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