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 <sha2.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <ctype.h>
40 #include <limits.h>
41 #include <imsg.h>
42 #include <time.h>
43 #include <uuid.h>
44 
45 #include "got_error.h"
46 #include "got_reference.h"
47 #include "got_repository.h"
48 #include "got_path.h"
49 #include "got_cancel.h"
50 #include "got_worktree.h"
51 #include "got_object.h"
52 #include "got_opentemp.h"
53 #include "got_fetch.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_hash.h"
62 #include "got_lib_privsep.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_dial.h"
66 #include "got_lib_pkt.h"
67 
68 #ifndef nitems
69 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
70 #endif
71 
72 #ifndef ssizeof
73 #define ssizeof(_x) ((ssize_t)(sizeof(_x)))
74 #endif
75 
76 #ifndef MIN
77 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
78 #endif
79 
80 const struct got_error *
81 got_fetch_connect(pid_t *fetchpid, int *fetchfd, const char *proto,
82     const char *host, const char *port, const char *server_path, int verbosity)
83 {
84 	const struct got_error *err = NULL;
85 
86 	*fetchpid = -1;
87 	*fetchfd = -1;
88 
89 	if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
90 		err = got_dial_ssh(fetchpid, fetchfd, host, port,
91 		    server_path, GOT_DIAL_CMD_FETCH, verbosity);
92 	else if (strcmp(proto, "git") == 0)
93 		err = got_dial_git(fetchfd, host, port, server_path,
94 		    GOT_DIAL_CMD_FETCH);
95 	else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
96 		err = got_error_path(proto, GOT_ERR_NOT_IMPL);
97 	else
98 		err = got_error_path(proto, GOT_ERR_BAD_PROTO);
99 	return err;
100 }
101 
102 const struct got_error *
103 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
104     struct got_pathlist_head *symrefs, const char *remote_name,
105     int mirror_references, int fetch_all_branches,
106     struct got_pathlist_head *wanted_branches,
107     struct got_pathlist_head *wanted_refs, int list_refs_only, int verbosity,
108     int fetchfd, struct got_repository *repo, const char *worktree_refname,
109     const char *remote_head, int no_head, got_fetch_progress_cb progress_cb,
110     void *progress_arg)
111 {
112 	size_t i;
113 	int imsg_fetchfds[2], imsg_idxfds[2];
114 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
115 	int tmpfds[3];
116 	int fetchstatus, idxstatus, done = 0;
117 	const struct got_error *err;
118 	struct imsgbuf fetchibuf, idxibuf;
119 	pid_t fetchpid, idxpid;
120 	char *tmppackpath = NULL, *tmpidxpath = NULL;
121 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
122 	const char *repo_path = NULL;
123 	struct got_pathlist_head have_refs;
124 	struct got_pathlist_entry *pe;
125 	struct got_reflist_head my_refs;
126 	struct got_reflist_entry *re;
127 	off_t packfile_size = 0;
128 	struct got_packfile_hdr pack_hdr;
129 	uint32_t nobj = 0;
130 	char *path;
131 	char *progress = NULL;
132 
133 	*pack_hash = NULL;
134 
135 	/*
136 	 * Prevent fetching of references that won't make any
137 	 * sense outside of the remote repository's context.
138 	 */
139 	TAILQ_FOREACH(pe, wanted_refs, entry) {
140 		const char *refname = pe->path;
141 		if (strncmp(refname, "refs/got/", 9) == 0 ||
142 		    strncmp(refname, "got/", 4) == 0 ||
143 		    strncmp(refname, "refs/remotes/", 13) == 0 ||
144 		    strncmp(refname, "remotes/", 8) == 0)
145 			return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
146 	}
147 
148 	if (!list_refs_only)
149 		repo_path = got_repo_get_path_git_dir(repo);
150 
151 	for (i = 0; i < nitems(tmpfds); i++)
152 		tmpfds[i] = -1;
153 
154 	TAILQ_INIT(&have_refs);
155 	TAILQ_INIT(&my_refs);
156 
157 	if (!list_refs_only) {
158 		err = got_ref_list(&my_refs, repo, NULL,
159 		    got_ref_cmp_by_name, NULL);
160 		if (err)
161 			goto done;
162 	}
163 
164 	TAILQ_FOREACH(re, &my_refs, entry) {
165 		struct got_object_id *id;
166 		const char *refname;
167 
168 		if (got_ref_is_symbolic(re->ref))
169 			continue;
170 
171 		err = got_ref_resolve(&id, repo, re->ref);
172 		if (err)
173 			goto done;
174 		refname = strdup(got_ref_get_name(re->ref));
175 		if (refname == NULL) {
176 			err = got_error_from_errno("strdup");
177 			goto done;
178 		}
179 		err = got_pathlist_append(&have_refs, refname, id);
180 		if (err)
181 			goto done;
182 	}
183 
184 	if (list_refs_only) {
185 		packfd = got_opentempfd();
186 		if (packfd == -1) {
187 			err = got_error_from_errno("got_opentempfd");
188 			goto done;
189 		}
190 	} else {
191 		if (asprintf(&path, "%s/%s/fetching.pack",
192 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
193 			err = got_error_from_errno("asprintf");
194 			goto done;
195 		}
196 		err = got_opentemp_named_fd(&tmppackpath, &packfd, path, "");
197 		free(path);
198 		if (err)
199 			goto done;
200 		if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
201 			err = got_error_from_errno2("fchmod", tmppackpath);
202 			goto done;
203 		}
204 	}
205 	if (list_refs_only) {
206 		idxfd = got_opentempfd();
207 		if (idxfd == -1) {
208 			err = got_error_from_errno("got_opentempfd");
209 			goto done;
210 		}
211 	} else {
212 		if (asprintf(&path, "%s/%s/fetching.idx",
213 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
214 			err = got_error_from_errno("asprintf");
215 			goto done;
216 		}
217 		err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
218 		free(path);
219 		if (err)
220 			goto done;
221 		if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
222 			err = got_error_from_errno2("fchmod", tmpidxpath);
223 			goto done;
224 		}
225 	}
226 	nidxfd = dup(idxfd);
227 	if (nidxfd == -1) {
228 		err = got_error_from_errno("dup");
229 		goto done;
230 	}
231 
232 	for (i = 0; i < nitems(tmpfds); i++) {
233 		tmpfds[i] = got_opentempfd();
234 		if (tmpfds[i] == -1) {
235 			err = got_error_from_errno("got_opentempfd");
236 			goto done;
237 		}
238 	}
239 
240 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
241 		err = got_error_from_errno("socketpair");
242 		goto done;
243 	}
244 
245 	fetchpid = fork();
246 	if (fetchpid == -1) {
247 		err = got_error_from_errno("fork");
248 		goto done;
249 	} else if (fetchpid == 0){
250 		got_privsep_exec_child(imsg_fetchfds,
251 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
252 	}
253 
254 	if (close(imsg_fetchfds[1]) == -1) {
255 		err = got_error_from_errno("close");
256 		goto done;
257 	}
258 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
259 	nfetchfd = dup(fetchfd);
260 	if (nfetchfd == -1) {
261 		err = got_error_from_errno("dup");
262 		goto done;
263 	}
264 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
265 	    fetch_all_branches, wanted_branches, wanted_refs,
266 	    list_refs_only, worktree_refname, remote_head, no_head, verbosity);
267 	if (err != NULL)
268 		goto done;
269 	nfetchfd = -1;
270 	npackfd = dup(packfd);
271 	if (npackfd == -1) {
272 		err = got_error_from_errno("dup");
273 		goto done;
274 	}
275 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
276 	if (err != NULL)
277 		goto done;
278 	npackfd = -1;
279 
280 	packfile_size = 0;
281 	progress = calloc(GOT_PKT_MAX, 1);
282 	if (progress == NULL) {
283 		err = got_error_from_errno("calloc");
284 		goto done;
285 	}
286 
287 	*pack_hash = calloc(1, sizeof(**pack_hash));
288 	if (*pack_hash == NULL) {
289 		err = got_error_from_errno("calloc");
290 		goto done;
291 	}
292 
293 	while (!done) {
294 		struct got_object_id *id = NULL;
295 		char *refname = NULL;
296 		char *server_progress = NULL;
297 		off_t packfile_size_cur = 0;
298 
299 		err = got_privsep_recv_fetch_progress(&done,
300 		    &id, &refname, symrefs, &server_progress,
301 		    &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
302 		if (err != NULL)
303 			goto done;
304 		/* Don't report size progress for an empty pack file. */
305 		if (packfile_size_cur <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
306 			packfile_size_cur = 0;
307 		if (!done && refname && id) {
308 			err = got_pathlist_insert(NULL, refs, refname, id);
309 			if (err)
310 				goto done;
311 		} else if (!done && server_progress) {
312 			char *p;
313 			/*
314 			 * XXX git-daemon tends to send batched output with
315 			 * lines spanning separate packets. Buffer progress
316 			 * output until we see a CR or LF to avoid giving
317 			 * partial lines of progress output to the callback.
318 			 */
319 			if (strlcat(progress, server_progress,
320 			    GOT_PKT_MAX) >= GOT_PKT_MAX) {
321 				progress[0] = '\0'; /* discard */
322 				continue;
323 			}
324 			while ((p = strchr(progress, '\r')) != NULL ||
325 			    (p = strchr(progress, '\n')) != NULL) {
326 				char *s;
327 				size_t n;
328 				char c = *p;
329 				*p = '\0';
330 				if (asprintf(&s, "%s%s", progress,
331 				    c == '\n' ? "\n" : "") == -1) {
332 					err = got_error_from_errno("asprintf");
333 					goto done;
334 				}
335 				err = progress_cb(progress_arg, s,
336 				    packfile_size_cur, 0, 0, 0, 0);
337 				free(s);
338 				if (err)
339 					break;
340 				n = strlen(progress);
341 				if (n < GOT_PKT_MAX - 1) {
342 					memmove(progress, &progress[n + 1],
343 					    GOT_PKT_MAX - n - 1);
344 				} else
345 					progress[0] = '\0';
346 			}
347 			free(server_progress);
348 			if (err)
349 				goto done;
350 		} else if (!done && packfile_size_cur != packfile_size) {
351 			err = progress_cb(progress_arg, NULL,
352 			    packfile_size_cur, 0, 0, 0, 0);
353 			if (err)
354 				break;
355 			packfile_size = packfile_size_cur;
356 		}
357 	}
358 	if (close(imsg_fetchfds[0]) == -1) {
359 		err = got_error_from_errno("close");
360 		goto done;
361 	}
362 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
363 		err = got_error_from_errno("waitpid");
364 		goto done;
365 	}
366 
367 	if (lseek(packfd, 0, SEEK_SET) == -1) {
368 		err = got_error_from_errno("lseek");
369 		goto done;
370 	}
371 
372 	/* If zero data was fetched without error we are already up-to-date. */
373 	if (packfile_size == 0) {
374 		free(*pack_hash);
375 		*pack_hash = NULL;
376 		goto done;
377 	} else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
378 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
379 		goto done;
380 	} else {
381 		ssize_t n;
382 
383 		n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
384 		if (n == -1) {
385 			err = got_error_from_errno("read");
386 			goto done;
387 		}
388 		if (n != ssizeof(pack_hdr)) {
389 			err = got_error(GOT_ERR_IO);
390 			goto done;
391 		}
392 		if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
393 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
394 			    "bad pack file signature");
395 			goto done;
396 		}
397 		if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
398 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
399 			    "bad pack file version");
400 			goto done;
401 		}
402 		nobj = be32toh(pack_hdr.nobjects);
403 		if (nobj == 0 &&
404 		    packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
405 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
406 			    "bad pack file with zero objects");
407 		if (nobj != 0 &&
408 		    packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
409 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
410 			    "empty pack file with non-zero object count");
411 	}
412 
413 	/*
414 	 * If the pack file contains no objects, we may only need to update
415 	 * references in our repository. The caller will take care of that.
416 	 */
417 	if (nobj == 0) {
418 		free(*pack_hash);
419 		*pack_hash = NULL;
420 		goto done;
421 	}
422 
423 	if (lseek(packfd, 0, SEEK_SET) == -1) {
424 		err = got_error_from_errno("lseek");
425 		goto done;
426 	}
427 
428 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
429 		err = got_error_from_errno("socketpair");
430 		goto done;
431 	}
432 	idxpid = fork();
433 	if (idxpid == -1) {
434 		err= got_error_from_errno("fork");
435 		goto done;
436 	} else if (idxpid == 0)
437 		got_privsep_exec_child(imsg_idxfds,
438 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
439 	if (close(imsg_idxfds[1]) == -1) {
440 		err = got_error_from_errno("close");
441 		goto done;
442 	}
443 	imsg_init(&idxibuf, imsg_idxfds[0]);
444 
445 	npackfd = dup(packfd);
446 	if (npackfd == -1) {
447 		err = got_error_from_errno("dup");
448 		goto done;
449 	}
450 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
451 	    npackfd);
452 	if (err != NULL)
453 		goto done;
454 	npackfd = -1;
455 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
456 	if (err != NULL)
457 		goto done;
458 	nidxfd = -1;
459 	for (i = 0; i < nitems(tmpfds); i++) {
460 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
461 		if (err != NULL)
462 			goto done;
463 		tmpfds[i] = -1;
464 	}
465 	done = 0;
466 	while (!done) {
467 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
468 
469 		err = got_privsep_recv_index_progress(&done, &nobj_total,
470 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
471 		    &idxibuf);
472 		if (err != NULL)
473 			goto done;
474 		if (nobj_indexed != 0) {
475 			err = progress_cb(progress_arg, NULL,
476 			    packfile_size, nobj_total,
477 			    nobj_indexed, nobj_loose, nobj_resolved);
478 			if (err)
479 				break;
480 		}
481 	}
482 	if (close(imsg_idxfds[0]) == -1) {
483 		err = got_error_from_errno("close");
484 		goto done;
485 	}
486 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
487 		err = got_error_from_errno("waitpid");
488 		goto done;
489 	}
490 
491 	err = got_object_id_str(&id_str, *pack_hash);
492 	if (err)
493 		goto done;
494 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
495 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
496 		err = got_error_from_errno("asprintf");
497 		goto done;
498 	}
499 
500 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
501 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
502 		err = got_error_from_errno("asprintf");
503 		goto done;
504 	}
505 	free(id_str);
506 	id_str = NULL;
507 
508 	if (rename(tmppackpath, packpath) == -1) {
509 		err = got_error_from_errno3("rename", tmppackpath, packpath);
510 		goto done;
511 	}
512 	free(tmppackpath);
513 	tmppackpath = NULL;
514 	if (rename(tmpidxpath, idxpath) == -1) {
515 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
516 		goto done;
517 	}
518 	free(tmpidxpath);
519 	tmpidxpath = NULL;
520 
521 done:
522 	if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
523 		err = got_error_from_errno2("unlink", tmppackpath);
524 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
525 		err = got_error_from_errno2("unlink", tmpidxpath);
526 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
527 		err = got_error_from_errno("close");
528 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
529 		err = got_error_from_errno("close");
530 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
531 		err = got_error_from_errno("close");
532 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
533 		err = got_error_from_errno("close");
534 	for (i = 0; i < nitems(tmpfds); i++) {
535 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
536 			err = got_error_from_errno("close");
537 	}
538 	free(tmppackpath);
539 	free(tmpidxpath);
540 	free(idxpath);
541 	free(id_str);
542 	free(packpath);
543 	free(progress);
544 
545 	got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
546 	got_ref_list_free(&my_refs);
547 	if (err) {
548 		free(*pack_hash);
549 		*pack_hash = NULL;
550 		got_pathlist_free(refs, GOT_PATHLIST_FREE_ALL);
551 		got_pathlist_free(symrefs, GOT_PATHLIST_FREE_ALL);
552 	}
553 	return err;
554 }