op public repos

Blob

Date:: Tue Jan 19 07:38:37 2021 UTC
Message:: fix bug where 'got up -c commit path' deleted unrelated files from work tree Problem reported by Timo Myyrä
Actions:: History | Blame | Raw File
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  */
16 
17 #include <sys/queue.h>
18 #include <sys/tree.h>
19 #include <sys/stat.h>
20 
21 #include <errno.h>
22 #include <dirent.h>
23 #include <fcntl.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <sha1.h>
28 #include <endian.h>
29 #include <limits.h>
30 #include <unistd.h>
31 #include <uuid.h>
32 
33 #include "got_error.h"
34 #include "got_object.h"
35 #include "got_path.h"
36 
37 #include "got_lib_fileindex.h"
38 #include "got_lib_worktree.h"
39 
40 /* got_fileindex_entry flags */
41 #define GOT_FILEIDX_F_PATH_LEN		0x00000fff
42 #define GOT_FILEIDX_F_STAGE		0x0000f000
43 #define GOT_FILEIDX_F_STAGE_SHIFT	12
44 #define GOT_FILEIDX_F_NOT_FLUSHED	0x00010000
45 #define GOT_FILEIDX_F_NO_BLOB		0x00020000
46 #define GOT_FILEIDX_F_NO_COMMIT		0x00040000
47 #define GOT_FILEIDX_F_NO_FILE_ON_DISK	0x00080000
48 #define GOT_FILEIDX_F_REMOVE_ON_FLUSH	0x00100000
49 
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 };
55 
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 }
62 
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 }
70 
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;
77 
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;
83 
84 	return (ftype | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
85 }
86 
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;
93 
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;
103 	}
104 
105 
106 	if ((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0) {
107 		if (update_timestamps) {
108 			ie->ctime_sec = sb.st_ctim.tv_sec;
109 			ie->ctime_nsec = sb.st_ctim.tv_nsec;
110 			ie->mtime_sec = sb.st_mtim.tv_sec;
111 			ie->mtime_nsec = sb.st_mtim.tv_nsec;
112 		}
113 		ie->uid = sb.st_uid;
114 		ie->gid = sb.st_gid;
115 		ie->size = (sb.st_size & 0xffffffff);
116 		if (S_ISLNK(sb.st_mode)) {
117 			got_fileindex_entry_filetype_set(ie,
118 			    GOT_FILEIDX_MODE_SYMLINK);
119 			fileindex_entry_perms_set(ie, 0);
120 		} else {
121 			got_fileindex_entry_filetype_set(ie,
122 			    GOT_FILEIDX_MODE_REGULAR_FILE);
123 			fileindex_entry_perms_set(ie,
124 			    sb.st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
125 		}
126 	}
127 
128 	if (blob_sha1) {
129 		memcpy(ie->blob_sha1, blob_sha1, SHA1_DIGEST_LENGTH);
130 		ie->flags &= ~GOT_FILEIDX_F_NO_BLOB;
131 	} else
132 		ie->flags |= GOT_FILEIDX_F_NO_BLOB;
133 
134 	if (commit_sha1) {
135 		memcpy(ie->commit_sha1, commit_sha1, SHA1_DIGEST_LENGTH);
136 		ie->flags &= ~GOT_FILEIDX_F_NO_COMMIT;
137 	} else
138 		ie->flags |= GOT_FILEIDX_F_NO_COMMIT;
139 
140 	return NULL;
141 }
142 
143 void
144 got_fileindex_entry_mark_deleted_from_disk(struct got_fileindex_entry *ie)
145 {
146 	ie->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
147 }
148 
149 const struct got_error *
150 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
151     const char *relpath)
152 {
153 	size_t len;
154 
155 	*ie = calloc(1, sizeof(**ie));
156 	if (*ie == NULL)
157 		return got_error_from_errno("calloc");
158 
159 	(*ie)->path = strdup(relpath);
160 	if ((*ie)->path == NULL) {
161 		const struct got_error *err = got_error_from_errno("strdup");
162 		free(*ie);
163 		*ie = NULL;
164 		return err;
165 	}
166 
167 	len = strlen(relpath);
168 	if (len > GOT_FILEIDX_F_PATH_LEN)
169 		len = GOT_FILEIDX_F_PATH_LEN;
170 	(*ie)->flags |= len;
171 
172 	return NULL;
173 }
174 
175 void
176 got_fileindex_entry_free(struct got_fileindex_entry *ie)
177 {
178 	free(ie->path);
179 	free(ie);
180 }
181 
182 size_t
183 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
184 {
185 	return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
186 }
187 
188 uint32_t
189 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
190 {
191 	return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
192 }
193 
194 void
195 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
196 {
197 	ie->flags &= ~GOT_FILEIDX_F_STAGE;
198 	ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
199 	    GOT_FILEIDX_F_STAGE);
200 }
201 
202 int
203 got_fileindex_entry_filetype_get(struct got_fileindex_entry *ie)
204 {
205 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
206 }
207 
208 void
209 got_fileindex_entry_filetype_set(struct got_fileindex_entry *ie, int type)
210 {
211 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_ONDISK;
212 	ie->mode |= (type & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
213 }
214 
215 void
216 got_fileindex_entry_staged_filetype_set(struct got_fileindex_entry *ie, int type)
217 {
218 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_STAGED;
219 	ie->mode |= ((type << GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT) &
220 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED);
221 }
222 
223 int
224 got_fileindex_entry_staged_filetype_get(struct got_fileindex_entry *ie)
225 {
226 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_STAGED) >>
227 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT;
228 }
229 
230 int
231 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
232 {
233 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
234 }
235 
236 int
237 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
238 {
239 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
240 }
241 
242 int
243 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
244 {
245 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
246 }
247 
248 static const struct got_error *
249 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
250 {
251 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
252 		return got_error(GOT_ERR_NO_SPACE);
253 
254 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
255 	fileindex->nentries++;
256 	return NULL;
257 }
258 
259 const struct got_error *
260 got_fileindex_entry_add(struct got_fileindex *fileindex,
261     struct got_fileindex_entry *ie)
262 {
263 	/* Flag this entry until it gets written out to disk. */
264 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
265 
266 	return add_entry(fileindex, ie);
267 }
268 
269 void
270 got_fileindex_entry_remove(struct got_fileindex *fileindex,
271     struct got_fileindex_entry *ie)
272 {
273 	/*
274 	 * Removing an entry from the RB tree immediately breaks
275 	 * in-progress iterations over file index entries.
276 	 * So flag this entry for removal and remove it once the index
277 	 * is written out to disk. Meanwhile, pretend this entry no longer
278 	 * exists if we get queried for it again before then.
279 	 */
280 	ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
281 	fileindex->nentries--;
282 }
283 
284 struct got_fileindex_entry *
285 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
286     size_t path_len)
287 {
288 	struct got_fileindex_entry *ie;
289 	struct got_fileindex_entry key;
290 	memset(&key, 0, sizeof(key));
291 	key.path = (char *)path;
292 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
293 	ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
294 	if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
295 		return NULL;
296 	return ie;
297 }
298 
299 const struct got_error *
300 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
301     got_fileindex_cb cb, void *cb_arg)
302 {
303 	const struct got_error *err;
304 	struct got_fileindex_entry *ie, *tmp;
305 
306 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
307 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
308 			continue;
309 		err = (*cb)(cb_arg, ie);
310 		if (err)
311 			return err;
312 	}
313 	return NULL;
314 }
315 
316 struct got_fileindex *
317 got_fileindex_alloc(void)
318 {
319 	struct got_fileindex *fileindex;
320 
321 	fileindex = calloc(1, sizeof(*fileindex));
322 	if (fileindex == NULL)
323 		return NULL;
324 
325 	RB_INIT(&fileindex->entries);
326 	return fileindex;
327 }
328 
329 void
330 got_fileindex_free(struct got_fileindex *fileindex)
331 {
332 	struct got_fileindex_entry *ie;
333 
334 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
335 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
336 		got_fileindex_entry_free(ie);
337 	}
338 	free(fileindex);
339 }
340 
341 static const struct got_error *
342 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
343 {
344 	size_t n;
345 
346 	val = htobe64(val);
347 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
348 	n = fwrite(&val, 1, sizeof(val), outfile);
349 	if (n != sizeof(val))
350 		return got_ferror(outfile, GOT_ERR_IO);
351 	return NULL;
352 }
353 
354 static const struct got_error *
355 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
356 {
357 	size_t n;
358 
359 	val = htobe32(val);
360 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
361 	n = fwrite(&val, 1, sizeof(val), outfile);
362 	if (n != sizeof(val))
363 		return got_ferror(outfile, GOT_ERR_IO);
364 	return NULL;
365 }
366 
367 static const struct got_error *
368 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
369 {
370 	size_t n;
371 
372 	val = htobe16(val);
373 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
374 	n = fwrite(&val, 1, sizeof(val), outfile);
375 	if (n != sizeof(val))
376 		return got_ferror(outfile, GOT_ERR_IO);
377 	return NULL;
378 }
379 
380 static const struct got_error *
381 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
382 {
383 	size_t n, len, pad = 0;
384 	static const uint8_t zero[8] = { 0 };
385 
386 	len = strlen(path);
387 	while ((len + pad) % 8 != 0)
388 		pad++;
389 	if (pad == 0)
390 		pad = 8; /* NUL-terminate */
391 
392 	SHA1Update(ctx, path, len);
393 	n = fwrite(path, 1, len, outfile);
394 	if (n != len)
395 		return got_ferror(outfile, GOT_ERR_IO);
396 	SHA1Update(ctx, zero, pad);
397 	n = fwrite(zero, 1, pad, outfile);
398 	if (n != pad)
399 		return got_ferror(outfile, GOT_ERR_IO);
400 	return NULL;
401 }
402 
403 static const struct got_error *
404 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
405     FILE *outfile)
406 {
407 	const struct got_error *err;
408 	size_t n;
409 	uint32_t stage;
410 
411 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
412 	if (err)
413 		return err;
414 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
415 	if (err)
416 		return err;
417 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
418 	if (err)
419 		return err;
420 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
421 	if (err)
422 		return err;
423 
424 	err = write_fileindex_val32(ctx, ie->uid, outfile);
425 	if (err)
426 		return err;
427 	err = write_fileindex_val32(ctx, ie->gid, outfile);
428 	if (err)
429 		return err;
430 	err = write_fileindex_val32(ctx, ie->size, outfile);
431 	if (err)
432 		return err;
433 
434 	err = write_fileindex_val16(ctx, ie->mode, outfile);
435 	if (err)
436 		return err;
437 
438 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
439 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
440 	if (n != SHA1_DIGEST_LENGTH)
441 		return got_ferror(outfile, GOT_ERR_IO);
442 
443 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
444 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
445 	if (n != SHA1_DIGEST_LENGTH)
446 		return got_ferror(outfile, GOT_ERR_IO);
447 
448 	err = write_fileindex_val32(ctx, ie->flags, outfile);
449 	if (err)
450 		return err;
451 
452 	err = write_fileindex_path(ctx, ie->path, outfile);
453 	if (err)
454 		return err;
455 
456 	stage = got_fileindex_entry_stage_get(ie);
457 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
458 	    stage == GOT_FILEIDX_STAGE_ADD) {
459 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
460 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
461 		    outfile);
462 		if (n != SHA1_DIGEST_LENGTH)
463 			return got_ferror(outfile, GOT_ERR_IO);
464 	}
465 
466 	return NULL;
467 }
468 
469 const struct got_error *
470 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
471 {
472 	const struct got_error *err = NULL;
473 	struct got_fileindex_hdr hdr;
474 	SHA1_CTX ctx;
475 	uint8_t sha1[SHA1_DIGEST_LENGTH];
476 	size_t n;
477 	struct got_fileindex_entry *ie, *tmp;
478 
479 	SHA1Init(&ctx);
480 
481 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
482 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
483 	hdr.nentries = htobe32(fileindex->nentries);
484 
485 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
486 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
487 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
488 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
489 	if (n != sizeof(hdr.signature))
490 		return got_ferror(outfile, GOT_ERR_IO);
491 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
492 	if (n != sizeof(hdr.version))
493 		return got_ferror(outfile, GOT_ERR_IO);
494 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
495 	if (n != sizeof(hdr.nentries))
496 		return got_ferror(outfile, GOT_ERR_IO);
497 
498 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
499 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
500 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
501 			RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
502 			got_fileindex_entry_free(ie);
503 			continue;
504 		}
505 		err = write_fileindex_entry(&ctx, ie, outfile);
506 		if (err)
507 			return err;
508 	}
509 
510 	SHA1Final(sha1, &ctx);
511 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
512 	if (n != sizeof(sha1))
513 		return got_ferror(outfile, GOT_ERR_IO);
514 
515 	if (fflush(outfile) != 0)
516 		return got_error_from_errno("fflush");
517 
518 	return NULL;
519 }
520 
521 static const struct got_error *
522 read_fileindex_val64(uint64_t *val, SHA1_CTX *ctx, FILE *infile)
523 {
524 	size_t n;
525 
526 	n = fread(val, 1, sizeof(*val), infile);
527 	if (n != sizeof(*val))
528 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
529 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
530 	*val = be64toh(*val);
531 	return NULL;
532 }
533 
534 static const struct got_error *
535 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
536 {
537 	size_t n;
538 
539 	n = fread(val, 1, sizeof(*val), infile);
540 	if (n != sizeof(*val))
541 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
542 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
543 	*val = be32toh(*val);
544 	return NULL;
545 }
546 
547 static const struct got_error *
548 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
549 {
550 	size_t n;
551 
552 	n = fread(val, 1, sizeof(*val), infile);
553 	if (n != sizeof(*val))
554 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
555 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
556 	*val = be16toh(*val);
557 	return NULL;
558 }
559 
560 static const struct got_error *
561 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
562 {
563 	const struct got_error *err = NULL;
564 	const size_t chunk_size = 8;
565 	size_t n, len = 0, totlen = chunk_size;
566 
567 	*path = malloc(totlen);
568 	if (*path == NULL)
569 		return got_error_from_errno("malloc");
570 
571 	do {
572 		if (len + chunk_size > totlen) {
573 			char *p = reallocarray(*path, totlen + chunk_size, 1);
574 			if (p == NULL) {
575 				err = got_error_from_errno("reallocarray");
576 				break;
577 			}
578 			totlen += chunk_size;
579 			*path = p;
580 		}
581 		n = fread(*path + len, 1, chunk_size, infile);
582 		if (n != chunk_size) {
583 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
584 			break;
585 		}
586 		SHA1Update(ctx, *path + len, chunk_size);
587 		len += chunk_size;
588 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
589 
590 	if (err) {
591 		free(*path);
592 		*path = NULL;
593 	}
594 	return err;
595 }
596 
597 static const struct got_error *
598 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
599     FILE *infile, uint32_t version)
600 {
601 	const struct got_error *err;
602 	struct got_fileindex_entry *ie;
603 	size_t n;
604 
605 	*iep = NULL;
606 
607 	ie = calloc(1, sizeof(*ie));
608 	if (ie == NULL)
609 		return got_error_from_errno("calloc");
610 
611 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
612 	if (err)
613 		goto done;
614 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
615 	if (err)
616 		goto done;
617 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
618 	if (err)
619 		goto done;
620 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
621 	if (err)
622 		goto done;
623 
624 	err = read_fileindex_val32(&ie->uid, ctx, infile);
625 	if (err)
626 		goto done;
627 	err = read_fileindex_val32(&ie->gid, ctx, infile);
628 	if (err)
629 		goto done;
630 	err = read_fileindex_val32(&ie->size, ctx, infile);
631 	if (err)
632 		goto done;
633 
634 	err = read_fileindex_val16(&ie->mode, ctx, infile);
635 	if (err)
636 		goto done;
637 
638 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
639 	if (n != SHA1_DIGEST_LENGTH) {
640 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
641 		goto done;
642 	}
643 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
644 
645 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
646 	if (n != SHA1_DIGEST_LENGTH) {
647 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
648 		goto done;
649 	}
650 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
651 
652 	err = read_fileindex_val32(&ie->flags, ctx, infile);
653 	if (err)
654 		goto done;
655 
656 	err = read_fileindex_path(&ie->path, ctx, infile);
657 	if (err)
658 		goto done;
659 
660 	if (version >= 2) {
661 		uint32_t stage = got_fileindex_entry_stage_get(ie);
662 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
663 		    stage == GOT_FILEIDX_STAGE_ADD) {
664 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
665 			    infile);
666 			if (n != SHA1_DIGEST_LENGTH) {
667 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
668 				goto done;
669 			}
670 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
671 		}
672 	} else {
673 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
674 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
675 	}
676 
677 done:
678 	if (err)
679 		got_fileindex_entry_free(ie);
680 	else
681 		*iep = ie;
682 	return err;
683 }
684 
685 const struct got_error *
686 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
687 {
688 	const struct got_error *err = NULL;
689 	struct got_fileindex_hdr hdr;
690 	SHA1_CTX ctx;
691 	struct got_fileindex_entry *ie;
692 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
693 	uint8_t sha1[SHA1_DIGEST_LENGTH];
694 	size_t n;
695 	int i;
696 
697 	SHA1Init(&ctx);
698 
699 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
700 	if (n != sizeof(hdr.signature)) {
701 		if (n == 0) /* EOF */
702 			return NULL;
703 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
704 	}
705 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
706 	if (n != sizeof(hdr.version)) {
707 		if (n == 0) /* EOF */
708 			return NULL;
709 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
710 	}
711 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
712 	if (n != sizeof(hdr.nentries)) {
713 		if (n == 0) /* EOF */
714 			return NULL;
715 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
716 	}
717 
718 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
719 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
720 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
721 
722 	hdr.signature = be32toh(hdr.signature);
723 	hdr.version = be32toh(hdr.version);
724 	hdr.nentries = be32toh(hdr.nentries);
725 
726 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
727 		return got_error(GOT_ERR_FILEIDX_SIG);
728 	if (hdr.version > GOT_FILE_INDEX_VERSION)
729 		return got_error(GOT_ERR_FILEIDX_VER);
730 
731 	for (i = 0; i < hdr.nentries; i++) {
732 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
733 		if (err)
734 			return err;
735 		err = add_entry(fileindex, ie);
736 		if (err)
737 			return err;
738 	}
739 
740 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
741 	if (n != sizeof(sha1_expected))
742 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
743 	SHA1Final(sha1, &ctx);
744 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
745 		return got_error(GOT_ERR_FILEIDX_CSUM);
746 
747 	return NULL;
748 }
749 
750 static struct got_fileindex_entry *
751 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
752 {
753 	struct got_fileindex_entry *next;
754 
755 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
756 
757 	/* Skip entries which were added or removed by diff callbacks. */
758 	while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
759 	    GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
760 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
761 
762 	return next;
763 }
764 
765 static const struct got_error *
766 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
767     struct got_tree_object *tree, const char *, const char *,
768     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
769 
770 static const struct got_error *
771 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
772     struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
773     const char *path, const char *entry_name, struct got_repository *repo,
774     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
775 {
776 	const struct got_error *err = NULL;
777 	struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
778 
779 	if (!got_object_tree_entry_is_submodule(te) &&
780 	    S_ISDIR(got_tree_entry_get_mode(te))) {
781 		char *subpath;
782 		struct got_tree_object *subtree;
783 
784 		if (asprintf(&subpath, "%s%s%s", path,
785 		    path[0] == '\0' ? "" : "/",
786 		    got_tree_entry_get_name(te)) == -1)
787 			return got_error_from_errno("asprintf");
788 
789 		err = got_object_open_as_tree(&subtree, repo,
790 		    got_tree_entry_get_id(te));
791 		if (err) {
792 			free(subpath);
793 			return err;
794 		}
795 
796 		err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
797 		    entry_name, repo, cb, cb_arg);
798 		free(subpath);
799 		got_object_tree_close(subtree);
800 		if (err)
801 			return err;
802 	}
803 
804 	(*tidx)++;
805 	*next = got_object_tree_get_entry(tree, *tidx);
806 	return NULL;
807 }
808 
809 static const struct got_error *
810 diff_fileindex_tree(struct got_fileindex *fileindex,
811     struct got_fileindex_entry **ie, struct got_tree_object *tree,
812     const char *path, const char *entry_name, struct got_repository *repo,
813     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
814 {
815 	const struct got_error *err = NULL;
816 	struct got_tree_entry *te = NULL;
817 	size_t path_len = strlen(path);
818 	struct got_fileindex_entry *next;
819 	int tidx = 0;
820 
821 	te = got_object_tree_get_entry(tree, tidx);
822 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
823 		if (te && *ie) {
824 			char *te_path;
825 			const char *te_name = got_tree_entry_get_name(te);
826 			int cmp;
827 			if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
828 				err = got_error_from_errno("asprintf");
829 				break;
830 			}
831 			cmp = got_path_cmp((*ie)->path, te_path,
832 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
833 			free(te_path);
834 			if (cmp == 0) {
835 				if (got_path_is_child((*ie)->path, path,
836 				    path_len) &&
837 				    !got_object_tree_entry_is_submodule(te) &&
838 				    (entry_name == NULL ||
839 				    strcmp(te_name, entry_name) == 0)) {
840 					err = cb->diff_old_new(cb_arg, *ie, te,
841 					    path);
842 					if (err || entry_name)
843 						break;
844 				}
845 				*ie = walk_fileindex(fileindex, *ie);
846 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
847 				    path, entry_name, repo, cb, cb_arg);
848 			} else if (cmp < 0) {
849 				next = walk_fileindex(fileindex, *ie);
850 				if (got_path_is_child((*ie)->path, path,
851 				    path_len) && entry_name == NULL) {
852 					err = cb->diff_old(cb_arg, *ie, path);
853 					if (err || entry_name)
854 						break;
855 				}
856 				*ie = next;
857 			} else {
858 				if ((entry_name == NULL ||
859 				    strcmp(te_name, entry_name) == 0)) {
860 					err = cb->diff_new(cb_arg, te, path);
861 					if (err || entry_name)
862 						break;
863 				}
864 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
865 				    path, entry_name, repo, cb, cb_arg);
866 			}
867 			if (err)
868 				break;
869 		} else if (*ie) {
870 			next = walk_fileindex(fileindex, *ie);
871 			if (got_path_is_child((*ie)->path, path, path_len) &&
872 			    (entry_name == NULL ||
873 			    (te && strcmp(got_tree_entry_get_name(te),
874 			    entry_name) == 0))) {
875 				err = cb->diff_old(cb_arg, *ie, path);
876 				if (err || entry_name)
877 					break;
878 			}
879 			*ie = next;
880 		} else if (te) {
881 			if (!got_object_tree_entry_is_submodule(te) &&
882 			    (entry_name == NULL ||
883 			    strcmp(got_tree_entry_get_name(te), entry_name)
884 			    == 0)) {
885 				err = cb->diff_new(cb_arg, te, path);
886 				if (err || entry_name)
887 					break;
888 			}
889 			err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
890 			    entry_name, repo, cb, cb_arg);
891 			if (err)
892 				break;
893 		}
894 	}
895 
896 	return err;
897 }
898 
899 const struct got_error *
900 got_fileindex_diff_tree(struct got_fileindex *fileindex,
901     struct got_tree_object *tree, const char *path, const char *entry_name,
902     struct got_repository *repo,
903     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
904 {
905 	struct got_fileindex_entry *ie;
906 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
907 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
908 		ie = walk_fileindex(fileindex, ie);
909 	return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
910 	    cb, cb_arg);
911 }
912 
913 static const struct got_error *
914 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
915     struct got_pathlist_head *, int, const char *, const char *,
916     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
917 
918 static const struct got_error *
919 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
920 {
921 	const struct got_error *err = NULL;
922 	struct got_pathlist_entry *new = NULL;
923 	struct dirent *dep = NULL;
924 	struct dirent *de = NULL;
925 
926 	for (;;) {
927 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
928 		if (de == NULL) {
929 			err = got_error_from_errno("malloc");
930 			break;
931 		}
932 
933 		if (readdir_r(dir, de, &dep) != 0) {
934 			err = got_error_from_errno("readdir_r");
935 			free(de);
936 			break;
937 		}
938 		if (dep == NULL) {
939 			free(de);
940 			break;
941 		}
942 
943 		if (strcmp(de->d_name, ".") == 0 ||
944 		    strcmp(de->d_name, "..") == 0 ||
945 		    (path[0] == '\0' &&
946 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
947 			free(de);
948 			continue;
949 		}
950 
951 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
952 		if (err) {
953 			free(de);
954 			break;
955 		}
956 		if (new == NULL) {
957 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
958 			free(de);
959 			break;
960 		}
961 	}
962 
963 	return err;
964 }
965 
966 void
967 free_dirlist(struct got_pathlist_head *dirlist)
968 {
969 	struct got_pathlist_entry *dle;
970 
971 	TAILQ_FOREACH(dle, dirlist, entry)
972 		free(dle->data);
973 	got_pathlist_free(dirlist);
974 }
975 
976 static const struct got_error *
977 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
978     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
979     const char *path, const char *rootpath, struct got_repository *repo,
980     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
981 {
982 	const struct got_error *err = NULL;
983 	struct dirent *de = dle->data;
984 	DIR *subdir = NULL;
985 	int subdirfd = -1;
986 	int type;
987 
988 	*next = NULL;
989 
990 	if (de->d_type == DT_UNKNOWN) {
991 		/* Occurs on NFS mounts without "readdir plus" RPC. */
992 		char *dir_path;
993 		if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
994 			return got_error_from_errno("asprintf");
995 		err = got_path_dirent_type(&type, dir_path, de);
996 		free(dir_path);
997 		if (err)
998 			return err;
999 	} else
1000 		type = de->d_type;
1001 
1002 	if (type == DT_DIR) {
1003 		char *subpath;
1004 		char *subdirpath;
1005 		struct got_pathlist_head subdirlist;
1006 
1007 		TAILQ_INIT(&subdirlist);
1008 
1009 		if (asprintf(&subpath, "%s%s%s", path,
1010 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1011 			return got_error_from_errno("asprintf");
1012 
1013 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
1014 			free(subpath);
1015 			return got_error_from_errno("asprintf");
1016 		}
1017 
1018 		subdirfd = openat(fd, de->d_name,
1019 		    O_RDONLY | O_NOFOLLOW | O_DIRECTORY);
1020 		if (subdirfd == -1) {
1021 			if (errno == EACCES) {
1022 				*next = TAILQ_NEXT(dle, entry);
1023 				return NULL;
1024 			}
1025 			err = got_error_from_errno2("openat", subdirpath);
1026 			free(subpath);
1027 			free(subdirpath);
1028 			return err;
1029 		}
1030 
1031 		subdir = fdopendir(subdirfd);
1032 		if (subdir == NULL)
1033 			return got_error_from_errno2("fdopendir", path);
1034 		subdirfd = -1;
1035 		err = read_dirlist(&subdirlist, subdir, subdirpath);
1036 		if (err) {
1037 			free(subpath);
1038 			free(subdirpath);
1039 			closedir(subdir);
1040 			return err;
1041 		}
1042 		err = diff_fileindex_dir(fileindex, ie, &subdirlist,
1043 		    dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
1044 		if (subdir && closedir(subdir) == -1 && err == NULL)
1045 			err = got_error_from_errno2("closedir", subdirpath);
1046 		free(subpath);
1047 		free(subdirpath);
1048 		free_dirlist(&subdirlist);
1049 		if (err)
1050 			return err;
1051 	}
1052 
1053 	*next = TAILQ_NEXT(dle, entry);
1054 	return NULL;
1055 }
1056 
1057 static const struct got_error *
1058 diff_fileindex_dir(struct got_fileindex *fileindex,
1059     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1060     int dirfd, const char *rootpath, const char *path,
1061     struct got_repository *repo,
1062     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1063 {
1064 	const struct got_error *err = NULL;
1065 	struct dirent *de = NULL;
1066 	size_t path_len = strlen(path);
1067 	struct got_pathlist_entry *dle;
1068 
1069 	if (cb->diff_traverse) {
1070 		err = cb->diff_traverse(cb_arg, path, dirfd);
1071 		if (err)
1072 			return err;
1073 	}
1074 
1075 	dle = TAILQ_FIRST(dirlist);
1076 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1077 		if (dle && *ie) {
1078 			char *de_path;
1079 			int cmp;
1080 			de = dle->data;
1081 			if (asprintf(&de_path, "%s/%s", path,
1082 			    de->d_name) == -1) {
1083 				err = got_error_from_errno("asprintf");
1084 				break;
1085 			}
1086 			cmp = got_path_cmp((*ie)->path, de_path,
1087 			    got_fileindex_entry_path_len(*ie),
1088 			    strlen(path) + 1 + de->d_namlen);
1089 			free(de_path);
1090 			if (cmp == 0) {
1091 				err = cb->diff_old_new(cb_arg, *ie, de, path,
1092 				    dirfd);
1093 				if (err)
1094 					break;
1095 				*ie = walk_fileindex(fileindex, *ie);
1096 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1097 				    path, rootpath, repo, cb, cb_arg);
1098 			} else if (cmp < 0 ) {
1099 				err = cb->diff_old(cb_arg, *ie, path);
1100 				if (err)
1101 					break;
1102 				*ie = walk_fileindex(fileindex, *ie);
1103 			} else {
1104 				err = cb->diff_new(cb_arg, de, path, dirfd);
1105 				if (err)
1106 					break;
1107 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1108 				    path, rootpath, repo, cb, cb_arg);
1109 			}
1110 			if (err)
1111 				break;
1112 		} else if (*ie) {
1113 			err = cb->diff_old(cb_arg, *ie, path);
1114 			if (err)
1115 				break;
1116 			*ie = walk_fileindex(fileindex, *ie);
1117 		} else if (dle) {
1118 			de = dle->data;
1119 			err = cb->diff_new(cb_arg, de, path, dirfd);
1120 			if (err)
1121 				break;
1122 			err = walk_dir(&dle, fileindex, ie, dle, dirfd, path,
1123 			    rootpath, repo, cb, cb_arg);
1124 			if (err)
1125 				break;
1126 		}
1127 	}
1128 
1129 	return err;
1130 }
1131 
1132 const struct got_error *
1133 got_fileindex_diff_dir(struct got_fileindex *fileindex, int fd,
1134     const char *rootpath, const char *path, struct got_repository *repo,
1135     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1136 {
1137 	const struct got_error *err;
1138 	struct got_fileindex_entry *ie;
1139 	struct got_pathlist_head dirlist;
1140 	int fd2;
1141 	DIR *dir;
1142 
1143 	TAILQ_INIT(&dirlist);
1144 
1145 	/* 
1146 	 * Duplicate the file descriptor so we can call closedir() below
1147 	 * without closing the file descriptor passed in by our caller.
1148 	 */
1149 	fd2 = dup(fd);
1150 	if (fd2 == -1)
1151 		return got_error_from_errno2("dup", path);
1152 	if (lseek(fd2, 0, SEEK_SET) == -1) {
1153 		err = got_error_from_errno2("lseek", path);
1154 		close(fd2);
1155 		return err;
1156 	}
1157 	dir = fdopendir(fd2);
1158 	if (dir == NULL) {
1159 		err = got_error_from_errno2("fdopendir", path);
1160 		close(fd2);
1161 		return err;
1162 	}
1163 	err = read_dirlist(&dirlist, dir, path);
1164 	if (err) {
1165 		closedir(dir);
1166 		return err;
1167 	}
1168 
1169 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1170 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1171 		ie = walk_fileindex(fileindex, ie);
1172 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, dirfd(dir),
1173 	    rootpath, path, repo, cb, cb_arg);
1174 
1175 	if (closedir(dir) == -1 && err == NULL)
1176 		err = got_error_from_errno2("closedir", path);
1177 	free_dirlist(&dirlist);
1178 	return err;
1179 }
1180 
1181 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);