op public repos

Blob

Date:: Thu Jul 23 14:22:34 2020 UTC
Message:: give bad symlinks a dedicated file type in the file index
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_filetype_get(struct got_fileindex_entry *ie)
182 {
183 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE);
184 }
185 
186 const struct got_error *
187 got_fileindex_entry_filetype_set(struct got_fileindex_entry *ie, int type)
188 {
189 	switch (type) {
190 	case GOT_FILEIDX_MODE_REGULAR_FILE:
191 	case GOT_FILEIDX_MODE_SYMLINK:
192 	case GOT_FILEIDX_MODE_BAD_SYMLINK:
193 		break;
194 	default:
195 		return got_error(GOT_ERR_BAD_FILETYPE);
196 	}
197 
198 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE;
199 	ie->mode |= type;
200 	return NULL;
201 }
202 
203 int
204 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
205 {
206 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
207 }
208 
209 int
210 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
211 {
212 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
213 }
214 
215 int
216 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
217 {
218 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
219 }
220 
221 static const struct got_error *
222 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
223 {
224 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
225 		return got_error(GOT_ERR_NO_SPACE);
226 
227 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
228 	fileindex->nentries++;
229 	return NULL;
230 }
231 
232 const struct got_error *
233 got_fileindex_entry_add(struct got_fileindex *fileindex,
234     struct got_fileindex_entry *ie)
235 {
236 	/* Flag this entry until it gets written out to disk. */
237 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
238 
239 	return add_entry(fileindex, ie);
240 }
241 
242 void
243 got_fileindex_entry_remove(struct got_fileindex *fileindex,
244     struct got_fileindex_entry *ie)
245 {
246 	/*
247 	 * Removing an entry from the RB tree immediately breaks
248 	 * in-progress iterations over file index entries.
249 	 * So flag this entry for removal and remove it once the index
250 	 * is written out to disk. Meanwhile, pretend this entry no longer
251 	 * exists if we get queried for it again before then.
252 	 */
253 	ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
254 	fileindex->nentries--;
255 }
256 
257 struct got_fileindex_entry *
258 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
259     size_t path_len)
260 {
261 	struct got_fileindex_entry *ie;
262 	struct got_fileindex_entry key;
263 	memset(&key, 0, sizeof(key));
264 	key.path = (char *)path;
265 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
266 	ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
267 	if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
268 		return NULL;
269 	return ie;
270 }
271 
272 const struct got_error *
273 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
274     got_fileindex_cb cb, void *cb_arg)
275 {
276 	const struct got_error *err;
277 	struct got_fileindex_entry *ie, *tmp;
278 
279 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
280 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
281 			continue;
282 		err = (*cb)(cb_arg, ie);
283 		if (err)
284 			return err;
285 	}
286 	return NULL;
287 }
288 
289 struct got_fileindex *
290 got_fileindex_alloc(void)
291 {
292 	struct got_fileindex *fileindex;
293 
294 	fileindex = calloc(1, sizeof(*fileindex));
295 	if (fileindex == NULL)
296 		return NULL;
297 
298 	RB_INIT(&fileindex->entries);
299 	return fileindex;
300 }
301 
302 void
303 got_fileindex_free(struct got_fileindex *fileindex)
304 {
305 	struct got_fileindex_entry *ie;
306 
307 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
308 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
309 		got_fileindex_entry_free(ie);
310 	}
311 	free(fileindex);
312 }
313 
314 static const struct got_error *
315 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
316 {
317 	size_t n;
318 
319 	val = htobe64(val);
320 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
321 	n = fwrite(&val, 1, sizeof(val), outfile);
322 	if (n != sizeof(val))
323 		return got_ferror(outfile, GOT_ERR_IO);
324 	return NULL;
325 }
326 
327 static const struct got_error *
328 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
329 {
330 	size_t n;
331 
332 	val = htobe32(val);
333 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
334 	n = fwrite(&val, 1, sizeof(val), outfile);
335 	if (n != sizeof(val))
336 		return got_ferror(outfile, GOT_ERR_IO);
337 	return NULL;
338 }
339 
340 static const struct got_error *
341 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
342 {
343 	size_t n;
344 
345 	val = htobe16(val);
346 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
347 	n = fwrite(&val, 1, sizeof(val), outfile);
348 	if (n != sizeof(val))
349 		return got_ferror(outfile, GOT_ERR_IO);
350 	return NULL;
351 }
352 
353 static const struct got_error *
354 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
355 {
356 	size_t n, len, pad = 0;
357 	static const uint8_t zero[8] = { 0 };
358 
359 	len = strlen(path);
360 	while ((len + pad) % 8 != 0)
361 		pad++;
362 	if (pad == 0)
363 		pad = 8; /* NUL-terminate */
364 
365 	SHA1Update(ctx, path, len);
366 	n = fwrite(path, 1, len, outfile);
367 	if (n != len)
368 		return got_ferror(outfile, GOT_ERR_IO);
369 	SHA1Update(ctx, zero, pad);
370 	n = fwrite(zero, 1, pad, outfile);
371 	if (n != pad)
372 		return got_ferror(outfile, GOT_ERR_IO);
373 	return NULL;
374 }
375 
376 static const struct got_error *
377 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
378     FILE *outfile)
379 {
380 	const struct got_error *err;
381 	size_t n;
382 	uint32_t stage;
383 
384 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
385 	if (err)
386 		return err;
387 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
388 	if (err)
389 		return err;
390 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
391 	if (err)
392 		return err;
393 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
394 	if (err)
395 		return err;
396 
397 	err = write_fileindex_val32(ctx, ie->uid, outfile);
398 	if (err)
399 		return err;
400 	err = write_fileindex_val32(ctx, ie->gid, outfile);
401 	if (err)
402 		return err;
403 	err = write_fileindex_val32(ctx, ie->size, outfile);
404 	if (err)
405 		return err;
406 
407 	err = write_fileindex_val16(ctx, ie->mode, outfile);
408 	if (err)
409 		return err;
410 
411 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
412 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
413 	if (n != SHA1_DIGEST_LENGTH)
414 		return got_ferror(outfile, GOT_ERR_IO);
415 
416 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
417 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
418 	if (n != SHA1_DIGEST_LENGTH)
419 		return got_ferror(outfile, GOT_ERR_IO);
420 
421 	err = write_fileindex_val32(ctx, ie->flags, outfile);
422 	if (err)
423 		return err;
424 
425 	err = write_fileindex_path(ctx, ie->path, outfile);
426 	if (err)
427 		return err;
428 
429 	stage = got_fileindex_entry_stage_get(ie);
430 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
431 	    stage == GOT_FILEIDX_STAGE_ADD) {
432 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
433 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
434 		    outfile);
435 		if (n != SHA1_DIGEST_LENGTH)
436 			return got_ferror(outfile, GOT_ERR_IO);
437 	}
438 
439 	return NULL;
440 }
441 
442 const struct got_error *
443 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
444 {
445 	const struct got_error *err = NULL;
446 	struct got_fileindex_hdr hdr;
447 	SHA1_CTX ctx;
448 	uint8_t sha1[SHA1_DIGEST_LENGTH];
449 	size_t n;
450 	struct got_fileindex_entry *ie, *tmp;
451 
452 	SHA1Init(&ctx);
453 
454 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
455 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
456 	hdr.nentries = htobe32(fileindex->nentries);
457 
458 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
459 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
460 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
461 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
462 	if (n != sizeof(hdr.signature))
463 		return got_ferror(outfile, GOT_ERR_IO);
464 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
465 	if (n != sizeof(hdr.version))
466 		return got_ferror(outfile, GOT_ERR_IO);
467 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
468 	if (n != sizeof(hdr.nentries))
469 		return got_ferror(outfile, GOT_ERR_IO);
470 
471 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
472 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
473 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
474 			RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
475 			got_fileindex_entry_free(ie);
476 			continue;
477 		}
478 		err = write_fileindex_entry(&ctx, ie, outfile);
479 		if (err)
480 			return err;
481 	}
482 
483 	SHA1Final(sha1, &ctx);
484 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
485 	if (n != sizeof(sha1))
486 		return got_ferror(outfile, GOT_ERR_IO);
487 
488 	if (fflush(outfile) != 0)
489 		return got_error_from_errno("fflush");
490 
491 	return NULL;
492 }
493 
494 static const struct got_error *
495 read_fileindex_val64(uint64_t *val, SHA1_CTX *ctx, FILE *infile)
496 {
497 	size_t n;
498 
499 	n = fread(val, 1, sizeof(*val), infile);
500 	if (n != sizeof(*val))
501 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
502 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
503 	*val = be64toh(*val);
504 	return NULL;
505 }
506 
507 static const struct got_error *
508 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
509 {
510 	size_t n;
511 
512 	n = fread(val, 1, sizeof(*val), infile);
513 	if (n != sizeof(*val))
514 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
515 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
516 	*val = be32toh(*val);
517 	return NULL;
518 }
519 
520 static const struct got_error *
521 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
522 {
523 	size_t n;
524 
525 	n = fread(val, 1, sizeof(*val), infile);
526 	if (n != sizeof(*val))
527 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
528 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
529 	*val = be16toh(*val);
530 	return NULL;
531 }
532 
533 static const struct got_error *
534 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
535 {
536 	const struct got_error *err = NULL;
537 	const size_t chunk_size = 8;
538 	size_t n, len = 0, totlen = chunk_size;
539 
540 	*path = malloc(totlen);
541 	if (*path == NULL)
542 		return got_error_from_errno("malloc");
543 
544 	do {
545 		if (len + chunk_size > totlen) {
546 			char *p = reallocarray(*path, totlen + chunk_size, 1);
547 			if (p == NULL) {
548 				err = got_error_from_errno("reallocarray");
549 				break;
550 			}
551 			totlen += chunk_size;
552 			*path = p;
553 		}
554 		n = fread(*path + len, 1, chunk_size, infile);
555 		if (n != chunk_size) {
556 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
557 			break;
558 		}
559 		SHA1Update(ctx, *path + len, chunk_size);
560 		len += chunk_size;
561 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
562 
563 	if (err) {
564 		free(*path);
565 		*path = NULL;
566 	}
567 	return err;
568 }
569 
570 static const struct got_error *
571 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
572     FILE *infile, uint32_t version)
573 {
574 	const struct got_error *err;
575 	struct got_fileindex_entry *ie;
576 	size_t n;
577 
578 	*iep = NULL;
579 
580 	ie = calloc(1, sizeof(*ie));
581 	if (ie == NULL)
582 		return got_error_from_errno("calloc");
583 
584 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
585 	if (err)
586 		goto done;
587 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
588 	if (err)
589 		goto done;
590 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
591 	if (err)
592 		goto done;
593 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
594 	if (err)
595 		goto done;
596 
597 	err = read_fileindex_val32(&ie->uid, ctx, infile);
598 	if (err)
599 		goto done;
600 	err = read_fileindex_val32(&ie->gid, ctx, infile);
601 	if (err)
602 		goto done;
603 	err = read_fileindex_val32(&ie->size, ctx, infile);
604 	if (err)
605 		goto done;
606 
607 	err = read_fileindex_val16(&ie->mode, ctx, infile);
608 	if (err)
609 		goto done;
610 
611 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
612 	if (n != SHA1_DIGEST_LENGTH) {
613 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
614 		goto done;
615 	}
616 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
617 
618 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
619 	if (n != SHA1_DIGEST_LENGTH) {
620 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
621 		goto done;
622 	}
623 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
624 
625 	err = read_fileindex_val32(&ie->flags, ctx, infile);
626 	if (err)
627 		goto done;
628 
629 	err = read_fileindex_path(&ie->path, ctx, infile);
630 	if (err)
631 		goto done;
632 
633 	if (version >= 2) {
634 		uint32_t stage = got_fileindex_entry_stage_get(ie);
635 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
636 		    stage == GOT_FILEIDX_STAGE_ADD) {
637 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
638 			    infile);
639 			if (n != SHA1_DIGEST_LENGTH) {
640 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
641 				goto done;
642 			}
643 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
644 		}
645 	} else {
646 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
647 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
648 	}
649 
650 done:
651 	if (err)
652 		got_fileindex_entry_free(ie);
653 	else
654 		*iep = ie;
655 	return err;
656 }
657 
658 const struct got_error *
659 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
660 {
661 	const struct got_error *err = NULL;
662 	struct got_fileindex_hdr hdr;
663 	SHA1_CTX ctx;
664 	struct got_fileindex_entry *ie;
665 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
666 	uint8_t sha1[SHA1_DIGEST_LENGTH];
667 	size_t n;
668 	int i;
669 
670 	SHA1Init(&ctx);
671 
672 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
673 	if (n != sizeof(hdr.signature)) {
674 		if (n == 0) /* EOF */
675 			return NULL;
676 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
677 	}
678 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
679 	if (n != sizeof(hdr.version)) {
680 		if (n == 0) /* EOF */
681 			return NULL;
682 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
683 	}
684 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
685 	if (n != sizeof(hdr.nentries)) {
686 		if (n == 0) /* EOF */
687 			return NULL;
688 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
689 	}
690 
691 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
692 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
693 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
694 
695 	hdr.signature = be32toh(hdr.signature);
696 	hdr.version = be32toh(hdr.version);
697 	hdr.nentries = be32toh(hdr.nentries);
698 
699 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
700 		return got_error(GOT_ERR_FILEIDX_SIG);
701 	if (hdr.version > GOT_FILE_INDEX_VERSION)
702 		return got_error(GOT_ERR_FILEIDX_VER);
703 
704 	for (i = 0; i < hdr.nentries; i++) {
705 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
706 		if (err)
707 			return err;
708 		err = add_entry(fileindex, ie);
709 		if (err)
710 			return err;
711 	}
712 
713 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
714 	if (n != sizeof(sha1_expected))
715 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
716 	SHA1Final(sha1, &ctx);
717 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
718 		return got_error(GOT_ERR_FILEIDX_CSUM);
719 
720 	return NULL;
721 }
722 
723 static struct got_fileindex_entry *
724 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
725 {
726 	struct got_fileindex_entry *next;
727 
728 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
729 
730 	/* Skip entries which were added or removed by diff callbacks. */
731 	while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
732 	    GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
733 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
734 
735 	return next;
736 }
737 
738 static const struct got_error *
739 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
740     struct got_tree_object *tree, const char *, const char *,
741     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
742 
743 static const struct got_error *
744 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
745     struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
746     const char *path, const char *entry_name, struct got_repository *repo,
747     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
748 {
749 	const struct got_error *err = NULL;
750 	struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
751 
752 	if (!got_object_tree_entry_is_submodule(te) &&
753 	    S_ISDIR(got_tree_entry_get_mode(te))) {
754 		char *subpath;
755 		struct got_tree_object *subtree;
756 
757 		if (asprintf(&subpath, "%s%s%s", path,
758 		    path[0] == '\0' ? "" : "/",
759 		    got_tree_entry_get_name(te)) == -1)
760 			return got_error_from_errno("asprintf");
761 
762 		err = got_object_open_as_tree(&subtree, repo,
763 		    got_tree_entry_get_id(te));
764 		if (err) {
765 			free(subpath);
766 			return err;
767 		}
768 
769 		err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
770 		    entry_name, repo, cb, cb_arg);
771 		free(subpath);
772 		got_object_tree_close(subtree);
773 		if (err)
774 			return err;
775 	}
776 
777 	(*tidx)++;
778 	*next = got_object_tree_get_entry(tree, *tidx);
779 	return NULL;
780 }
781 
782 static const struct got_error *
783 diff_fileindex_tree(struct got_fileindex *fileindex,
784     struct got_fileindex_entry **ie, struct got_tree_object *tree,
785     const char *path, const char *entry_name, struct got_repository *repo,
786     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
787 {
788 	const struct got_error *err = NULL;
789 	struct got_tree_entry *te = NULL;
790 	size_t path_len = strlen(path);
791 	struct got_fileindex_entry *next;
792 	int tidx = 0;
793 
794 	te = got_object_tree_get_entry(tree, tidx);
795 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
796 		if (te && *ie) {
797 			char *te_path;
798 			const char *te_name = got_tree_entry_get_name(te);
799 			int cmp;
800 			if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
801 				err = got_error_from_errno("asprintf");
802 				break;
803 			}
804 			cmp = got_path_cmp((*ie)->path, te_path,
805 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
806 			free(te_path);
807 			if (cmp == 0) {
808 				if (got_path_is_child((*ie)->path, path,
809 				    path_len) &&
810 				    !got_object_tree_entry_is_submodule(te) &&
811 				    (entry_name == NULL ||
812 				    strcmp(te_name, entry_name) == 0)) {
813 					err = cb->diff_old_new(cb_arg, *ie, te,
814 					    path);
815 					if (err || entry_name)
816 						break;
817 				}
818 				*ie = walk_fileindex(fileindex, *ie);
819 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
820 				    path, entry_name, repo, cb, cb_arg);
821 			} else if (cmp < 0) {
822 				next = walk_fileindex(fileindex, *ie);
823 				if (got_path_is_child((*ie)->path, path,
824 				    path_len) && (entry_name == NULL ||
825 				    strcmp(te_name, 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 {
832 				if ((entry_name == NULL ||
833 				    strcmp(te_name, entry_name) == 0)) {
834 					err = cb->diff_new(cb_arg, te, path);
835 					if (err || entry_name)
836 						break;
837 				}
838 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
839 				    path, entry_name, repo, cb, cb_arg);
840 			}
841 			if (err)
842 				break;
843 		} else if (*ie) {
844 			next = walk_fileindex(fileindex, *ie);
845 			if (got_path_is_child((*ie)->path, path, path_len) &&
846 			    (entry_name == NULL ||
847 			    (te && strcmp(got_tree_entry_get_name(te),
848 			    entry_name) == 0))) {
849 				err = cb->diff_old(cb_arg, *ie, path);
850 				if (err || entry_name)
851 					break;
852 			}
853 			*ie = next;
854 		} else if (te) {
855 			if (!got_object_tree_entry_is_submodule(te) &&
856 			    (entry_name == NULL ||
857 			    strcmp(got_tree_entry_get_name(te), entry_name)
858 			    == 0)) {
859 				err = cb->diff_new(cb_arg, te, path);
860 				if (err || entry_name)
861 					break;
862 			}
863 			err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
864 			    entry_name, repo, cb, cb_arg);
865 			if (err)
866 				break;
867 		}
868 	}
869 
870 	return err;
871 }
872 
873 const struct got_error *
874 got_fileindex_diff_tree(struct got_fileindex *fileindex,
875     struct got_tree_object *tree, const char *path, const char *entry_name,
876     struct got_repository *repo,
877     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
878 {
879 	struct got_fileindex_entry *ie;
880 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
881 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
882 		ie = walk_fileindex(fileindex, ie);
883 	return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
884 	    cb, cb_arg);
885 }
886 
887 static const struct got_error *
888 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
889     struct got_pathlist_head *, int, const char *, const char *,
890     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
891 
892 static const struct got_error *
893 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
894 {
895 	const struct got_error *err = NULL;
896 	struct got_pathlist_entry *new = NULL;
897 	struct dirent *dep = NULL;
898 	struct dirent *de = NULL;
899 
900 	for (;;) {
901 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
902 		if (de == NULL) {
903 			err = got_error_from_errno("malloc");
904 			break;
905 		}
906 
907 		if (readdir_r(dir, de, &dep) != 0) {
908 			err = got_error_from_errno("readdir_r");
909 			free(de);
910 			break;
911 		}
912 		if (dep == NULL) {
913 			free(de);
914 			break;
915 		}
916 
917 		if (strcmp(de->d_name, ".") == 0 ||
918 		    strcmp(de->d_name, "..") == 0 ||
919 		    (path[0] == '\0' &&
920 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
921 			free(de);
922 			continue;
923 		}
924 
925 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
926 		if (err) {
927 			free(de);
928 			break;
929 		}
930 		if (new == NULL) {
931 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
932 			free(de);
933 			break;
934 		}
935 	}
936 
937 	return err;
938 }
939 
940 void
941 free_dirlist(struct got_pathlist_head *dirlist)
942 {
943 	struct got_pathlist_entry *dle;
944 
945 	TAILQ_FOREACH(dle, dirlist, entry)
946 		free(dle->data);
947 	got_pathlist_free(dirlist);
948 }
949 
950 static const struct got_error *
951 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
952     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
953     const char *path, const char *rootpath, struct got_repository *repo,
954     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
955 {
956 	const struct got_error *err = NULL;
957 	struct dirent *de = dle->data;
958 	DIR *subdir = NULL;
959 	int subdirfd = -1;
960 	int type;
961 
962 	*next = NULL;
963 
964 	if (de->d_type == DT_UNKNOWN) {
965 		/* Occurs on NFS mounts without "readdir plus" RPC. */
966 		char *dir_path;
967 		if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
968 			return got_error_from_errno("asprintf");
969 		err = got_path_dirent_type(&type, dir_path, de);
970 		free(dir_path);
971 		if (err)
972 			return err;
973 	} else
974 		type = de->d_type;
975 
976 	if (type == DT_DIR) {
977 		char *subpath;
978 		char *subdirpath;
979 		struct got_pathlist_head subdirlist;
980 
981 		TAILQ_INIT(&subdirlist);
982 
983 		if (asprintf(&subpath, "%s%s%s", path,
984 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
985 			return got_error_from_errno("asprintf");
986 
987 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
988 			free(subpath);
989 			return got_error_from_errno("asprintf");
990 		}
991 
992 		subdirfd = openat(fd, de->d_name,
993 		    O_RDONLY | O_NOFOLLOW | O_DIRECTORY);
994 		if (subdirfd == -1) {
995 			if (errno == EACCES) {
996 				*next = TAILQ_NEXT(dle, entry);
997 				return NULL;
998 			}
999 			err = got_error_from_errno2("openat", subdirpath);
1000 			free(subpath);
1001 			free(subdirpath);
1002 			return err;
1003 		}
1004 
1005 		subdir = fdopendir(subdirfd);
1006 		if (subdir == NULL)
1007 			return got_error_from_errno2("fdopendir", path);
1008 		subdirfd = -1;
1009 		err = read_dirlist(&subdirlist, subdir, subdirpath);
1010 		if (err) {
1011 			free(subpath);
1012 			free(subdirpath);
1013 			closedir(subdir);
1014 			return err;
1015 		}
1016 		err = diff_fileindex_dir(fileindex, ie, &subdirlist,
1017 		    dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
1018 		if (subdir && closedir(subdir) == -1 && err == NULL)
1019 			err = got_error_from_errno2("closedir", subdirpath);
1020 		free(subpath);
1021 		free(subdirpath);
1022 		free_dirlist(&subdirlist);
1023 		if (err)
1024 			return err;
1025 	}
1026 
1027 	*next = TAILQ_NEXT(dle, entry);
1028 	return NULL;
1029 }
1030 
1031 static const struct got_error *
1032 diff_fileindex_dir(struct got_fileindex *fileindex,
1033     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1034     int dirfd, const char *rootpath, const char *path,
1035     struct got_repository *repo,
1036     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1037 {
1038 	const struct got_error *err = NULL;
1039 	struct dirent *de = NULL;
1040 	size_t path_len = strlen(path);
1041 	struct got_pathlist_entry *dle;
1042 
1043 	if (cb->diff_traverse) {
1044 		err = cb->diff_traverse(cb_arg, path, dirfd);
1045 		if (err)
1046 			return err;
1047 	}
1048 
1049 	dle = TAILQ_FIRST(dirlist);
1050 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1051 		if (dle && *ie) {
1052 			char *de_path;
1053 			int cmp;
1054 			de = dle->data;
1055 			if (asprintf(&de_path, "%s/%s", path,
1056 			    de->d_name) == -1) {
1057 				err = got_error_from_errno("asprintf");
1058 				break;
1059 			}
1060 			cmp = got_path_cmp((*ie)->path, de_path,
1061 			    got_fileindex_entry_path_len(*ie),
1062 			    strlen(path) + 1 + de->d_namlen);
1063 			free(de_path);
1064 			if (cmp == 0) {
1065 				err = cb->diff_old_new(cb_arg, *ie, de, path,
1066 				    dirfd);
1067 				if (err)
1068 					break;
1069 				*ie = walk_fileindex(fileindex, *ie);
1070 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1071 				    path, rootpath, repo, cb, cb_arg);
1072 			} else if (cmp < 0 ) {
1073 				err = cb->diff_old(cb_arg, *ie, path);
1074 				if (err)
1075 					break;
1076 				*ie = walk_fileindex(fileindex, *ie);
1077 			} else {
1078 				err = cb->diff_new(cb_arg, de, path, dirfd);
1079 				if (err)
1080 					break;
1081 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1082 				    path, rootpath, repo, cb, cb_arg);
1083 			}
1084 			if (err)
1085 				break;
1086 		} else if (*ie) {
1087 			err = cb->diff_old(cb_arg, *ie, path);
1088 			if (err)
1089 				break;
1090 			*ie = walk_fileindex(fileindex, *ie);
1091 		} else if (dle) {
1092 			de = dle->data;
1093 			err = cb->diff_new(cb_arg, de, path, dirfd);
1094 			if (err)
1095 				break;
1096 			err = walk_dir(&dle, fileindex, ie, dle, dirfd, path,
1097 			    rootpath, repo, cb, cb_arg);
1098 			if (err)
1099 				break;
1100 		}
1101 	}
1102 
1103 	return err;
1104 }
1105 
1106 const struct got_error *
1107 got_fileindex_diff_dir(struct got_fileindex *fileindex, int fd,
1108     const char *rootpath, const char *path, struct got_repository *repo,
1109     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1110 {
1111 	const struct got_error *err;
1112 	struct got_fileindex_entry *ie;
1113 	struct got_pathlist_head dirlist;
1114 	int fd2;
1115 	DIR *dir;
1116 
1117 	TAILQ_INIT(&dirlist);
1118 
1119 	/* 
1120 	 * Duplicate the file descriptor so we can call closedir() below
1121 	 * without closing the file descriptor passed in by our caller.
1122 	 */
1123 	fd2 = dup(fd);
1124 	if (fd2 == -1)
1125 		return got_error_from_errno2("dup", path);
1126 	if (lseek(fd2, 0, SEEK_SET) == -1) {
1127 		err = got_error_from_errno2("lseek", path);
1128 		close(fd2);
1129 		return err;
1130 	}
1131 	dir = fdopendir(fd2);
1132 	if (dir == NULL) {
1133 		err = got_error_from_errno2("fdopendir", path);
1134 		close(fd2);
1135 		return err;
1136 	}
1137 	err = read_dirlist(&dirlist, dir, path);
1138 	if (err) {
1139 		closedir(dir);
1140 		return err;
1141 	}
1142 
1143 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1144 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1145 		ie = walk_fileindex(fileindex, ie);
1146 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, dirfd(dir),
1147 	    rootpath, path, repo, cb, cb_arg);
1148 
1149 	if (closedir(dir) == -1 && err == NULL)
1150 		err = got_error_from_errno2("closedir", path);
1151 	free_dirlist(&dirlist);
1152 	return err;
1153 }
1154 
1155 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);