Gotweb

Blob

Date:: Thu Jul 23 14:22:34 2020 UTC
Message:: keep file index entry permission bits clear for symlinks
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 uint16_t
57 got_fileindex_perms_from_st(struct stat *sb)
58 {
59 	uint16_t perms = (sb->st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
60 	return (perms << GOT_FILEIDX_MODE_PERMS_SHIFT);
61 }
62 
63 mode_t
64 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
65 {
66 	mode_t perms = (ie->mode >> GOT_FILEIDX_MODE_PERMS_SHIFT);
67 	return (S_IFREG | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
68 }
69 
70 const struct got_error *
71 got_fileindex_entry_update(struct got_fileindex_entry *ie,
72     const char *ondisk_path, uint8_t *blob_sha1, uint8_t *commit_sha1,
73     int update_timestamps)
74 {
75 	struct stat sb;
76 
77 	if (lstat(ondisk_path, &sb) != 0) {
78 		if (!((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) &&
79 		    errno == ENOENT))
80 			return got_error_from_errno2("lstat", ondisk_path);
81 	} else {
82 		if (sb.st_mode & S_IFDIR)
83 			return got_error_set_errno(EISDIR, ondisk_path);
84 		ie->flags &= ~GOT_FILEIDX_F_NO_FILE_ON_DISK;
85 	}
86 
87 
88 	if ((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0) {
89 		if (update_timestamps) {
90 			ie->ctime_sec = sb.st_ctime;
91 			ie->ctime_nsec = sb.st_ctimensec;
92 			ie->mtime_sec = sb.st_mtime;
93 			ie->mtime_nsec = sb.st_mtimensec;
94 		}
95 		ie->uid = sb.st_uid;
96 		ie->gid = sb.st_gid;
97 		ie->size = (sb.st_size & 0xffffffff);
98 		if (S_ISLNK(sb.st_mode))
99 			ie->mode = GOT_FILEIDX_MODE_SYMLINK;
100 		else {
101 			ie->mode = GOT_FILEIDX_MODE_REGULAR_FILE;
102 			ie->mode |= got_fileindex_perms_from_st(&sb);
103 		}
104 	}
105 
106 	if (blob_sha1) {
107 		memcpy(ie->blob_sha1, blob_sha1, SHA1_DIGEST_LENGTH);
108 		ie->flags &= ~GOT_FILEIDX_F_NO_BLOB;
109 	} else
110 		ie->flags |= GOT_FILEIDX_F_NO_BLOB;
111 
112 	if (commit_sha1) {
113 		memcpy(ie->commit_sha1, commit_sha1, SHA1_DIGEST_LENGTH);
114 		ie->flags &= ~GOT_FILEIDX_F_NO_COMMIT;
115 	} else
116 		ie->flags |= GOT_FILEIDX_F_NO_COMMIT;
117 
118 	return NULL;
119 }
120 
121 void
122 got_fileindex_entry_mark_deleted_from_disk(struct got_fileindex_entry *ie)
123 {
124 	ie->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
125 }
126 
127 const struct got_error *
128 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
129     const char *relpath)
130 {
131 	size_t len;
132 
133 	*ie = calloc(1, sizeof(**ie));
134 	if (*ie == NULL)
135 		return got_error_from_errno("calloc");
136 
137 	(*ie)->path = strdup(relpath);
138 	if ((*ie)->path == NULL) {
139 		const struct got_error *err = got_error_from_errno("strdup");
140 		free(*ie);
141 		*ie = NULL;
142 		return err;
143 	}
144 
145 	len = strlen(relpath);
146 	if (len > GOT_FILEIDX_F_PATH_LEN)
147 		len = GOT_FILEIDX_F_PATH_LEN;
148 	(*ie)->flags |= len;
149 
150 	return NULL;
151 }
152 
153 void
154 got_fileindex_entry_free(struct got_fileindex_entry *ie)
155 {
156 	free(ie->path);
157 	free(ie);
158 }
159 
160 size_t
161 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
162 {
163 	return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
164 }
165 
166 uint32_t
167 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
168 {
169 	return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
170 }
171 
172 void
173 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
174 {
175 	ie->flags &= ~GOT_FILEIDX_F_STAGE;
176 	ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
177 	    GOT_FILEIDX_F_STAGE);
178 }
179 
180 int
181 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
182 {
183 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
184 }
185 
186 int
187 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
188 {
189 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
190 }
191 
192 int
193 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
194 {
195 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
196 }
197 
198 static const struct got_error *
199 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
200 {
201 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
202 		return got_error(GOT_ERR_NO_SPACE);
203 
204 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
205 	fileindex->nentries++;
206 	return NULL;
207 }
208 
209 const struct got_error *
210 got_fileindex_entry_add(struct got_fileindex *fileindex,
211     struct got_fileindex_entry *ie)
212 {
213 	/* Flag this entry until it gets written out to disk. */
214 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
215 
216 	return add_entry(fileindex, ie);
217 }
218 
219 void
220 got_fileindex_entry_remove(struct got_fileindex *fileindex,
221     struct got_fileindex_entry *ie)
222 {
223 	/*
224 	 * Removing an entry from the RB tree immediately breaks
225 	 * in-progress iterations over file index entries.
226 	 * So flag this entry for removal and remove it once the index
227 	 * is written out to disk. Meanwhile, pretend this entry no longer
228 	 * exists if we get queried for it again before then.
229 	 */
230 	ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
231 	fileindex->nentries--;
232 }
233 
234 struct got_fileindex_entry *
235 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
236     size_t path_len)
237 {
238 	struct got_fileindex_entry *ie;
239 	struct got_fileindex_entry key;
240 	memset(&key, 0, sizeof(key));
241 	key.path = (char *)path;
242 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
243 	ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
244 	if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
245 		return NULL;
246 	return ie;
247 }
248 
249 const struct got_error *
250 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
251     got_fileindex_cb cb, void *cb_arg)
252 {
253 	const struct got_error *err;
254 	struct got_fileindex_entry *ie, *tmp;
255 
256 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
257 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
258 			continue;
259 		err = (*cb)(cb_arg, ie);
260 		if (err)
261 			return err;
262 	}
263 	return NULL;
264 }
265 
266 struct got_fileindex *
267 got_fileindex_alloc(void)
268 {
269 	struct got_fileindex *fileindex;
270 
271 	fileindex = calloc(1, sizeof(*fileindex));
272 	if (fileindex == NULL)
273 		return NULL;
274 
275 	RB_INIT(&fileindex->entries);
276 	return fileindex;
277 }
278 
279 void
280 got_fileindex_free(struct got_fileindex *fileindex)
281 {
282 	struct got_fileindex_entry *ie;
283 
284 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
285 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
286 		got_fileindex_entry_free(ie);
287 	}
288 	free(fileindex);
289 }
290 
291 static const struct got_error *
292 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
293 {
294 	size_t n;
295 
296 	val = htobe64(val);
297 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
298 	n = fwrite(&val, 1, sizeof(val), outfile);
299 	if (n != sizeof(val))
300 		return got_ferror(outfile, GOT_ERR_IO);
301 	return NULL;
302 }
303 
304 static const struct got_error *
305 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
306 {
307 	size_t n;
308 
309 	val = htobe32(val);
310 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
311 	n = fwrite(&val, 1, sizeof(val), outfile);
312 	if (n != sizeof(val))
313 		return got_ferror(outfile, GOT_ERR_IO);
314 	return NULL;
315 }
316 
317 static const struct got_error *
318 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
319 {
320 	size_t n;
321 
322 	val = htobe16(val);
323 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
324 	n = fwrite(&val, 1, sizeof(val), outfile);
325 	if (n != sizeof(val))
326 		return got_ferror(outfile, GOT_ERR_IO);
327 	return NULL;
328 }
329 
330 static const struct got_error *
331 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
332 {
333 	size_t n, len, pad = 0;
334 	static const uint8_t zero[8] = { 0 };
335 
336 	len = strlen(path);
337 	while ((len + pad) % 8 != 0)
338 		pad++;
339 	if (pad == 0)
340 		pad = 8; /* NUL-terminate */
341 
342 	SHA1Update(ctx, path, len);
343 	n = fwrite(path, 1, len, outfile);
344 	if (n != len)
345 		return got_ferror(outfile, GOT_ERR_IO);
346 	SHA1Update(ctx, zero, pad);
347 	n = fwrite(zero, 1, pad, outfile);
348 	if (n != pad)
349 		return got_ferror(outfile, GOT_ERR_IO);
350 	return NULL;
351 }
352 
353 static const struct got_error *
354 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
355     FILE *outfile)
356 {
357 	const struct got_error *err;
358 	size_t n;
359 	uint32_t stage;
360 
361 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
362 	if (err)
363 		return err;
364 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
365 	if (err)
366 		return err;
367 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
368 	if (err)
369 		return err;
370 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
371 	if (err)
372 		return err;
373 
374 	err = write_fileindex_val32(ctx, ie->uid, outfile);
375 	if (err)
376 		return err;
377 	err = write_fileindex_val32(ctx, ie->gid, outfile);
378 	if (err)
379 		return err;
380 	err = write_fileindex_val32(ctx, ie->size, outfile);
381 	if (err)
382 		return err;
383 
384 	err = write_fileindex_val16(ctx, ie->mode, outfile);
385 	if (err)
386 		return err;
387 
388 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
389 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
390 	if (n != SHA1_DIGEST_LENGTH)
391 		return got_ferror(outfile, GOT_ERR_IO);
392 
393 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
394 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
395 	if (n != SHA1_DIGEST_LENGTH)
396 		return got_ferror(outfile, GOT_ERR_IO);
397 
398 	err = write_fileindex_val32(ctx, ie->flags, outfile);
399 	if (err)
400 		return err;
401 
402 	err = write_fileindex_path(ctx, ie->path, outfile);
403 	if (err)
404 		return err;
405 
406 	stage = got_fileindex_entry_stage_get(ie);
407 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
408 	    stage == GOT_FILEIDX_STAGE_ADD) {
409 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
410 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
411 		    outfile);
412 		if (n != SHA1_DIGEST_LENGTH)
413 			return got_ferror(outfile, GOT_ERR_IO);
414 	}
415 
416 	return NULL;
417 }
418 
419 const struct got_error *
420 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
421 {
422 	const struct got_error *err = NULL;
423 	struct got_fileindex_hdr hdr;
424 	SHA1_CTX ctx;
425 	uint8_t sha1[SHA1_DIGEST_LENGTH];
426 	size_t n;
427 	struct got_fileindex_entry *ie, *tmp;
428 
429 	SHA1Init(&ctx);
430 
431 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
432 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
433 	hdr.nentries = htobe32(fileindex->nentries);
434 
435 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
436 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
437 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
438 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
439 	if (n != sizeof(hdr.signature))
440 		return got_ferror(outfile, GOT_ERR_IO);
441 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
442 	if (n != sizeof(hdr.version))
443 		return got_ferror(outfile, GOT_ERR_IO);
444 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
445 	if (n != sizeof(hdr.nentries))
446 		return got_ferror(outfile, GOT_ERR_IO);
447 
448 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
449 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
450 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
451 			RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
452 			got_fileindex_entry_free(ie);
453 			continue;
454 		}
455 		err = write_fileindex_entry(&ctx, ie, outfile);
456 		if (err)
457 			return err;
458 	}
459 
460 	SHA1Final(sha1, &ctx);
461 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
462 	if (n != sizeof(sha1))
463 		return got_ferror(outfile, GOT_ERR_IO);
464 
465 	if (fflush(outfile) != 0)
466 		return got_error_from_errno("fflush");
467 
468 	return NULL;
469 }
470 
471 static const struct got_error *
472 read_fileindex_val64(uint64_t *val, SHA1_CTX *ctx, FILE *infile)
473 {
474 	size_t n;
475 
476 	n = fread(val, 1, sizeof(*val), infile);
477 	if (n != sizeof(*val))
478 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
479 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
480 	*val = be64toh(*val);
481 	return NULL;
482 }
483 
484 static const struct got_error *
485 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
486 {
487 	size_t n;
488 
489 	n = fread(val, 1, sizeof(*val), infile);
490 	if (n != sizeof(*val))
491 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
492 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
493 	*val = be32toh(*val);
494 	return NULL;
495 }
496 
497 static const struct got_error *
498 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
499 {
500 	size_t n;
501 
502 	n = fread(val, 1, sizeof(*val), infile);
503 	if (n != sizeof(*val))
504 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
505 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
506 	*val = be16toh(*val);
507 	return NULL;
508 }
509 
510 static const struct got_error *
511 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
512 {
513 	const struct got_error *err = NULL;
514 	const size_t chunk_size = 8;
515 	size_t n, len = 0, totlen = chunk_size;
516 
517 	*path = malloc(totlen);
518 	if (*path == NULL)
519 		return got_error_from_errno("malloc");
520 
521 	do {
522 		if (len + chunk_size > totlen) {
523 			char *p = reallocarray(*path, totlen + chunk_size, 1);
524 			if (p == NULL) {
525 				err = got_error_from_errno("reallocarray");
526 				break;
527 			}
528 			totlen += chunk_size;
529 			*path = p;
530 		}
531 		n = fread(*path + len, 1, chunk_size, infile);
532 		if (n != chunk_size) {
533 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
534 			break;
535 		}
536 		SHA1Update(ctx, *path + len, chunk_size);
537 		len += chunk_size;
538 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
539 
540 	if (err) {
541 		free(*path);
542 		*path = NULL;
543 	}
544 	return err;
545 }
546 
547 static const struct got_error *
548 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
549     FILE *infile, uint32_t version)
550 {
551 	const struct got_error *err;
552 	struct got_fileindex_entry *ie;
553 	size_t n;
554 
555 	*iep = NULL;
556 
557 	ie = calloc(1, sizeof(*ie));
558 	if (ie == NULL)
559 		return got_error_from_errno("calloc");
560 
561 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
562 	if (err)
563 		goto done;
564 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
565 	if (err)
566 		goto done;
567 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
568 	if (err)
569 		goto done;
570 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
571 	if (err)
572 		goto done;
573 
574 	err = read_fileindex_val32(&ie->uid, ctx, infile);
575 	if (err)
576 		goto done;
577 	err = read_fileindex_val32(&ie->gid, ctx, infile);
578 	if (err)
579 		goto done;
580 	err = read_fileindex_val32(&ie->size, ctx, infile);
581 	if (err)
582 		goto done;
583 
584 	err = read_fileindex_val16(&ie->mode, ctx, infile);
585 	if (err)
586 		goto done;
587 
588 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
589 	if (n != SHA1_DIGEST_LENGTH) {
590 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
591 		goto done;
592 	}
593 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
594 
595 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
596 	if (n != SHA1_DIGEST_LENGTH) {
597 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
598 		goto done;
599 	}
600 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
601 
602 	err = read_fileindex_val32(&ie->flags, ctx, infile);
603 	if (err)
604 		goto done;
605 
606 	err = read_fileindex_path(&ie->path, ctx, infile);
607 	if (err)
608 		goto done;
609 
610 	if (version >= 2) {
611 		uint32_t stage = got_fileindex_entry_stage_get(ie);
612 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
613 		    stage == GOT_FILEIDX_STAGE_ADD) {
614 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
615 			    infile);
616 			if (n != SHA1_DIGEST_LENGTH) {
617 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
618 				goto done;
619 			}
620 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
621 		}
622 	} else {
623 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
624 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
625 	}
626 
627 done:
628 	if (err)
629 		got_fileindex_entry_free(ie);
630 	else
631 		*iep = ie;
632 	return err;
633 }
634 
635 const struct got_error *
636 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
637 {
638 	const struct got_error *err = NULL;
639 	struct got_fileindex_hdr hdr;
640 	SHA1_CTX ctx;
641 	struct got_fileindex_entry *ie;
642 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
643 	uint8_t sha1[SHA1_DIGEST_LENGTH];
644 	size_t n;
645 	int i;
646 
647 	SHA1Init(&ctx);
648 
649 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
650 	if (n != sizeof(hdr.signature)) {
651 		if (n == 0) /* EOF */
652 			return NULL;
653 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
654 	}
655 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
656 	if (n != sizeof(hdr.version)) {
657 		if (n == 0) /* EOF */
658 			return NULL;
659 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
660 	}
661 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
662 	if (n != sizeof(hdr.nentries)) {
663 		if (n == 0) /* EOF */
664 			return NULL;
665 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
666 	}
667 
668 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
669 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
670 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
671 
672 	hdr.signature = be32toh(hdr.signature);
673 	hdr.version = be32toh(hdr.version);
674 	hdr.nentries = be32toh(hdr.nentries);
675 
676 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
677 		return got_error(GOT_ERR_FILEIDX_SIG);
678 	if (hdr.version > GOT_FILE_INDEX_VERSION)
679 		return got_error(GOT_ERR_FILEIDX_VER);
680 
681 	for (i = 0; i < hdr.nentries; i++) {
682 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
683 		if (err)
684 			return err;
685 		err = add_entry(fileindex, ie);
686 		if (err)
687 			return err;
688 	}
689 
690 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
691 	if (n != sizeof(sha1_expected))
692 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
693 	SHA1Final(sha1, &ctx);
694 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
695 		return got_error(GOT_ERR_FILEIDX_CSUM);
696 
697 	return NULL;
698 }
699 
700 static struct got_fileindex_entry *
701 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
702 {
703 	struct got_fileindex_entry *next;
704 
705 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
706 
707 	/* Skip entries which were added or removed by diff callbacks. */
708 	while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
709 	    GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
710 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
711 
712 	return next;
713 }
714 
715 static const struct got_error *
716 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
717     struct got_tree_object *tree, const char *, const char *,
718     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
719 
720 static const struct got_error *
721 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
722     struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
723     const char *path, const char *entry_name, struct got_repository *repo,
724     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
725 {
726 	const struct got_error *err = NULL;
727 	struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
728 
729 	if (!got_object_tree_entry_is_submodule(te) &&
730 	    S_ISDIR(got_tree_entry_get_mode(te))) {
731 		char *subpath;
732 		struct got_tree_object *subtree;
733 
734 		if (asprintf(&subpath, "%s%s%s", path,
735 		    path[0] == '\0' ? "" : "/",
736 		    got_tree_entry_get_name(te)) == -1)
737 			return got_error_from_errno("asprintf");
738 
739 		err = got_object_open_as_tree(&subtree, repo,
740 		    got_tree_entry_get_id(te));
741 		if (err) {
742 			free(subpath);
743 			return err;
744 		}
745 
746 		err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
747 		    entry_name, repo, cb, cb_arg);
748 		free(subpath);
749 		got_object_tree_close(subtree);
750 		if (err)
751 			return err;
752 	}
753 
754 	(*tidx)++;
755 	*next = got_object_tree_get_entry(tree, *tidx);
756 	return NULL;
757 }
758 
759 static const struct got_error *
760 diff_fileindex_tree(struct got_fileindex *fileindex,
761     struct got_fileindex_entry **ie, struct got_tree_object *tree,
762     const char *path, const char *entry_name, struct got_repository *repo,
763     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
764 {
765 	const struct got_error *err = NULL;
766 	struct got_tree_entry *te = NULL;
767 	size_t path_len = strlen(path);
768 	struct got_fileindex_entry *next;
769 	int tidx = 0;
770 
771 	te = got_object_tree_get_entry(tree, tidx);
772 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
773 		if (te && *ie) {
774 			char *te_path;
775 			const char *te_name = got_tree_entry_get_name(te);
776 			int cmp;
777 			if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
778 				err = got_error_from_errno("asprintf");
779 				break;
780 			}
781 			cmp = got_path_cmp((*ie)->path, te_path,
782 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
783 			free(te_path);
784 			if (cmp == 0) {
785 				if (got_path_is_child((*ie)->path, path,
786 				    path_len) &&
787 				    !got_object_tree_entry_is_submodule(te) &&
788 				    (entry_name == NULL ||
789 				    strcmp(te_name, entry_name) == 0)) {
790 					err = cb->diff_old_new(cb_arg, *ie, te,
791 					    path);
792 					if (err || entry_name)
793 						break;
794 				}
795 				*ie = walk_fileindex(fileindex, *ie);
796 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
797 				    path, entry_name, repo, cb, cb_arg);
798 			} else if (cmp < 0) {
799 				next = walk_fileindex(fileindex, *ie);
800 				if (got_path_is_child((*ie)->path, path,
801 				    path_len) && (entry_name == NULL ||
802 				    strcmp(te_name, entry_name) == 0)) {
803 					err = cb->diff_old(cb_arg, *ie, path);
804 					if (err || entry_name)
805 						break;
806 				}
807 				*ie = next;
808 			} else {
809 				if ((entry_name == NULL ||
810 				    strcmp(te_name, entry_name) == 0)) {
811 					err = cb->diff_new(cb_arg, te, path);
812 					if (err || entry_name)
813 						break;
814 				}
815 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
816 				    path, entry_name, repo, cb, cb_arg);
817 			}
818 			if (err)
819 				break;
820 		} else if (*ie) {
821 			next = walk_fileindex(fileindex, *ie);
822 			if (got_path_is_child((*ie)->path, path, path_len) &&
823 			    (entry_name == NULL ||
824 			    (te && strcmp(got_tree_entry_get_name(te),
825 			    entry_name) == 0))) {
826 				err = cb->diff_old(cb_arg, *ie, path);
827 				if (err || entry_name)
828 					break;
829 			}
830 			*ie = next;
831 		} else if (te) {
832 			if (!got_object_tree_entry_is_submodule(te) &&
833 			    (entry_name == NULL ||
834 			    strcmp(got_tree_entry_get_name(te), entry_name)
835 			    == 0)) {
836 				err = cb->diff_new(cb_arg, te, path);
837 				if (err || entry_name)
838 					break;
839 			}
840 			err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
841 			    entry_name, repo, cb, cb_arg);
842 			if (err)
843 				break;
844 		}
845 	}
846 
847 	return err;
848 }
849 
850 const struct got_error *
851 got_fileindex_diff_tree(struct got_fileindex *fileindex,
852     struct got_tree_object *tree, const char *path, const char *entry_name,
853     struct got_repository *repo,
854     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
855 {
856 	struct got_fileindex_entry *ie;
857 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
858 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
859 		ie = walk_fileindex(fileindex, ie);
860 	return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
861 	    cb, cb_arg);
862 }
863 
864 static const struct got_error *
865 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
866     struct got_pathlist_head *, int, const char *, const char *,
867     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
868 
869 static const struct got_error *
870 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
871 {
872 	const struct got_error *err = NULL;
873 	struct got_pathlist_entry *new = NULL;
874 	struct dirent *dep = NULL;
875 	struct dirent *de = NULL;
876 
877 	for (;;) {
878 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
879 		if (de == NULL) {
880 			err = got_error_from_errno("malloc");
881 			break;
882 		}
883 
884 		if (readdir_r(dir, de, &dep) != 0) {
885 			err = got_error_from_errno("readdir_r");
886 			free(de);
887 			break;
888 		}
889 		if (dep == NULL) {
890 			free(de);
891 			break;
892 		}
893 
894 		if (strcmp(de->d_name, ".") == 0 ||
895 		    strcmp(de->d_name, "..") == 0 ||
896 		    (path[0] == '\0' &&
897 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
898 			free(de);
899 			continue;
900 		}
901 
902 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
903 		if (err) {
904 			free(de);
905 			break;
906 		}
907 		if (new == NULL) {
908 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
909 			free(de);
910 			break;
911 		}
912 	}
913 
914 	return err;
915 }
916 
917 void
918 free_dirlist(struct got_pathlist_head *dirlist)
919 {
920 	struct got_pathlist_entry *dle;
921 
922 	TAILQ_FOREACH(dle, dirlist, entry)
923 		free(dle->data);
924 	got_pathlist_free(dirlist);
925 }
926 
927 static const struct got_error *
928 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
929     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
930     const char *path, const char *rootpath, struct got_repository *repo,
931     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
932 {
933 	const struct got_error *err = NULL;
934 	struct dirent *de = dle->data;
935 	DIR *subdir = NULL;
936 	int subdirfd = -1;
937 	int type;
938 
939 	*next = NULL;
940 
941 	if (de->d_type == DT_UNKNOWN) {
942 		/* Occurs on NFS mounts without "readdir plus" RPC. */
943 		char *dir_path;
944 		if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
945 			return got_error_from_errno("asprintf");
946 		err = got_path_dirent_type(&type, dir_path, de);
947 		free(dir_path);
948 		if (err)
949 			return err;
950 	} else
951 		type = de->d_type;
952 
953 	if (type == DT_DIR) {
954 		char *subpath;
955 		char *subdirpath;
956 		struct got_pathlist_head subdirlist;
957 
958 		TAILQ_INIT(&subdirlist);
959 
960 		if (asprintf(&subpath, "%s%s%s", path,
961 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
962 			return got_error_from_errno("asprintf");
963 
964 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
965 			free(subpath);
966 			return got_error_from_errno("asprintf");
967 		}
968 
969 		subdirfd = openat(fd, de->d_name,
970 		    O_RDONLY | O_NOFOLLOW | O_DIRECTORY);
971 		if (subdirfd == -1) {
972 			if (errno == EACCES) {
973 				*next = TAILQ_NEXT(dle, entry);
974 				return NULL;
975 			}
976 			err = got_error_from_errno2("openat", subdirpath);
977 			free(subpath);
978 			free(subdirpath);
979 			return err;
980 		}
981 
982 		subdir = fdopendir(subdirfd);
983 		if (subdir == NULL)
984 			return got_error_from_errno2("fdopendir", path);
985 		subdirfd = -1;
986 		err = read_dirlist(&subdirlist, subdir, subdirpath);
987 		if (err) {
988 			free(subpath);
989 			free(subdirpath);
990 			closedir(subdir);
991 			return err;
992 		}
993 		err = diff_fileindex_dir(fileindex, ie, &subdirlist,
994 		    dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
995 		if (subdir && closedir(subdir) == -1 && err == NULL)
996 			err = got_error_from_errno2("closedir", subdirpath);
997 		free(subpath);
998 		free(subdirpath);
999 		free_dirlist(&subdirlist);
1000 		if (err)
1001 			return err;
1002 	}
1003 
1004 	*next = TAILQ_NEXT(dle, entry);
1005 	return NULL;
1006 }
1007 
1008 static const struct got_error *
1009 diff_fileindex_dir(struct got_fileindex *fileindex,
1010     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1011     int dirfd, const char *rootpath, const char *path,
1012     struct got_repository *repo,
1013     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1014 {
1015 	const struct got_error *err = NULL;
1016 	struct dirent *de = NULL;
1017 	size_t path_len = strlen(path);
1018 	struct got_pathlist_entry *dle;
1019 
1020 	if (cb->diff_traverse) {
1021 		err = cb->diff_traverse(cb_arg, path, dirfd);
1022 		if (err)
1023 			return err;
1024 	}
1025 
1026 	dle = TAILQ_FIRST(dirlist);
1027 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1028 		if (dle && *ie) {
1029 			char *de_path;
1030 			int cmp;
1031 			de = dle->data;
1032 			if (asprintf(&de_path, "%s/%s", path,
1033 			    de->d_name) == -1) {
1034 				err = got_error_from_errno("asprintf");
1035 				break;
1036 			}
1037 			cmp = got_path_cmp((*ie)->path, de_path,
1038 			    got_fileindex_entry_path_len(*ie),
1039 			    strlen(path) + 1 + de->d_namlen);
1040 			free(de_path);
1041 			if (cmp == 0) {
1042 				err = cb->diff_old_new(cb_arg, *ie, de, path,
1043 				    dirfd);
1044 				if (err)
1045 					break;
1046 				*ie = walk_fileindex(fileindex, *ie);
1047 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1048 				    path, rootpath, repo, cb, cb_arg);
1049 			} else if (cmp < 0 ) {
1050 				err = cb->diff_old(cb_arg, *ie, path);
1051 				if (err)
1052 					break;
1053 				*ie = walk_fileindex(fileindex, *ie);
1054 			} else {
1055 				err = cb->diff_new(cb_arg, de, path, dirfd);
1056 				if (err)
1057 					break;
1058 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1059 				    path, rootpath, repo, cb, cb_arg);
1060 			}
1061 			if (err)
1062 				break;
1063 		} else if (*ie) {
1064 			err = cb->diff_old(cb_arg, *ie, path);
1065 			if (err)
1066 				break;
1067 			*ie = walk_fileindex(fileindex, *ie);
1068 		} else if (dle) {
1069 			de = dle->data;
1070 			err = cb->diff_new(cb_arg, de, path, dirfd);
1071 			if (err)
1072 				break;
1073 			err = walk_dir(&dle, fileindex, ie, dle, dirfd, path,
1074 			    rootpath, repo, cb, cb_arg);
1075 			if (err)
1076 				break;
1077 		}
1078 	}
1079 
1080 	return err;
1081 }
1082 
1083 const struct got_error *
1084 got_fileindex_diff_dir(struct got_fileindex *fileindex, int fd,
1085     const char *rootpath, const char *path, struct got_repository *repo,
1086     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1087 {
1088 	const struct got_error *err;
1089 	struct got_fileindex_entry *ie;
1090 	struct got_pathlist_head dirlist;
1091 	int fd2;
1092 	DIR *dir;
1093 
1094 	TAILQ_INIT(&dirlist);
1095 
1096 	/* 
1097 	 * Duplicate the file descriptor so we can call closedir() below
1098 	 * without closing the file descriptor passed in by our caller.
1099 	 */
1100 	fd2 = dup(fd);
1101 	if (fd2 == -1)
1102 		return got_error_from_errno2("dup", path);
1103 	if (lseek(fd2, 0, SEEK_SET) == -1) {
1104 		err = got_error_from_errno2("lseek", path);
1105 		close(fd2);
1106 		return err;
1107 	}
1108 	dir = fdopendir(fd2);
1109 	if (dir == NULL) {
1110 		err = got_error_from_errno2("fdopendir", path);
1111 		close(fd2);
1112 		return err;
1113 	}
1114 	err = read_dirlist(&dirlist, dir, path);
1115 	if (err) {
1116 		closedir(dir);
1117 		return err;
1118 	}
1119 
1120 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1121 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1122 		ie = walk_fileindex(fileindex, ie);
1123 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, dirfd(dir),
1124 	    rootpath, path, repo, cb, cb_arg);
1125 
1126 	if (closedir(dir) == -1 && err == NULL)
1127 		err = got_error_from_errno2("closedir", path);
1128 	free_dirlist(&dirlist);
1129 	return err;
1130 }
1131 
1132 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);