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/uio.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 #include <sys/resource.h>
24 #include <sys/socket.h>
25 
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 <sha2.h>
35 #include <unistd.h>
36 #include <zlib.h>
37 #include <ctype.h>
38 #include <limits.h>
39 #include <time.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, const char *worktree_refname,
105     const char *remote_head, int no_head, got_fetch_progress_cb progress_cb,
106     void *progress_arg)
107 {
108 	size_t i;
109 	int imsg_fetchfds[2], imsg_idxfds[2];
110 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
111 	int tmpfds[3];
112 	int fetchstatus, idxstatus, done = 0;
113 	const struct got_error *err;
114 	struct imsgbuf fetchibuf, idxibuf;
115 	pid_t fetchpid, idxpid;
116 	char *tmppackpath = NULL, *tmpidxpath = NULL;
117 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
118 	const char *repo_path = NULL;
119 	struct got_pathlist_head have_refs;
120 	struct got_pathlist_entry *pe;
121 	struct got_reflist_head my_refs;
122 	struct got_reflist_entry *re;
123 	off_t packfile_size = 0;
124 	struct got_packfile_hdr pack_hdr;
125 	uint32_t nobj = 0;
126 	char *path;
127 	char *progress = NULL;
128 
129 	*pack_hash = NULL;
130 
131 	/*
132 	 * Prevent fetching of references that won't make any
133 	 * sense outside of the remote repository's context.
134 	 */
135 	TAILQ_FOREACH(pe, wanted_refs, entry) {
136 		const char *refname = pe->path;
137 		if (strncmp(refname, "refs/got/", 9) == 0 ||
138 		    strncmp(refname, "got/", 4) == 0 ||
139 		    strncmp(refname, "refs/remotes/", 13) == 0 ||
140 		    strncmp(refname, "remotes/", 8) == 0)
141 			return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
142 	}
143 
144 	if (!list_refs_only)
145 		repo_path = got_repo_get_path_git_dir(repo);
146 
147 	for (i = 0; i < nitems(tmpfds); i++)
148 		tmpfds[i] = -1;
149 
150 	TAILQ_INIT(&have_refs);
151 	TAILQ_INIT(&my_refs);
152 
153 	if (!list_refs_only) {
154 		err = got_ref_list(&my_refs, repo, NULL,
155 		    got_ref_cmp_by_name, NULL);
156 		if (err)
157 			goto done;
158 	}
159 
160 	TAILQ_FOREACH(re, &my_refs, entry) {
161 		struct got_object_id *id;
162 		const char *refname;
163 
164 		if (got_ref_is_symbolic(re->ref))
165 			continue;
166 
167 		err = got_ref_resolve(&id, repo, re->ref);
168 		if (err)
169 			goto done;
170 		refname = strdup(got_ref_get_name(re->ref));
171 		if (refname == NULL) {
172 			err = got_error_from_errno("strdup");
173 			goto done;
174 		}
175 		err = got_pathlist_append(&have_refs, refname, id);
176 		if (err)
177 			goto done;
178 	}
179 
180 	if (list_refs_only) {
181 		packfd = got_opentempfd();
182 		if (packfd == -1) {
183 			err = got_error_from_errno("got_opentempfd");
184 			goto done;
185 		}
186 	} else {
187 		if (asprintf(&path, "%s/%s/fetching.pack",
188 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
189 			err = got_error_from_errno("asprintf");
190 			goto done;
191 		}
192 		err = got_opentemp_named_fd(&tmppackpath, &packfd, path, "");
193 		free(path);
194 		if (err)
195 			goto done;
196 		if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
197 			err = got_error_from_errno2("fchmod", tmppackpath);
198 			goto done;
199 		}
200 	}
201 	if (list_refs_only) {
202 		idxfd = got_opentempfd();
203 		if (idxfd == -1) {
204 			err = got_error_from_errno("got_opentempfd");
205 			goto done;
206 		}
207 	} else {
208 		if (asprintf(&path, "%s/%s/fetching.idx",
209 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
210 			err = got_error_from_errno("asprintf");
211 			goto done;
212 		}
213 		err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
214 		free(path);
215 		if (err)
216 			goto done;
217 		if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
218 			err = got_error_from_errno2("fchmod", tmpidxpath);
219 			goto done;
220 		}
221 	}
222 	nidxfd = dup(idxfd);
223 	if (nidxfd == -1) {
224 		err = got_error_from_errno("dup");
225 		goto done;
226 	}
227 
228 	for (i = 0; i < nitems(tmpfds); i++) {
229 		tmpfds[i] = got_opentempfd();
230 		if (tmpfds[i] == -1) {
231 			err = got_error_from_errno("got_opentempfd");
232 			goto done;
233 		}
234 	}
235 
236 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
237 		err = got_error_from_errno("socketpair");
238 		goto done;
239 	}
240 
241 	fetchpid = fork();
242 	if (fetchpid == -1) {
243 		err = got_error_from_errno("fork");
244 		goto done;
245 	} else if (fetchpid == 0){
246 		got_privsep_exec_child(imsg_fetchfds,
247 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
248 	}
249 
250 	if (close(imsg_fetchfds[1]) == -1) {
251 		err = got_error_from_errno("close");
252 		goto done;
253 	}
254 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
255 	nfetchfd = dup(fetchfd);
256 	if (nfetchfd == -1) {
257 		err = got_error_from_errno("dup");
258 		goto done;
259 	}
260 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
261 	    fetch_all_branches, wanted_branches, wanted_refs,
262 	    list_refs_only, worktree_refname, remote_head, no_head, verbosity);
263 	if (err != NULL)
264 		goto done;
265 	nfetchfd = -1;
266 	npackfd = dup(packfd);
267 	if (npackfd == -1) {
268 		err = got_error_from_errno("dup");
269 		goto done;
270 	}
271 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
272 	if (err != NULL)
273 		goto done;
274 	npackfd = -1;
275 
276 	packfile_size = 0;
277 	progress = calloc(GOT_PKT_MAX, 1);
278 	if (progress == NULL) {
279 		err = got_error_from_errno("calloc");
280 		goto done;
281 	}
282 
283 	*pack_hash = calloc(1, sizeof(**pack_hash));
284 	if (*pack_hash == NULL) {
285 		err = got_error_from_errno("calloc");
286 		goto done;
287 	}
288 
289 	while (!done) {
290 		struct got_object_id *id = NULL;
291 		char *refname = NULL;
292 		char *server_progress = NULL;
293 		off_t packfile_size_cur = 0;
294 
295 		err = got_privsep_recv_fetch_progress(&done,
296 		    &id, &refname, symrefs, &server_progress,
297 		    &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
298 		if (err != NULL)
299 			goto done;
300 		/* Don't report size progress for an empty pack file. */
301 		if (packfile_size_cur <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
302 			packfile_size_cur = 0;
303 		if (!done && refname && id) {
304 			err = got_pathlist_insert(NULL, refs, refname, id);
305 			if (err)
306 				goto done;
307 		} else if (!done && server_progress) {
308 			char *p;
309 			/*
310 			 * XXX git-daemon tends to send batched output with
311 			 * lines spanning separate packets. Buffer progress
312 			 * output until we see a CR or LF to avoid giving
313 			 * partial lines of progress output to the callback.
314 			 */
315 			if (strlcat(progress, server_progress,
316 			    GOT_PKT_MAX) >= GOT_PKT_MAX) {
317 				progress[0] = '\0'; /* discard */
318 				continue;
319 			}
320 			while ((p = strchr(progress, '\r')) != NULL ||
321 			    (p = strchr(progress, '\n')) != NULL) {
322 				char *s;
323 				size_t n;
324 				char c = *p;
325 				*p = '\0';
326 				if (asprintf(&s, "%s%s", progress,
327 				    c == '\n' ? "\n" : "") == -1) {
328 					err = got_error_from_errno("asprintf");
329 					goto done;
330 				}
331 				err = progress_cb(progress_arg, s,
332 				    packfile_size_cur, 0, 0, 0, 0);
333 				free(s);
334 				if (err)
335 					break;
336 				n = strlen(progress);
337 				if (n < GOT_PKT_MAX - 1) {
338 					memmove(progress, &progress[n + 1],
339 					    GOT_PKT_MAX - n - 1);
340 				} else
341 					progress[0] = '\0';
342 			}
343 			free(server_progress);
344 			if (err)
345 				goto done;
346 		} else if (!done && packfile_size_cur != packfile_size) {
347 			err = progress_cb(progress_arg, NULL,
348 			    packfile_size_cur, 0, 0, 0, 0);
349 			if (err)
350 				break;
351 			packfile_size = packfile_size_cur;
352 		}
353 	}
354 	if (close(imsg_fetchfds[0]) == -1) {
355 		err = got_error_from_errno("close");
356 		goto done;
357 	}
358 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
359 		err = got_error_from_errno("waitpid");
360 		goto done;
361 	}
362 
363 	if (lseek(packfd, 0, SEEK_SET) == -1) {
364 		err = got_error_from_errno("lseek");
365 		goto done;
366 	}
367 
368 	/* If zero data was fetched without error we are already up-to-date. */
369 	if (packfile_size == 0) {
370 		free(*pack_hash);
371 		*pack_hash = NULL;
372 		goto done;
373 	} else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
374 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
375 		goto done;
376 	} else {
377 		ssize_t n;
378 
379 		n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
380 		if (n == -1) {
381 			err = got_error_from_errno("read");
382 			goto done;
383 		}
384 		if (n != ssizeof(pack_hdr)) {
385 			err = got_error(GOT_ERR_IO);
386 			goto done;
387 		}
388 		if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
389 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
390 			    "bad pack file signature");
391 			goto done;
392 		}
393 		if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
394 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
395 			    "bad pack file version");
396 			goto done;
397 		}
398 		nobj = be32toh(pack_hdr.nobjects);
399 		if (nobj == 0 &&
400 		    packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
401 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
402 			    "bad pack file with zero objects");
403 		if (nobj != 0 &&
404 		    packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
405 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
406 			    "empty pack file with non-zero object count");
407 	}
408 
409 	/*
410 	 * If the pack file contains no objects, we may only need to update
411 	 * references in our repository. The caller will take care of that.
412 	 */
413 	if (nobj == 0) {
414 		free(*pack_hash);
415 		*pack_hash = NULL;
416 		goto done;
417 	}
418 
419 	if (lseek(packfd, 0, SEEK_SET) == -1) {
420 		err = got_error_from_errno("lseek");
421 		goto done;
422 	}
423 
424 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
425 		err = got_error_from_errno("socketpair");
426 		goto done;
427 	}
428 	idxpid = fork();
429 	if (idxpid == -1) {
430 		err= got_error_from_errno("fork");
431 		goto done;
432 	} else if (idxpid == 0)
433 		got_privsep_exec_child(imsg_idxfds,
434 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
435 	if (close(imsg_idxfds[1]) == -1) {
436 		err = got_error_from_errno("close");
437 		goto done;
438 	}
439 	imsg_init(&idxibuf, imsg_idxfds[0]);
440 
441 	npackfd = dup(packfd);
442 	if (npackfd == -1) {
443 		err = got_error_from_errno("dup");
444 		goto done;
445 	}
446 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
447 	    npackfd);
448 	if (err != NULL)
449 		goto done;
450 	npackfd = -1;
451 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
452 	if (err != NULL)
453 		goto done;
454 	nidxfd = -1;
455 	for (i = 0; i < nitems(tmpfds); i++) {
456 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
457 		if (err != NULL)
458 			goto done;
459 		tmpfds[i] = -1;
460 	}
461 	done = 0;
462 	while (!done) {
463 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
464 
465 		err = got_privsep_recv_index_progress(&done, &nobj_total,
466 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
467 		    &idxibuf);
468 		if (err != NULL)
469 			goto done;
470 		if (nobj_indexed != 0) {
471 			err = progress_cb(progress_arg, NULL,
472 			    packfile_size, nobj_total,
473 			    nobj_indexed, nobj_loose, nobj_resolved);
474 			if (err)
475 				break;
476 		}
477 	}
478 	if (close(imsg_idxfds[0]) == -1) {
479 		err = got_error_from_errno("close");
480 		goto done;
481 	}
482 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
483 		err = got_error_from_errno("waitpid");
484 		goto done;
485 	}
486 
487 	err = got_object_id_str(&id_str, *pack_hash);
488 	if (err)
489 		goto done;
490 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
491 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
492 		err = got_error_from_errno("asprintf");
493 		goto done;
494 	}
495 
496 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
497 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
498 		err = got_error_from_errno("asprintf");
499 		goto done;
500 	}
501 	free(id_str);
502 	id_str = NULL;
503 
504 	if (rename(tmppackpath, packpath) == -1) {
505 		err = got_error_from_errno3("rename", tmppackpath, packpath);
506 		goto done;
507 	}
508 	free(tmppackpath);
509 	tmppackpath = NULL;
510 	if (rename(tmpidxpath, idxpath) == -1) {
511 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
512 		goto done;
513 	}
514 	free(tmpidxpath);
515 	tmpidxpath = NULL;
516 
517 done:
518 	if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
519 		err = got_error_from_errno2("unlink", tmppackpath);
520 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
521 		err = got_error_from_errno2("unlink", tmpidxpath);
522 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
523 		err = got_error_from_errno("close");
524 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
525 		err = got_error_from_errno("close");
526 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
527 		err = got_error_from_errno("close");
528 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
529 		err = got_error_from_errno("close");
530 	for (i = 0; i < nitems(tmpfds); i++) {
531 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
532 			err = got_error_from_errno("close");
533 	}
534 	free(tmppackpath);
535 	free(tmpidxpath);
536 	free(idxpath);
537 	free(id_str);
538 	free(packpath);
539 	free(progress);
540 
541 	got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
542 	got_ref_list_free(&my_refs);
543 	if (err) {
544 		free(*pack_hash);
545 		*pack_hash = NULL;
546 		got_pathlist_free(refs, GOT_PATHLIST_FREE_ALL);
547 		got_pathlist_free(symrefs, GOT_PATHLIST_FREE_ALL);
548 	}
549 	return err;
550 }