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 #include "got_compat.h"
17 
18 #include <sys/types.h>
19 #include <sys/stat.h>
20 #include <sys/queue.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 <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 <unistd.h>
35 #include <zlib.h>
36 #include <ctype.h>
37 #include <limits.h>
38 #include <time.h>
39 
40 #include "got_error.h"
41 #include "got_reference.h"
42 #include "got_repository.h"
43 #include "got_path.h"
44 #include "got_cancel.h"
45 #include "got_worktree.h"
46 #include "got_object.h"
47 #include "got_opentemp.h"
48 #include "got_fetch.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_hash.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_object_cache.h"
59 #include "got_lib_repository.h"
60 #include "got_lib_dial.h"
61 #include "got_lib_pkt.h"
62 
63 #ifndef nitems
64 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
65 #endif
66 
67 #ifndef ssizeof
68 #define ssizeof(_x) ((ssize_t)(sizeof(_x)))
69 #endif
70 
71 #ifndef MIN
72 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
73 #endif
74 
75 const struct got_error *
76 got_fetch_connect(pid_t *fetchpid, int *fetchfd, const char *proto,
77     const char *host, const char *port, const char *server_path, int verbosity)
78 {
79 	const struct got_error *err = NULL;
80 
81 	*fetchpid = -1;
82 	*fetchfd = -1;
83 
84 	if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
85 		err = got_dial_ssh(fetchpid, fetchfd, host, port,
86 		    server_path, GOT_DIAL_CMD_FETCH, verbosity);
87 	else if (strcmp(proto, "git") == 0)
88 		err = got_dial_git(fetchfd, host, port, server_path,
89 		    GOT_DIAL_CMD_FETCH);
90 	else if (strcmp(proto, "http") == 0 ||
91 	    strcmp(proto, "git+http") == 0 ||
92 	    strcmp(proto, "https") == 0 ||
93 	    strcmp(proto, "git+https") == 0)
94 		err = got_dial_http(fetchpid, fetchfd, host, port,
95 		    server_path, verbosity, strstr(proto, "https") != NULL);
96 	else
97 		err = got_error_path(proto, GOT_ERR_BAD_PROTO);
98 	return err;
99 }
100 
101 const struct got_error *
102 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
103     struct got_pathlist_head *symrefs, const char *remote_name,
104     int mirror_references, int fetch_all_branches,
105     struct got_pathlist_head *wanted_branches,
106     struct got_pathlist_head *wanted_refs, int list_refs_only, int verbosity,
107     int fetchfd, struct got_repository *repo, const char *worktree_refname,
108     const char *remote_head, int no_head, got_fetch_progress_cb progress_cb,
109     void *progress_arg)
110 {
111 	size_t i;
112 	int imsg_fetchfds[2], imsg_idxfds[2];
113 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
114 	int tmpfds[3];
115 	int fetchstatus, idxstatus, done = 0;
116 	const struct got_error *err;
117 	struct imsgbuf fetchibuf, idxibuf;
118 	pid_t fetchpid, idxpid;
119 	char *tmppackpath = NULL, *tmpidxpath = NULL;
120 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
121 	const char *repo_path = NULL;
122 	struct got_pathlist_head have_refs;
123 	struct got_pathlist_entry *pe;
124 	struct got_reflist_head my_refs;
125 	struct got_reflist_entry *re;
126 	off_t packfile_size = 0;
127 	struct got_packfile_hdr pack_hdr;
128 	uint32_t nobj = 0;
129 	char *path;
130 	char *progress = NULL;
131 
132 	*pack_hash = NULL;
133 
134 	/*
135 	 * Prevent fetching of references that won't make any
136 	 * sense outside of the remote repository's context.
137 	 */
138 	TAILQ_FOREACH(pe, wanted_refs, entry) {
139 		const char *refname = pe->path;
140 		if (strncmp(refname, "refs/got/", 9) == 0 ||
141 		    strncmp(refname, "got/", 4) == 0 ||
142 		    strncmp(refname, "refs/remotes/", 13) == 0 ||
143 		    strncmp(refname, "remotes/", 8) == 0)
144 			return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
145 	}
146 
147 	if (!list_refs_only)
148 		repo_path = got_repo_get_path_git_dir(repo);
149 
150 	for (i = 0; i < nitems(tmpfds); i++)
151 		tmpfds[i] = -1;
152 
153 	TAILQ_INIT(&have_refs);
154 	TAILQ_INIT(&my_refs);
155 
156 	if (!list_refs_only) {
157 		err = got_ref_list(&my_refs, repo, NULL,
158 		    got_ref_cmp_by_name, NULL);
159 		if (err)
160 			goto done;
161 	}
162 
163 	TAILQ_FOREACH(re, &my_refs, entry) {
164 		struct got_object_id *id;
165 		const char *refname;
166 
167 		if (got_ref_is_symbolic(re->ref))
168 			continue;
169 
170 		err = got_ref_resolve(&id, repo, re->ref);
171 		if (err)
172 			goto done;
173 		refname = strdup(got_ref_get_name(re->ref));
174 		if (refname == NULL) {
175 			err = got_error_from_errno("strdup");
176 			goto done;
177 		}
178 		err = got_pathlist_append(&have_refs, refname, id);
179 		if (err)
180 			goto done;
181 	}
182 
183 	if (list_refs_only) {
184 		packfd = got_opentempfd();
185 		if (packfd == -1) {
186 			err = got_error_from_errno("got_opentempfd");
187 			goto done;
188 		}
189 	} else {
190 		if (asprintf(&path, "%s/%s/fetching.pack",
191 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
192 			err = got_error_from_errno("asprintf");
193 			goto done;
194 		}
195 		err = got_opentemp_named_fd(&tmppackpath, &packfd, path, "");
196 		free(path);
197 		if (err)
198 			goto done;
199 		if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
200 			err = got_error_from_errno2("fchmod", tmppackpath);
201 			goto done;
202 		}
203 	}
204 	if (list_refs_only) {
205 		idxfd = got_opentempfd();
206 		if (idxfd == -1) {
207 			err = got_error_from_errno("got_opentempfd");
208 			goto done;
209 		}
210 	} else {
211 		if (asprintf(&path, "%s/%s/fetching.idx",
212 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
213 			err = got_error_from_errno("asprintf");
214 			goto done;
215 		}
216 		err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
217 		free(path);
218 		if (err)
219 			goto done;
220 		if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
221 			err = got_error_from_errno2("fchmod", tmpidxpath);
222 			goto done;
223 		}
224 	}
225 	nidxfd = dup(idxfd);
226 	if (nidxfd == -1) {
227 		err = got_error_from_errno("dup");
228 		goto done;
229 	}
230 
231 	for (i = 0; i < nitems(tmpfds); i++) {
232 		tmpfds[i] = got_opentempfd();
233 		if (tmpfds[i] == -1) {
234 			err = got_error_from_errno("got_opentempfd");
235 			goto done;
236 		}
237 	}
238 
239 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
240 		err = got_error_from_errno("socketpair");
241 		goto done;
242 	}
243 
244 	fetchpid = fork();
245 	if (fetchpid == -1) {
246 		err = got_error_from_errno("fork");
247 		goto done;
248 	} else if (fetchpid == 0){
249 		got_privsep_exec_child(imsg_fetchfds,
250 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
251 	}
252 
253 	if (close(imsg_fetchfds[1]) == -1) {
254 		err = got_error_from_errno("close");
255 		goto done;
256 	}
257 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
258 	nfetchfd = dup(fetchfd);
259 	if (nfetchfd == -1) {
260 		err = got_error_from_errno("dup");
261 		goto done;
262 	}
263 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
264 	    fetch_all_branches, wanted_branches, wanted_refs,
265 	    list_refs_only, worktree_refname, remote_head, no_head, verbosity);
266 	if (err != NULL)
267 		goto done;
268 	nfetchfd = -1;
269 	npackfd = dup(packfd);
270 	if (npackfd == -1) {
271 		err = got_error_from_errno("dup");
272 		goto done;
273 	}
274 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
275 	if (err != NULL)
276 		goto done;
277 	npackfd = -1;
278 
279 	packfile_size = 0;
280 	progress = calloc(GOT_PKT_MAX, 1);
281 	if (progress == NULL) {
282 		err = got_error_from_errno("calloc");
283 		goto done;
284 	}
285 
286 	*pack_hash = calloc(1, sizeof(**pack_hash));
287 	if (*pack_hash == NULL) {
288 		err = got_error_from_errno("calloc");
289 		goto done;
290 	}
291 
292 	while (!done) {
293 		struct got_object_id *id = NULL;
294 		char *refname = NULL;
295 		char *server_progress = NULL;
296 		off_t packfile_size_cur = 0;
297 
298 		err = got_privsep_recv_fetch_progress(&done,
299 		    &id, &refname, symrefs, &server_progress,
300 		    &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
301 		if (err != NULL)
302 			goto done;
303 		/* Don't report size progress for an empty pack file. */
304 		if (packfile_size_cur <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
305 			packfile_size_cur = 0;
306 		if (!done && refname && id) {
307 			err = got_pathlist_insert(NULL, refs, refname, id);
308 			if (err)
309 				goto done;
310 		} else if (!done && server_progress) {
311 			char *p;
312 			/*
313 			 * XXX git-daemon tends to send batched output with
314 			 * lines spanning separate packets. Buffer progress
315 			 * output until we see a CR or LF to avoid giving
316 			 * partial lines of progress output to the callback.
317 			 */
318 			if (strlcat(progress, server_progress,
319 			    GOT_PKT_MAX) >= GOT_PKT_MAX) {
320 				progress[0] = '\0'; /* discard */
321 				continue;
322 			}
323 			while ((p = strchr(progress, '\r')) != NULL ||
324 			    (p = strchr(progress, '\n')) != NULL) {
325 				char *s;
326 				size_t n;
327 				char c = *p;
328 				*p = '\0';
329 				if (asprintf(&s, "%s%s", progress,
330 				    c == '\n' ? "\n" : "") == -1) {
331 					err = got_error_from_errno("asprintf");
332 					goto done;
333 				}
334 				err = progress_cb(progress_arg, s,
335 				    packfile_size_cur, 0, 0, 0, 0);
336 				free(s);
337 				if (err)
338 					break;
339 				n = strlen(progress);
340 				if (n < GOT_PKT_MAX - 1) {
341 					memmove(progress, &progress[n + 1],
342 					    GOT_PKT_MAX - n - 1);
343 				} else
344 					progress[0] = '\0';
345 			}
346 			free(server_progress);
347 			if (err)
348 				goto done;
349 		} else if (!done && packfile_size_cur != packfile_size) {
350 			err = progress_cb(progress_arg, NULL,
351 			    packfile_size_cur, 0, 0, 0, 0);
352 			if (err)
353 				break;
354 			packfile_size = packfile_size_cur;
355 		}
356 	}
357 	if (close(imsg_fetchfds[0]) == -1) {
358 		err = got_error_from_errno("close");
359 		goto done;
360 	}
361 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
362 		err = got_error_from_errno("waitpid");
363 		goto done;
364 	}
365 
366 	if (lseek(packfd, 0, SEEK_SET) == -1) {
367 		err = got_error_from_errno("lseek");
368 		goto done;
369 	}
370 
371 	/* If zero data was fetched without error we are already up-to-date. */
372 	if (packfile_size == 0) {
373 		free(*pack_hash);
374 		*pack_hash = NULL;
375 		goto done;
376 	} else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
377 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
378 		goto done;
379 	} else {
380 		ssize_t n;
381 
382 		n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
383 		if (n == -1) {
384 			err = got_error_from_errno("read");
385 			goto done;
386 		}
387 		if (n != ssizeof(pack_hdr)) {
388 			err = got_error(GOT_ERR_IO);
389 			goto done;
390 		}
391 		if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
392 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
393 			    "bad pack file signature");
394 			goto done;
395 		}
396 		if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
397 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
398 			    "bad pack file version");
399 			goto done;
400 		}
401 		nobj = be32toh(pack_hdr.nobjects);
402 		if (nobj == 0 &&
403 		    packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
404 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
405 			    "bad pack file with zero objects");
406 			goto done;
407 		}
408 		if (nobj != 0 &&
409 		    packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
410 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
411 			    "empty pack file with non-zero object count");
412 			goto done;
413 		}
414 	}
415 
416 	/*
417 	 * If the pack file contains no objects, we may only need to update
418 	 * references in our repository. The caller will take care of that.
419 	 */
420 	if (nobj == 0) {
421 		free(*pack_hash);
422 		*pack_hash = NULL;
423 		goto done;
424 	}
425 
426 	if (lseek(packfd, 0, SEEK_SET) == -1) {
427 		err = got_error_from_errno("lseek");
428 		goto done;
429 	}
430 
431 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
432 		err = got_error_from_errno("socketpair");
433 		goto done;
434 	}
435 	idxpid = fork();
436 	if (idxpid == -1) {
437 		err= got_error_from_errno("fork");
438 		goto done;
439 	} else if (idxpid == 0)
440 		got_privsep_exec_child(imsg_idxfds,
441 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
442 	if (close(imsg_idxfds[1]) == -1) {
443 		err = got_error_from_errno("close");
444 		goto done;
445 	}
446 	imsg_init(&idxibuf, imsg_idxfds[0]);
447 
448 	npackfd = dup(packfd);
449 	if (npackfd == -1) {
450 		err = got_error_from_errno("dup");
451 		goto done;
452 	}
453 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
454 	    npackfd);
455 	if (err != NULL)
456 		goto done;
457 	npackfd = -1;
458 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
459 	if (err != NULL)
460 		goto done;
461 	nidxfd = -1;
462 	for (i = 0; i < nitems(tmpfds); i++) {
463 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
464 		if (err != NULL)
465 			goto done;
466 		tmpfds[i] = -1;
467 	}
468 	done = 0;
469 	while (!done) {
470 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
471 
472 		err = got_privsep_recv_index_progress(&done, &nobj_total,
473 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
474 		    &idxibuf);
475 		if (err != NULL)
476 			goto done;
477 		if (nobj_indexed != 0) {
478 			err = progress_cb(progress_arg, NULL,
479 			    packfile_size, nobj_total,
480 			    nobj_indexed, nobj_loose, nobj_resolved);
481 			if (err)
482 				break;
483 		}
484 	}
485 	if (close(imsg_idxfds[0]) == -1) {
486 		err = got_error_from_errno("close");
487 		goto done;
488 	}
489 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
490 		err = got_error_from_errno("waitpid");
491 		goto done;
492 	}
493 
494 	err = got_object_id_str(&id_str, *pack_hash);
495 	if (err)
496 		goto done;
497 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
498 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
499 		err = got_error_from_errno("asprintf");
500 		goto done;
501 	}
502 
503 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
504 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
505 		err = got_error_from_errno("asprintf");
506 		goto done;
507 	}
508 	free(id_str);
509 	id_str = NULL;
510 
511 	if (rename(tmppackpath, packpath) == -1) {
512 		err = got_error_from_errno3("rename", tmppackpath, packpath);
513 		goto done;
514 	}
515 	free(tmppackpath);
516 	tmppackpath = NULL;
517 	if (rename(tmpidxpath, idxpath) == -1) {
518 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
519 		goto done;
520 	}
521 	free(tmpidxpath);
522 	tmpidxpath = NULL;
523 
524 done:
525 	if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
526 		err = got_error_from_errno2("unlink", tmppackpath);
527 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
528 		err = got_error_from_errno2("unlink", tmpidxpath);
529 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
530 		err = got_error_from_errno("close");
531 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
532 		err = got_error_from_errno("close");
533 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
534 		err = got_error_from_errno("close");
535 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
536 		err = got_error_from_errno("close");
537 	for (i = 0; i < nitems(tmpfds); i++) {
538 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
539 			err = got_error_from_errno("close");
540 	}
541 	free(tmppackpath);
542 	free(tmpidxpath);
543 	free(idxpath);
544 	free(id_str);
545 	free(packpath);
546 	free(progress);
547 
548 	got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
549 	got_ref_list_free(&my_refs);
550 	if (err) {
551 		free(*pack_hash);
552 		*pack_hash = NULL;
553 		got_pathlist_free(refs, GOT_PATHLIST_FREE_ALL);
554 		got_pathlist_free(symrefs, GOT_PATHLIST_FREE_ALL);
555 	}
556 	return err;
557 }