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