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/uio.h>
20 #include <sys/socket.h>
21 #include <sys/wait.h>
22 #include <sys/resource.h>
23 #include <sys/socket.h>
24 
25 #include <errno.h>
26 #include <err.h>
27 #include <fcntl.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <stdint.h>
32 #include <unistd.h>
33 #include <zlib.h>
34 #include <ctype.h>
35 #include <limits.h>
36 #include <time.h>
37 #include <uuid.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,
103     got_fetch_progress_cb progress_cb, void *progress_arg)
104 {
105 	size_t i;
106 	int imsg_fetchfds[2], imsg_idxfds[2];
107 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
108 	int tmpfds[3];
109 	int fetchstatus, idxstatus, done = 0;
110 	const struct got_error *err;
111 	struct imsgbuf fetchibuf, idxibuf;
112 	pid_t fetchpid, idxpid;
113 	char *tmppackpath = NULL, *tmpidxpath = NULL;
114 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
115 	const char *repo_path = NULL;
116 	struct got_pathlist_head have_refs;
117 	struct got_pathlist_entry *pe;
118 	struct got_reflist_head my_refs;
119 	struct got_reflist_entry *re;
120 	off_t packfile_size = 0;
121 	struct got_packfile_hdr pack_hdr;
122 	uint32_t nobj = 0;
123 	char *path;
124 	char *progress = NULL;
125 
126 	*pack_hash = NULL;
127 
128 	/*
129 	 * Prevent fetching of references that won't make any
130 	 * sense outside of the remote repository's context.
131 	 */
132 	TAILQ_FOREACH(pe, wanted_refs, entry) {
133 		const char *refname = pe->path;
134 		if (strncmp(refname, "refs/got/", 9) == 0 ||
135 		    strncmp(refname, "got/", 4) == 0 ||
136 		    strncmp(refname, "refs/remotes/", 13) == 0 ||
137 		    strncmp(refname, "remotes/", 8) == 0)
138 			return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
139 	}
140 
141 	if (!list_refs_only)
142 		repo_path = got_repo_get_path_git_dir(repo);
143 
144 	for (i = 0; i < nitems(tmpfds); i++)
145 		tmpfds[i] = -1;
146 
147 	TAILQ_INIT(&have_refs);
148 	TAILQ_INIT(&my_refs);
149 
150 	if (!list_refs_only) {
151 		err = got_ref_list(&my_refs, repo, NULL,
152 		    got_ref_cmp_by_name, NULL);
153 		if (err)
154 			goto done;
155 	}
156 
157 	TAILQ_FOREACH(re, &my_refs, entry) {
158 		struct got_object_id *id;
159 		const char *refname;
160 
161 		if (got_ref_is_symbolic(re->ref))
162 			continue;
163 
164 		err = got_ref_resolve(&id, repo, re->ref);
165 		if (err)
166 			goto done;
167 		refname = strdup(got_ref_get_name(re->ref));
168 		if (refname == NULL) {
169 			err = got_error_from_errno("strdup");
170 			goto done;
171 		}
172 		err = got_pathlist_append(&have_refs, refname, id);
173 		if (err)
174 			goto done;
175 	}
176 
177 	if (list_refs_only) {
178 		packfd = got_opentempfd();
179 		if (packfd == -1) {
180 			err = got_error_from_errno("got_opentempfd");
181 			goto done;
182 		}
183 	} else {
184 		if (asprintf(&path, "%s/%s/fetching.pack",
185 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
186 			err = got_error_from_errno("asprintf");
187 			goto done;
188 		}
189 		err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
190 		free(path);
191 		if (err)
192 			goto done;
193 		if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
194 			err = got_error_from_errno2("fchmod", tmppackpath);
195 			goto done;
196 		}
197 	}
198 	if (list_refs_only) {
199 		idxfd = got_opentempfd();
200 		if (idxfd == -1) {
201 			err = got_error_from_errno("got_opentempfd");
202 			goto done;
203 		}
204 	} else {
205 		if (asprintf(&path, "%s/%s/fetching.idx",
206 		    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
207 			err = got_error_from_errno("asprintf");
208 			goto done;
209 		}
210 		err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
211 		free(path);
212 		if (err)
213 			goto done;
214 		if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
215 			err = got_error_from_errno2("fchmod", tmpidxpath);
216 			goto done;
217 		}
218 	}
219 	nidxfd = dup(idxfd);
220 	if (nidxfd == -1) {
221 		err = got_error_from_errno("dup");
222 		goto done;
223 	}
224 
225 	for (i = 0; i < nitems(tmpfds); i++) {
226 		tmpfds[i] = got_opentempfd();
227 		if (tmpfds[i] == -1) {
228 			err = got_error_from_errno("got_opentempfd");
229 			goto done;
230 		}
231 	}
232 
233 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
234 		err = got_error_from_errno("socketpair");
235 		goto done;
236 	}
237 
238 	fetchpid = fork();
239 	if (fetchpid == -1) {
240 		err = got_error_from_errno("fork");
241 		goto done;
242 	} else if (fetchpid == 0){
243 		got_privsep_exec_child(imsg_fetchfds,
244 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
245 	}
246 
247 	if (close(imsg_fetchfds[1]) == -1) {
248 		err = got_error_from_errno("close");
249 		goto done;
250 	}
251 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
252 	nfetchfd = dup(fetchfd);
253 	if (nfetchfd == -1) {
254 		err = got_error_from_errno("dup");
255 		goto done;
256 	}
257 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
258 	    fetch_all_branches, wanted_branches, wanted_refs,
259 	    list_refs_only, verbosity);
260 	if (err != NULL)
261 		goto done;
262 	nfetchfd = -1;
263 	npackfd = dup(packfd);
264 	if (npackfd == -1) {
265 		err = got_error_from_errno("dup");
266 		goto done;
267 	}
268 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
269 	if (err != NULL)
270 		goto done;
271 	npackfd = -1;
272 
273 	packfile_size = 0;
274 	progress = calloc(GOT_PKT_MAX, 1);
275 	if (progress == NULL) {
276 		err = got_error_from_errno("calloc");
277 		goto done;
278 	}
279 
280 	*pack_hash = calloc(1, sizeof(**pack_hash));
281 	if (*pack_hash == NULL) {
282 		err = got_error_from_errno("calloc");
283 		goto done;
284 	}
285 
286 	while (!done) {
287 		struct got_object_id *id = NULL;
288 		char *refname = NULL;
289 		char *server_progress = NULL;
290 		off_t packfile_size_cur = 0;
291 
292 		err = got_privsep_recv_fetch_progress(&done,
293 		    &id, &refname, symrefs, &server_progress,
294 		    &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
295 		if (err != NULL)
296 			goto done;
297 		if (!done && refname && id) {
298 			err = got_pathlist_insert(NULL, refs, refname, id);
299 			if (err)
300 				goto done;
301 		} else if (!done && server_progress) {
302 			char *p;
303 			/*
304 			 * XXX git-daemon tends to send batched output with
305 			 * lines spanning separate packets. Buffer progress
306 			 * output until we see a CR or LF to avoid giving
307 			 * partial lines of progress output to the callback.
308 			 */
309 			if (strlcat(progress, server_progress,
310 			    GOT_PKT_MAX) >= GOT_PKT_MAX) {
311 				progress[0] = '\0'; /* discard */
312 				continue;
313 			}
314 			while ((p = strchr(progress, '\r')) != NULL ||
315 			    (p = strchr(progress, '\n')) != NULL) {
316 				char *s;
317 				size_t n;
318 				char c = *p;
319 				*p = '\0';
320 				if (asprintf(&s, "%s%s", progress,
321 				    c == '\n' ? "\n" : "") == -1) {
322 					err = got_error_from_errno("asprintf");
323 					goto done;
324 				}
325 				err = progress_cb(progress_arg, s,
326 				    packfile_size_cur, 0, 0, 0, 0);
327 				free(s);
328 				if (err)
329 					break;
330 				n = strlen(progress);
331 				if (n < GOT_PKT_MAX - 1) {
332 					memmove(progress, &progress[n + 1],
333 					    GOT_PKT_MAX - n - 1);
334 				} else
335 					progress[0] = '\0';
336 			}
337 			free(server_progress);
338 			if (err)
339 				goto done;
340 		} else if (!done && packfile_size_cur != packfile_size) {
341 			err = progress_cb(progress_arg, NULL,
342 			    packfile_size_cur, 0, 0, 0, 0);
343 			if (err)
344 				break;
345 			packfile_size = packfile_size_cur;
346 		}
347 	}
348 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
349 		err = got_error_from_errno("waitpid");
350 		goto done;
351 	}
352 
353 	if (lseek(packfd, 0, SEEK_SET) == -1) {
354 		err = got_error_from_errno("lseek");
355 		goto done;
356 	}
357 
358 	/* If zero data was fetched without error we are already up-to-date. */
359 	if (packfile_size == 0) {
360 		free(*pack_hash);
361 		*pack_hash = NULL;
362 		goto done;
363 	} else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
364 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
365 		goto done;
366 	} else {
367 		ssize_t n;
368 
369 		n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
370 		if (n == -1) {
371 			err = got_error_from_errno("read");
372 			goto done;
373 		}
374 		if (n != ssizeof(pack_hdr)) {
375 			err = got_error(GOT_ERR_IO);
376 			goto done;
377 		}
378 		if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
379 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
380 			    "bad pack file signature");
381 			goto done;
382 		}
383 		if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
384 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
385 			    "bad pack file version");
386 			goto done;
387 		}
388 		nobj = be32toh(pack_hdr.nobjects);
389 		if (nobj == 0 &&
390 		    packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
391 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
392 			    "bad pack file with zero objects");
393 		if (nobj != 0 &&
394 		    packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
395 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
396 			    "empty pack file with non-zero object count");
397 	}
398 
399 	/*
400 	 * If the pack file contains no objects, we may only need to update
401 	 * references in our repository. The caller will take care of that.
402 	 */
403 	if (nobj == 0)
404 		goto done;
405 
406 	if (lseek(packfd, 0, SEEK_SET) == -1) {
407 		err = got_error_from_errno("lseek");
408 		goto done;
409 	}
410  
411 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
412 		err = got_error_from_errno("socketpair");
413 		goto done;
414 	}
415 	idxpid = fork();
416 	if (idxpid == -1) {
417 		err= got_error_from_errno("fork");
418 		goto done;
419 	} else if (idxpid == 0)
420 		got_privsep_exec_child(imsg_idxfds,
421 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
422 	if (close(imsg_idxfds[1]) == -1) {
423 		err = got_error_from_errno("close");
424 		goto done;
425 	}
426 	imsg_init(&idxibuf, imsg_idxfds[0]);
427 
428 	npackfd = dup(packfd);
429 	if (npackfd == -1) {
430 		err = got_error_from_errno("dup");
431 		goto done;
432 	}
433 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
434 	    npackfd);
435 	if (err != NULL)
436 		goto done;
437 	npackfd = -1;
438 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
439 	if (err != NULL)
440 		goto done;
441 	nidxfd = -1;
442 	for (i = 0; i < nitems(tmpfds); i++) {
443 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
444 		if (err != NULL)
445 			goto done;
446 		tmpfds[i] = -1;
447 	}
448 	done = 0;
449 	while (!done) {
450 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
451 
452 		err = got_privsep_recv_index_progress(&done, &nobj_total,
453 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
454 		    &idxibuf);
455 		if (err != NULL)
456 			goto done;
457 		if (nobj_indexed != 0) {
458 			err = progress_cb(progress_arg, NULL,
459 			    packfile_size, nobj_total,
460 			    nobj_indexed, nobj_loose, nobj_resolved);
461 			if (err)
462 				break;
463 		}
464 		imsg_clear(&idxibuf);
465 	}
466 	if (close(imsg_idxfds[0]) == -1) {
467 		err = got_error_from_errno("close");
468 		goto done;
469 	}
470 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
471 		err = got_error_from_errno("waitpid");
472 		goto done;
473 	}
474 
475 	err = got_object_id_str(&id_str, *pack_hash);
476 	if (err)
477 		goto done;
478 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
479 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
480 		err = got_error_from_errno("asprintf");
481 		goto done;
482 	}
483 
484 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
485 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
486 		err = got_error_from_errno("asprintf");
487 		goto done;
488 	}
489 
490 	if (rename(tmppackpath, packpath) == -1) {
491 		err = got_error_from_errno3("rename", tmppackpath, packpath);
492 		goto done;
493 	}
494 	free(tmppackpath);
495 	tmppackpath = NULL;
496 	if (rename(tmpidxpath, idxpath) == -1) {
497 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
498 		goto done;
499 	}
500 	free(tmpidxpath);
501 	tmpidxpath = NULL;
502 
503 done:
504 	if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
505 		err = got_error_from_errno2("unlink", tmppackpath);
506 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
507 		err = got_error_from_errno2("unlink", tmpidxpath);
508 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
509 		err = got_error_from_errno("close");
510 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
511 		err = got_error_from_errno("close");
512 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
513 		err = got_error_from_errno("close");
514 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
515 		err = got_error_from_errno("close");
516 	for (i = 0; i < nitems(tmpfds); i++) {
517 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
518 			err = got_error_from_errno("close");
519 	}
520 	free(tmppackpath);
521 	free(tmpidxpath);
522 	free(idxpath);
523 	free(packpath);
524 	free(progress);
525 
526 	TAILQ_FOREACH(pe, &have_refs, entry) {
527 		free((char *)pe->path);
528 		free(pe->data);
529 	}
530 	got_pathlist_free(&have_refs);
531 	got_ref_list_free(&my_refs);
532 	if (err) {
533 		free(*pack_hash);
534 		*pack_hash = NULL;
535 		TAILQ_FOREACH(pe, refs, entry) {
536 			free((void *)pe->path);
537 			free(pe->data);
538 		}
539 		got_pathlist_free(refs);
540 		TAILQ_FOREACH(pe, symrefs, entry) {
541 			free((void *)pe->path);
542 			free(pe->data);
543 		}
544 		got_pathlist_free(symrefs);
545 	}
546 	return err;
547 }