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