Gotweb

Blob

Date:: Fri Aug 2 13:42:42 2019 UTC
Message:: eliminate strlen() calls during fileindex rb tree searches
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 <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <sha1.h>
27 #include <endian.h>
28 #include <limits.h>
29 #include <uuid.h>
30 
31 #include "got_error.h"
32 #include "got_object.h"
33 #include "got_path.h"
34 
35 #include "got_lib_fileindex.h"
36 #include "got_lib_worktree.h"
37 
38 /* got_fileindex_entry flags */
39 #define GOT_FILEIDX_F_PATH_LEN		0x00000fff
40 #define GOT_FILEIDX_F_STAGE		0x00003000
41 #define GOT_FILEIDX_F_EXTENDED		0x00004000
42 #define GOT_FILEIDX_F_ASSUME_VALID	0x00008000
43 #define GOT_FILEIDX_F_NOT_FLUSHED	0x00010000
44 #define GOT_FILEIDX_F_NO_BLOB		0x00020000
45 #define GOT_FILEIDX_F_NO_COMMIT		0x00040000
46 #define GOT_FILEIDX_F_NO_FILE_ON_DISK	0x00080000
47 
48 struct got_fileindex {
49 	struct got_fileindex_tree entries;
50 	int nentries;
51 #define GOT_FILEIDX_MAX_ENTRIES INT_MAX
52 };
53 
54 uint16_t
55 got_fileindex_perms_from_st(struct stat *sb)
56 {
57 	uint16_t perms = (sb->st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
58 	return (perms << GOT_FILEIDX_MODE_PERMS_SHIFT);
59 }
60 
61 mode_t
62 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
63 {
64 	mode_t perms = (ie->mode >> GOT_FILEIDX_MODE_PERMS_SHIFT);
65 	return (perms & (S_IRWXU | S_IRWXG | S_IRWXO));
66 }
67 
68 const struct got_error *
69 got_fileindex_entry_update(struct got_fileindex_entry *entry,
70     const char *ondisk_path, uint8_t *blob_sha1, uint8_t *commit_sha1,
71     int update_timestamps)
72 {
73 	struct stat sb;
74 
75 	if (lstat(ondisk_path, &sb) != 0) {
76 		if ((entry->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0)
77 			return got_error_from_errno2("lstat", ondisk_path);
78 	} else {
79 		if (sb.st_mode & S_IFDIR)
80 			return got_error_set_errno(EISDIR, ondisk_path);
81 		entry->flags &= ~GOT_FILEIDX_F_NO_FILE_ON_DISK;
82 	}
83 
84 
85 	if ((entry->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0) {
86 		if (update_timestamps) {
87 			entry->ctime_sec = sb.st_ctime;
88 			entry->ctime_nsec = sb.st_ctimensec;
89 			entry->mtime_sec = sb.st_mtime;
90 			entry->mtime_nsec = sb.st_mtimensec;
91 		}
92 		entry->uid = sb.st_uid;
93 		entry->gid = sb.st_gid;
94 		entry->size = (sb.st_size & 0xffffffff);
95 		if (sb.st_mode & S_IFLNK)
96 			entry->mode = GOT_FILEIDX_MODE_SYMLINK;
97 		else
98 			entry->mode = GOT_FILEIDX_MODE_REGULAR_FILE;
99 		entry->mode |= got_fileindex_perms_from_st(&sb);
100 	}
101 
102 	if (blob_sha1) {
103 		memcpy(entry->blob_sha1, blob_sha1, SHA1_DIGEST_LENGTH);
104 		entry->flags &= ~GOT_FILEIDX_F_NO_BLOB;
105 	} else
106 		entry->flags |= GOT_FILEIDX_F_NO_BLOB;
107 
108 	if (commit_sha1) {
109 		memcpy(entry->commit_sha1, commit_sha1, SHA1_DIGEST_LENGTH);
110 		entry->flags &= ~GOT_FILEIDX_F_NO_COMMIT;
111 	} else
112 		entry->flags |= GOT_FILEIDX_F_NO_COMMIT;
113 
114 	return NULL;
115 }
116 
117 void
118 got_fileindex_entry_mark_deleted_from_disk(struct got_fileindex_entry *entry)
119 {
120 	entry->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
121 }
122 
123 const struct got_error *
124 got_fileindex_entry_alloc(struct got_fileindex_entry **entry,
125     const char *ondisk_path, const char *relpath, uint8_t *blob_sha1,
126     uint8_t *commit_sha1)
127 {
128 	size_t len;
129 
130 	*entry = calloc(1, sizeof(**entry));
131 	if (*entry == NULL)
132 		return got_error_from_errno("calloc");
133 
134 	(*entry)->path = strdup(relpath);
135 	if ((*entry)->path == NULL) {
136 		const struct got_error *err = got_error_from_errno("strdup");
137 		free(*entry);
138 		*entry = NULL;
139 		return err;
140 	}
141 
142 	len = strlen(relpath);
143 	if (len > GOT_FILEIDX_F_PATH_LEN)
144 		len = GOT_FILEIDX_F_PATH_LEN;
145 	(*entry)->flags |= len;
146 
147 	return got_fileindex_entry_update(*entry, ondisk_path, blob_sha1,
148 	    commit_sha1, 1);
149 }
150 
151 void
152 got_fileindex_entry_free(struct got_fileindex_entry *entry)
153 {
154 	free(entry->path);
155 	free(entry);
156 }
157 
158 int
159 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
160 {
161 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
162 }
163 
164 int
165 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
166 {
167 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
168 }
169 
170 int
171 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
172 {
173 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
174 }
175 
176 static const struct got_error *
177 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *entry)
178 {
179 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
180 		return got_error(GOT_ERR_NO_SPACE);
181 
182 	RB_INSERT(got_fileindex_tree, &fileindex->entries, entry);
183 	fileindex->nentries++;
184 	return NULL;
185 }
186 
187 const struct got_error *
188 got_fileindex_entry_add(struct got_fileindex *fileindex,
189     struct got_fileindex_entry *entry)
190 {
191 	/* Flag this entry until it gets written out to disk. */
192 	entry->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
193 
194 	return add_entry(fileindex, entry);
195 }
196 
197 void
198 got_fileindex_entry_remove(struct got_fileindex *fileindex,
199     struct got_fileindex_entry *entry)
200 {
201 	RB_REMOVE(got_fileindex_tree, &fileindex->entries, entry);
202 	fileindex->nentries--;
203 }
204 
205 struct got_fileindex_entry *
206 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path)
207 {
208 	struct got_fileindex_entry key;
209 	memset(&key, 0, sizeof(key));
210 	key.path = (char *)path;
211 	return RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
212 }
213 
214 const struct got_error *
215 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
216     got_fileindex_cb cb, void *cb_arg)
217 {
218 	const struct got_error *err;
219 	struct got_fileindex_entry *entry, *tmp;
220 
221 	RB_FOREACH_SAFE(entry, got_fileindex_tree, &fileindex->entries, tmp) {
222 		err = (*cb)(cb_arg, entry);
223 		if (err)
224 			return err;
225 	}
226 	return NULL;
227 }
228 
229 struct got_fileindex *
230 got_fileindex_alloc(void)
231 {
232 	struct got_fileindex *fileindex;
233 
234 	fileindex = calloc(1, sizeof(*fileindex));
235 	if (fileindex == NULL)
236 		return NULL;
237 
238 	RB_INIT(&fileindex->entries);
239 	return fileindex;
240 }
241 
242 void
243 got_fileindex_free(struct got_fileindex *fileindex)
244 {
245 	struct got_fileindex_entry *entry;
246 
247 	while ((entry = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
248 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, entry);
249 		got_fileindex_entry_free(entry);
250 	}
251 	free(fileindex);
252 }
253 
254 static const struct got_error *
255 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
256 {
257 	size_t n;
258 
259 	val = htobe64(val);
260 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
261 	n = fwrite(&val, 1, sizeof(val), outfile);
262 	if (n != sizeof(val))
263 		return got_ferror(outfile, GOT_ERR_IO);
264 	return NULL;
265 }
266 
267 static const struct got_error *
268 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
269 {
270 	size_t n;
271 
272 	val = htobe32(val);
273 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
274 	n = fwrite(&val, 1, sizeof(val), outfile);
275 	if (n != sizeof(val))
276 		return got_ferror(outfile, GOT_ERR_IO);
277 	return NULL;
278 }
279 
280 static const struct got_error *
281 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
282 {
283 	size_t n;
284 
285 	val = htobe16(val);
286 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
287 	n = fwrite(&val, 1, sizeof(val), outfile);
288 	if (n != sizeof(val))
289 		return got_ferror(outfile, GOT_ERR_IO);
290 	return NULL;
291 }
292 
293 static const struct got_error *
294 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
295 {
296 	size_t n, len, pad = 0;
297 	static const uint8_t zero[8] = { 0 };
298 
299 	len = strlen(path);
300 	while ((len + pad) % 8 != 0)
301 		pad++;
302 	if (pad == 0)
303 		pad = 8; /* NUL-terminate */
304 
305 	SHA1Update(ctx, path, len);
306 	n = fwrite(path, 1, len, outfile);
307 	if (n != len)
308 		return got_ferror(outfile, GOT_ERR_IO);
309 	SHA1Update(ctx, zero, pad);
310 	n = fwrite(zero, 1, pad, outfile);
311 	if (n != pad)
312 		return got_ferror(outfile, GOT_ERR_IO);
313 	return NULL;
314 }
315 
316 static const struct got_error *
317 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *entry,
318     FILE *outfile)
319 {
320 	const struct got_error *err;
321 	size_t n;
322 
323 	err = write_fileindex_val64(ctx, entry->ctime_sec, outfile);
324 	if (err)
325 		return err;
326 	err = write_fileindex_val64(ctx, entry->ctime_nsec, outfile);
327 	if (err)
328 		return err;
329 	err = write_fileindex_val64(ctx, entry->mtime_sec, outfile);
330 	if (err)
331 		return err;
332 	err = write_fileindex_val64(ctx, entry->mtime_nsec, outfile);
333 	if (err)
334 		return err;
335 
336 	err = write_fileindex_val32(ctx, entry->uid, outfile);
337 	if (err)
338 		return err;
339 	err = write_fileindex_val32(ctx, entry->gid, outfile);
340 	if (err)
341 		return err;
342 	err = write_fileindex_val32(ctx, entry->size, outfile);
343 	if (err)
344 		return err;
345 
346 	err = write_fileindex_val16(ctx, entry->mode, outfile);
347 	if (err)
348 		return err;
349 
350 	SHA1Update(ctx, entry->blob_sha1, SHA1_DIGEST_LENGTH);
351 	n = fwrite(entry->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
352 	if (n != SHA1_DIGEST_LENGTH)
353 		return got_ferror(outfile, GOT_ERR_IO);
354 
355 	SHA1Update(ctx, entry->commit_sha1, SHA1_DIGEST_LENGTH);
356 	n = fwrite(entry->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
357 	if (n != SHA1_DIGEST_LENGTH)
358 		return got_ferror(outfile, GOT_ERR_IO);
359 
360 	err = write_fileindex_val32(ctx, entry->flags, outfile);
361 	if (err)
362 		return err;
363 
364 	err = write_fileindex_path(ctx, entry->path, outfile);
365 	return err;
366 }
367 
368 const struct got_error *
369 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
370 {
371 	const struct got_error *err = NULL;
372 	struct got_fileindex_hdr hdr;
373 	SHA1_CTX ctx;
374 	uint8_t sha1[SHA1_DIGEST_LENGTH];
375 	size_t n;
376 	struct got_fileindex_entry *entry;
377 
378 	SHA1Init(&ctx);
379 
380 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
381 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
382 	hdr.nentries = htobe32(fileindex->nentries);
383 
384 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
385 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
386 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
387 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
388 	if (n != sizeof(hdr.signature))
389 		return got_ferror(outfile, GOT_ERR_IO);
390 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
391 	if (n != sizeof(hdr.version))
392 		return got_ferror(outfile, GOT_ERR_IO);
393 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
394 	if (n != sizeof(hdr.nentries))
395 		return got_ferror(outfile, GOT_ERR_IO);
396 
397 	RB_FOREACH(entry, got_fileindex_tree, &fileindex->entries) {
398 		entry->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
399 		err = write_fileindex_entry(&ctx, entry, outfile);
400 		if (err)
401 			return err;
402 	}
403 
404 	SHA1Final(sha1, &ctx);
405 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
406 	if (n != sizeof(sha1))
407 		return got_ferror(outfile, GOT_ERR_IO);
408 
409 	if (fflush(outfile) != 0)
410 		return got_error_from_errno("fflush");
411 
412 	return NULL;
413 }
414 
415 static const struct got_error *
416 read_fileindex_val64(uint64_t *val, SHA1_CTX *ctx, FILE *infile)
417 {
418 	size_t n;
419 
420 	n = fread(val, 1, sizeof(*val), infile);
421 	if (n != sizeof(*val))
422 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
423 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
424 	*val = be64toh(*val);
425 	return NULL;
426 }
427 
428 static const struct got_error *
429 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
430 {
431 	size_t n;
432 
433 	n = fread(val, 1, sizeof(*val), infile);
434 	if (n != sizeof(*val))
435 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
436 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
437 	*val = be32toh(*val);
438 	return NULL;
439 }
440 
441 static const struct got_error *
442 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
443 {
444 	size_t n;
445 
446 	n = fread(val, 1, sizeof(*val), infile);
447 	if (n != sizeof(*val))
448 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
449 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
450 	*val = be16toh(*val);
451 	return NULL;
452 }
453 
454 static const struct got_error *
455 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
456 {
457 	const struct got_error *err = NULL;
458 	const size_t chunk_size = 8;
459 	size_t n, len = 0, totlen = chunk_size;
460 
461 	*path = malloc(totlen);
462 	if (*path == NULL)
463 		return got_error_from_errno("malloc");
464 
465 	do {
466 		if (len + chunk_size > totlen) {
467 			char *p = reallocarray(*path, totlen + chunk_size, 1);
468 			if (p == NULL) {
469 				err = got_error_from_errno("reallocarray");
470 				break;
471 			}
472 			totlen += chunk_size;
473 			*path = p;
474 		}
475 		n = fread(*path + len, 1, chunk_size, infile);
476 		if (n != chunk_size) {
477 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
478 			break;
479 		}
480 		SHA1Update(ctx, *path + len, chunk_size);
481 		len += chunk_size;
482 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
483 
484 	if (err) {
485 		free(*path);
486 		*path = NULL;
487 	}
488 	return err;
489 }
490 
491 static const struct got_error *
492 read_fileindex_entry(struct got_fileindex_entry **entryp, SHA1_CTX *ctx,
493     FILE *infile)
494 {
495 	const struct got_error *err;
496 	struct got_fileindex_entry *entry;
497 	size_t n;
498 
499 	*entryp = NULL;
500 
501 	entry = calloc(1, sizeof(*entry));
502 	if (entry == NULL)
503 		return got_error_from_errno("calloc");
504 
505 	err = read_fileindex_val64(&entry->ctime_sec, ctx, infile);
506 	if (err)
507 		goto done;
508 	err = read_fileindex_val64(&entry->ctime_nsec, ctx, infile);
509 	if (err)
510 		goto done;
511 	err = read_fileindex_val64(&entry->mtime_sec, ctx, infile);
512 	if (err)
513 		goto done;
514 	err = read_fileindex_val64(&entry->mtime_nsec, ctx, infile);
515 	if (err)
516 		goto done;
517 
518 	err = read_fileindex_val32(&entry->uid, ctx, infile);
519 	if (err)
520 		goto done;
521 	err = read_fileindex_val32(&entry->gid, ctx, infile);
522 	if (err)
523 		goto done;
524 	err = read_fileindex_val32(&entry->size, ctx, infile);
525 	if (err)
526 		goto done;
527 
528 	err = read_fileindex_val16(&entry->mode, ctx, infile);
529 	if (err)
530 		goto done;
531 
532 	n = fread(entry->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
533 	if (n != SHA1_DIGEST_LENGTH) {
534 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
535 		goto done;
536 	}
537 	SHA1Update(ctx, entry->blob_sha1, SHA1_DIGEST_LENGTH);
538 
539 	n = fread(entry->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
540 	if (n != SHA1_DIGEST_LENGTH) {
541 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
542 		goto done;
543 	}
544 	SHA1Update(ctx, entry->commit_sha1, SHA1_DIGEST_LENGTH);
545 
546 	err = read_fileindex_val32(&entry->flags, ctx, infile);
547 	if (err)
548 		goto done;
549 
550 	err = read_fileindex_path(&entry->path, ctx, infile);
551 	if (err == NULL)
552 		entry->path_len = strlen(entry->path);
553 done:
554 	if (err)
555 		got_fileindex_entry_free(entry);
556 	else
557 		*entryp = entry;
558 	return err;
559 }
560 
561 const struct got_error *
562 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
563 {
564 	const struct got_error *err = NULL;
565 	struct got_fileindex_hdr hdr;
566 	SHA1_CTX ctx;
567 	struct got_fileindex_entry *entry;
568 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
569 	uint8_t sha1[SHA1_DIGEST_LENGTH];
570 	size_t n;
571 	int i;
572 
573 	SHA1Init(&ctx);
574 
575 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
576 	if (n != sizeof(hdr.signature)) {
577 		if (n == 0) /* EOF */
578 			return NULL;
579 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
580 	}
581 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
582 	if (n != sizeof(hdr.version)) {
583 		if (n == 0) /* EOF */
584 			return NULL;
585 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
586 	}
587 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
588 	if (n != sizeof(hdr.nentries)) {
589 		if (n == 0) /* EOF */
590 			return NULL;
591 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
592 	}
593 
594 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
595 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
596 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
597 
598 	hdr.signature = be32toh(hdr.signature);
599 	hdr.version = be32toh(hdr.version);
600 	hdr.nentries = be32toh(hdr.nentries);
601 
602 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
603 		return got_error(GOT_ERR_FILEIDX_SIG);
604 	if (hdr.version != GOT_FILE_INDEX_VERSION)
605 		return got_error(GOT_ERR_FILEIDX_VER);
606 
607 	for (i = 0; i < hdr.nentries; i++) {
608 		err = read_fileindex_entry(&entry, &ctx, infile);
609 		if (err)
610 			return err;
611 		err = add_entry(fileindex, entry);
612 		if (err)
613 			return err;
614 	}
615 
616 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
617 	if (n != sizeof(sha1_expected))
618 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
619 	SHA1Final(sha1, &ctx);
620 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
621 		return got_error(GOT_ERR_FILEIDX_CSUM);
622 
623 	return NULL;
624 }
625 
626 static struct got_fileindex_entry *
627 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
628 {
629 	struct got_fileindex_entry *next;
630 
631 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
632 
633 	/* Skip entries which were newly added by diff callbacks. */
634 	while (next && (next->flags & GOT_FILEIDX_F_NOT_FLUSHED))
635 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
636 
637 	return next;
638 }
639 
640 static const struct got_error *
641 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **,
642     const struct got_tree_entries *, const char *, const char *,
643     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
644 
645 static const struct got_error *
646 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
647     struct got_fileindex_entry **ie, struct got_tree_entry *te,
648     const char *path, const char *entry_name, struct got_repository *repo,
649     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
650 {
651 	const struct got_error *err = NULL;
652 
653 	if (S_ISDIR(te->mode)) {
654 		char *subpath;
655 		struct got_tree_object *subtree;
656 
657 		if (asprintf(&subpath, "%s%s%s", path,
658 		    path[0] == '\0' ? "" : "/", te->name) == -1)
659 			return got_error_from_errno("asprintf");
660 
661 		err = got_object_open_as_tree(&subtree, repo, te->id);
662 		if (err) {
663 			free(subpath);
664 			return err;
665 		}
666 
667 		err = diff_fileindex_tree(fileindex, ie,
668 		    got_object_tree_get_entries(subtree), subpath, entry_name,
669 		    repo, cb, cb_arg);
670 		free(subpath);
671 		got_object_tree_close(subtree);
672 		if (err)
673 			return err;
674 	}
675 
676 	*next = SIMPLEQ_NEXT(te, entry);
677 	return NULL;
678 }
679 
680 static const struct got_error *
681 diff_fileindex_tree(struct got_fileindex *fileindex,
682     struct got_fileindex_entry **ie, const struct got_tree_entries *entries,
683     const char *path, const char *entry_name, struct got_repository *repo,
684     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
685 {
686 	const struct got_error *err = NULL;
687 	struct got_tree_entry *te = NULL;
688 	size_t path_len = strlen(path);
689 	struct got_fileindex_entry *next;
690 
691 	te = SIMPLEQ_FIRST(&entries->head);
692 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
693 		if (te && *ie) {
694 			char *te_path;
695 			int cmp;
696 			if (asprintf(&te_path, "%s/%s", path, te->name) == -1) {
697 				err = got_error_from_errno("asprintf");
698 				break;
699 			}
700 			cmp = got_path_cmp((*ie)->path, te_path,
701 			    (*ie)->path_len, strlen(te_path));
702 			free(te_path);
703 			if (cmp == 0) {
704 				if (got_path_is_child((*ie)->path, path,
705 				    path_len) && (entry_name == NULL ||
706 				    strcmp(te->name, entry_name) == 0)) {
707 					err = cb->diff_old_new(cb_arg, *ie, te,
708 					    path);
709 					if (err || entry_name)
710 						break;
711 				}
712 				*ie = walk_fileindex(fileindex, *ie);
713 				err = walk_tree(&te, fileindex, ie, te,
714 				    path, entry_name, repo, cb, cb_arg);
715 			} else if (cmp < 0) {
716 				next = walk_fileindex(fileindex, *ie);
717 				if (got_path_is_child((*ie)->path, path,
718 				    path_len) && (entry_name == NULL ||
719 				    strcmp(te->name, entry_name) == 0)) {
720 					err = cb->diff_old(cb_arg, *ie, path);
721 					if (err || entry_name)
722 						break;
723 				}
724 				*ie = next;
725 			} else {
726 				if ((entry_name == NULL ||
727 				    strcmp(te->name, entry_name) == 0)) {
728 					err = cb->diff_new(cb_arg, te, path);
729 					if (err || entry_name)
730 						break;
731 				}
732 				err = walk_tree(&te, fileindex, ie, te,
733 				    path, entry_name, repo, cb, cb_arg);
734 			}
735 			if (err)
736 				break;
737 		} else if (*ie) {
738 			next = walk_fileindex(fileindex, *ie);
739 			if (got_path_is_child((*ie)->path, path, path_len) &&
740 			    (entry_name == NULL ||
741 			    strcmp(te->name, entry_name) == 0)) {
742 				err = cb->diff_old(cb_arg, *ie, path);
743 				if (err || entry_name)
744 					break;
745 			}
746 			*ie = next;
747 		} else if (te) {
748 			if (entry_name == NULL ||
749 			    strcmp(te->name, entry_name) == 0) {
750 				err = cb->diff_new(cb_arg, te, path);
751 				if (err || entry_name)
752 					break;
753 			}
754 			err = walk_tree(&te, fileindex, ie, te, path,
755 			    entry_name, repo, cb, cb_arg);
756 			if (err)
757 				break;
758 		}
759 	}
760 
761 	return err;
762 }
763 
764 const struct got_error *
765 got_fileindex_diff_tree(struct got_fileindex *fileindex,
766     struct got_tree_object *tree, const char *path, const char *entry_name,
767     struct got_repository *repo,
768     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
769 {
770 	struct got_fileindex_entry *ie;
771 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
772 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
773 		ie = walk_fileindex(fileindex, ie);
774 	return diff_fileindex_tree(fileindex, &ie,
775 	    got_object_tree_get_entries(tree), path, entry_name, repo,
776 	    cb, cb_arg);
777 }
778 
779 static const struct got_error *
780 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
781     struct got_pathlist_head *, const char *, const char *,
782     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
783 
784 static const struct got_error *
785 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
786 {
787 	const struct got_error *err = NULL;
788 	struct got_pathlist_entry *new = NULL;
789 	struct dirent *dep = NULL;
790 	struct dirent *de = NULL;
791 
792 	for (;;) {
793 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
794 		if (de == NULL) {
795 			err = got_error_from_errno("malloc");
796 			break;
797 		}
798 
799 		if (readdir_r(dir, de, &dep) != 0) {
800 			err = got_error_from_errno("readdir_r");
801 			free(de);
802 			break;
803 		}
804 		if (dep == NULL) {
805 			free(de);
806 			break;
807 		}
808 
809 		if (strcmp(de->d_name, ".") == 0 ||
810 		    strcmp(de->d_name, "..") == 0 ||
811 		    (path[0] == '\0' &&
812 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
813 			free(de);
814 			continue;
815 		}
816 
817 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
818 		if (err) {
819 			free(de);
820 			break;
821 		}
822 		if (new == NULL) {
823 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
824 			free(de);
825 			break;
826 		}
827 	}
828 
829 	return err;
830 }
831 
832 void
833 free_dirlist(struct got_pathlist_head *dirlist)
834 {
835 	struct got_pathlist_entry *dle;
836 
837 	TAILQ_FOREACH(dle, dirlist, entry)
838 		free(dle->data);
839 	got_pathlist_free(dirlist);
840 }
841 
842 static const struct got_error *
843 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
844     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle,
845     const char *path, const char *rootpath, struct got_repository *repo,
846     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
847 {
848 	const struct got_error *err = NULL;
849 	struct dirent *de = dle->data;
850 
851 	*next = NULL;
852 
853 	if (de->d_type == DT_DIR) {
854 		char *subpath;
855 		char *subdirpath;
856 		DIR *subdir;
857 		struct got_pathlist_head subdirlist;
858 
859 		TAILQ_INIT(&subdirlist);
860 
861 		if (asprintf(&subpath, "%s%s%s", path,
862 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
863 			return got_error_from_errno("asprintf");
864 
865 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
866 			free(subpath);
867 			return got_error_from_errno("asprintf");
868 		}
869 
870 		subdir = opendir(subdirpath);
871 		if (subdir == NULL) {
872 			free(subpath);
873 			free(subdirpath);
874 			return got_error_from_errno2("opendir", subdirpath);
875 		}
876 
877 		err = read_dirlist(&subdirlist, subdir, subdirpath);
878 		if (err) {
879 			free(subpath);
880 			free(subdirpath);
881 			closedir(subdir);
882 			return err;
883 		}
884 		err = diff_fileindex_dir(fileindex, ie, &subdirlist, rootpath,
885 		    subpath, repo, cb, cb_arg);
886 		free(subpath);
887 		free(subdirpath);
888 		closedir(subdir);
889 		free_dirlist(&subdirlist);
890 		if (err)
891 			return err;
892 	}
893 
894 	*next = TAILQ_NEXT(dle, entry);
895 	return NULL;
896 }
897 
898 static const struct got_error *
899 diff_fileindex_dir(struct got_fileindex *fileindex,
900     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
901     const char *rootpath, const char *path, struct got_repository *repo,
902     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
903 {
904 	const struct got_error *err = NULL;
905 	struct dirent *de = NULL;
906 	size_t path_len = strlen(path);
907 	struct got_pathlist_entry *dle;
908 
909 	dle = TAILQ_FIRST(dirlist);
910 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
911 		if (dle && *ie) {
912 			char *de_path;
913 			int cmp;
914 			de = dle->data;
915 			if (asprintf(&de_path, "%s/%s", path,
916 			    de->d_name) == -1) {
917 				err = got_error_from_errno("asprintf");
918 				break;
919 			}
920 			cmp = got_path_cmp((*ie)->path, de_path,
921 			    (*ie)->path_len, strlen(path) + 1 + de->d_namlen);
922 			free(de_path);
923 			if (cmp == 0) {
924 				err = cb->diff_old_new(cb_arg, *ie, de, path);
925 				if (err)
926 					break;
927 				*ie = walk_fileindex(fileindex, *ie);
928 				err = walk_dir(&dle, fileindex, ie, dle, path,
929 				    rootpath, repo, cb, cb_arg);
930 			} else if (cmp < 0 ) {
931 				err = cb->diff_old(cb_arg, *ie, path);
932 				if (err)
933 					break;
934 				*ie = walk_fileindex(fileindex, *ie);
935 			} else {
936 				err = cb->diff_new(cb_arg, de, path);
937 				if (err)
938 					break;
939 				err = walk_dir(&dle, fileindex, ie, dle, path,
940 				    rootpath, repo, cb, cb_arg);
941 			}
942 			if (err)
943 				break;
944 		} else if (*ie) {
945 			err = cb->diff_old(cb_arg, *ie, path);
946 			if (err)
947 				break;
948 			*ie = walk_fileindex(fileindex, *ie);
949 		} else if (dle) {
950 			de = dle->data;
951 			err = cb->diff_new(cb_arg, de, path);
952 			if (err)
953 				break;
954 			err = walk_dir(&dle, fileindex, ie, dle, path,
955 			    rootpath, repo, cb, cb_arg);
956 			if (err)
957 				break;
958 		}
959 	}
960 
961 	return err;
962 }
963 
964 const struct got_error *
965 got_fileindex_diff_dir(struct got_fileindex *fileindex, DIR *rootdir,
966     const char *rootpath, const char *path, struct got_repository *repo,
967     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
968 {
969 	const struct got_error *err;
970 	struct got_fileindex_entry *ie;
971 	struct got_pathlist_head dirlist;
972 
973 	TAILQ_INIT(&dirlist);
974 	err = read_dirlist(&dirlist, rootdir, path);
975 	if (err)
976 		return err;
977 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
978 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
979 		ie = walk_fileindex(fileindex, ie);
980 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, rootpath, path,
981 	    repo, cb, cb_arg);
982 	free_dirlist(&dirlist);
983 	return err;
984 }
985 
986 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);