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/stat.h>
18 
19 #include <errno.h>
20 #include <dirent.h>
21 #include <fcntl.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include <limits.h>
26 #include <unistd.h>
27 #include <uuid.h>
28 
29 #include "got_compat.h"
30 
31 #include "got_error.h"
32 #include "got_object.h"
33 #include "got_path.h"
34 
35 #include "got_lib_fileindex.h"
36 #include "got_lib_worktree.h"
37 
38 /* got_fileindex_entry flags */
39 #define GOT_FILEIDX_F_PATH_LEN		0x00000fff
40 #define GOT_FILEIDX_F_STAGE		0x0000f000
41 #define GOT_FILEIDX_F_STAGE_SHIFT	12
42 #define GOT_FILEIDX_F_NOT_FLUSHED	0x00010000
43 #define GOT_FILEIDX_F_NO_BLOB		0x00020000
44 #define GOT_FILEIDX_F_NO_COMMIT		0x00040000
45 #define GOT_FILEIDX_F_NO_FILE_ON_DISK	0x00080000
46 #define GOT_FILEIDX_F_REMOVE_ON_FLUSH	0x00100000
47 #define GOT_FILEIDX_F_SKIPPED		0x00200000
48 
49 struct got_fileindex {
50 	struct got_fileindex_tree entries;
51 	int nentries; /* Does not include entries marked for removal. */
52 #define GOT_FILEIDX_MAX_ENTRIES INT_MAX
53 };
54 
55 mode_t
56 got_fileindex_entry_perms_get(struct got_fileindex_entry *ie)
57 {
58 	return ((ie->mode & GOT_FILEIDX_MODE_PERMS) >>
59 	    GOT_FILEIDX_MODE_PERMS_SHIFT);
60 }
61 
62 static void
63 fileindex_entry_perms_set(struct got_fileindex_entry *ie, mode_t mode)
64 {
65 	ie->mode &= ~GOT_FILEIDX_MODE_PERMS;
66 	ie->mode |= ((mode << GOT_FILEIDX_MODE_PERMS_SHIFT) &
67 	    GOT_FILEIDX_MODE_PERMS);
68 }
69 
70 mode_t
71 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
72 {
73 	mode_t perms = got_fileindex_entry_perms_get(ie);
74 	int type = got_fileindex_entry_filetype_get(ie);
75 	uint32_t ftype;
76 
77 	if (type == GOT_FILEIDX_MODE_REGULAR_FILE ||
78 	    type == GOT_FILEIDX_MODE_BAD_SYMLINK)
79 		ftype = S_IFREG;
80 	else
81 		ftype = S_IFLNK;
82 
83 	return (ftype | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
84 }
85 
86 const struct got_error *
87 got_fileindex_entry_update(struct got_fileindex_entry *ie,
88     int wt_fd, const char *ondisk_path, uint8_t *blob_sha1,
89     uint8_t *commit_sha1, int update_timestamps)
90 {
91 	struct stat sb;
92 
93 	if (fstatat(wt_fd, ondisk_path, &sb, AT_SYMLINK_NOFOLLOW) != 0) {
94 		if (!((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) &&
95 		    errno == ENOENT))
96 			return got_error_from_errno2("fstatat", ondisk_path);
97 		sb.st_mode = GOT_DEFAULT_FILE_MODE;
98 	} else {
99 		if (sb.st_mode & S_IFDIR)
100 			return got_error_set_errno(EISDIR, ondisk_path);
101 		ie->flags &= ~GOT_FILEIDX_F_NO_FILE_ON_DISK;
102 	}
103 
104 
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;
111 		}
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));
124 		}
125 	}
126 
127 	if (blob_sha1) {
128 		memcpy(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;
132 
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;
138 
139 	return NULL;
140 }
141 
142 void
143 got_fileindex_entry_mark_deleted_from_disk(struct got_fileindex_entry *ie)
144 {
145 	ie->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
146 }
147 
148 void
149 got_fileindex_entry_mark_skipped(struct got_fileindex_entry *ie)
150 {
151 	ie->flags |= GOT_FILEIDX_F_SKIPPED;
152 }
153 
154 const struct got_error *
155 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
156     const char *relpath)
157 {
158 	size_t len;
159 
160 	*ie = calloc(1, sizeof(**ie));
161 	if (*ie == NULL)
162 		return got_error_from_errno("calloc");
163 
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;
170 	}
171 
172 	len = strlen(relpath);
173 	if (len > GOT_FILEIDX_F_PATH_LEN)
174 		len = GOT_FILEIDX_F_PATH_LEN;
175 	(*ie)->flags |= len;
176 
177 	return NULL;
178 }
179 
180 void
181 got_fileindex_entry_free(struct got_fileindex_entry *ie)
182 {
183 	free(ie->path);
184 	free(ie);
185 }
186 
187 size_t
188 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
189 {
190 	return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
191 }
192 
193 uint32_t
194 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
195 {
196 	return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
197 }
198 
199 void
200 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
201 {
202 	ie->flags &= ~GOT_FILEIDX_F_STAGE;
203 	ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
204 	    GOT_FILEIDX_F_STAGE);
205 }
206 
207 int
208 got_fileindex_entry_filetype_get(struct got_fileindex_entry *ie)
209 {
210 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
211 }
212 
213 void
214 got_fileindex_entry_filetype_set(struct got_fileindex_entry *ie, int type)
215 {
216 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_ONDISK;
217 	ie->mode |= (type & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
218 }
219 
220 void
221 got_fileindex_entry_staged_filetype_set(struct got_fileindex_entry *ie,
222     int type)
223 {
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);
227 }
228 
229 int
230 got_fileindex_entry_staged_filetype_get(struct got_fileindex_entry *ie)
231 {
232 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_STAGED) >>
233 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT;
234 }
235 
236 int
237 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
238 {
239 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
240 }
241 
242 int
243 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
244 {
245 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
246 }
247 
248 int
249 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
250 {
251 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
252 }
253 
254 int
255 got_fileindex_entry_was_skipped(struct got_fileindex_entry *ie)
256 {
257 	return (ie->flags & GOT_FILEIDX_F_SKIPPED) != 0;
258 }
259 
260 static const struct got_error *
261 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
262 {
263 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
264 		return got_error(GOT_ERR_NO_SPACE);
265 
266 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
267 	fileindex->nentries++;
268 	return NULL;
269 }
270 
271 const struct got_error *
272 got_fileindex_entry_add(struct got_fileindex *fileindex,
273     struct got_fileindex_entry *ie)
274 {
275 	/* Flag this entry until it gets written out to disk. */
276 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
277 
278 	return add_entry(fileindex, ie);
279 }
280 
281 void
282 got_fileindex_entry_remove(struct got_fileindex *fileindex,
283     struct got_fileindex_entry *ie)
284 {
285 	/*
286 	 * Removing an entry from the RB tree immediately breaks
287 	 * in-progress iterations over file index entries.
288 	 * So flag this entry for removal and remove it once the index
289 	 * is written out to disk. Meanwhile, pretend this entry no longer
290 	 * exists if we get queried for it again before then.
291 	 */
292 	ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
293 	fileindex->nentries--;
294 }
295 
296 struct got_fileindex_entry *
297 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
298     size_t path_len)
299 {
300 	struct got_fileindex_entry *ie;
301 	struct got_fileindex_entry key;
302 	memset(&key, 0, sizeof(key));
303 	key.path = (char *)path;
304 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
305 	ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
306 	if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
307 		return NULL;
308 	return ie;
309 }
310 
311 const struct got_error *
312 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
313     got_fileindex_cb cb, void *cb_arg)
314 {
315 	const struct got_error *err;
316 	struct got_fileindex_entry *ie, *tmp;
317 
318 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
319 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
320 			continue;
321 		err = (*cb)(cb_arg, ie);
322 		if (err)
323 			return err;
324 	}
325 	return NULL;
326 }
327 
328 struct got_fileindex *
329 got_fileindex_alloc(void)
330 {
331 	struct got_fileindex *fileindex;
332 
333 	fileindex = calloc(1, sizeof(*fileindex));
334 	if (fileindex == NULL)
335 		return NULL;
336 
337 	RB_INIT(&fileindex->entries);
338 	return fileindex;
339 }
340 
341 void
342 got_fileindex_free(struct got_fileindex *fileindex)
343 {
344 	struct got_fileindex_entry *ie;
345 
346 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
347 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
348 		got_fileindex_entry_free(ie);
349 	}
350 	free(fileindex);
351 }
352 
353 static const struct got_error *
354 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
355 {
356 	size_t n;
357 
358 	val = htobe64(val);
359 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
360 	n = fwrite(&val, 1, sizeof(val), outfile);
361 	if (n != sizeof(val))
362 		return got_ferror(outfile, GOT_ERR_IO);
363 	return NULL;
364 }
365 
366 static const struct got_error *
367 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
368 {
369 	size_t n;
370 
371 	val = htobe32(val);
372 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
373 	n = fwrite(&val, 1, sizeof(val), outfile);
374 	if (n != sizeof(val))
375 		return got_ferror(outfile, GOT_ERR_IO);
376 	return NULL;
377 }
378 
379 static const struct got_error *
380 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
381 {
382 	size_t n;
383 
384 	val = htobe16(val);
385 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
386 	n = fwrite(&val, 1, sizeof(val), outfile);
387 	if (n != sizeof(val))
388 		return got_ferror(outfile, GOT_ERR_IO);
389 	return NULL;
390 }
391 
392 static const struct got_error *
393 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
394 {
395 	size_t n, len, pad = 0;
396 	static const uint8_t zero[8] = { 0 };
397 
398 	len = strlen(path);
399 	while ((len + pad) % 8 != 0)
400 		pad++;
401 	if (pad == 0)
402 		pad = 8; /* NUL-terminate */
403 
404 	SHA1Update(ctx, path, len);
405 	n = fwrite(path, 1, len, outfile);
406 	if (n != len)
407 		return got_ferror(outfile, GOT_ERR_IO);
408 	SHA1Update(ctx, zero, pad);
409 	n = fwrite(zero, 1, pad, outfile);
410 	if (n != pad)
411 		return got_ferror(outfile, GOT_ERR_IO);
412 	return NULL;
413 }
414 
415 static const struct got_error *
416 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
417     FILE *outfile)
418 {
419 	const struct got_error *err;
420 	size_t n;
421 	uint32_t stage;
422 
423 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
424 	if (err)
425 		return err;
426 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
427 	if (err)
428 		return err;
429 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
430 	if (err)
431 		return err;
432 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
433 	if (err)
434 		return err;
435 
436 	err = write_fileindex_val32(ctx, ie->uid, outfile);
437 	if (err)
438 		return err;
439 	err = write_fileindex_val32(ctx, ie->gid, outfile);
440 	if (err)
441 		return err;
442 	err = write_fileindex_val32(ctx, ie->size, outfile);
443 	if (err)
444 		return err;
445 
446 	err = write_fileindex_val16(ctx, ie->mode, outfile);
447 	if (err)
448 		return err;
449 
450 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
451 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
452 	if (n != SHA1_DIGEST_LENGTH)
453 		return got_ferror(outfile, GOT_ERR_IO);
454 
455 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
456 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
457 	if (n != SHA1_DIGEST_LENGTH)
458 		return got_ferror(outfile, GOT_ERR_IO);
459 
460 	err = write_fileindex_val32(ctx, ie->flags, outfile);
461 	if (err)
462 		return err;
463 
464 	err = write_fileindex_path(ctx, ie->path, outfile);
465 	if (err)
466 		return err;
467 
468 	stage = got_fileindex_entry_stage_get(ie);
469 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
470 	    stage == GOT_FILEIDX_STAGE_ADD) {
471 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
472 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
473 		    outfile);
474 		if (n != SHA1_DIGEST_LENGTH)
475 			return got_ferror(outfile, GOT_ERR_IO);
476 	}
477 
478 	return NULL;
479 }
480 
481 const struct got_error *
482 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
483 {
484 	const struct got_error *err = NULL;
485 	struct got_fileindex_hdr hdr;
486 	SHA1_CTX ctx;
487 	uint8_t sha1[SHA1_DIGEST_LENGTH];
488 	size_t n;
489 	struct got_fileindex_entry *ie, *tmp;
490 
491 	SHA1Init(&ctx);
492 
493 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
494 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
495 	hdr.nentries = htobe32(fileindex->nentries);
496 
497 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
498 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
499 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
500 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
501 	if (n != sizeof(hdr.signature))
502 		return got_ferror(outfile, GOT_ERR_IO);
503 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
504 	if (n != sizeof(hdr.version))
505 		return got_ferror(outfile, GOT_ERR_IO);
506 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
507 	if (n != sizeof(hdr.nentries))
508 		return got_ferror(outfile, GOT_ERR_IO);
509 
510 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
511 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
512 		ie->flags &= ~GOT_FILEIDX_F_SKIPPED;
513 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
514 			RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
515 			got_fileindex_entry_free(ie);
516 			continue;
517 		}
518 		err = write_fileindex_entry(&ctx, ie, outfile);
519 		if (err)
520 			return err;
521 	}
522 
523 	SHA1Final(sha1, &ctx);
524 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
525 	if (n != sizeof(sha1))
526 		return got_ferror(outfile, GOT_ERR_IO);
527 
528 	if (fflush(outfile) != 0)
529 		return got_error_from_errno("fflush");
530 
531 	return NULL;
532 }
533 
534 static const struct got_error *
535 read_fileindex_val64(uint64_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 = be64toh(*val);
544 	return NULL;
545 }
546 
547 static const struct got_error *
548 read_fileindex_val32(uint32_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 = be32toh(*val);
557 	return NULL;
558 }
559 
560 static const struct got_error *
561 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
562 {
563 	size_t n;
564 
565 	n = fread(val, 1, sizeof(*val), infile);
566 	if (n != sizeof(*val))
567 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
568 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
569 	*val = be16toh(*val);
570 	return NULL;
571 }
572 
573 static const struct got_error *
574 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
575 {
576 	const struct got_error *err = NULL;
577 	const size_t chunk_size = 8;
578 	size_t n, len = 0, totlen = chunk_size;
579 
580 	*path = malloc(totlen);
581 	if (*path == NULL)
582 		return got_error_from_errno("malloc");
583 
584 	do {
585 		if (len + chunk_size > totlen) {
586 			char *p = reallocarray(*path, totlen + chunk_size, 1);
587 			if (p == NULL) {
588 				err = got_error_from_errno("reallocarray");
589 				break;
590 			}
591 			totlen += chunk_size;
592 			*path = p;
593 		}
594 		n = fread(*path + len, 1, chunk_size, infile);
595 		if (n != chunk_size) {
596 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
597 			break;
598 		}
599 		SHA1Update(ctx, *path + len, chunk_size);
600 		len += chunk_size;
601 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
602 
603 	if (err) {
604 		free(*path);
605 		*path = NULL;
606 	}
607 	return err;
608 }
609 
610 static const struct got_error *
611 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
612     FILE *infile, uint32_t version)
613 {
614 	const struct got_error *err;
615 	struct got_fileindex_entry *ie;
616 	size_t n;
617 
618 	*iep = NULL;
619 
620 	ie = calloc(1, sizeof(*ie));
621 	if (ie == NULL)
622 		return got_error_from_errno("calloc");
623 
624 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
625 	if (err)
626 		goto done;
627 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
628 	if (err)
629 		goto done;
630 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
631 	if (err)
632 		goto done;
633 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
634 	if (err)
635 		goto done;
636 
637 	err = read_fileindex_val32(&ie->uid, ctx, infile);
638 	if (err)
639 		goto done;
640 	err = read_fileindex_val32(&ie->gid, ctx, infile);
641 	if (err)
642 		goto done;
643 	err = read_fileindex_val32(&ie->size, ctx, infile);
644 	if (err)
645 		goto done;
646 
647 	err = read_fileindex_val16(&ie->mode, ctx, infile);
648 	if (err)
649 		goto done;
650 
651 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
652 	if (n != SHA1_DIGEST_LENGTH) {
653 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
654 		goto done;
655 	}
656 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
657 
658 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
659 	if (n != SHA1_DIGEST_LENGTH) {
660 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
661 		goto done;
662 	}
663 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
664 
665 	err = read_fileindex_val32(&ie->flags, ctx, infile);
666 	if (err)
667 		goto done;
668 
669 	err = read_fileindex_path(&ie->path, ctx, infile);
670 	if (err)
671 		goto done;
672 
673 	if (version >= 2) {
674 		uint32_t stage = got_fileindex_entry_stage_get(ie);
675 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
676 		    stage == GOT_FILEIDX_STAGE_ADD) {
677 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
678 			    infile);
679 			if (n != SHA1_DIGEST_LENGTH) {
680 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
681 				goto done;
682 			}
683 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
684 		}
685 	} else {
686 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
687 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
688 	}
689 
690 done:
691 	if (err)
692 		got_fileindex_entry_free(ie);
693 	else
694 		*iep = ie;
695 	return err;
696 }
697 
698 const struct got_error *
699 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
700 {
701 	const struct got_error *err = NULL;
702 	struct got_fileindex_hdr hdr;
703 	SHA1_CTX ctx;
704 	struct got_fileindex_entry *ie;
705 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
706 	uint8_t sha1[SHA1_DIGEST_LENGTH];
707 	size_t n;
708 	int i;
709 
710 	SHA1Init(&ctx);
711 
712 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
713 	if (n != sizeof(hdr.signature)) {
714 		if (n == 0) /* EOF */
715 			return NULL;
716 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
717 	}
718 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
719 	if (n != sizeof(hdr.version)) {
720 		if (n == 0) /* EOF */
721 			return NULL;
722 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
723 	}
724 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
725 	if (n != sizeof(hdr.nentries)) {
726 		if (n == 0) /* EOF */
727 			return NULL;
728 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
729 	}
730 
731 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
732 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
733 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
734 
735 	hdr.signature = be32toh(hdr.signature);
736 	hdr.version = be32toh(hdr.version);
737 	hdr.nentries = be32toh(hdr.nentries);
738 
739 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
740 		return got_error(GOT_ERR_FILEIDX_SIG);
741 	if (hdr.version > GOT_FILE_INDEX_VERSION)
742 		return got_error(GOT_ERR_FILEIDX_VER);
743 
744 	for (i = 0; i < hdr.nentries; i++) {
745 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
746 		if (err)
747 			return err;
748 		err = add_entry(fileindex, ie);
749 		if (err)
750 			return err;
751 	}
752 
753 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
754 	if (n != sizeof(sha1_expected))
755 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
756 	SHA1Final(sha1, &ctx);
757 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
758 		return got_error(GOT_ERR_FILEIDX_CSUM);
759 
760 	return NULL;
761 }
762 
763 static struct got_fileindex_entry *
764 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
765 {
766 	struct got_fileindex_entry *next;
767 
768 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
769 
770 	/* Skip entries which were added or removed by diff callbacks. */
771 	while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
772 	    GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
773 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
774 
775 	return next;
776 }
777 
778 static const struct got_error *
779 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
780     struct got_tree_object *tree, const char *, const char *,
781     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
782 
783 static const struct got_error *
784 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
785     struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
786     const char *path, const char *entry_name, struct got_repository *repo,
787     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
788 {
789 	const struct got_error *err = NULL;
790 	struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
791 
792 	if (!got_object_tree_entry_is_submodule(te) &&
793 	    S_ISDIR(got_tree_entry_get_mode(te))) {
794 		char *subpath;
795 		struct got_tree_object *subtree;
796 
797 		if (asprintf(&subpath, "%s%s%s", path,
798 		    path[0] == '\0' ? "" : "/",
799 		    got_tree_entry_get_name(te)) == -1)
800 			return got_error_from_errno("asprintf");
801 
802 		err = got_object_open_as_tree(&subtree, repo,
803 		    got_tree_entry_get_id(te));
804 		if (err) {
805 			free(subpath);
806 			return err;
807 		}
808 
809 		err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
810 		    entry_name, repo, cb, cb_arg);
811 		free(subpath);
812 		got_object_tree_close(subtree);
813 		if (err)
814 			return err;
815 	}
816 
817 	(*tidx)++;
818 	*next = got_object_tree_get_entry(tree, *tidx);
819 	return NULL;
820 }
821 
822 static const struct got_error *
823 diff_fileindex_tree(struct got_fileindex *fileindex,
824     struct got_fileindex_entry **ie, struct got_tree_object *tree,
825     const char *path, const char *entry_name, struct got_repository *repo,
826     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
827 {
828 	const struct got_error *err = NULL;
829 	struct got_tree_entry *te = NULL;
830 	size_t path_len = strlen(path);
831 	struct got_fileindex_entry *next;
832 	int tidx = 0;
833 
834 	te = got_object_tree_get_entry(tree, tidx);
835 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
836 		if (te && *ie) {
837 			char *te_path;
838 			const char *te_name = got_tree_entry_get_name(te);
839 			int cmp;
840 			if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
841 				err = got_error_from_errno("asprintf");
842 				break;
843 			}
844 			cmp = got_path_cmp((*ie)->path, te_path,
845 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
846 			free(te_path);
847 			if (cmp == 0) {
848 				if (got_path_is_child((*ie)->path, path,
849 				    path_len) &&
850 				    !got_object_tree_entry_is_submodule(te) &&
851 				    (entry_name == NULL ||
852 				    strcmp(te_name, entry_name) == 0)) {
853 					err = cb->diff_old_new(cb_arg, *ie, te,
854 					    path);
855 					if (err || entry_name)
856 						break;
857 				}
858 				*ie = walk_fileindex(fileindex, *ie);
859 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
860 				    path, entry_name, repo, cb, cb_arg);
861 			} else if (cmp < 0) {
862 				next = walk_fileindex(fileindex, *ie);
863 				if (got_path_is_child((*ie)->path, path,
864 				    path_len) && entry_name == NULL) {
865 					err = cb->diff_old(cb_arg, *ie, path);
866 					if (err || entry_name)
867 						break;
868 				}
869 				*ie = next;
870 			} else {
871 				if ((entry_name == NULL ||
872 				    strcmp(te_name, entry_name) == 0)) {
873 					err = cb->diff_new(cb_arg, te, path);
874 					if (err || entry_name)
875 						break;
876 				}
877 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
878 				    path, entry_name, repo, cb, cb_arg);
879 			}
880 			if (err)
881 				break;
882 		} else if (*ie) {
883 			next = walk_fileindex(fileindex, *ie);
884 			if (got_path_is_child((*ie)->path, path, path_len) &&
885 			    (entry_name == NULL ||
886 			    (te && strcmp(got_tree_entry_get_name(te),
887 			    entry_name) == 0))) {
888 				err = cb->diff_old(cb_arg, *ie, path);
889 				if (err || entry_name)
890 					break;
891 			}
892 			*ie = next;
893 		} else if (te) {
894 			if (!got_object_tree_entry_is_submodule(te) &&
895 			    (entry_name == NULL ||
896 			    strcmp(got_tree_entry_get_name(te), entry_name)
897 			    == 0)) {
898 				err = cb->diff_new(cb_arg, te, path);
899 				if (err || entry_name)
900 					break;
901 			}
902 			err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
903 			    entry_name, repo, cb, cb_arg);
904 			if (err)
905 				break;
906 		}
907 	}
908 
909 	return err;
910 }
911 
912 const struct got_error *
913 got_fileindex_diff_tree(struct got_fileindex *fileindex,
914     struct got_tree_object *tree, const char *path, const char *entry_name,
915     struct got_repository *repo,
916     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
917 {
918 	struct got_fileindex_entry *ie;
919 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
920 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
921 		ie = walk_fileindex(fileindex, ie);
922 	return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
923 	    cb, cb_arg);
924 }
925 
926 static const struct got_error *
927 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
928     struct got_pathlist_head *, int, const char *, const char *,
929     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
930 
931 static const struct got_error *
932 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
933 {
934 	const struct got_error *err = NULL;
935 	struct got_pathlist_entry *new = NULL;
936 	struct dirent *dep = NULL;
937 	struct dirent *de = NULL;
938 
939 	for (;;) {
940 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
941 		if (de == NULL) {
942 			err = got_error_from_errno("malloc");
943 			break;
944 		}
945 
946 		if (readdir_r(dir, de, &dep) != 0) {
947 			err = got_error_from_errno("readdir_r");
948 			free(de);
949 			break;
950 		}
951 		if (dep == NULL) {
952 			free(de);
953 			break;
954 		}
955 
956 		if (strcmp(de->d_name, ".") == 0 ||
957 		    strcmp(de->d_name, "..") == 0 ||
958 		    (path[0] == '\0' &&
959 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
960 			free(de);
961 			continue;
962 		}
963 
964 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
965 		if (err) {
966 			free(de);
967 			break;
968 		}
969 		if (new == NULL) {
970 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
971 			free(de);
972 			break;
973 		}
974 	}
975 
976 	return err;
977 }
978 
979 void
980 free_dirlist(struct got_pathlist_head *dirlist)
981 {
982 	struct got_pathlist_entry *dle;
983 
984 	TAILQ_FOREACH(dle, dirlist, entry)
985 		free(dle->data);
986 	got_pathlist_free(dirlist);
987 }
988 
989 static int
990 have_tracked_file_in_dir(struct got_fileindex *fileindex, const char *path)
991 {
992 	struct got_fileindex_entry *ie;
993 	size_t path_len = strlen(path);
994 	int cmp;
995 
996 	ie = RB_ROOT(&fileindex->entries);
997 	while (ie) {
998 		if (got_path_is_child(ie->path, path, path_len))
999 			return 1;
1000 		cmp = got_path_cmp(path, ie->path, path_len,
1001 		    got_fileindex_entry_path_len(ie));
1002 		if (cmp < 0)
1003 			ie = RB_LEFT(ie, entry);
1004 		else if (cmp > 0)
1005 			ie = RB_RIGHT(ie, entry);
1006 		else
1007 			break;
1008 	}
1009 
1010 	return 0;
1011 }
1012 
1013 static const struct got_error *
1014 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
1015     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
1016     const char *path, const char *rootpath, struct got_repository *repo,
1017     int ignore, struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1018 {
1019 	const struct got_error *err = NULL;
1020 	struct dirent *de = dle->data;
1021 	DIR *subdir = NULL;
1022 	int subdirfd = -1;
1023 	int type;
1024 
1025 	*next = NULL;
1026 
1027 	if (de->d_type == DT_UNKNOWN) {
1028 		/* Occurs on NFS mounts without "readdir plus" RPC. */
1029 		char *dir_path;
1030 		if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
1031 			return got_error_from_errno("asprintf");
1032 		err = got_path_dirent_type(&type, dir_path, de);
1033 		free(dir_path);
1034 		if (err)
1035 			return err;
1036 	} else
1037 		type = de->d_type;
1038 
1039 	/* Must traverse ignored directories if they contain tracked files. */
1040 	if (type == DT_DIR && ignore &&
1041 	    have_tracked_file_in_dir(fileindex, path))
1042 		ignore = 0;
1043 
1044 	if (type == DT_DIR && !ignore) {
1045 		char *subpath;
1046 		char *subdirpath;
1047 		struct got_pathlist_head subdirlist;
1048 
1049 		TAILQ_INIT(&subdirlist);
1050 
1051 		if (asprintf(&subpath, "%s%s%s", path,
1052 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1053 			return got_error_from_errno("asprintf");
1054 
1055 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
1056 			free(subpath);
1057 			return got_error_from_errno("asprintf");
1058 		}
1059 
1060 		subdirfd = openat(fd, de->d_name,
1061 		    O_RDONLY | O_NOFOLLOW | O_DIRECTORY | O_CLOEXEC);
1062 		if (subdirfd == -1) {
1063 			if (errno == EACCES) {
1064 				*next = TAILQ_NEXT(dle, entry);
1065 				return NULL;
1066 			}
1067 			err = got_error_from_errno2("openat", subdirpath);
1068 			free(subpath);
1069 			free(subdirpath);
1070 			return err;
1071 		}
1072 
1073 		subdir = fdopendir(subdirfd);
1074 		if (subdir == NULL)
1075 			return got_error_from_errno2("fdopendir", path);
1076 		subdirfd = -1;
1077 		err = read_dirlist(&subdirlist, subdir, subdirpath);
1078 		if (err) {
1079 			free(subpath);
1080 			free(subdirpath);
1081 			closedir(subdir);
1082 			return err;
1083 		}
1084 		err = diff_fileindex_dir(fileindex, ie, &subdirlist,
1085 		    dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
1086 		if (subdir && closedir(subdir) == -1 && err == NULL)
1087 			err = got_error_from_errno2("closedir", subdirpath);
1088 		free(subpath);
1089 		free(subdirpath);
1090 		free_dirlist(&subdirlist);
1091 		if (err)
1092 			return err;
1093 	}
1094 
1095 	*next = TAILQ_NEXT(dle, entry);
1096 	return NULL;
1097 }
1098 
1099 static const struct got_error *
1100 diff_fileindex_dir(struct got_fileindex *fileindex,
1101     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1102     int dirfd, const char *rootpath, const char *path,
1103     struct got_repository *repo,
1104     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1105 {
1106 	const struct got_error *err = NULL;
1107 	struct dirent *de = NULL;
1108 	size_t path_len = strlen(path);
1109 	struct got_pathlist_entry *dle;
1110 	int ignore;
1111 
1112 	if (cb->diff_traverse) {
1113 		err = cb->diff_traverse(cb_arg, path, dirfd);
1114 		if (err)
1115 			return err;
1116 	}
1117 
1118 	dle = TAILQ_FIRST(dirlist);
1119 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1120 		if (dle && *ie) {
1121 			char *de_path;
1122 			int cmp;
1123 			de = dle->data;
1124 			if (asprintf(&de_path, "%s/%s", path,
1125 			    de->d_name) == -1) {
1126 				err = got_error_from_errno("asprintf");
1127 				break;
1128 			}
1129 			cmp = got_path_cmp((*ie)->path, de_path,
1130 			    got_fileindex_entry_path_len(*ie),
1131 			    strlen(path) + 1 + strlen(de->d_name));
1132 			free(de_path);
1133 			if (cmp == 0) {
1134 				err = cb->diff_old_new(cb_arg, *ie, de, path,
1135 				    dirfd);
1136 				if (err)
1137 					break;
1138 				*ie = walk_fileindex(fileindex, *ie);
1139 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1140 				    path, rootpath, repo, 0, cb, cb_arg);
1141 			} else if (cmp < 0 ) {
1142 				err = cb->diff_old(cb_arg, *ie, path);
1143 				if (err)
1144 					break;
1145 				*ie = walk_fileindex(fileindex, *ie);
1146 			} else {
1147 				err = cb->diff_new(&ignore, cb_arg, de, path,
1148 				    dirfd);
1149 				if (err)
1150 					break;
1151 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1152 				    path, rootpath, repo, ignore, cb, cb_arg);
1153 			}
1154 			if (err)
1155 				break;
1156 		} else if (*ie) {
1157 			err = cb->diff_old(cb_arg, *ie, path);
1158 			if (err)
1159 				break;
1160 			*ie = walk_fileindex(fileindex, *ie);
1161 		} else if (dle) {
1162 			de = dle->data;
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;
1170 		}
1171 	}
1172 
1173 	return err;
1174 }
1175 
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)
1180 {
1181 	const struct got_error *err;
1182 	struct got_fileindex_entry *ie;
1183 	struct got_pathlist_head dirlist;
1184 	int fd2;
1185 	DIR *dir;
1186 
1187 	TAILQ_INIT(&dirlist);
1188 
1189 	/* 
1190 	 * Duplicate the file descriptor so we can call closedir() below
1191 	 * without closing the file descriptor passed in by our caller.
1192 	 */
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;
1200 	}
1201 	dir = fdopendir(fd2);
1202 	if (dir == NULL) {
1203 		err = got_error_from_errno2("fdopendir", path);
1204 		close(fd2);
1205 		return err;
1206 	}
1207 	err = read_dirlist(&dirlist, dir, path);
1208 	if (err) {
1209 		closedir(dir);
1210 		return err;
1211 	}
1212 
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);
1218 
1219 	if (closedir(dir) == -1 && err == NULL)
1220 		err = got_error_from_errno2("closedir", path);
1221 	free_dirlist(&dirlist);
1222 	return err;
1223 }
1224 
1225 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);