1 /*
2  * Copyright (c) 2018, 2019 Ori Bernstein <ori@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/stat.h>
19 #include <sys/uio.h>
20 #include <sys/socket.h>
21 #include <sys/wait.h>
22 #include <sys/resource.h>
23 #include <sys/socket.h>
24 
25 #include <endian.h>
26 #include <errno.h>
27 #include <err.h>
28 #include <fcntl.h>
29 #include <stdio.h>
30 #include <stdlib.h>
31 #include <string.h>
32 #include <stdint.h>
33 #include <sha1.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <ctype.h>
37 #include <limits.h>
38 #include <time.h>
39 #include <uuid.h>
40 
41 #include "got_error.h"
42 #include "got_reference.h"
43 #include "got_repository.h"
44 #include "got_path.h"
45 #include "got_cancel.h"
46 #include "got_worktree.h"
47 #include "got_object.h"
48 #include "got_opentemp.h"
49 #include "got_fetch.h"
50 
51 #include "got_lib_delta.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_sha1.h"
58 #include "got_lib_privsep.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 #include "got_lib_dial.h"
62 #include "got_lib_pkt.h"
63 
64 #ifndef nitems
65 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
66 #endif
67 
68 #ifndef ssizeof
69 #define ssizeof(_x) ((ssize_t)(sizeof(_x)))
70 #endif
71 
72 #ifndef MIN
73 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
74 #endif
75 
76 const struct got_error *
77 got_fetch_connect(pid_t *fetchpid, int *fetchfd, const char *proto,
78     const char *host, const char *port, const char *server_path, int verbosity)
79 {
80 	const struct got_error *err = NULL;
81 
82 	*fetchpid = -1;
83 	*fetchfd = -1;
84 
85 	if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
86 		err = got_dial_ssh(fetchpid, fetchfd, host, port,
87 		    server_path, GOT_DIAL_DIRECTION_FETCH, verbosity);
88 	else if (strcmp(proto, "git") == 0)
89 		err = got_dial_git(fetchfd, host, port, server_path,
90 		    GOT_DIAL_DIRECTION_FETCH);
91 	else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
92 		err = got_error_path(proto, GOT_ERR_NOT_IMPL);
93 	else
94 		err = got_error_path(proto, GOT_ERR_BAD_PROTO);
95 	return err;
96 }
97 
98 const struct got_error*
99 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
100     struct got_pathlist_head *symrefs, const char *remote_name,
101     int mirror_references, int fetch_all_branches,
102     struct got_pathlist_head *wanted_branches,
103     struct got_pathlist_head *wanted_refs, int list_refs_only, int verbosity,
104     int fetchfd, struct got_repository *repo,
105     got_fetch_progress_cb progress_cb, void *progress_arg)
106 {
107 	size_t i;
108 	int imsg_fetchfds[2], imsg_idxfds[2];
109 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
110 	int tmpfds[3];
111 	int fetchstatus, idxstatus, done = 0;
112 	const struct got_error *err;
113 	struct imsgbuf fetchibuf, idxibuf;
114 	pid_t fetchpid, idxpid;
115 	char *tmppackpath = NULL, *tmpidxpath = NULL;
116 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
117 	const char *repo_path = NULL;
118 	struct got_pathlist_head have_refs;
119 	struct got_pathlist_entry *pe;
120 	struct got_reflist_head my_refs;
121 	struct got_reflist_entry *re;
122 	off_t packfile_size = 0;
123 	struct got_packfile_hdr pack_hdr;
124 	uint32_t nobj = 0;
125 	char *ref_prefix = NULL;
126 	size_t ref_prefixlen = 0;
127 	char *path;
128 	char *progress = NULL;
129 
130 	*pack_hash = NULL;
131 
132 	/*
133 	 * Prevent fetching of references that won't make any
134 	 * sense outside of the remote repository's context.
135 	 */
136 	TAILQ_FOREACH(pe, wanted_refs, entry) {
137 		const char *refname = pe->path;
138 		if (strncmp(refname, "refs/got/", 9) == 0 ||
139 		    strncmp(refname, "got/", 4) == 0 ||
140 		    strncmp(refname, "refs/remotes/", 13) == 0 ||
141 		    strncmp(refname, "remotes/", 8) == 0)
142 			return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
143 	}
144 
145 	if (!list_refs_only)
146 		repo_path = got_repo_get_path_git_dir(repo);
147 
148 	for (i = 0; i < nitems(tmpfds); i++)
149 		tmpfds[i] = -1;
150 
151 	TAILQ_INIT(&have_refs);
152 	TAILQ_INIT(&my_refs);
153 
154 	if (!mirror_references) {
155 		if (asprintf(&ref_prefix, "refs/remotes/%s/",
156 		    remote_name) == -1)
157 			return got_error_from_errno("asprintf");
158 		ref_prefixlen = strlen(ref_prefix);
159 	}
160 
161 	if (!list_refs_only) {
162 		err = got_ref_list(&my_refs, repo, NULL,
163 		    got_ref_cmp_by_name, NULL);
164 		if (err)
165 			goto done;
166 	}
167 
168 	TAILQ_FOREACH(re, &my_refs, entry) {
169 		struct got_object_id *id;
170 		const char *refname;
171 
172 		if (got_ref_is_symbolic(re->ref))
173 			continue;
174 
175 		refname = got_ref_get_name(re->ref);
176 
177 		if (mirror_references) {
178 			char *name;
179 			err = got_ref_resolve(&id, repo, re->ref);
180 			if (err)
181 				goto done;
182 			name = strdup(refname);
183 			if (name == NULL) {
184 				err = got_error_from_errno("strdup");
185 				goto done;
186 			}
187 			err = got_pathlist_append(&have_refs, name, id);
188 			if (err)
189 				goto done;
190 			continue;
191 		}
192 
193 		if (strncmp("refs/tags/", refname, 10) == 0) {
194 			char *tagname;
195 
196 			err = got_ref_resolve(&id, repo, re->ref);
197 			if (err)
198 				goto done;
199 			tagname = strdup(refname);
200 			if (tagname == NULL) {
201 				err = got_error_from_errno("strdup");
202 				goto done;
203 			}
204 			err = got_pathlist_append(&have_refs, tagname, id);
205 			if (err) {
206 				free(tagname);
207 				goto done;
208 			}
209 		}
210 
211 		if (strncmp(ref_prefix, refname, ref_prefixlen) == 0) {
212 			char *branchname;
213 
214 			err = got_ref_resolve(&id, repo, re->ref);
215 			if (err)
216 				goto done;
217 
218 			if (asprintf(&branchname, "refs/heads/%s",
219 			    refname + ref_prefixlen) == -1) {
220 				err = got_error_from_errno("asprintf");
221 				goto done;
222 			}
223 			err = got_pathlist_append(&have_refs, branchname, id);
224 			if (err) {
225 				free(branchname);
226 				goto done;
227 			}
228 		}
229 	}
230 
231 	if (list_refs_only) {
232 		packfd = got_opentempfd();
233 		if (packfd == -1) {
234 			err = got_error_from_errno("got_opentempfd");
235 			goto done;
236 		}
237 	} else {
238 		if (asprintf(&path, "%s/%s/fetching.pack",
239 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
240 			err = got_error_from_errno("asprintf");
241 			goto done;
242 		}
243 		err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
244 		free(path);
245 		if (err)
246 			goto done;
247 		if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
248 			err = got_error_from_errno2("fchmod", tmppackpath);
249 			goto done;
250 		}
251 	}
252 	if (list_refs_only) {
253 		idxfd = got_opentempfd();
254 		if (idxfd == -1) {
255 			err = got_error_from_errno("got_opentempfd");
256 			goto done;
257 		}
258 	} else {
259 		if (asprintf(&path, "%s/%s/fetching.idx",
260 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
261 			err = got_error_from_errno("asprintf");
262 			goto done;
263 		}
264 		err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
265 		free(path);
266 		if (err)
267 			goto done;
268 		if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
269 			err = got_error_from_errno2("fchmod", tmpidxpath);
270 			goto done;
271 		}
272 	}
273 	nidxfd = dup(idxfd);
274 	if (nidxfd == -1) {
275 		err = got_error_from_errno("dup");
276 		goto done;
277 	}
278 
279 	for (i = 0; i < nitems(tmpfds); i++) {
280 		tmpfds[i] = got_opentempfd();
281 		if (tmpfds[i] == -1) {
282 			err = got_error_from_errno("got_opentempfd");
283 			goto done;
284 		}
285 	}
286 
287 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
288 		err = got_error_from_errno("socketpair");
289 		goto done;
290 	}
291 
292 	fetchpid = fork();
293 	if (fetchpid == -1) {
294 		err = got_error_from_errno("fork");
295 		goto done;
296 	} else if (fetchpid == 0){
297 		got_privsep_exec_child(imsg_fetchfds,
298 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
299 	}
300 
301 	if (close(imsg_fetchfds[1]) == -1) {
302 		err = got_error_from_errno("close");
303 		goto done;
304 	}
305 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
306 	nfetchfd = dup(fetchfd);
307 	if (nfetchfd == -1) {
308 		err = got_error_from_errno("dup");
309 		goto done;
310 	}
311 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
312 	    fetch_all_branches, wanted_branches, wanted_refs,
313 	    list_refs_only, verbosity);
314 	if (err != NULL)
315 		goto done;
316 	nfetchfd = -1;
317 	npackfd = dup(packfd);
318 	if (npackfd == -1) {
319 		err = got_error_from_errno("dup");
320 		goto done;
321 	}
322 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
323 	if (err != NULL)
324 		goto done;
325 	npackfd = -1;
326 
327 	packfile_size = 0;
328 	progress = calloc(GOT_PKT_MAX, 1);
329 	if (progress == NULL) {
330 		err = got_error_from_errno("calloc");
331 		goto done;
332 	}
333 
334 	*pack_hash = calloc(1, sizeof(**pack_hash));
335 	if (*pack_hash == NULL) {
336 		err = got_error_from_errno("calloc");
337 		goto done;
338 	}
339 
340 	while (!done) {
341 		struct got_object_id *id = NULL;
342 		char *refname = NULL;
343 		char *server_progress = NULL;
344 		off_t packfile_size_cur = 0;
345 
346 		err = got_privsep_recv_fetch_progress(&done,
347 		    &id, &refname, symrefs, &server_progress,
348 		    &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
349 		if (err != NULL)
350 			goto done;
351 		if (!done && refname && id) {
352 			err = got_pathlist_insert(NULL, refs, refname, id);
353 			if (err)
354 				goto done;
355 		} else if (!done && server_progress) {
356 			char *p;
357 			/*
358 			 * XXX git-daemon tends to send batched output with
359 			 * lines spanning separate packets. Buffer progress
360 			 * output until we see a CR or LF to avoid giving
361 			 * partial lines of progress output to the callback.
362 			 */
363 			if (strlcat(progress, server_progress,
364 			    GOT_PKT_MAX) >= GOT_PKT_MAX) {
365 				progress[0] = '\0'; /* discard */
366 				continue;
367 			}
368 			while ((p = strchr(progress, '\r')) != NULL ||
369 			    (p = strchr(progress, '\n')) != NULL) {
370 				char *s;
371 				size_t n;
372 				char c = *p;
373 				*p = '\0';
374 				if (asprintf(&s, "%s%s", progress,
375 				    c == '\n' ? "\n" : "") == -1) {
376 					err = got_error_from_errno("asprintf");
377 					goto done;
378 				}
379 				err = progress_cb(progress_arg, s,
380 				    packfile_size_cur, 0, 0, 0, 0);
381 				free(s);
382 				if (err)
383 					break;
384 				n = strlen(progress);
385 				if (n < GOT_PKT_MAX - 1) {
386 					memmove(progress, &progress[n + 1],
387 					    GOT_PKT_MAX - n - 1);
388 				} else
389 					progress[0] = '\0';
390 			}
391 			free(server_progress);
392 			if (err)
393 				goto done;
394 		} else if (!done && packfile_size_cur != packfile_size) {
395 			err = progress_cb(progress_arg, NULL,
396 			    packfile_size_cur, 0, 0, 0, 0);
397 			if (err)
398 				break;
399 			packfile_size = packfile_size_cur;
400 		}
401 	}
402 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
403 		err = got_error_from_errno("waitpid");
404 		goto done;
405 	}
406 
407 	if (lseek(packfd, 0, SEEK_SET) == -1) {
408 		err = got_error_from_errno("lseek");
409 		goto done;
410 	}
411 
412 	/* If zero data was fetched without error we are already up-to-date. */
413 	if (packfile_size == 0) {
414 		free(*pack_hash);
415 		*pack_hash = NULL;
416 		goto done;
417 	} else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
418 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
419 		goto done;
420 	} else {
421 		ssize_t n;
422 
423 		n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
424 		if (n == -1) {
425 			err = got_error_from_errno("read");
426 			goto done;
427 		}
428 		if (n != ssizeof(pack_hdr)) {
429 			err = got_error(GOT_ERR_IO);
430 			goto done;
431 		}
432 		if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
433 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
434 			    "bad pack file signature");
435 			goto done;
436 		}
437 		if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
438 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
439 			    "bad pack file version");
440 			goto done;
441 		}
442 		nobj = be32toh(pack_hdr.nobjects);
443 		if (nobj == 0 &&
444 		    packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
445 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
446 			    "bad pack file with zero objects");
447 		if (nobj != 0 &&
448 		    packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
449 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
450 			    "empty pack file with non-zero object count");
451 	}
452 
453 	/*
454 	 * If the pack file contains no objects, we may only need to update
455 	 * references in our repository. The caller will take care of that.
456 	 */
457 	if (nobj == 0)
458 		goto done;
459 
460 	if (lseek(packfd, 0, SEEK_SET) == -1) {
461 		err = got_error_from_errno("lseek");
462 		goto done;
463 	}
464  
465 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
466 		err = got_error_from_errno("socketpair");
467 		goto done;
468 	}
469 	idxpid = fork();
470 	if (idxpid == -1) {
471 		err= got_error_from_errno("fork");
472 		goto done;
473 	} else if (idxpid == 0)
474 		got_privsep_exec_child(imsg_idxfds,
475 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
476 	if (close(imsg_idxfds[1]) == -1) {
477 		err = got_error_from_errno("close");
478 		goto done;
479 	}
480 	imsg_init(&idxibuf, imsg_idxfds[0]);
481 
482 	npackfd = dup(packfd);
483 	if (npackfd == -1) {
484 		err = got_error_from_errno("dup");
485 		goto done;
486 	}
487 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
488 	    npackfd);
489 	if (err != NULL)
490 		goto done;
491 	npackfd = -1;
492 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
493 	if (err != NULL)
494 		goto done;
495 	nidxfd = -1;
496 	for (i = 0; i < nitems(tmpfds); i++) {
497 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
498 		if (err != NULL)
499 			goto done;
500 		tmpfds[i] = -1;
501 	}
502 	done = 0;
503 	while (!done) {
504 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
505 
506 		err = got_privsep_recv_index_progress(&done, &nobj_total,
507 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
508 		    &idxibuf);
509 		if (err != NULL)
510 			goto done;
511 		if (nobj_indexed != 0) {
512 			err = progress_cb(progress_arg, NULL,
513 			    packfile_size, nobj_total,
514 			    nobj_indexed, nobj_loose, nobj_resolved);
515 			if (err)
516 				break;
517 		}
518 		imsg_clear(&idxibuf);
519 	}
520 	if (close(imsg_idxfds[0]) == -1) {
521 		err = got_error_from_errno("close");
522 		goto done;
523 	}
524 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
525 		err = got_error_from_errno("waitpid");
526 		goto done;
527 	}
528 
529 	err = got_object_id_str(&id_str, *pack_hash);
530 	if (err)
531 		goto done;
532 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
533 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
534 		err = got_error_from_errno("asprintf");
535 		goto done;
536 	}
537 
538 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
539 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
540 		err = got_error_from_errno("asprintf");
541 		goto done;
542 	}
543 
544 	if (rename(tmppackpath, packpath) == -1) {
545 		err = got_error_from_errno3("rename", tmppackpath, packpath);
546 		goto done;
547 	}
548 	free(tmppackpath);
549 	tmppackpath = NULL;
550 	if (rename(tmpidxpath, idxpath) == -1) {
551 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
552 		goto done;
553 	}
554 	free(tmpidxpath);
555 	tmpidxpath = NULL;
556 
557 done:
558 	if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
559 		err = got_error_from_errno2("unlink", tmppackpath);
560 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
561 		err = got_error_from_errno2("unlink", tmpidxpath);
562 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
563 		err = got_error_from_errno("close");
564 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
565 		err = got_error_from_errno("close");
566 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
567 		err = got_error_from_errno("close");
568 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
569 		err = got_error_from_errno("close");
570 	for (i = 0; i < nitems(tmpfds); i++) {
571 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
572 			err = got_error_from_errno("close");
573 	}
574 	free(tmppackpath);
575 	free(tmpidxpath);
576 	free(idxpath);
577 	free(packpath);
578 	free(ref_prefix);
579 	free(progress);
580 
581 	TAILQ_FOREACH(pe, &have_refs, entry) {
582 		free((char *)pe->path);
583 		free(pe->data);
584 	}
585 	got_pathlist_free(&have_refs);
586 	got_ref_list_free(&my_refs);
587 	if (err) {
588 		free(*pack_hash);
589 		*pack_hash = NULL;
590 		TAILQ_FOREACH(pe, refs, entry) {
591 			free((void *)pe->path);
592 			free(pe->data);
593 		}
594 		got_pathlist_free(refs);
595 		TAILQ_FOREACH(pe, symrefs, entry) {
596 			free((void *)pe->path);
597 			free(pe->data);
598 		}
599 		got_pathlist_free(symrefs);
600 	}
601 	return err;
602 }