1 /*
2  * Copyright (c) 2018, 2019 Ori Bernstein <ori@openbsd.org>
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  * Copyright (c) 2023 Josh Rickmar <jrick@zettaport.com>
5  *
6  * Permission to use, copy, modify, and distribute this software for any
7  * purpose with or without fee is hereby granted, provided that the above
8  * copyright notice and this permission notice appear in all copies.
9  *
10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17  */
18 
19 #include "got_compat.h"
20 
21 #include <sys/types.h>
22 #include <sys/stat.h>
23 #include <sys/queue.h>
24 #include <sys/uio.h>
25 #include <sys/socket.h>
26 #include <sys/wait.h>
27 #include <sys/resource.h>
28 #include <sys/socket.h>
29 
30 #include <errno.h>
31 #include <err.h>
32 #include <fcntl.h>
33 #include <stdio.h>
34 #include <stdlib.h>
35 #include <string.h>
36 #include <stdint.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <ctype.h>
40 #include <limits.h>
41 #include <time.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_send.h"
52 #include "got_repository_admin.h"
53 #include "got_commit_graph.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_ratelimit.h"
66 #include "got_lib_pack_create.h"
67 #include "got_lib_dial.h"
68 
69 #ifndef nitems
70 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
71 #endif
72 
73 #ifndef ssizeof
74 #define ssizeof(_x) ((ssize_t)(sizeof(_x)))
75 #endif
76 
77 #ifndef MIN
78 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
79 #endif
80 
81 const struct got_error *
82 got_send_connect(pid_t *sendpid, int *sendfd, const char *proto,
83     const char *host, const char *port, const char *server_path, int verbosity)
84 {
85 	const struct got_error *err = NULL;
86 
87 	*sendpid = -1;
88 	*sendfd = -1;
89 
90 	if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
91 		err = got_dial_ssh(sendpid, sendfd, host, port, server_path,
92 		    GOT_DIAL_CMD_SEND, verbosity);
93 	else if (strcmp(proto, "git") == 0)
94 		err = got_dial_git(sendfd, host, port, server_path,
95 		    GOT_DIAL_CMD_SEND);
96 	else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
97 		err = got_error_path(proto, GOT_ERR_NOT_IMPL);
98 	else
99 		err = got_error_path(proto, GOT_ERR_BAD_PROTO);
100 	return err;
101 }
102 
103 struct pack_progress_arg {
104     got_send_progress_cb progress_cb;
105     void *progress_arg;
106 
107     int ncolored;
108     int nfound;
109     int ntrees;
110     off_t packfile_size;
111     int ncommits;
112     int nobj_total;
113     int nobj_deltify;
114     int nobj_written;
115 };
116 
117 static const struct got_error *
118 pack_progress(void *arg, int ncolored, int nfound, int ntrees,
119     off_t packfile_size, int ncommits, int nobj_total, int nobj_deltify,
120     int nobj_written)
121 {
122 	const struct got_error *err;
123 	struct pack_progress_arg *a = arg;
124 
125 	err = a->progress_cb(a->progress_arg, ncolored, nfound, ntrees,
126 	    packfile_size, ncommits, nobj_total, nobj_deltify,
127 	    nobj_written, 0, NULL, NULL, 0);
128 	if (err)
129 		return err;
130 
131 	a->ncolored= ncolored;
132 	a->nfound = nfound;
133 	a->ntrees = ntrees;
134 	a->packfile_size = packfile_size;
135 	a->ncommits = ncommits;
136 	a->nobj_total = nobj_total;
137 	a->nobj_deltify = nobj_deltify;
138 	a->nobj_written = nobj_written;
139 	return NULL;
140 }
141 
142 static const struct got_error *
143 insert_sendable_ref(struct got_pathlist_head *refs, const char *refname,
144     struct got_repository *repo)
145 {
146 	const struct got_error *err;
147 	struct got_reference *ref;
148 	struct got_object_id *id = NULL;
149 	int obj_type;
150 
151 	err = got_ref_open(&ref, repo, refname, 0);
152 	if (err)
153 		return err;
154 
155 	if (got_ref_is_symbolic(ref)) {
156 		err = got_error_fmt(GOT_ERR_BAD_REF_TYPE,
157 		    "cannot send symbolic reference %s", refname);
158 		goto done;
159 	}
160 
161 	err = got_ref_resolve(&id, repo, ref);
162 	if (err)
163 		goto done;
164 	err = got_object_get_type(&obj_type, repo, id);
165 	if (err)
166 		goto done;
167 	switch (obj_type) {
168 	case GOT_OBJ_TYPE_COMMIT:
169 	case GOT_OBJ_TYPE_TAG:
170 		break;
171 	default:
172 		err = got_error_fmt(GOT_ERR_OBJ_TYPE," cannot send %s",
173 		    refname);
174 		goto done;
175 	}
176 
177 	err = got_pathlist_insert(NULL, refs, refname, id);
178 done:
179 	if (ref)
180 		got_ref_close(ref);
181 	if (err)
182 		free(id);
183 	return err;
184 }
185 
186 static const struct got_error *
187 check_common_ancestry(const char *refname, struct got_object_id *my_id,
188     struct got_object_id *their_id, struct got_repository *repo,
189     got_cancel_cb cancel_cb, void *cancel_arg)
190 {
191 	const struct got_error *err = NULL;
192 	struct got_object_id *yca_id;
193 	int obj_type;
194 
195 	err = got_object_get_type(&obj_type, repo, their_id);
196 	if (err)
197 		return err;
198 	if (obj_type != GOT_OBJ_TYPE_COMMIT)
199 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
200 		    "bad object type on server for %s", refname);
201 
202 	err = got_commit_graph_find_youngest_common_ancestor(&yca_id,
203 	    my_id, their_id, 0, repo, cancel_cb, cancel_arg);
204 	if (err)
205 		return err;
206 	if (yca_id == NULL)
207 		return got_error_fmt(GOT_ERR_SEND_ANCESTRY, "%s", refname);
208 
209 	if (got_object_id_cmp(their_id, yca_id) != 0)
210 		err = got_error_fmt(GOT_ERR_SEND_ANCESTRY, "%s", refname);
211 
212 	free(yca_id);
213 	return err;
214 }
215 
216 static const struct got_error *
217 realloc_ids(struct got_object_id ***ids, size_t *nalloc, size_t n)
218 {
219 	struct got_object_id **new;
220 	const size_t alloc_chunksz = 256;
221 
222 	if (*nalloc >= n)
223 		return NULL;
224 
225 	new = recallocarray(*ids, *nalloc, *nalloc + alloc_chunksz,
226 	    sizeof(struct got_object_id));
227 	if (new == NULL)
228 		return got_error_from_errno("recallocarray");
229 
230 	*ids = new;
231 	*nalloc += alloc_chunksz;
232 	return NULL;
233 }
234 
235 static struct got_pathlist_entry *
236 find_ref(struct got_pathlist_head *refs, const char *refname)
237 {
238 	struct got_pathlist_entry *pe;
239 
240 	TAILQ_FOREACH(pe, refs, entry) {
241 		if (got_path_cmp(pe->path, refname, strlen(pe->path),
242 		    strlen(refname)) == 0) {
243 			return pe;
244 		}
245 	}
246 
247 	return NULL;
248 }
249 
250 static const struct got_error *
251 get_remote_refname(char **remote_refname, const char *remote_name,
252     const char *refname)
253 {
254 	if (strncmp(refname, "refs/", 5) == 0)
255 		refname += 5;
256 	if (strncmp(refname, "heads/", 6) == 0)
257 		refname += 6;
258 
259 	if (asprintf(remote_refname, "refs/remotes/%s/%s",
260 	    remote_name, refname) == -1)
261 		return got_error_from_errno("asprintf");
262 
263 	return NULL;
264 }
265 
266 static const struct got_error *
267 update_remote_ref(struct got_pathlist_entry *my_ref, const char *remote_name,
268     struct got_repository *repo)
269 {
270 	const struct got_error *err, *unlock_err;
271 	const char *refname = my_ref->path;
272 	struct got_object_id *my_id = my_ref->data;
273 	struct got_reference *ref = NULL;
274 	char *remote_refname = NULL;
275 	int ref_locked = 0;
276 
277 	err = get_remote_refname(&remote_refname, remote_name, refname);
278 	if (err)
279 		goto done;
280 
281 	err = got_ref_open(&ref, repo, remote_refname, 1 /* lock */);
282 	if (err) {
283 		if (err->code != GOT_ERR_NOT_REF)
284 			goto done;
285 		err = got_ref_alloc(&ref, remote_refname, my_id);
286 		if (err)
287 			goto done;
288 	} else {
289 		ref_locked = 1;
290 		err = got_ref_change_ref(ref, my_id);
291 		if (err)
292 			goto done;
293 	}
294 
295 	err = got_ref_write(ref, repo);
296 done:
297 	if (ref) {
298 		if (ref_locked) {
299 			unlock_err = got_ref_unlock(ref);
300 			if (unlock_err && err == NULL)
301 				err = unlock_err;
302 		}
303 		got_ref_close(ref);
304 	}
305 	free(remote_refname);
306 	return err;
307 }
308 
309 const struct got_error*
310 got_send_pack(const char *remote_name, struct got_pathlist_head *branch_names,
311     struct got_pathlist_head *tag_names,
312     struct got_pathlist_head *delete_branches,
313     int verbosity, int overwrite_refs, int sendfd,
314     struct got_repository *repo, got_send_progress_cb progress_cb,
315     void *progress_arg, got_cancel_cb cancel_cb, void *cancel_arg)
316 {
317 	int imsg_sendfds[2];
318 	int npackfd = -1, nsendfd = -1;
319 	int sendstatus, done = 0;
320 	const struct got_error *err;
321 	struct imsgbuf sendibuf;
322 	pid_t sendpid = -1;
323 	struct got_pathlist_head have_refs;
324 	struct got_pathlist_head their_refs;
325 	struct got_pathlist_entry *pe;
326 	struct got_object_id **our_ids = NULL;
327 	struct got_object_id **their_ids = NULL;
328 	int nours = 0, ntheirs = 0;
329 	size_t nalloc_ours = 0, nalloc_theirs = 0;
330 	int refs_to_send = 0, refs_to_delete = 0;
331 	off_t bytes_sent = 0, bytes_sent_cur = 0;
332 	struct pack_progress_arg ppa;
333 	uint8_t packsha1[SHA1_DIGEST_LENGTH];
334 	int packfd = -1;
335 	FILE *delta_cache = NULL;
336 	char *s = NULL;
337 
338 	TAILQ_INIT(&have_refs);
339 	TAILQ_INIT(&their_refs);
340 
341 	TAILQ_FOREACH(pe, branch_names, entry) {
342 		const char *branchname = pe->path;
343 		if (strncmp(branchname, "refs/heads/", 11) != 0) {
344 			if (asprintf(&s, "refs/heads/%s", branchname) == -1) {
345 				err = got_error_from_errno("asprintf");
346 				goto done;
347 			}
348 		} else {
349 			if ((s = strdup(branchname)) == NULL) {
350 				err = got_error_from_errno("strdup");
351 				goto done;
352 			}
353 		}
354 		err = insert_sendable_ref(&have_refs, s, repo);
355 		if (err)
356 			goto done;
357 		s = NULL;
358 	}
359 
360 	TAILQ_FOREACH(pe, delete_branches, entry) {
361 		const char *branchname = pe->path;
362 		struct got_pathlist_entry *ref;
363 		if (strncmp(branchname, "refs/heads/", 11) != 0) {
364 			err = got_error_fmt(GOT_ERR_SEND_DELETE_REF, "%s",
365 			    branchname);
366 			goto done;
367 		}
368 		ref = find_ref(&have_refs, branchname);
369 		if (ref) {
370 			err = got_error_fmt(GOT_ERR_SEND_DELETE_REF,
371 			    "changes on %s will be sent to server",
372 			    branchname);
373 			goto done;
374 		}
375 	}
376 
377 	TAILQ_FOREACH(pe, tag_names, entry) {
378 		const char *tagname = pe->path;
379 		if (strncmp(tagname, "refs/tags/", 10) != 0) {
380 			if (asprintf(&s, "refs/tags/%s", tagname) == -1) {
381 				err = got_error_from_errno("asprintf");
382 				goto done;
383 			}
384 		} else {
385 			if ((s = strdup(pe->path)) == NULL) {
386 				err = got_error_from_errno("strdup");
387 				goto done;
388 			}
389 		}
390 		err = insert_sendable_ref(&have_refs, s, repo);
391 		if (err)
392 			goto done;
393 		s = NULL;
394 	}
395 
396 	if (TAILQ_EMPTY(&have_refs) && TAILQ_EMPTY(delete_branches)) {
397 		err = got_error(GOT_ERR_SEND_EMPTY);
398 		goto done;
399 	}
400 
401 	packfd = got_opentempfd();
402 	if (packfd == -1) {
403 		err = got_error_from_errno("got_opentempfd");
404 		goto done;
405 	}
406 
407 	delta_cache = got_opentemp();
408 	if (delta_cache == NULL) {
409 		err = got_error_from_errno("got_opentemp");
410 		goto done;
411 	}
412 
413 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_sendfds) == -1) {
414 		err = got_error_from_errno("socketpair");
415 		goto done;
416 	}
417 
418 	sendpid = fork();
419 	if (sendpid == -1) {
420 		err = got_error_from_errno("fork");
421 		goto done;
422 	} else if (sendpid == 0){
423 		got_privsep_exec_child(imsg_sendfds,
424 		    GOT_PATH_PROG_SEND_PACK, got_repo_get_path(repo));
425 	}
426 
427 	if (close(imsg_sendfds[1]) == -1) {
428 		err = got_error_from_errno("close");
429 		goto done;
430 	}
431 	imsg_init(&sendibuf, imsg_sendfds[0]);
432 	nsendfd = dup(sendfd);
433 	if (nsendfd == -1) {
434 		err = got_error_from_errno("dup");
435 		goto done;
436 	}
437 
438 	/*
439 	 * Prepare the array of our object IDs which
440 	 * will be needed for generating a pack file.
441 	 */
442 	TAILQ_FOREACH(pe, &have_refs, entry) {
443 		struct got_object_id *id = pe->data;
444 
445 		err = realloc_ids(&our_ids, &nalloc_ours, nours + 1);
446 		if (err)
447 			goto done;
448 		our_ids[nours] = id;
449 		nours++;
450 	}
451 
452 	err = got_privsep_send_send_req(&sendibuf, nsendfd, &have_refs,
453 	    delete_branches, verbosity);
454 	if (err)
455 		goto done;
456 	nsendfd = -1;
457 
458 	err = got_privsep_recv_send_remote_refs(&their_refs, &sendibuf);
459 	if (err)
460 		goto done;
461 
462 	/*
463 	 * Process references reported by the server.
464 	 * Push appropriate object IDs onto the "their IDs" array.
465 	 * This array will be used to exclude objects which already
466 	 * exist on the server from our pack file.
467 	 */
468 	TAILQ_FOREACH(pe, &their_refs, entry) {
469 		const char *refname = pe->path;
470 		struct got_object_id *their_id = pe->data;
471 		int have_their_id;
472 		struct got_object *obj;
473 		struct got_pathlist_entry *my_ref = NULL;
474 		int is_tag = 0;
475 
476 		/* Don't blindly trust the server to send us valid names. */
477 		if (!got_ref_name_is_valid(refname))
478 			continue;
479 
480 		if (strncmp(refname, "refs/tags/", 10) == 0)
481 			is_tag = 1;
482 		/*
483 		 * Find out whether this is a reference we want to upload.
484 		 * Otherwise we can still use this reference as a hint to
485 		 * avoid uploading any objects the server already has.
486 		 */
487 		my_ref = find_ref(&have_refs, refname);
488 		if (my_ref) {
489 			struct got_object_id *my_id = my_ref->data;
490 			if (got_object_id_cmp(my_id, their_id) != 0) {
491 				if (!overwrite_refs && is_tag) {
492 					err = got_error_fmt(
493 					    GOT_ERR_SEND_TAG_EXISTS,
494 					    "%s", refname);
495 					goto done;
496 				}
497 				refs_to_send++;
498 			}
499 		}
500 
501 		/* Check if their object exists locally. */
502 		err = got_object_open(&obj, repo, their_id);
503 		if (err) {
504 			if (err->code != GOT_ERR_NO_OBJ)
505 				goto done;
506 			if (!overwrite_refs && my_ref != NULL) {
507 				err = got_error_fmt(GOT_ERR_SEND_ANCESTRY,
508 				    "%s", refname);
509 				goto done;
510 			}
511 			have_their_id = 0;
512 		} else {
513 			got_object_close(obj);
514 			have_their_id = 1;
515 		}
516 
517 		err = realloc_ids(&their_ids, &nalloc_theirs, ntheirs + 1);
518 		if (err)
519 			goto done;
520 
521 		if (have_their_id) {
522 			/* Enforce linear ancestry if required. */
523 			if (!overwrite_refs && my_ref && !is_tag) {
524 				struct got_object_id *my_id = my_ref->data;
525 				err = check_common_ancestry(refname, my_id,
526 				    their_id, repo, cancel_cb, cancel_arg);
527 				if (err)
528 					goto done;
529 			}
530 			/* Exclude any objects reachable via their ID. */
531 			their_ids[ntheirs] = their_id;
532 			ntheirs++;
533 		} else if (!is_tag) {
534 			char *remote_refname;
535 			struct got_reference *ref;
536 			/*
537 			 * Exclude any objects which exist on the server
538 			 * according to a locally cached remote reference.
539 			 */
540 			err = get_remote_refname(&remote_refname,
541 			    remote_name, refname);
542 			if (err)
543 				goto done;
544 			err = got_ref_open(&ref, repo, remote_refname, 0);
545 			free(remote_refname);
546 			if (err) {
547 				if (err->code != GOT_ERR_NOT_REF)
548 					goto done;
549 			} else {
550 				err = got_ref_resolve(&their_ids[ntheirs],
551 				    repo, ref);
552 				got_ref_close(ref);
553 				if (err)
554 					goto done;
555 				ntheirs++;
556 			}
557 		}
558 	}
559 
560 	/* Account for any new references we are going to upload. */
561 	TAILQ_FOREACH(pe, &have_refs, entry) {
562 		const char *refname = pe->path;
563 		if (find_ref(&their_refs, refname) == NULL)
564 			refs_to_send++;
565 	}
566 
567 	/* Account for any existing references we are going to delete. */
568 	TAILQ_FOREACH(pe, delete_branches, entry) {
569 		const char *branchname = pe->path;
570 		if (find_ref(&their_refs, branchname))
571 			refs_to_delete++;
572 	}
573 
574 	if (refs_to_send == 0 && refs_to_delete == 0) {
575 		got_privsep_send_stop(imsg_sendfds[0]);
576 		goto done;
577 	}
578 
579 	if (refs_to_send > 0) {
580 		struct got_ratelimit rl;
581 		got_ratelimit_init(&rl, 0, 500);
582 		memset(&ppa, 0, sizeof(ppa));
583 		ppa.progress_cb = progress_cb;
584 		ppa.progress_arg = progress_arg;
585 		err = got_pack_create(packsha1, packfd, delta_cache,
586 		    their_ids, ntheirs, our_ids, nours, repo, 0, 1, 0,
587 		    pack_progress, &ppa, &rl, cancel_cb, cancel_arg);
588 		if (err)
589 			goto done;
590 
591 		npackfd = dup(packfd);
592 		if (npackfd == -1) {
593 			err = got_error_from_errno("dup");
594 			goto done;
595 		}
596 		err = got_privsep_send_packfd(&sendibuf, npackfd);
597 		if (err != NULL)
598 			goto done;
599 		npackfd = -1;
600 	} else {
601 		err = got_privsep_send_packfd(&sendibuf, -1);
602 		if (err != NULL)
603 			goto done;
604 	}
605 
606 	while (!done) {
607 		int success = 0;
608 		char *refname = NULL;
609 		char *errmsg = NULL;
610 
611 		if (cancel_cb) {
612 			err = (*cancel_cb)(cancel_arg);
613 			if (err)
614 				goto done;
615 		}
616 		err = got_privsep_recv_send_progress(&done, &bytes_sent,
617 		    &success, &refname, &errmsg, &sendibuf);
618 		if (err)
619 			goto done;
620 		if (refname && got_ref_name_is_valid(refname) && success &&
621 		    strncmp(refname, "refs/tags/", 10) != 0) {
622 			struct got_pathlist_entry *my_ref;
623 			/*
624 			 * The server has accepted our changes.
625 			 * Update our reference in refs/remotes/ accordingly.
626 			 */
627 			my_ref = find_ref(&have_refs, refname);
628 			if (my_ref) {
629 				err = update_remote_ref(my_ref, remote_name,
630 				    repo);
631 				if (err)
632 					goto done;
633 			}
634 		}
635 		if (refname != NULL ||
636 		    bytes_sent_cur != bytes_sent) {
637 			err = progress_cb(progress_arg, ppa.ncolored,
638 			    ppa.nfound, ppa.ntrees, ppa.packfile_size,
639 			    ppa.ncommits, ppa.nobj_total, ppa.nobj_deltify,
640 			    ppa.nobj_written, bytes_sent,
641 			    refname, errmsg, success);
642 			if (err) {
643 				free(refname);
644 				free(errmsg);
645 				goto done;
646 			}
647 			bytes_sent_cur = bytes_sent;
648 		}
649 		free(refname);
650 		free(errmsg);
651 	}
652 done:
653 	if (sendpid != -1) {
654 		if (err)
655 			got_privsep_send_stop(imsg_sendfds[0]);
656 		if (waitpid(sendpid, &sendstatus, 0) == -1 && err == NULL)
657 			err = got_error_from_errno("waitpid");
658 	}
659 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
660 		err = got_error_from_errno("close");
661 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
662 		err = got_error_from_errno("fclose");
663 	if (nsendfd != -1 && close(nsendfd) == -1 && err == NULL)
664 		err = got_error_from_errno("close");
665 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
666 		err = got_error_from_errno("close");
667 
668 	got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
669 	got_pathlist_free(&their_refs, GOT_PATHLIST_FREE_ALL);
670 	/*
671 	 * Object ids are owned by have_refs/their_refs and are already freed;
672 	 * Only the arrays must be freed.
673 	 */
674 	free(our_ids);
675 	free(their_ids);
676 	free(s);
677 	return err;
678 }