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