op public repos

Blob

Date:: Sun Sep 22 11:28:06 2019 UTC
Message:: avoid a NULL deref in diff_fileindex_tree(); found by jj@ with scan-build
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		0x0000f000
41 #define GOT_FILEIDX_F_STAGE_SHIFT	12
42 #define GOT_FILEIDX_F_NOT_FLUSHED	0x00010000
43 #define GOT_FILEIDX_F_NO_BLOB		0x00020000
44 #define GOT_FILEIDX_F_NO_COMMIT		0x00040000
45 #define GOT_FILEIDX_F_NO_FILE_ON_DISK	0x00080000
46 
47 struct got_fileindex {
48 	struct got_fileindex_tree entries;
49 	int nentries;
50 #define GOT_FILEIDX_MAX_ENTRIES INT_MAX
51 };
52 
53 uint16_t
54 got_fileindex_perms_from_st(struct stat *sb)
55 {
56 	uint16_t perms = (sb->st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
57 	return (perms << GOT_FILEIDX_MODE_PERMS_SHIFT);
58 }
59 
60 mode_t
61 got_fileindex_perms_to_st(struct got_fileindex_entry *ie)
62 {
63 	mode_t perms = (ie->mode >> GOT_FILEIDX_MODE_PERMS_SHIFT);
64 	return (S_IFREG | (perms & (S_IRWXU | S_IRWXG | S_IRWXO)));
65 }
66 
67 const struct got_error *
68 got_fileindex_entry_update(struct got_fileindex_entry *ie,
69     const char *ondisk_path, uint8_t *blob_sha1, uint8_t *commit_sha1,
70     int update_timestamps)
71 {
72 	struct stat sb;
73 
74 	if (lstat(ondisk_path, &sb) != 0) {
75 		if (!((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) &&
76 		    errno == ENOENT))
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 		ie->flags &= ~GOT_FILEIDX_F_NO_FILE_ON_DISK;
82 	}
83 
84 
85 	if ((ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0) {
86 		if (update_timestamps) {
87 			ie->ctime_sec = sb.st_ctime;
88 			ie->ctime_nsec = sb.st_ctimensec;
89 			ie->mtime_sec = sb.st_mtime;
90 			ie->mtime_nsec = sb.st_mtimensec;
91 		}
92 		ie->uid = sb.st_uid;
93 		ie->gid = sb.st_gid;
94 		ie->size = (sb.st_size & 0xffffffff);
95 		if (sb.st_mode & S_IFLNK)
96 			ie->mode = GOT_FILEIDX_MODE_SYMLINK;
97 		else
98 			ie->mode = GOT_FILEIDX_MODE_REGULAR_FILE;
99 		ie->mode |= got_fileindex_perms_from_st(&sb);
100 	}
101 
102 	if (blob_sha1) {
103 		memcpy(ie->blob_sha1, blob_sha1, SHA1_DIGEST_LENGTH);
104 		ie->flags &= ~GOT_FILEIDX_F_NO_BLOB;
105 	} else
106 		ie->flags |= GOT_FILEIDX_F_NO_BLOB;
107 
108 	if (commit_sha1) {
109 		memcpy(ie->commit_sha1, commit_sha1, SHA1_DIGEST_LENGTH);
110 		ie->flags &= ~GOT_FILEIDX_F_NO_COMMIT;
111 	} else
112 		ie->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 *ie)
119 {
120 	ie->flags |= GOT_FILEIDX_F_NO_FILE_ON_DISK;
121 }
122 
123 const struct got_error *
124 got_fileindex_entry_alloc(struct got_fileindex_entry **ie,
125     const char *ondisk_path, const char *relpath, uint8_t *blob_sha1,
126     uint8_t *commit_sha1)
127 {
128 	size_t len;
129 
130 	*ie = calloc(1, sizeof(**ie));
131 	if (*ie == NULL)
132 		return got_error_from_errno("calloc");
133 
134 	(*ie)->path = strdup(relpath);
135 	if ((*ie)->path == NULL) {
136 		const struct got_error *err = got_error_from_errno("strdup");
137 		free(*ie);
138 		*ie = 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 	(*ie)->flags |= len;
146 
147 	return got_fileindex_entry_update(*ie, ondisk_path, blob_sha1,
148 	    commit_sha1, 1);
149 }
150 
151 void
152 got_fileindex_entry_free(struct got_fileindex_entry *ie)
153 {
154 	free(ie->path);
155 	free(ie);
156 }
157 
158 size_t
159 got_fileindex_entry_path_len(const struct got_fileindex_entry *ie)
160 {
161 	return (size_t)(ie->flags & GOT_FILEIDX_F_PATH_LEN);
162 }
163 
164 uint32_t
165 got_fileindex_entry_stage_get(const struct got_fileindex_entry *ie)
166 {
167 	return ((ie->flags & GOT_FILEIDX_F_STAGE) >> GOT_FILEIDX_F_STAGE_SHIFT);
168 }
169 
170 void
171 got_fileindex_entry_stage_set(struct got_fileindex_entry *ie, uint32_t stage)
172 {
173 	ie->flags &= ~GOT_FILEIDX_F_STAGE;
174 	ie->flags |= ((stage << GOT_FILEIDX_F_STAGE_SHIFT) &
175 	    GOT_FILEIDX_F_STAGE);
176 }
177 
178 int
179 got_fileindex_entry_has_blob(struct got_fileindex_entry *ie)
180 {
181 	return (ie->flags & GOT_FILEIDX_F_NO_BLOB) == 0;
182 }
183 
184 int
185 got_fileindex_entry_has_commit(struct got_fileindex_entry *ie)
186 {
187 	return (ie->flags & GOT_FILEIDX_F_NO_COMMIT) == 0;
188 }
189 
190 int
191 got_fileindex_entry_has_file_on_disk(struct got_fileindex_entry *ie)
192 {
193 	return (ie->flags & GOT_FILEIDX_F_NO_FILE_ON_DISK) == 0;
194 }
195 
196 static const struct got_error *
197 add_entry(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
198 {
199 	if (fileindex->nentries >= GOT_FILEIDX_MAX_ENTRIES)
200 		return got_error(GOT_ERR_NO_SPACE);
201 
202 	RB_INSERT(got_fileindex_tree, &fileindex->entries, ie);
203 	fileindex->nentries++;
204 	return NULL;
205 }
206 
207 const struct got_error *
208 got_fileindex_entry_add(struct got_fileindex *fileindex,
209     struct got_fileindex_entry *ie)
210 {
211 	/* Flag this entry until it gets written out to disk. */
212 	ie->flags |= GOT_FILEIDX_F_NOT_FLUSHED;
213 
214 	return add_entry(fileindex, ie);
215 }
216 
217 void
218 got_fileindex_entry_remove(struct got_fileindex *fileindex,
219     struct got_fileindex_entry *ie)
220 {
221 	RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
222 	fileindex->nentries--;
223 }
224 
225 struct got_fileindex_entry *
226 got_fileindex_entry_get(struct got_fileindex *fileindex, const char *path,
227     size_t path_len)
228 {
229 	struct got_fileindex_entry key;
230 	memset(&key, 0, sizeof(key));
231 	key.path = (char *)path;
232 	key.flags = (path_len & GOT_FILEIDX_F_PATH_LEN);
233 	return RB_FIND(got_fileindex_tree, &fileindex->entries, &key);
234 }
235 
236 const struct got_error *
237 got_fileindex_for_each_entry_safe(struct got_fileindex *fileindex,
238     got_fileindex_cb cb, void *cb_arg)
239 {
240 	const struct got_error *err;
241 	struct got_fileindex_entry *ie, *tmp;
242 
243 	RB_FOREACH_SAFE(ie, got_fileindex_tree, &fileindex->entries, tmp) {
244 		err = (*cb)(cb_arg, ie);
245 		if (err)
246 			return err;
247 	}
248 	return NULL;
249 }
250 
251 struct got_fileindex *
252 got_fileindex_alloc(void)
253 {
254 	struct got_fileindex *fileindex;
255 
256 	fileindex = calloc(1, sizeof(*fileindex));
257 	if (fileindex == NULL)
258 		return NULL;
259 
260 	RB_INIT(&fileindex->entries);
261 	return fileindex;
262 }
263 
264 void
265 got_fileindex_free(struct got_fileindex *fileindex)
266 {
267 	struct got_fileindex_entry *ie;
268 
269 	while ((ie = RB_MIN(got_fileindex_tree, &fileindex->entries))) {
270 		RB_REMOVE(got_fileindex_tree, &fileindex->entries, ie);
271 		got_fileindex_entry_free(ie);
272 	}
273 	free(fileindex);
274 }
275 
276 static const struct got_error *
277 write_fileindex_val64(SHA1_CTX *ctx, uint64_t val, FILE *outfile)
278 {
279 	size_t n;
280 
281 	val = htobe64(val);
282 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
283 	n = fwrite(&val, 1, sizeof(val), outfile);
284 	if (n != sizeof(val))
285 		return got_ferror(outfile, GOT_ERR_IO);
286 	return NULL;
287 }
288 
289 static const struct got_error *
290 write_fileindex_val32(SHA1_CTX *ctx, uint32_t val, FILE *outfile)
291 {
292 	size_t n;
293 
294 	val = htobe32(val);
295 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
296 	n = fwrite(&val, 1, sizeof(val), outfile);
297 	if (n != sizeof(val))
298 		return got_ferror(outfile, GOT_ERR_IO);
299 	return NULL;
300 }
301 
302 static const struct got_error *
303 write_fileindex_val16(SHA1_CTX *ctx, uint16_t val, FILE *outfile)
304 {
305 	size_t n;
306 
307 	val = htobe16(val);
308 	SHA1Update(ctx, (uint8_t *)&val, sizeof(val));
309 	n = fwrite(&val, 1, sizeof(val), outfile);
310 	if (n != sizeof(val))
311 		return got_ferror(outfile, GOT_ERR_IO);
312 	return NULL;
313 }
314 
315 static const struct got_error *
316 write_fileindex_path(SHA1_CTX *ctx, const char *path, FILE *outfile)
317 {
318 	size_t n, len, pad = 0;
319 	static const uint8_t zero[8] = { 0 };
320 
321 	len = strlen(path);
322 	while ((len + pad) % 8 != 0)
323 		pad++;
324 	if (pad == 0)
325 		pad = 8; /* NUL-terminate */
326 
327 	SHA1Update(ctx, path, len);
328 	n = fwrite(path, 1, len, outfile);
329 	if (n != len)
330 		return got_ferror(outfile, GOT_ERR_IO);
331 	SHA1Update(ctx, zero, pad);
332 	n = fwrite(zero, 1, pad, outfile);
333 	if (n != pad)
334 		return got_ferror(outfile, GOT_ERR_IO);
335 	return NULL;
336 }
337 
338 static const struct got_error *
339 write_fileindex_entry(SHA1_CTX *ctx, struct got_fileindex_entry *ie,
340     FILE *outfile)
341 {
342 	const struct got_error *err;
343 	size_t n;
344 	uint32_t stage;
345 
346 	err = write_fileindex_val64(ctx, ie->ctime_sec, outfile);
347 	if (err)
348 		return err;
349 	err = write_fileindex_val64(ctx, ie->ctime_nsec, outfile);
350 	if (err)
351 		return err;
352 	err = write_fileindex_val64(ctx, ie->mtime_sec, outfile);
353 	if (err)
354 		return err;
355 	err = write_fileindex_val64(ctx, ie->mtime_nsec, outfile);
356 	if (err)
357 		return err;
358 
359 	err = write_fileindex_val32(ctx, ie->uid, outfile);
360 	if (err)
361 		return err;
362 	err = write_fileindex_val32(ctx, ie->gid, outfile);
363 	if (err)
364 		return err;
365 	err = write_fileindex_val32(ctx, ie->size, outfile);
366 	if (err)
367 		return err;
368 
369 	err = write_fileindex_val16(ctx, ie->mode, outfile);
370 	if (err)
371 		return err;
372 
373 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
374 	n = fwrite(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
375 	if (n != SHA1_DIGEST_LENGTH)
376 		return got_ferror(outfile, GOT_ERR_IO);
377 
378 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
379 	n = fwrite(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, outfile);
380 	if (n != SHA1_DIGEST_LENGTH)
381 		return got_ferror(outfile, GOT_ERR_IO);
382 
383 	err = write_fileindex_val32(ctx, ie->flags, outfile);
384 	if (err)
385 		return err;
386 
387 	err = write_fileindex_path(ctx, ie->path, outfile);
388 	if (err)
389 		return err;
390 
391 	stage = got_fileindex_entry_stage_get(ie);
392 	if (stage == GOT_FILEIDX_STAGE_MODIFY ||
393 	    stage == GOT_FILEIDX_STAGE_ADD) {
394 		SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
395 		n = fwrite(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
396 		    outfile);
397 		if (n != SHA1_DIGEST_LENGTH)
398 			return got_ferror(outfile, GOT_ERR_IO);
399 	}
400 
401 	return NULL;
402 }
403 
404 const struct got_error *
405 got_fileindex_write(struct got_fileindex *fileindex, FILE *outfile)
406 {
407 	const struct got_error *err = NULL;
408 	struct got_fileindex_hdr hdr;
409 	SHA1_CTX ctx;
410 	uint8_t sha1[SHA1_DIGEST_LENGTH];
411 	size_t n;
412 	struct got_fileindex_entry *ie;
413 
414 	SHA1Init(&ctx);
415 
416 	hdr.signature = htobe32(GOT_FILE_INDEX_SIGNATURE);
417 	hdr.version = htobe32(GOT_FILE_INDEX_VERSION);
418 	hdr.nentries = htobe32(fileindex->nentries);
419 
420 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
421 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
422 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
423 	n = fwrite(&hdr.signature, 1, sizeof(hdr.signature), outfile);
424 	if (n != sizeof(hdr.signature))
425 		return got_ferror(outfile, GOT_ERR_IO);
426 	n = fwrite(&hdr.version, 1, sizeof(hdr.version), outfile);
427 	if (n != sizeof(hdr.version))
428 		return got_ferror(outfile, GOT_ERR_IO);
429 	n = fwrite(&hdr.nentries, 1, sizeof(hdr.nentries), outfile);
430 	if (n != sizeof(hdr.nentries))
431 		return got_ferror(outfile, GOT_ERR_IO);
432 
433 	RB_FOREACH(ie, got_fileindex_tree, &fileindex->entries) {
434 		ie->flags &= ~GOT_FILEIDX_F_NOT_FLUSHED;
435 		err = write_fileindex_entry(&ctx, ie, outfile);
436 		if (err)
437 			return err;
438 	}
439 
440 	SHA1Final(sha1, &ctx);
441 	n = fwrite(sha1, 1, sizeof(sha1), outfile);
442 	if (n != sizeof(sha1))
443 		return got_ferror(outfile, GOT_ERR_IO);
444 
445 	if (fflush(outfile) != 0)
446 		return got_error_from_errno("fflush");
447 
448 	return NULL;
449 }
450 
451 static const struct got_error *
452 read_fileindex_val64(uint64_t *val, SHA1_CTX *ctx, FILE *infile)
453 {
454 	size_t n;
455 
456 	n = fread(val, 1, sizeof(*val), infile);
457 	if (n != sizeof(*val))
458 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
459 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
460 	*val = be64toh(*val);
461 	return NULL;
462 }
463 
464 static const struct got_error *
465 read_fileindex_val32(uint32_t *val, SHA1_CTX *ctx, FILE *infile)
466 {
467 	size_t n;
468 
469 	n = fread(val, 1, sizeof(*val), infile);
470 	if (n != sizeof(*val))
471 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
472 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
473 	*val = be32toh(*val);
474 	return NULL;
475 }
476 
477 static const struct got_error *
478 read_fileindex_val16(uint16_t *val, SHA1_CTX *ctx, FILE *infile)
479 {
480 	size_t n;
481 
482 	n = fread(val, 1, sizeof(*val), infile);
483 	if (n != sizeof(*val))
484 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
485 	SHA1Update(ctx, (uint8_t *)val, sizeof(*val));
486 	*val = be16toh(*val);
487 	return NULL;
488 }
489 
490 static const struct got_error *
491 read_fileindex_path(char **path, SHA1_CTX *ctx, FILE *infile)
492 {
493 	const struct got_error *err = NULL;
494 	const size_t chunk_size = 8;
495 	size_t n, len = 0, totlen = chunk_size;
496 
497 	*path = malloc(totlen);
498 	if (*path == NULL)
499 		return got_error_from_errno("malloc");
500 
501 	do {
502 		if (len + chunk_size > totlen) {
503 			char *p = reallocarray(*path, totlen + chunk_size, 1);
504 			if (p == NULL) {
505 				err = got_error_from_errno("reallocarray");
506 				break;
507 			}
508 			totlen += chunk_size;
509 			*path = p;
510 		}
511 		n = fread(*path + len, 1, chunk_size, infile);
512 		if (n != chunk_size) {
513 			err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
514 			break;
515 		}
516 		SHA1Update(ctx, *path + len, chunk_size);
517 		len += chunk_size;
518 	} while (memchr(*path + len - chunk_size, '\0', chunk_size) == NULL);
519 
520 	if (err) {
521 		free(*path);
522 		*path = NULL;
523 	}
524 	return err;
525 }
526 
527 static const struct got_error *
528 read_fileindex_entry(struct got_fileindex_entry **iep, SHA1_CTX *ctx,
529     FILE *infile, uint32_t version)
530 {
531 	const struct got_error *err;
532 	struct got_fileindex_entry *ie;
533 	size_t n;
534 
535 	*iep = NULL;
536 
537 	ie = calloc(1, sizeof(*ie));
538 	if (ie == NULL)
539 		return got_error_from_errno("calloc");
540 
541 	err = read_fileindex_val64(&ie->ctime_sec, ctx, infile);
542 	if (err)
543 		goto done;
544 	err = read_fileindex_val64(&ie->ctime_nsec, ctx, infile);
545 	if (err)
546 		goto done;
547 	err = read_fileindex_val64(&ie->mtime_sec, ctx, infile);
548 	if (err)
549 		goto done;
550 	err = read_fileindex_val64(&ie->mtime_nsec, ctx, infile);
551 	if (err)
552 		goto done;
553 
554 	err = read_fileindex_val32(&ie->uid, ctx, infile);
555 	if (err)
556 		goto done;
557 	err = read_fileindex_val32(&ie->gid, ctx, infile);
558 	if (err)
559 		goto done;
560 	err = read_fileindex_val32(&ie->size, ctx, infile);
561 	if (err)
562 		goto done;
563 
564 	err = read_fileindex_val16(&ie->mode, ctx, infile);
565 	if (err)
566 		goto done;
567 
568 	n = fread(ie->blob_sha1, 1, SHA1_DIGEST_LENGTH, infile);
569 	if (n != SHA1_DIGEST_LENGTH) {
570 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
571 		goto done;
572 	}
573 	SHA1Update(ctx, ie->blob_sha1, SHA1_DIGEST_LENGTH);
574 
575 	n = fread(ie->commit_sha1, 1, SHA1_DIGEST_LENGTH, infile);
576 	if (n != SHA1_DIGEST_LENGTH) {
577 		err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
578 		goto done;
579 	}
580 	SHA1Update(ctx, ie->commit_sha1, SHA1_DIGEST_LENGTH);
581 
582 	err = read_fileindex_val32(&ie->flags, ctx, infile);
583 	if (err)
584 		goto done;
585 
586 	err = read_fileindex_path(&ie->path, ctx, infile);
587 	if (err)
588 		goto done;
589 
590 	if (version >= 2) {
591 		uint32_t stage = got_fileindex_entry_stage_get(ie);
592 		if (stage == GOT_FILEIDX_STAGE_MODIFY ||
593 		    stage == GOT_FILEIDX_STAGE_ADD) {
594 			n = fread(ie->staged_blob_sha1, 1, SHA1_DIGEST_LENGTH,
595 			    infile);
596 			if (n != SHA1_DIGEST_LENGTH) {
597 				err = got_ferror(infile, GOT_ERR_FILEIDX_BAD);
598 				goto done;
599 			}
600 			SHA1Update(ctx, ie->staged_blob_sha1, SHA1_DIGEST_LENGTH);
601 		}
602 	} else {
603 		/* GOT_FILE_INDEX_VERSION 1 does not support staging. */
604 		ie->flags &= ~GOT_FILEIDX_F_STAGE;
605 	}
606 
607 done:
608 	if (err)
609 		got_fileindex_entry_free(ie);
610 	else
611 		*iep = ie;
612 	return err;
613 }
614 
615 const struct got_error *
616 got_fileindex_read(struct got_fileindex *fileindex, FILE *infile)
617 {
618 	const struct got_error *err = NULL;
619 	struct got_fileindex_hdr hdr;
620 	SHA1_CTX ctx;
621 	struct got_fileindex_entry *ie;
622 	uint8_t sha1_expected[SHA1_DIGEST_LENGTH];
623 	uint8_t sha1[SHA1_DIGEST_LENGTH];
624 	size_t n;
625 	int i;
626 
627 	SHA1Init(&ctx);
628 
629 	n = fread(&hdr.signature, 1, sizeof(hdr.signature), infile);
630 	if (n != sizeof(hdr.signature)) {
631 		if (n == 0) /* EOF */
632 			return NULL;
633 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
634 	}
635 	n = fread(&hdr.version, 1, sizeof(hdr.version), infile);
636 	if (n != sizeof(hdr.version)) {
637 		if (n == 0) /* EOF */
638 			return NULL;
639 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
640 	}
641 	n = fread(&hdr.nentries, 1, sizeof(hdr.nentries), infile);
642 	if (n != sizeof(hdr.nentries)) {
643 		if (n == 0) /* EOF */
644 			return NULL;
645 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
646 	}
647 
648 	SHA1Update(&ctx, (uint8_t *)&hdr.signature, sizeof(hdr.signature));
649 	SHA1Update(&ctx, (uint8_t *)&hdr.version, sizeof(hdr.version));
650 	SHA1Update(&ctx, (uint8_t *)&hdr.nentries, sizeof(hdr.nentries));
651 
652 	hdr.signature = be32toh(hdr.signature);
653 	hdr.version = be32toh(hdr.version);
654 	hdr.nentries = be32toh(hdr.nentries);
655 
656 	if (hdr.signature != GOT_FILE_INDEX_SIGNATURE)
657 		return got_error(GOT_ERR_FILEIDX_SIG);
658 	if (hdr.version > GOT_FILE_INDEX_VERSION)
659 		return got_error(GOT_ERR_FILEIDX_VER);
660 
661 	for (i = 0; i < hdr.nentries; i++) {
662 		err = read_fileindex_entry(&ie, &ctx, infile, hdr.version);
663 		if (err)
664 			return err;
665 		err = add_entry(fileindex, ie);
666 		if (err)
667 			return err;
668 	}
669 
670 	n = fread(sha1_expected, 1, sizeof(sha1_expected), infile);
671 	if (n != sizeof(sha1_expected))
672 		return got_ferror(infile, GOT_ERR_FILEIDX_BAD);
673 	SHA1Final(sha1, &ctx);
674 	if (memcmp(sha1, sha1_expected, SHA1_DIGEST_LENGTH) != 0)
675 		return got_error(GOT_ERR_FILEIDX_CSUM);
676 
677 	return NULL;
678 }
679 
680 static struct got_fileindex_entry *
681 walk_fileindex(struct got_fileindex *fileindex, struct got_fileindex_entry *ie)
682 {
683 	struct got_fileindex_entry *next;
684 
685 	next = RB_NEXT(got_fileindex_tree, &fileindex->entries, ie);
686 
687 	/* Skip entries which were newly added by diff callbacks. */
688 	while (next && (next->flags & GOT_FILEIDX_F_NOT_FLUSHED))
689 		next = RB_NEXT(got_fileindex_tree, &fileindex->entries, next);
690 
691 	return next;
692 }
693 
694 static const struct got_error *
695 diff_fileindex_tree(struct got_fileindex *, struct got_fileindex_entry **,
696     const struct got_tree_entries *, const char *, const char *,
697     struct got_repository *, struct got_fileindex_diff_tree_cb *, void *);
698 
699 static const struct got_error *
700 walk_tree(struct got_tree_entry **next, struct got_fileindex *fileindex,
701     struct got_fileindex_entry **ie, struct got_tree_entry *te,
702     const char *path, const char *entry_name, struct got_repository *repo,
703     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
704 {
705 	const struct got_error *err = NULL;
706 
707 	if (!got_object_tree_entry_is_submodule(te) && S_ISDIR(te->mode)) {
708 		char *subpath;
709 		struct got_tree_object *subtree;
710 
711 		if (asprintf(&subpath, "%s%s%s", path,
712 		    path[0] == '\0' ? "" : "/", te->name) == -1)
713 			return got_error_from_errno("asprintf");
714 
715 		err = got_object_open_as_tree(&subtree, repo, te->id);
716 		if (err) {
717 			free(subpath);
718 			return err;
719 		}
720 
721 		err = diff_fileindex_tree(fileindex, ie,
722 		    got_object_tree_get_entries(subtree), subpath, entry_name,
723 		    repo, cb, cb_arg);
724 		free(subpath);
725 		got_object_tree_close(subtree);
726 		if (err)
727 			return err;
728 	}
729 
730 	*next = SIMPLEQ_NEXT(te, entry);
731 	return NULL;
732 }
733 
734 static const struct got_error *
735 diff_fileindex_tree(struct got_fileindex *fileindex,
736     struct got_fileindex_entry **ie, const struct got_tree_entries *entries,
737     const char *path, const char *entry_name, struct got_repository *repo,
738     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
739 {
740 	const struct got_error *err = NULL;
741 	struct got_tree_entry *te = NULL;
742 	size_t path_len = strlen(path);
743 	struct got_fileindex_entry *next;
744 
745 	te = SIMPLEQ_FIRST(&entries->head);
746 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || te) {
747 		if (te && *ie) {
748 			char *te_path;
749 			int cmp;
750 			if (asprintf(&te_path, "%s/%s", path, te->name) == -1) {
751 				err = got_error_from_errno("asprintf");
752 				break;
753 			}
754 			cmp = got_path_cmp((*ie)->path, te_path,
755 			    got_fileindex_entry_path_len(*ie), strlen(te_path));
756 			free(te_path);
757 			if (cmp == 0) {
758 				if (got_path_is_child((*ie)->path, path,
759 				    path_len) &&
760 				    !got_object_tree_entry_is_submodule(te) &&
761 				    (entry_name == NULL ||
762 				    strcmp(te->name, entry_name) == 0)) {
763 					err = cb->diff_old_new(cb_arg, *ie, te,
764 					    path);
765 					if (err || entry_name)
766 						break;
767 				}
768 				*ie = walk_fileindex(fileindex, *ie);
769 				err = walk_tree(&te, fileindex, ie, te,
770 				    path, entry_name, repo, cb, cb_arg);
771 			} else if (cmp < 0) {
772 				next = walk_fileindex(fileindex, *ie);
773 				if (got_path_is_child((*ie)->path, path,
774 				    path_len) && (entry_name == NULL ||
775 				    strcmp(te->name, entry_name) == 0)) {
776 					err = cb->diff_old(cb_arg, *ie, path);
777 					if (err || entry_name)
778 						break;
779 				}
780 				*ie = next;
781 			} else {
782 				if ((entry_name == NULL ||
783 				    strcmp(te->name, entry_name) == 0)) {
784 					err = cb->diff_new(cb_arg, te, path);
785 					if (err || entry_name)
786 						break;
787 				}
788 				err = walk_tree(&te, fileindex, ie, te,
789 				    path, entry_name, repo, cb, cb_arg);
790 			}
791 			if (err)
792 				break;
793 		} else if (*ie) {
794 			next = walk_fileindex(fileindex, *ie);
795 			if (got_path_is_child((*ie)->path, path, path_len) &&
796 			    (entry_name == NULL ||
797 			    (te && strcmp(te->name, entry_name) == 0))) {
798 				err = cb->diff_old(cb_arg, *ie, path);
799 				if (err || entry_name)
800 					break;
801 			}
802 			*ie = next;
803 		} else if (te) {
804 			if (!got_object_tree_entry_is_submodule(te) &&
805 			    (entry_name == NULL ||
806 			    strcmp(te->name, entry_name) == 0)) {
807 				err = cb->diff_new(cb_arg, te, path);
808 				if (err || entry_name)
809 					break;
810 			}
811 			err = walk_tree(&te, fileindex, ie, te, path,
812 			    entry_name, repo, cb, cb_arg);
813 			if (err)
814 				break;
815 		}
816 	}
817 
818 	return err;
819 }
820 
821 const struct got_error *
822 got_fileindex_diff_tree(struct got_fileindex *fileindex,
823     struct got_tree_object *tree, const char *path, const char *entry_name,
824     struct got_repository *repo,
825     struct got_fileindex_diff_tree_cb *cb, void *cb_arg)
826 {
827 	struct got_fileindex_entry *ie;
828 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
829 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
830 		ie = walk_fileindex(fileindex, ie);
831 	return diff_fileindex_tree(fileindex, &ie,
832 	    got_object_tree_get_entries(tree), path, entry_name, repo,
833 	    cb, cb_arg);
834 }
835 
836 static const struct got_error *
837 diff_fileindex_dir(struct got_fileindex *, struct got_fileindex_entry **,
838     struct got_pathlist_head *, const char *, const char *,
839     struct got_repository *, struct got_fileindex_diff_dir_cb *, void *);
840 
841 static const struct got_error *
842 read_dirlist(struct got_pathlist_head *dirlist, DIR *dir, const char *path)
843 {
844 	const struct got_error *err = NULL;
845 	struct got_pathlist_entry *new = NULL;
846 	struct dirent *dep = NULL;
847 	struct dirent *de = NULL;
848 
849 	for (;;) {
850 		de = malloc(sizeof(struct dirent) + NAME_MAX + 1);
851 		if (de == NULL) {
852 			err = got_error_from_errno("malloc");
853 			break;
854 		}
855 
856 		if (readdir_r(dir, de, &dep) != 0) {
857 			err = got_error_from_errno("readdir_r");
858 			free(de);
859 			break;
860 		}
861 		if (dep == NULL) {
862 			free(de);
863 			break;
864 		}
865 
866 		if (strcmp(de->d_name, ".") == 0 ||
867 		    strcmp(de->d_name, "..") == 0 ||
868 		    (path[0] == '\0' &&
869 		    strcmp(de->d_name, GOT_WORKTREE_GOT_DIR) == 0)) {
870 			free(de);
871 			continue;
872 		}
873 
874 		err = got_pathlist_insert(&new, dirlist, de->d_name, de);
875 		if (err) {
876 			free(de);
877 			break;
878 		}
879 		if (new == NULL) {
880 			err = got_error(GOT_ERR_DIR_DUP_ENTRY);
881 			free(de);
882 			break;
883 		}
884 	}
885 
886 	return err;
887 }
888 
889 void
890 free_dirlist(struct got_pathlist_head *dirlist)
891 {
892 	struct got_pathlist_entry *dle;
893 
894 	TAILQ_FOREACH(dle, dirlist, entry)
895 		free(dle->data);
896 	got_pathlist_free(dirlist);
897 }
898 
899 static const struct got_error *
900 walk_dir(struct got_pathlist_entry **next, struct got_fileindex *fileindex,
901     struct got_fileindex_entry **ie, struct got_pathlist_entry *dle,
902     const char *path, const char *rootpath, struct got_repository *repo,
903     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
904 {
905 	const struct got_error *err = NULL;
906 	struct dirent *de = dle->data;
907 
908 	*next = NULL;
909 
910 	if (de->d_type == DT_DIR) {
911 		char *subpath;
912 		char *subdirpath;
913 		DIR *subdir;
914 		struct got_pathlist_head subdirlist;
915 
916 		TAILQ_INIT(&subdirlist);
917 
918 		if (asprintf(&subpath, "%s%s%s", path,
919 		    path[0] == '\0' ? "" : "/", de->d_name) == -1)
920 			return got_error_from_errno("asprintf");
921 
922 		if (asprintf(&subdirpath, "%s/%s", rootpath, subpath) == -1) {
923 			free(subpath);
924 			return got_error_from_errno("asprintf");
925 		}
926 
927 		subdir = opendir(subdirpath);
928 		if (subdir == NULL) {
929 			if (errno == EACCES) {
930 				*next = TAILQ_NEXT(dle, entry);
931 				return NULL;
932 			}
933 			err = got_error_from_errno2("opendir", subdirpath);
934 			free(subpath);
935 			free(subdirpath);
936 			return err;
937 		}
938 
939 		err = read_dirlist(&subdirlist, subdir, subdirpath);
940 		if (err) {
941 			free(subpath);
942 			free(subdirpath);
943 			closedir(subdir);
944 			return err;
945 		}
946 		err = diff_fileindex_dir(fileindex, ie, &subdirlist, rootpath,
947 		    subpath, repo, cb, cb_arg);
948 		free(subpath);
949 		free(subdirpath);
950 		closedir(subdir);
951 		free_dirlist(&subdirlist);
952 		if (err)
953 			return err;
954 	}
955 
956 	*next = TAILQ_NEXT(dle, entry);
957 	return NULL;
958 }
959 
960 static const struct got_error *
961 diff_fileindex_dir(struct got_fileindex *fileindex,
962     struct got_fileindex_entry **ie, struct got_pathlist_head *dirlist,
963     const char *rootpath, const char *path, struct got_repository *repo,
964     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
965 {
966 	const struct got_error *err = NULL;
967 	struct dirent *de = NULL;
968 	size_t path_len = strlen(path);
969 	struct got_pathlist_entry *dle;
970 
971 	dle = TAILQ_FIRST(dirlist);
972 	while ((*ie && got_path_is_child((*ie)->path, path, path_len)) || dle) {
973 		if (dle && *ie) {
974 			char *de_path;
975 			int cmp;
976 			de = dle->data;
977 			if (asprintf(&de_path, "%s/%s", path,
978 			    de->d_name) == -1) {
979 				err = got_error_from_errno("asprintf");
980 				break;
981 			}
982 			cmp = got_path_cmp((*ie)->path, de_path,
983 			    got_fileindex_entry_path_len(*ie),
984 			    strlen(path) + 1 + de->d_namlen);
985 			free(de_path);
986 			if (cmp == 0) {
987 				err = cb->diff_old_new(cb_arg, *ie, de, path);
988 				if (err)
989 					break;
990 				*ie = walk_fileindex(fileindex, *ie);
991 				err = walk_dir(&dle, fileindex, ie, dle, path,
992 				    rootpath, repo, cb, cb_arg);
993 			} else if (cmp < 0 ) {
994 				err = cb->diff_old(cb_arg, *ie, path);
995 				if (err)
996 					break;
997 				*ie = walk_fileindex(fileindex, *ie);
998 			} else {
999 				err = cb->diff_new(cb_arg, de, path);
1000 				if (err)
1001 					break;
1002 				err = walk_dir(&dle, fileindex, ie, dle, path,
1003 				    rootpath, repo, cb, cb_arg);
1004 			}
1005 			if (err)
1006 				break;
1007 		} else if (*ie) {
1008 			err = cb->diff_old(cb_arg, *ie, path);
1009 			if (err)
1010 				break;
1011 			*ie = walk_fileindex(fileindex, *ie);
1012 		} else if (dle) {
1013 			de = dle->data;
1014 			err = cb->diff_new(cb_arg, de, path);
1015 			if (err)
1016 				break;
1017 			err = walk_dir(&dle, fileindex, ie, dle, path,
1018 			    rootpath, repo, cb, cb_arg);
1019 			if (err)
1020 				break;
1021 		}
1022 	}
1023 
1024 	return err;
1025 }
1026 
1027 const struct got_error *
1028 got_fileindex_diff_dir(struct got_fileindex *fileindex, DIR *rootdir,
1029     const char *rootpath, const char *path, struct got_repository *repo,
1030     struct got_fileindex_diff_dir_cb *cb, void *cb_arg)
1031 {
1032 	const struct got_error *err;
1033 	struct got_fileindex_entry *ie;
1034 	struct got_pathlist_head dirlist;
1035 
1036 	TAILQ_INIT(&dirlist);
1037 	err = read_dirlist(&dirlist, rootdir, path);
1038 	if (err)
1039 		return err;
1040 	ie = RB_MIN(got_fileindex_tree, &fileindex->entries);
1041 	while (ie && !got_path_is_child(ie->path, path, strlen(path)))
1042 		ie = walk_fileindex(fileindex, ie);
1043 	err = diff_fileindex_dir(fileindex, &ie, &dirlist, rootpath, path,
1044 	    repo, cb, cb_arg);
1045 	free_dirlist(&dirlist);
1046 	return err;
1047 }
1048 
1049 RB_GENERATE(got_fileindex_tree, got_fileindex_entry, entry, got_fileindex_cmp);