rootpath, repo, ignore, cb, cb_arg);
1151 if (err)
1152 break;
1153 } else if (*ie) {
1154 err = cb->diff_old(cb_arg, *ie, path);
1155 if (err)
1156 break;
1157 *ie = walk_fileindex(fileindex, *ie);
1158 } else if (dle) {
1159 de = dle->data;
1160 err = dirent_type_fixup(de, rootpath, path);
1161 if (err)
1162 break;
1163 err = cb->diff_new(&ignore, cb_arg, de, path, dirfd);
1164 if (err)
1165 break;
1166 err = walk_dir(&dle, fileindex, ie, dle, dirfd, path,
1167 rootpath, repo, ignore, cb, cb_arg);
1168 if (err)
1169 break;
1173 return err;
1176 const struct got_error *
1177 got_fileindex_diff_dir(struct got_fileindex *fileindex, int fd,
1178 const char *rootpath, const char *path, struct got_repository *repo,
1179 struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1181 const struct got_error *err;
1182 struct got_fileindex_entry *ie;
1183 struct got_pathlist_head dirlist;
1184 int fd2;
1185 DIR *dir;
1187 TAILQ_INIT(&dirlist);
1190 * Duplicate the file descriptor so we can call closedir() below
1191 * without closing the file descriptor passed in by our caller.
1193 fd2 = dup(fd);
1194 if (fd2 == -1)
1195 return got_error_from_errno2("dup", path);
1196 if (lseek(fd2, 0, SEEK_SET) == -1) {
1197 err = got_error_from_errno2("lseek", path);
1198 close(fd2);
1199 return err;
1201 dir = fdopendir(fd2);
1202 if (dir == NULL) {
1203 err = got_error_from_errno2("fdopendir", path);
1204 close(fd2);
1205 return err;
1207 err = read_dirlist(&dirlist, dir, path);
1208 if (err) {
1209 closedir(dir);
1210 return err;
1213 ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1214 while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1215 ie = walk_fileindex(fileindex, ie);
1216 err = diff_fileindex_dir(fileindex, &ie, &dirlist, dirfd(dir),
1217 rootpath, path, repo, cb, cb_arg);
1219 if (closedir(dir) == -1 && err == NULL)
1220 err = got_error_from_errno2("closedir", path);
1221 got_pathlist_free(&dirlist, GOT_PATHLIST_FREE_DATA);
1222 return err;
1225 struct got_object_id *
1226 got_fileindex_entry_get_staged_blob_id(struct got_object_id *id,
1227 struct got_fileindex_entry *ie)
1229 memset(id, 0, sizeof(*id));
1230 memcpy(id->sha1, ie->staged_blob_sha1, sizeof(ie->staged_blob_sha1));
1231 return id;
1234 struct got_object_id *
1235 got_fileindex_entry_get_blob_id(struct got_object_id *id,
1236 struct got_fileindex_entry *ie)
1238 memset(id, 0, sizeof(*id));
1239 memcpy(id->sha1, ie->blob_sha1, sizeof(ie->blob_sha1));
1240 return id;
1243 struct got_object_id *
1244 got_fileindex_entry_get_commit_id(struct got_object_id *id,
1245 struct got_fileindex_entry *ie)
1247 memset(id, 0, sizeof(*id));
1248 memcpy(id->sha1, ie->commit_sha1, sizeof(ie->commit_sha1));
1249 return id;
1252 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);