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