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