1 /*
2  * Copyright (c) 2020 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/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/stat.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 
24 #include <dirent.h>
25 #include <endian.h>
26 #include <errno.h>
27 #include <fcntl.h>
28 #include <stdint.h>
29 #include <sha1.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <limits.h>
34 #include <unistd.h>
35 #include <imsg.h>
36 
37 #include "got_error.h"
38 #include "got_cancel.h"
39 #include "got_object.h"
40 #include "got_reference.h"
41 #include "got_repository.h"
42 #include "got_repository_admin.h"
43 #include "got_opentemp.h"
44 #include "got_path.h"
45 
46 #include "got_lib_delta.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_cache.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_repository.h"
52 #include "got_lib_pack_create.h"
53 #include "got_lib_sha1.h"
54 
55 #ifndef nitems
56 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
57 #endif
58 
59 static const struct got_error *
60 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
61     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
62     struct got_repository *repo,
63     got_cancel_cb cancel_cb, void *cancel_arg)
64 {
65 	const struct got_error *err = NULL;
66 	const size_t alloc_chunksz = 256;
67 	size_t nalloc;
68 	struct got_reflist_entry *re;
69 	int i;
70 
71 	*ids = NULL;
72 	*nobjects = 0;
73 
74 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
75 	if (*ids == NULL)
76 		return got_error_from_errno("reallocarray");
77 	nalloc = alloc_chunksz;
78 
79 	TAILQ_FOREACH(re, refs, entry) {
80 		struct got_object_id *id;
81 
82 		if (cancel_cb) {
83 			err = cancel_cb(cancel_arg);
84 			if (err)
85 				goto done;
86 		}
87 
88 		err = got_ref_resolve(&id, repo, re->ref);
89 		if (err)
90 			goto done;
91 
92 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
93 			int obj_type;
94 			err = got_object_get_type(&obj_type, repo, id);
95 			if (err)
96 				goto done;
97 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
98 				free(id);
99 				id = NULL;
100 				continue;
101 			}
102 		}
103 
104 		if (nalloc >= *nobjects) {
105 			struct got_object_id **new;
106 			new = recallocarray(*ids, nalloc,
107 			    nalloc + alloc_chunksz,
108 			    sizeof(struct got_object_id *));
109 			if (new == NULL) {
110 				err = got_error_from_errno(
111 				    "recallocarray");
112 				goto done;
113 			}
114 			*ids = new;
115 			nalloc += alloc_chunksz;
116 		}
117 		(*ids)[*nobjects] = id;
118 		if ((*ids)[*nobjects] == NULL) {
119 			err = got_error_from_errno("got_object_id_dup");
120 			goto done;
121 		}
122 		(*nobjects)++;
123 	}
124 done:
125 	if (err) {
126 		for (i = 0; i < *nobjects; i++)
127 			free((*ids)[i]);
128 		free(*ids);
129 		*ids = NULL;
130 		*nobjects = 0;
131 	}
132 	return err;
133 }
134 
135 const struct got_error *
136 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
137     struct got_reflist_head *include_refs,
138     struct got_reflist_head *exclude_refs, struct got_repository *repo,
139     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
140     got_cancel_cb cancel_cb, void *cancel_arg)
141 {
142 	const struct got_error *err = NULL;
143 	struct got_object_id **ours = NULL, **theirs = NULL;
144 	int nours = 0, ntheirs = 0, packfd = -1, i;
145 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
146 	char *sha1_str = NULL;
147 
148 	*packfile = NULL;
149 	*pack_hash = NULL;
150 
151 	if (asprintf(&path, "%s/%s/packing.pack",
152 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
153 		err = got_error_from_errno("asprintf");
154 		goto done;
155 	}
156 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
157 	if (err)
158 		goto done;
159 
160 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
161 		err = got_error_from_errno2("fchmod", tmpfile_path);
162 		goto done;
163 	}
164 
165 	*packfile = fdopen(packfd, "w");
166 	if (*packfile == NULL) {
167 		err = got_error_from_errno2("fdopen", tmpfile_path);
168 		goto done;
169 	}
170 	packfd = -1;
171 
172 	err = get_reflist_object_ids(&ours, &nours,
173 	   (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
174 	   include_refs, repo, cancel_cb, cancel_arg);
175 	if (err)
176 		goto done;
177 
178 	if (nours == 0) {
179 		err = got_error(GOT_ERR_CANNOT_PACK);
180 		goto done;
181 	}
182 
183 	if (!TAILQ_EMPTY(exclude_refs)) {
184 		err = get_reflist_object_ids(&theirs, &ntheirs,
185 		   (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
186 		   exclude_refs, repo,
187 		   cancel_cb, cancel_arg);
188 		if (err)
189 			goto done;
190 	}
191 
192 	*pack_hash = calloc(1, sizeof(**pack_hash));
193 	if (*pack_hash == NULL) {
194 		err = got_error_from_errno("calloc");
195 		goto done;
196 	}
197 
198 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
199 	    ours, nours, repo, loose_obj_only, progress_cb, progress_arg,
200 	    cancel_cb, cancel_arg);
201 	if (err)
202 		goto done;
203 
204 	err = got_object_id_str(&sha1_str, *pack_hash);
205 	if (err)
206 		goto done;
207 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
208 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
209 	    sha1_str) == -1) {
210 		err = got_error_from_errno("asprintf");
211 		goto done;
212 	}
213 
214 	if (fflush(*packfile) == -1) {
215 		err = got_error_from_errno("fflush");
216 		goto done;
217 	}
218 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
219 		err = got_error_from_errno("fseek");
220 		goto done;
221 	}
222 	if (rename(tmpfile_path, packfile_path) == -1) {
223 		err = got_error_from_errno3("rename", tmpfile_path,
224 		    packfile_path);
225 		goto done;
226 	}
227 	free(tmpfile_path);
228 	tmpfile_path = NULL;
229 done:
230 	for (i = 0; i < nours; i++)
231 		free(ours[i]);
232 	free(ours);
233 	for (i = 0; i < ntheirs; i++)
234 		free(theirs[i]);
235 	free(theirs);
236 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
237 		err = got_error_from_errno2("close", packfile_path);
238 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
239 		err = got_error_from_errno2("unlink", tmpfile_path);
240 	free(tmpfile_path);
241 	free(packfile_path);
242 	free(sha1_str);
243 	free(path);
244 	if (err) {
245 		free(*pack_hash);
246 		*pack_hash = NULL;
247 		if (*packfile)
248 			fclose(*packfile);
249 		*packfile = NULL;
250 	}
251 	return err;
252 }
253 
254 const struct got_error *
255 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
256     struct got_repository *repo,
257     got_pack_index_progress_cb progress_cb, void *progress_arg,
258     got_cancel_cb cancel_cb, void *cancel_arg)
259 {
260 	size_t i;
261 	char *path;
262 	int imsg_idxfds[2];
263 	int npackfd = -1, idxfd = -1, nidxfd = -1;
264 	int tmpfds[3];
265 	int idxstatus, done = 0;
266 	const struct got_error *err;
267 	struct imsgbuf idxibuf;
268 	pid_t idxpid;
269 	char *tmpidxpath = NULL;
270 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
271 	const char *repo_path = got_repo_get_path_git_dir(repo);
272 	struct stat sb;
273 
274 	for (i = 0; i < nitems(tmpfds); i++)
275 		tmpfds[i] = -1;
276 
277 	if (asprintf(&path, "%s/%s/indexing.idx",
278 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
279 		err = got_error_from_errno("asprintf");
280 		goto done;
281 	}
282 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
283 	free(path);
284 	if (err)
285 		goto done;
286 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
287 		err = got_error_from_errno2("fchmod", tmpidxpath);
288 		goto done;
289 	}
290 
291 	nidxfd = dup(idxfd);
292 	if (nidxfd == -1) {
293 		err = got_error_from_errno("dup");
294 		goto done;
295 	}
296 
297 	for (i = 0; i < nitems(tmpfds); i++) {
298 		tmpfds[i] = got_opentempfd();
299 		if (tmpfds[i] == -1) {
300 			err = got_error_from_errno("got_opentempfd");
301 			goto done;
302 		}
303 	}
304 
305 	err = got_object_id_str(&id_str, pack_hash);
306 	if (err)
307 		goto done;
308 
309 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
310 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
311 		err = got_error_from_errno("asprintf");
312 		goto done;
313 	}
314 
315 	if (fstat(fileno(packfile), &sb) == -1) {
316 		err = got_error_from_errno2("fstat", packfile_path);
317 		goto done;
318 	}
319 
320 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
321 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
322 		err = got_error_from_errno("asprintf");
323 		goto done;
324 	}
325 
326 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
327 		err = got_error_from_errno("socketpair");
328 		goto done;
329 	}
330 	idxpid = fork();
331 	if (idxpid == -1) {
332 		err= got_error_from_errno("fork");
333 		goto done;
334 	} else if (idxpid == 0)
335 		got_privsep_exec_child(imsg_idxfds,
336 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
337 	if (close(imsg_idxfds[1]) == -1) {
338 		err = got_error_from_errno("close");
339 		goto done;
340 	}
341 	imsg_init(&idxibuf, imsg_idxfds[0]);
342 
343 	npackfd = dup(fileno(packfile));
344 	if (npackfd == -1) {
345 		err = got_error_from_errno("dup");
346 		goto done;
347 	}
348 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
349 	    npackfd);
350 	if (err != NULL)
351 		goto done;
352 	npackfd = -1;
353 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
354 	if (err != NULL)
355 		goto done;
356 	nidxfd = -1;
357 	for (i = 0; i < nitems(tmpfds); i++) {
358 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
359 		if (err != NULL)
360 			goto done;
361 		tmpfds[i] = -1;
362 	}
363 	done = 0;
364 	while (!done) {
365 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
366 
367 		if (cancel_cb) {
368 			err = cancel_cb(cancel_arg);
369 			if (err)
370 				goto done;
371 		}
372 
373 		err = got_privsep_recv_index_progress(&done, &nobj_total,
374 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
375 		    &idxibuf);
376 		if (err != NULL)
377 			goto done;
378 		if (nobj_indexed != 0) {
379 			err = progress_cb(progress_arg, sb.st_size,
380 			    nobj_total, nobj_indexed, nobj_loose,
381 			    nobj_resolved);
382 			if (err)
383 				break;
384 		}
385 		imsg_clear(&idxibuf);
386 	}
387 	if (close(imsg_idxfds[0]) == -1) {
388 		err = got_error_from_errno("close");
389 		goto done;
390 	}
391 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
392 		err = got_error_from_errno("waitpid");
393 		goto done;
394 	}
395 
396 	if (rename(tmpidxpath, idxpath) == -1) {
397 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
398 		goto done;
399 	}
400 	free(tmpidxpath);
401 	tmpidxpath = NULL;
402 
403 done:
404 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
405 		err = got_error_from_errno2("unlink", tmpidxpath);
406 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
407 		err = got_error_from_errno("close");
408 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
409 		err = got_error_from_errno("close");
410 	for (i = 0; i < nitems(tmpfds); i++) {
411 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
412 			err = got_error_from_errno("close");
413 	}
414 	free(tmpidxpath);
415 	free(idxpath);
416 	free(packfile_path);
417 	return err;
418 }
419 
420 const struct got_error *
421 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
422     struct got_repository *repo, const char *packfile_path)
423 {
424 	const struct got_error *err = NULL;
425 	const char *packdir_path = NULL;
426 	char *packfile_name = NULL, *p, *dot;
427 	struct got_object_id id;
428 	int packfd = -1;
429 
430 	*packfile = NULL;
431 	*pack_hash = NULL;
432 
433 	packdir_path = got_repo_get_path_objects_pack(repo);
434 	if (packdir_path == NULL)
435 		return got_error_from_errno("got_repo_get_path_objects_pack");
436 
437 	if (!got_path_is_child(packfile_path, packdir_path,
438 	    strlen(packdir_path))) {
439 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
440 		goto done;
441 
442 	}
443 
444 	err = got_path_basename(&packfile_name, packfile_path);
445 	if (err)
446 		goto done;
447 	p = packfile_name;
448 
449 	if (strncmp(p, "pack-", 5) != 0) {
450 		err = got_error_fmt(GOT_ERR_BAD_PATH,
451 		   "'%s' is not a valid pack file name",
452 		   packfile_name);
453 		goto done;
454 	}
455 	p += 5;
456 	dot = strchr(p, '.');
457 	if (dot == NULL) {
458 		err = got_error_fmt(GOT_ERR_BAD_PATH,
459 		   "'%s' is not a valid pack file name",
460 		   packfile_name);
461 		goto done;
462 	}
463 	if (strcmp(dot + 1, "pack") != 0) {
464 		err = got_error_fmt(GOT_ERR_BAD_PATH,
465 		   "'%s' is not a valid pack file name",
466 		   packfile_name);
467 		goto done;
468 	}
469 	*dot = '\0';
470 	if (!got_parse_sha1_digest(id.sha1, p)) {
471 		err = got_error_fmt(GOT_ERR_BAD_PATH,
472 		   "'%s' is not a valid pack file name",
473 		   packfile_name);
474 		goto done;
475 	}
476 
477 	*pack_hash = got_object_id_dup(&id);
478 	if (*pack_hash == NULL) {
479 		err = got_error_from_errno("got_object_id_dup");
480 		goto done;
481 	}
482 
483 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW);
484 	if (packfd == -1) {
485 		err = got_error_from_errno2("open", packfile_path);
486 		goto done;
487 	}
488 
489 	*packfile = fdopen(packfd, "r");
490 	if (*packfile == NULL) {
491 		err = got_error_from_errno2("fdopen", packfile_path);
492 		goto done;
493 	}
494 	packfd = -1;
495 done:
496 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
497 		err = got_error_from_errno2("close", packfile_path);
498 	free(packfile_name);
499 	if (err) {
500 		free(*pack_hash);
501 		*pack_hash = NULL;
502 	}
503 	return err;
504 }
505 
506 const struct got_error *
507 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
508     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
509     got_cancel_cb cancel_cb, void *cancel_arg)
510 {
511 	const struct got_error *err = NULL;
512 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
513 	struct got_packidx *packidx = NULL;
514 	struct got_pack *pack = NULL;
515 	uint32_t nobj, i;
516 
517 	err = got_object_id_str(&id_str, pack_hash);
518 	if (err)
519 		goto done;
520 
521 	if (asprintf(&packpath, "%s/pack-%s.pack",
522 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
523 		err = got_error_from_errno("asprintf");
524 		goto done;
525 	}
526 	if (asprintf(&idxpath, "%s/pack-%s.idx",
527 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
528 		err = got_error_from_errno("asprintf");
529 		goto done;
530 	}
531 
532 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
533 	if (err)
534 		goto done;
535 
536 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
537 	if (err)
538 		goto done;
539 
540 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
541 	for (i = 0; i < nobj; i++) {
542 		struct got_packidx_object_id *oid;
543 		struct got_object_id id, base_id;
544 		off_t offset, base_offset = 0;
545 		uint8_t type;
546 		uint64_t size;
547 		size_t tslen, len;
548 
549 		if (cancel_cb) {
550 			err = cancel_cb(cancel_arg);
551 			if (err)
552 				break;
553 		}
554 		oid = &packidx->hdr.sorted_ids[i];
555 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
556 
557 		offset = got_packidx_get_object_offset(packidx, i);
558 		if (offset == -1) {
559 			err = got_error(GOT_ERR_BAD_PACKIDX);
560 			goto done;
561 		}
562 
563 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
564 		    pack, offset);
565 		if (err)
566 			goto done;
567 
568 		switch (type) {
569 		case GOT_OBJ_TYPE_OFFSET_DELTA:
570 			err = got_pack_parse_offset_delta(&base_offset, &len,
571 			    pack, offset, tslen);
572 			if (err)
573 				goto done;
574 			break;
575 		case GOT_OBJ_TYPE_REF_DELTA:
576 			err = got_pack_parse_ref_delta(&base_id,
577 			    pack, offset, tslen);
578 			if (err)
579 				goto done;
580 			break;
581 		}
582 		err = (*list_cb)(list_arg, &id, type, offset, size,
583 		    base_offset, &base_id);
584 		if (err)
585 			goto done;
586 	}
587 
588 done:
589 	free(id_str);
590 	free(idxpath);
591 	free(packpath);
592 	if (packidx)
593 		got_packidx_close(packidx);
594 	return err;
595 }