op public repos

Blob

Date:: Thu Jul 23 14:22:37 2020 UTC
Message:: set a staged file type and handle it separately from the on-disk file type
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/queue.h>
18 #include <sys/tree.h>
19 #include <sys/stat.h>
20 
21 #include <errno.h>
22 #include <dirent.h>
23 #include <fcntl.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <sha1.h>
28 #include <endian.h>
29 #include <limits.h>
30 #include <unistd.h>
31 #include <uuid.h>
32 
33 #include "got_error.h"
34 #include "got_object.h"
35 #include "got_path.h"
36 
37 #include "got_lib_fileindex.h"
38 #include "got_lib_worktree.h"
39 
40 /* got_fileindex_entry flags */
41 #define GOT_FILEIDX_F_PATH_LEN		0x00000fff
42 #define GOT_FILEIDX_F_STAGE		0x0000f000
43 #define GOT_FILEIDX_F_STAGE_SHIFT	12
44 #define GOT_FILEIDX_F_NOT_FLUSHED	0x00010000
45 #define GOT_FILEIDX_F_NO_BLOB		0x00020000
46 #define GOT_FILEIDX_F_NO_COMMIT		0x00040000
47 #define GOT_FILEIDX_F_NO_FILE_ON_DISK	0x00080000
48 #define GOT_FILEIDX_F_REMOVE_ON_FLUSH	0x00100000
49 
50 struct got_fileindex {
51 	struct got_fileindex_tree entries;
52 	int nentries; /* Does not include entries marked for removal. */
53 #define GOT_FILEIDX_MAX_ENTRIES INT_MAX
54 };
55 
56 mode_t
57 got_fileindex_entry_perms_get(struct got_fileindex_entry *ie)
58 {
59 	return ((ie->mode & GOT_FILEIDX_MODE_PERMS) >>
60 	    GOT_FILEIDX_MODE_PERMS_SHIFT);
61 }
62 
63 static void
64 fileindex_entry_perms_set(struct got_fileindex_entry *ie, mode_t mode)
65 {
66 	ie->mode &= ~GOT_FILEIDX_MODE_PERMS;
67 	ie->mode |= ((mode << GOT_FILEIDX_MODE_PERMS_SHIFT) &
68 	    GOT_FILEIDX_MODE_PERMS);
69 }
70 
71 mode_t
72 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
73 {
74 	mode_t perms = got_fileindex_entry_perms_get(ie);
75 	int type = got_fileindex_entry_filetype_get(ie);
76 	uint32_t ftype;
77 
78 	if (type == GOT_FILEIDX_MODE_REGULAR_FILE ||
79 	    type == GOT_FILEIDX_MODE_BAD_SYMLINK)
80 		ftype = S_IFREG;
81 	else
82 		ftype = S_IFLNK;
83 
84 	return (ftype | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
85 }
86 
87 const struct got_error *
88 got_fileindex_entry_update(struct got_fileindex_entry *ie,
89     const char *ondisk_path, uint8_t *blob_sha1, uint8_t *commit_sha1,
90     int update_timestamps)
91 {
92 	struct stat sb;
93 
94 	if (lstat(ondisk_path, &sb) != 0) {
95 		if (!((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) &&
96 		    errno == ENOENT))
97 			return got_error_from_errno2("lstat", ondisk_path);
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_ctime;
108 			ie->ctime_nsec = sb.st_ctimensec;
109 			ie->mtime_sec = sb.st_mtime;
110 			ie->mtime_nsec = sb.st_mtimensec;
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 const struct got_error *
149 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
150     const char *relpath)
151 {
152 	size_t len;
153 
154 	*ie = calloc(1, sizeof(**ie));
155 	if (*ie == NULL)
156 		return got_error_from_errno("calloc");
157 
158 	(*ie)->path = strdup(relpath);
159 	if ((*ie)->path == NULL) {
160 		const struct got_error *err = got_error_from_errno("strdup");
161 		free(*ie);
162 		*ie = NULL;
163 		return err;
164 	}
165 
166 	len = strlen(relpath);
167 	if (len > GOT_FILEIDX_F_PATH_LEN)
168 		len = GOT_FILEIDX_F_PATH_LEN;
169 	(*ie)->flags |= len;
170 
171 	return NULL;
172 }
173 
174 void
175 got_fileindex_entry_free(struct got_fileindex_entry *ie)
176 {
177 	free(ie->path);
178 	free(ie);
179 }
180 
181 size_t
182 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
183 {
184 	return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
185 }
186 
187 uint32_t
188 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
189 {
190 	return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
191 }
192 
193 void
194 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
195 {
196 	ie->flags &= ~GOT_FILEIDX_F_STAGE;
197 	ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
198 	    GOT_FILEIDX_F_STAGE);
199 }
200 
201 int
202 got_fileindex_entry_filetype_get(struct got_fileindex_entry *ie)
203 {
204 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
205 }
206 
207 void
208 got_fileindex_entry_filetype_set(struct got_fileindex_entry *ie, int type)
209 {
210 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_ONDISK;
211 	ie->mode |= (type & GOT_FILEIDX_MODE_FILE_TYPE_ONDISK);
212 }
213 
214 void
215 got_fileindex_entry_staged_filetype_set(struct got_fileindex_entry *ie, int type)
216 {
217 	ie->mode &= ~GOT_FILEIDX_MODE_FILE_TYPE_STAGED;
218 	ie->mode |= ((type << GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT) &
219 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED);
220 }
221 
222 int
223 got_fileindex_entry_staged_filetype_get(struct got_fileindex_entry *ie)
224 {
225 	return (ie->mode & GOT_FILEIDX_MODE_FILE_TYPE_STAGED) >>
226 	    GOT_FILEIDX_MODE_FILE_TYPE_STAGED_SHIFT;
227 }
228 
229 int
230 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
231 {
232 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
233 }
234 
235 int
236 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
237 {
238 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
239 }
240 
241 int
242 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
243 {
244 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
245 }
246 
247 static const struct got_error *
248 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
249 {
250 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
251 		return got_error(GOT_ERR_NO_SPACE);
252 
253 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
254 	fileindex->nentries++;
255 	return NULL;
256 }
257 
258 const struct got_error *
259 got_fileindex_entry_add(struct got_fileindex *fileindex,
260     struct got_fileindex_entry *ie)
261 {
262 	/* Flag this entry until it gets written out to disk. */
263 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
264 
265 	return add_entry(fileindex, ie);
266 }
267 
268 void
269 got_fileindex_entry_remove(struct got_fileindex *fileindex,
270     struct got_fileindex_entry *ie)
271 {
272 	/*
273 	 * Removing an entry from the RB tree immediately breaks
274 	 * in-progress iterations over file index entries.
275 	 * So flag this entry for removal and remove it once the index
276 	 * is written out to disk. Meanwhile, pretend this entry no longer
277 	 * exists if we get queried for it again before then.
278 	 */
279 	ie->flags |= GOT_FILEIDX_F_REMOVE_ON_FLUSH;
280 	fileindex->nentries--;
281 }
282 
283 struct got_fileindex_entry *
284 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
285     size_t path_len)
286 {
287 	struct got_fileindex_entry *ie;
288 	struct got_fileindex_entry key;
289 	memset(&key, 0, sizeof(key));
290 	key.path = (char *)path;
291 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
292 	ie = RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
293 	if (ie && (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH))
294 		return NULL;
295 	return ie;
296 }
297 
298 const struct got_error *
299 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
300     got_fileindex_cb cb, void *cb_arg)
301 {
302 	const struct got_error *err;
303 	struct got_fileindex_entry *ie, *tmp;
304 
305 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
306 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH)
307 			continue;
308 		err = (*cb)(cb_arg, ie);
309 		if (err)
310 			return err;
311 	}
312 	return NULL;
313 }
314 
315 struct got_fileindex *
316 got_fileindex_alloc(void)
317 {
318 	struct got_fileindex *fileindex;
319 
320 	fileindex = calloc(1, sizeof(*fileindex));
321 	if (fileindex == NULL)
322 		return NULL;
323 
324 	RB_INIT(&fileindex->entries);
325 	return fileindex;
326 }
327 
328 void
329 got_fileindex_free(struct got_fileindex *fileindex)
330 {
331 	struct got_fileindex_entry *ie;
332 
333 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
334 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
335 		got_fileindex_entry_free(ie);
336 	}
337 	free(fileindex);
338 }
339 
340 static const struct got_error *
341 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
342 {
343 	size_t n;
344 
345 	val = htobe64(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_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
355 {
356 	size_t n;
357 
358 	val = htobe32(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_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
368 {
369 	size_t n;
370 
371 	val = htobe16(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_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
381 {
382 	size_t n, len, pad = 0;
383 	static const uint8_t zero[8] = { 0 };
384 
385 	len = strlen(path);
386 	while ((len + pad) % 8 != 0)
387 		pad++;
388 	if (pad == 0)
389 		pad = 8; /* NUL-terminate */
390 
391 	SHA1Update(ctx, path, len);
392 	n = fwrite(path, 1, len, outfile);
393 	if (n != len)
394 		return got_ferror(outfile, GOT_ERR_IO);
395 	SHA1Update(ctx, zero, pad);
396 	n = fwrite(zero, 1, pad, outfile);
397 	if (n != pad)
398 		return got_ferror(outfile, GOT_ERR_IO);
399 	return NULL;
400 }
401 
402 static const struct got_error *
403 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
404     FILE *outfile)
405 {
406 	const struct got_error *err;
407 	size_t n;
408 	uint32_t stage;
409 
410 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
411 	if (err)
412 		return err;
413 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
414 	if (err)
415 		return err;
416 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
417 	if (err)
418 		return err;
419 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
420 	if (err)
421 		return err;
422 
423 	err = write_fileindex_val32(ctx, ie->uid, outfile);
424 	if (err)
425 		return err;
426 	err = write_fileindex_val32(ctx, ie->gid, outfile);
427 	if (err)
428 		return err;
429 	err = write_fileindex_val32(ctx, ie->size, outfile);
430 	if (err)
431 		return err;
432 
433 	err = write_fileindex_val16(ctx, ie->mode, outfile);
434 	if (err)
435 		return err;
436 
437 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
438 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
439 	if (n != SHA1_DIGEST_LENGTH)
440 		return got_ferror(outfile, GOT_ERR_IO);
441 
442 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
443 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
444 	if (n != SHA1_DIGEST_LENGTH)
445 		return got_ferror(outfile, GOT_ERR_IO);
446 
447 	err = write_fileindex_val32(ctx, ie->flags, outfile);
448 	if (err)
449 		return err;
450 
451 	err = write_fileindex_path(ctx, ie->path, outfile);
452 	if (err)
453 		return err;
454 
455 	stage = got_fileindex_entry_stage_get(ie);
456 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
457 	    stage == GOT_FILEIDX_STAGE_ADD) {
458 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
459 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
460 		    outfile);
461 		if (n != SHA1_DIGEST_LENGTH)
462 			return got_ferror(outfile, GOT_ERR_IO);
463 	}
464 
465 	return NULL;
466 }
467 
468 const struct got_error *
469 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
470 {
471 	const struct got_error *err = NULL;
472 	struct got_fileindex_hdr hdr;
473 	SHA1_CTX ctx;
474 	uint8_t sha1[SHA1_DIGEST_LENGTH];
475 	size_t n;
476 	struct got_fileindex_entry *ie, *tmp;
477 
478 	SHA1Init(&ctx);
479 
480 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
481 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
482 	hdr.nentries = htobe32(fileindex->nentries);
483 
484 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
485 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
486 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
487 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
488 	if (n != sizeof(hdr.signature))
489 		return got_ferror(outfile, GOT_ERR_IO);
490 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
491 	if (n != sizeof(hdr.version))
492 		return got_ferror(outfile, GOT_ERR_IO);
493 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
494 	if (n != sizeof(hdr.nentries))
495 		return got_ferror(outfile, GOT_ERR_IO);
496 
497 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
498 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
499 		if (ie->flags & GOT_FILEIDX_F_REMOVE_ON_FLUSH) {
500 			RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
501 			got_fileindex_entry_free(ie);
502 			continue;
503 		}
504 		err = write_fileindex_entry(&ctx, ie, outfile);
505 		if (err)
506 			return err;
507 	}
508 
509 	SHA1Final(sha1, &ctx);
510 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
511 	if (n != sizeof(sha1))
512 		return got_ferror(outfile, GOT_ERR_IO);
513 
514 	if (fflush(outfile) != 0)
515 		return got_error_from_errno("fflush");
516 
517 	return NULL;
518 }
519 
520 static const struct got_error *
521 read_fileindex_val64(uint64_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 = be64toh(*val);
530 	return NULL;
531 }
532 
533 static const struct got_error *
534 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
535 {
536 	size_t n;
537 
538 	n = fread(val, 1, sizeof(*val), infile);
539 	if (n != sizeof(*val))
540 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
541 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
542 	*val = be32toh(*val);
543 	return NULL;
544 }
545 
546 static const struct got_error *
547 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
548 {
549 	size_t n;
550 
551 	n = fread(val, 1, sizeof(*val), infile);
552 	if (n != sizeof(*val))
553 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
554 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
555 	*val = be16toh(*val);
556 	return NULL;
557 }
558 
559 static const struct got_error *
560 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
561 {
562 	const struct got_error *err = NULL;
563 	const size_t chunk_size = 8;
564 	size_t n, len = 0, totlen = chunk_size;
565 
566 	*path = malloc(totlen);
567 	if (*path == NULL)
568 		return got_error_from_errno("malloc");
569 
570 	do {
571 		if (len + chunk_size > totlen) {
572 			char *p = reallocarray(*path, totlen + chunk_size, 1);
573 			if (p == NULL) {
574 				err = got_error_from_errno("reallocarray");
575 				break;
576 			}
577 			totlen += chunk_size;
578 			*path = p;
579 		}
580 		n = fread(*path + len, 1, chunk_size, infile);
581 		if (n != chunk_size) {
582 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
583 			break;
584 		}
585 		SHA1Update(ctx, *path + len, chunk_size);
586 		len += chunk_size;
587 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
588 
589 	if (err) {
590 		free(*path);
591 		*path = NULL;
592 	}
593 	return err;
594 }
595 
596 static const struct got_error *
597 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
598     FILE *infile, uint32_t version)
599 {
600 	const struct got_error *err;
601 	struct got_fileindex_entry *ie;
602 	size_t n;
603 
604 	*iep = NULL;
605 
606 	ie = calloc(1, sizeof(*ie));
607 	if (ie == NULL)
608 		return got_error_from_errno("calloc");
609 
610 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
611 	if (err)
612 		goto done;
613 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
614 	if (err)
615 		goto done;
616 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
617 	if (err)
618 		goto done;
619 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
620 	if (err)
621 		goto done;
622 
623 	err = read_fileindex_val32(&ie->uid, ctx, infile);
624 	if (err)
625 		goto done;
626 	err = read_fileindex_val32(&ie->gid, ctx, infile);
627 	if (err)
628 		goto done;
629 	err = read_fileindex_val32(&ie->size, ctx, infile);
630 	if (err)
631 		goto done;
632 
633 	err = read_fileindex_val16(&ie->mode, ctx, infile);
634 	if (err)
635 		goto done;
636 
637 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
638 	if (n != SHA1_DIGEST_LENGTH) {
639 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
640 		goto done;
641 	}
642 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
643 
644 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
645 	if (n != SHA1_DIGEST_LENGTH) {
646 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
647 		goto done;
648 	}
649 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
650 
651 	err = read_fileindex_val32(&ie->flags, ctx, infile);
652 	if (err)
653 		goto done;
654 
655 	err = read_fileindex_path(&ie->path, ctx, infile);
656 	if (err)
657 		goto done;
658 
659 	if (version >= 2) {
660 		uint32_t stage = got_fileindex_entry_stage_get(ie);
661 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
662 		    stage == GOT_FILEIDX_STAGE_ADD) {
663 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
664 			    infile);
665 			if (n != SHA1_DIGEST_LENGTH) {
666 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
667 				goto done;
668 			}
669 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
670 		}
671 	} else {
672 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
673 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
674 	}
675 
676 done:
677 	if (err)
678 		got_fileindex_entry_free(ie);
679 	else
680 		*iep = ie;
681 	return err;
682 }
683 
684 const struct got_error *
685 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
686 {
687 	const struct got_error *err = NULL;
688 	struct got_fileindex_hdr hdr;
689 	SHA1_CTX ctx;
690 	struct got_fileindex_entry *ie;
691 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
692 	uint8_t sha1[SHA1_DIGEST_LENGTH];
693 	size_t n;
694 	int i;
695 
696 	SHA1Init(&ctx);
697 
698 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
699 	if (n != sizeof(hdr.signature)) {
700 		if (n == 0) /* EOF */
701 			return NULL;
702 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
703 	}
704 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
705 	if (n != sizeof(hdr.version)) {
706 		if (n == 0) /* EOF */
707 			return NULL;
708 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
709 	}
710 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
711 	if (n != sizeof(hdr.nentries)) {
712 		if (n == 0) /* EOF */
713 			return NULL;
714 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
715 	}
716 
717 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
718 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
719 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
720 
721 	hdr.signature = be32toh(hdr.signature);
722 	hdr.version = be32toh(hdr.version);
723 	hdr.nentries = be32toh(hdr.nentries);
724 
725 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
726 		return got_error(GOT_ERR_FILEIDX_SIG);
727 	if (hdr.version > GOT_FILE_INDEX_VERSION)
728 		return got_error(GOT_ERR_FILEIDX_VER);
729 
730 	for (i = 0; i < hdr.nentries; i++) {
731 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
732 		if (err)
733 			return err;
734 		err = add_entry(fileindex, ie);
735 		if (err)
736 			return err;
737 	}
738 
739 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
740 	if (n != sizeof(sha1_expected))
741 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
742 	SHA1Final(sha1, &ctx);
743 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
744 		return got_error(GOT_ERR_FILEIDX_CSUM);
745 
746 	return NULL;
747 }
748 
749 static struct got_fileindex_entry *
750 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
751 {
752 	struct got_fileindex_entry *next;
753 
754 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
755 
756 	/* Skip entries which were added or removed by diff callbacks. */
757 	while (next && (next->flags & (GOT_FILEIDX_F_NOT_FLUSHED |
758 	    GOT_FILEIDX_F_REMOVE_ON_FLUSH)))
759 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
760 
761 	return next;
762 }
763 
764 static const struct got_error *
765 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **ie,
766     struct got_tree_object *tree, const char *, const char *,
767     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
768 
769 static const struct got_error *
770 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
771     struct got_fileindex_entry **ie, struct got_tree_object *tree, int *tidx,
772     const char *path, const char *entry_name, struct got_repository *repo,
773     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
774 {
775 	const struct got_error *err = NULL;
776 	struct got_tree_entry *te = got_object_tree_get_entry(tree, *tidx);
777 
778 	if (!got_object_tree_entry_is_submodule(te) &&
779 	    S_ISDIR(got_tree_entry_get_mode(te))) {
780 		char *subpath;
781 		struct got_tree_object *subtree;
782 
783 		if (asprintf(&subpath, "%s%s%s", path,
784 		    path[0] == '\0' ? "" : "/",
785 		    got_tree_entry_get_name(te)) == -1)
786 			return got_error_from_errno("asprintf");
787 
788 		err = got_object_open_as_tree(&subtree, repo,
789 		    got_tree_entry_get_id(te));
790 		if (err) {
791 			free(subpath);
792 			return err;
793 		}
794 
795 		err = diff_fileindex_tree(fileindex, ie, subtree, subpath,
796 		    entry_name, repo, cb, cb_arg);
797 		free(subpath);
798 		got_object_tree_close(subtree);
799 		if (err)
800 			return err;
801 	}
802 
803 	(*tidx)++;
804 	*next = got_object_tree_get_entry(tree, *tidx);
805 	return NULL;
806 }
807 
808 static const struct got_error *
809 diff_fileindex_tree(struct got_fileindex *fileindex,
810     struct got_fileindex_entry **ie, struct got_tree_object *tree,
811     const char *path, const char *entry_name, struct got_repository *repo,
812     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
813 {
814 	const struct got_error *err = NULL;
815 	struct got_tree_entry *te = NULL;
816 	size_t path_len = strlen(path);
817 	struct got_fileindex_entry *next;
818 	int tidx = 0;
819 
820 	te = got_object_tree_get_entry(tree, tidx);
821 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
822 		if (te && *ie) {
823 			char *te_path;
824 			const char *te_name = got_tree_entry_get_name(te);
825 			int cmp;
826 			if (asprintf(&te_path, "%s/%s", path, te_name) == -1) {
827 				err = got_error_from_errno("asprintf");
828 				break;
829 			}
830 			cmp = got_path_cmp((*ie)->path, te_path,
831 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
832 			free(te_path);
833 			if (cmp == 0) {
834 				if (got_path_is_child((*ie)->path, path,
835 				    path_len) &&
836 				    !got_object_tree_entry_is_submodule(te) &&
837 				    (entry_name == NULL ||
838 				    strcmp(te_name, entry_name) == 0)) {
839 					err = cb->diff_old_new(cb_arg, *ie, te,
840 					    path);
841 					if (err || entry_name)
842 						break;
843 				}
844 				*ie = walk_fileindex(fileindex, *ie);
845 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
846 				    path, entry_name, repo, cb, cb_arg);
847 			} else if (cmp < 0) {
848 				next = walk_fileindex(fileindex, *ie);
849 				if (got_path_is_child((*ie)->path, path,
850 				    path_len) && (entry_name == NULL ||
851 				    strcmp(te_name, entry_name) == 0)) {
852 					err = cb->diff_old(cb_arg, *ie, path);
853 					if (err || entry_name)
854 						break;
855 				}
856 				*ie = next;
857 			} else {
858 				if ((entry_name == NULL ||
859 				    strcmp(te_name, entry_name) == 0)) {
860 					err = cb->diff_new(cb_arg, te, path);
861 					if (err || entry_name)
862 						break;
863 				}
864 				err = walk_tree(&te, fileindex, ie, tree, &tidx,
865 				    path, entry_name, repo, cb, cb_arg);
866 			}
867 			if (err)
868 				break;
869 		} else if (*ie) {
870 			next = walk_fileindex(fileindex, *ie);
871 			if (got_path_is_child((*ie)->path, path, path_len) &&
872 			    (entry_name == NULL ||
873 			    (te && strcmp(got_tree_entry_get_name(te),
874 			    entry_name) == 0))) {
875 				err = cb->diff_old(cb_arg, *ie, path);
876 				if (err || entry_name)
877 					break;
878 			}
879 			*ie = next;
880 		} else if (te) {
881 			if (!got_object_tree_entry_is_submodule(te) &&
882 			    (entry_name == NULL ||
883 			    strcmp(got_tree_entry_get_name(te), entry_name)
884 			    == 0)) {
885 				err = cb->diff_new(cb_arg, te, path);
886 				if (err || entry_name)
887 					break;
888 			}
889 			err = walk_tree(&te, fileindex, ie, tree, &tidx, path,
890 			    entry_name, repo, cb, cb_arg);
891 			if (err)
892 				break;
893 		}
894 	}
895 
896 	return err;
897 }
898 
899 const struct got_error *
900 got_fileindex_diff_tree(struct got_fileindex *fileindex,
901     struct got_tree_object *tree, const char *path, const char *entry_name,
902     struct got_repository *repo,
903     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
904 {
905 	struct got_fileindex_entry *ie;
906 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
907 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
908 		ie = walk_fileindex(fileindex, ie);
909 	return diff_fileindex_tree(fileindex, &ie, tree, path, entry_name, repo,
910 	    cb, cb_arg);
911 }
912 
913 static const struct got_error *
914 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
915     struct got_pathlist_head *, int, const char *, const char *,
916     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
917 
918 static const struct got_error *
919 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
920 {
921 	const struct got_error *err = NULL;
922 	struct got_pathlist_entry *new = NULL;
923 	struct dirent *dep = NULL;
924 	struct dirent *de = NULL;
925 
926 	for (;;) {
927 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
928 		if (de == NULL) {
929 			err = got_error_from_errno("malloc");
930 			break;
931 		}
932 
933 		if (readdir_r(dir, de, &dep) != 0) {
934 			err = got_error_from_errno("readdir_r");
935 			free(de);
936 			break;
937 		}
938 		if (dep == NULL) {
939 			free(de);
940 			break;
941 		}
942 
943 		if (strcmp(de->d_name, ".") == 0 ||
944 		    strcmp(de->d_name, "..") == 0 ||
945 		    (path[0] == '\0' &&
946 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
947 			free(de);
948 			continue;
949 		}
950 
951 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
952 		if (err) {
953 			free(de);
954 			break;
955 		}
956 		if (new == NULL) {
957 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
958 			free(de);
959 			break;
960 		}
961 	}
962 
963 	return err;
964 }
965 
966 void
967 free_dirlist(struct got_pathlist_head *dirlist)
968 {
969 	struct got_pathlist_entry *dle;
970 
971 	TAILQ_FOREACH(dle, dirlist, entry)
972 		free(dle->data);
973 	got_pathlist_free(dirlist);
974 }
975 
976 static const struct got_error *
977 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
978     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle, int fd,
979     const char *path, const char *rootpath, struct got_repository *repo,
980     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
981 {
982 	const struct got_error *err = NULL;
983 	struct dirent *de = dle->data;
984 	DIR *subdir = NULL;
985 	int subdirfd = -1;
986 	int type;
987 
988 	*next = NULL;
989 
990 	if (de->d_type == DT_UNKNOWN) {
991 		/* Occurs on NFS mounts without "readdir plus" RPC. */
992 		char *dir_path;
993 		if (asprintf(&dir_path, "%s/%s", rootpath, path) == -1)
994 			return got_error_from_errno("asprintf");
995 		err = got_path_dirent_type(&type, dir_path, de);
996 		free(dir_path);
997 		if (err)
998 			return err;
999 	} else
1000 		type = de->d_type;
1001 
1002 	if (type == DT_DIR) {
1003 		char *subpath;
1004 		char *subdirpath;
1005 		struct got_pathlist_head subdirlist;
1006 
1007 		TAILQ_INIT(&subdirlist);
1008 
1009 		if (asprintf(&subpath, "%s%s%s", path,
1010 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1011 			return got_error_from_errno("asprintf");
1012 
1013 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
1014 			free(subpath);
1015 			return got_error_from_errno("asprintf");
1016 		}
1017 
1018 		subdirfd = openat(fd, de->d_name,
1019 		    O_RDONLY | O_NOFOLLOW | O_DIRECTORY);
1020 		if (subdirfd == -1) {
1021 			if (errno == EACCES) {
1022 				*next = TAILQ_NEXT(dle, entry);
1023 				return NULL;
1024 			}
1025 			err = got_error_from_errno2("openat", subdirpath);
1026 			free(subpath);
1027 			free(subdirpath);
1028 			return err;
1029 		}
1030 
1031 		subdir = fdopendir(subdirfd);
1032 		if (subdir == NULL)
1033 			return got_error_from_errno2("fdopendir", path);
1034 		subdirfd = -1;
1035 		err = read_dirlist(&subdirlist, subdir, subdirpath);
1036 		if (err) {
1037 			free(subpath);
1038 			free(subdirpath);
1039 			closedir(subdir);
1040 			return err;
1041 		}
1042 		err = diff_fileindex_dir(fileindex, ie, &subdirlist,
1043 		    dirfd(subdir), rootpath, subpath, repo, cb, cb_arg);
1044 		if (subdir && closedir(subdir) == -1 && err == NULL)
1045 			err = got_error_from_errno2("closedir", subdirpath);
1046 		free(subpath);
1047 		free(subdirpath);
1048 		free_dirlist(&subdirlist);
1049 		if (err)
1050 			return err;
1051 	}
1052 
1053 	*next = TAILQ_NEXT(dle, entry);
1054 	return NULL;
1055 }
1056 
1057 static const struct got_error *
1058 diff_fileindex_dir(struct got_fileindex *fileindex,
1059     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
1060     int dirfd, const char *rootpath, const char *path,
1061     struct got_repository *repo,
1062     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1063 {
1064 	const struct got_error *err = NULL;
1065 	struct dirent *de = NULL;
1066 	size_t path_len = strlen(path);
1067 	struct got_pathlist_entry *dle;
1068 
1069 	if (cb->diff_traverse) {
1070 		err = cb->diff_traverse(cb_arg, path, dirfd);
1071 		if (err)
1072 			return err;
1073 	}
1074 
1075 	dle = TAILQ_FIRST(dirlist);
1076 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
1077 		if (dle && *ie) {
1078 			char *de_path;
1079 			int cmp;
1080 			de = dle->data;
1081 			if (asprintf(&de_path, "%s/%s", path,
1082 			    de->d_name) == -1) {
1083 				err = got_error_from_errno("asprintf");
1084 				break;
1085 			}
1086 			cmp = got_path_cmp((*ie)->path, de_path,
1087 			    got_fileindex_entry_path_len(*ie),
1088 			    strlen(path) + 1 + de->d_namlen);
1089 			free(de_path);
1090 			if (cmp == 0) {
1091 				err = cb->diff_old_new(cb_arg, *ie, de, path,
1092 				    dirfd);
1093 				if (err)
1094 					break;
1095 				*ie = walk_fileindex(fileindex, *ie);
1096 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1097 				    path, rootpath, repo, cb, cb_arg);
1098 			} else if (cmp < 0 ) {
1099 				err = cb->diff_old(cb_arg, *ie, path);
1100 				if (err)
1101 					break;
1102 				*ie = walk_fileindex(fileindex, *ie);
1103 			} else {
1104 				err = cb->diff_new(cb_arg, de, path, dirfd);
1105 				if (err)
1106 					break;
1107 				err = walk_dir(&dle, fileindex, ie, dle, dirfd,
1108 				    path, rootpath, repo, cb, cb_arg);
1109 			}
1110 			if (err)
1111 				break;
1112 		} else if (*ie) {
1113 			err = cb->diff_old(cb_arg, *ie, path);
1114 			if (err)
1115 				break;
1116 			*ie = walk_fileindex(fileindex, *ie);
1117 		} else if (dle) {
1118 			de = dle->data;
1119 			err = cb->diff_new(cb_arg, de, path, dirfd);
1120 			if (err)
1121 				break;
1122 			err = walk_dir(&dle, fileindex, ie, dle, dirfd, path,
1123 			    rootpath, repo, cb, cb_arg);
1124 			if (err)
1125 				break;
1126 		}
1127 	}
1128 
1129 	return err;
1130 }
1131 
1132 const struct got_error *
1133 got_fileindex_diff_dir(struct got_fileindex *fileindex, int fd,
1134     const char *rootpath, const char *path, struct got_repository *repo,
1135     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1136 {
1137 	const struct got_error *err;
1138 	struct got_fileindex_entry *ie;
1139 	struct got_pathlist_head dirlist;
1140 	int fd2;
1141 	DIR *dir;
1142 
1143 	TAILQ_INIT(&dirlist);
1144 
1145 	/* 
1146 	 * Duplicate the file descriptor so we can call closedir() below
1147 	 * without closing the file descriptor passed in by our caller.
1148 	 */
1149 	fd2 = dup(fd);
1150 	if (fd2 == -1)
1151 		return got_error_from_errno2("dup", path);
1152 	if (lseek(fd2, 0, SEEK_SET) == -1) {
1153 		err = got_error_from_errno2("lseek", path);
1154 		close(fd2);
1155 		return err;
1156 	}
1157 	dir = fdopendir(fd2);
1158 	if (dir == NULL) {
1159 		err = got_error_from_errno2("fdopendir", path);
1160 		close(fd2);
1161 		return err;
1162 	}
1163 	err = read_dirlist(&dirlist, dir, path);
1164 	if (err) {
1165 		closedir(dir);
1166 		return err;
1167 	}
1168 
1169 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1170 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1171 		ie = walk_fileindex(fileindex, ie);
1172 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, dirfd(dir),
1173 	    rootpath, path, repo, cb, cb_arg);
1174 
1175 	if (closedir(dir) == -1 && err == NULL)
1176 		err = got_error_from_errno2("closedir", path);
1177 	free_dirlist(&dirlist);
1178 	return err;
1179 }
1180 
1181 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);