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/uio.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 #include <sys/syslimits.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 <sha1.h>
35 #include <zlib.h>
36 #include <ctype.h>
37 #include <limits.h>
38 #include <imsg.h>
39 #include <time.h>
40 #include <uuid.h>
41 #include <netdb.h>
42 #include <netinet/in.h>
43 
44 #include "got_error.h"
45 #include "got_reference.h"
46 #include "got_repository.h"
47 #include "got_path.h"
48 #include "got_cancel.h"
49 #include "got_worktree.h"
50 #include "got_object.h"
51 #include "got_opentemp.h"
52 #include "got_fetch.h"
53 
54 #include "got_lib_delta.h"
55 #include "got_lib_inflate.h"
56 #include "got_lib_object.h"
57 #include "got_lib_object_parse.h"
58 #include "got_lib_object_create.h"
59 #include "got_lib_pack.h"
60 #include "got_lib_sha1.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_object_cache.h"
63 #include "got_lib_repository.h"
64 
65 #ifndef nitems
66 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
67 #endif
68 
69 #ifndef MIN
70 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
71 #endif
72 
73 static int
74 hassuffix(char *base, char *suf)
75 {
76 	int nb, ns;
77 
78 	nb = strlen(base);
79 	ns = strlen(suf);
80 	if (ns <= nb && strcmp(base + (nb - ns), suf) == 0)
81 		return 1;
82 	return 0;
83 }
84 
85 static const struct got_error *
86 dial_ssh(int *fetchfd, const char *host, const char *port, const char *path,
87     const char *direction, int verbosity)
88 {
89 	const struct got_error *error = NULL;
90 	int pid, pfd[2];
91 	char cmd[64];
92 	char *argv[9];
93 	int i = 0;
94 
95 	*fetchfd = -1;
96 
97 	argv[0] = GOT_FETCH_PATH_SSH;
98 	if (verbosity == -1) {
99 		argv[1 + i++] = "-q";
100 	} else {
101 		/* ssh(1) allows up to 3 "-v" options. */
102 		for (i = 0; i < MIN(3, verbosity); i++)
103 			argv[1 + i] = "-v";
104 	}
105 	argv[1 + i] = "--";
106 	argv[2 + i] = (char *)host;
107 	argv[3 + i] = (char *)cmd;
108 	argv[4 + i] = (char *)path;
109 	argv[5 + i] = NULL;
110 
111 	if (pipe(pfd) == -1)
112 		return got_error_from_errno("pipe");
113 
114 	pid = fork();
115 	if (pid == -1) {
116 		error = got_error_from_errno("fork");
117 		close(pfd[0]);
118 		close(pfd[1]);
119 		return error;
120 	} else if (pid == 0) {
121 		int n;
122 		close(pfd[1]);
123 		dup2(pfd[0], 0);
124 		dup2(pfd[0], 1);
125 		n = snprintf(cmd, sizeof(cmd), "git-%s-pack", direction);
126 		if (n < 0 || n >= sizeof(cmd))
127 			err(1, "snprintf");
128 		if (execv(GOT_FETCH_PATH_SSH, argv) == -1)
129 			err(1, "execl");
130 		abort(); /* not reached */
131 	} else {
132 		close(pfd[0]);
133 		*fetchfd = pfd[1];
134 		return NULL;
135 	}
136 }
137 
138 static const struct got_error *
139 dial_git(int *fetchfd, const char *host, const char *port, const char *path,
140     const char *direction)
141 {
142 	const struct got_error *err = NULL;
143 	struct addrinfo hints, *servinfo, *p;
144 	char *cmd = NULL, *pkt = NULL;
145 	int fd = -1, totlen, r, eaicode;
146 
147 	*fetchfd = -1;
148 
149 	memset(&hints, 0, sizeof hints);
150 	hints.ai_family = AF_UNSPEC;
151 	hints.ai_socktype = SOCK_STREAM;
152 	eaicode = getaddrinfo(host, port, &hints, &servinfo);
153 	if (eaicode) {
154 		char msg[512];
155 		snprintf(msg, sizeof(msg), "%s: %s", host,
156 		    gai_strerror(eaicode));
157 		return got_error_msg(GOT_ERR_ADDRINFO, msg);
158 	}
159 
160 	for (p = servinfo; p != NULL; p = p->ai_next) {
161 		if ((fd = socket(p->ai_family, p->ai_socktype,
162 		    p->ai_protocol)) == -1)
163 			continue;
164 		if (connect(fd, p->ai_addr, p->ai_addrlen) == 0)
165 			break;
166 		err = got_error_from_errno("connect");
167 		close(fd);
168 	}
169 	if (p == NULL)
170 		goto done;
171 
172 	if (asprintf(&cmd, "git-%s-pack %s", direction, path) == -1) {
173 		err = got_error_from_errno("asprintf");
174 		goto done;
175 	}
176 	totlen = 4 + strlen(cmd) + 1 + strlen("host=") + strlen(host) + 1;
177 	if (asprintf(&pkt, "%04x%s", totlen, cmd) == -1) {
178 		err = got_error_from_errno("asprintf");
179 		goto done;
180 	}
181 	r = write(fd, pkt, strlen(pkt) + 1);
182 	if (r == -1) {
183 		err = got_error_from_errno("write");
184 		goto done;
185 	}
186 	if (asprintf(&pkt, "host=%s", host) == -1) {
187 		err = got_error_from_errno("asprintf");
188 		goto done;
189 	}
190 	r = write(fd, pkt, strlen(pkt) + 1);
191 	if (r == -1) {
192 		err = got_error_from_errno("write");
193 		goto done;
194 	}
195 done:
196 	free(cmd);
197 	free(pkt);
198 	if (err) {
199 		if (fd != -1)
200 			close(fd);
201 	} else
202 		*fetchfd = fd;
203 	return err;
204 }
205 
206 const struct got_error *
207 got_fetch_connect(int *fetchfd, const char *proto, const char *host,
208     const char *port, const char *server_path, int verbosity)
209 {
210 	const struct got_error *err = NULL;
211 
212 	*fetchfd = -1;
213 
214 	if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
215 		err = dial_ssh(fetchfd, host, port, server_path, "upload",
216 		    verbosity);
217 	else if (strcmp(proto, "git") == 0)
218 		err = dial_git(fetchfd, host, port, server_path, "upload");
219 	else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
220 		err = got_error_path(proto, GOT_ERR_NOT_IMPL);
221 	else
222 		err = got_error_path(proto, GOT_ERR_BAD_PROTO);
223 	return err;
224 }
225 
226 const struct got_error *
227 got_fetch_parse_uri(char **proto, char **host, char **port,
228     char **server_path, char **repo_name, const char *uri)
229 {
230 	const struct got_error *err = NULL;
231 	char *s, *p, *q;
232 	int n;
233 
234 	*proto = *host = *port = *server_path = *repo_name = NULL;
235 
236 	p = strstr(uri, "://");
237 	if (!p) {
238 		/* Try parsing Git's "scp" style URL syntax. */
239 		*proto = strdup("ssh");
240 		if (proto == NULL) {
241 			err = got_error_from_errno("strdup");
242 			goto done;
243 		}
244 		*port = strdup("22");
245 		if (*port == NULL) {
246 			err = got_error_from_errno("strdup");
247 			goto done;
248 		}
249 		s = (char *)uri;
250 		q = strchr(s, ':');
251 		if (q == NULL) {
252 			err = got_error(GOT_ERR_PARSE_URI);
253 			goto done;
254 		}
255 		/* No slashes allowed before first colon. */
256 		p = strchr(s, '/');
257 		if (p && q > p) {
258 			err = got_error(GOT_ERR_PARSE_URI);
259 			goto done;
260 		}
261 		*host = strndup(s, q - s);
262 		if (*host == NULL) {
263 			err = got_error_from_errno("strndup");
264 			goto done;
265 		}
266 		p = q + 1;
267 	} else {
268 		*proto = strndup(uri, p - uri);
269 		if (proto == NULL) {
270 			err = got_error_from_errno("strndup");
271 			goto done;
272 		}
273 		s = p + 3;
274 
275 		p = strstr(s, "/");
276 		if (p == NULL || strlen(p) == 1) {
277 			err = got_error(GOT_ERR_PARSE_URI);
278 			goto done;
279 		}
280 
281 		q = memchr(s, ':', p - s);
282 		if (q) {
283 			*host = strndup(s, q - s);
284 			if (*host == NULL) {
285 				err = got_error_from_errno("strndup");
286 				goto done;
287 			}
288 			*port = strndup(q + 1, p - (q + 1));
289 			if (*port == NULL) {
290 				err = got_error_from_errno("strndup");
291 				goto done;
292 			}
293 		} else {
294 			*host = strndup(s, p - s);
295 			if (*host == NULL) {
296 				err = got_error_from_errno("strndup");
297 				goto done;
298 			}
299 			if (asprintf(port, "%u", GOT_DEFAULT_GIT_PORT) == -1) {
300 				err = got_error_from_errno("asprintf");
301 				goto done;
302 			}
303 		}
304 	}
305 
306 	*server_path = strdup(p);
307 	if (*server_path == NULL) {
308 		err = got_error_from_errno("strdup");
309 		goto done;
310 	}
311 
312 	p = strrchr(p, '/');
313 	if (!p || strlen(p) <= 1) {
314 		err = got_error(GOT_ERR_PARSE_URI);
315 		goto done;
316 	}
317 	p++;
318 	n = strlen(p);
319 	if (n == 0) {
320 		err = got_error(GOT_ERR_PARSE_URI);
321 		goto done;
322 	}
323 	if (hassuffix(p, ".git"))
324 		n -= 4;
325 	*repo_name = strndup(p, (p + n) - p);
326 	if (*repo_name == NULL) {
327 		err = got_error_from_errno("strndup");
328 		goto done;
329 	}
330 done:
331 	if (err) {
332 		free(*proto);
333 		*proto = NULL;
334 		free(*host);
335 		*host = NULL;
336 		free(*port);
337 		*port = NULL;
338 		free(*server_path);
339 		*server_path = NULL;
340 		free(*repo_name);
341 		*repo_name = NULL;
342 	}
343 	return err;
344 }
345 
346 static const struct got_error *
347 check_pack_hash(int fd, size_t sz, uint8_t *hcomp)
348 {
349 	SHA1_CTX ctx;
350 	uint8_t hexpect[SHA1_DIGEST_LENGTH];
351 	uint8_t buf[32 * 1024];
352 	ssize_t n, r, nr;
353 
354 	if (sz < sizeof(struct got_packfile_hdr) + SHA1_DIGEST_LENGTH)
355 		return got_error_msg(GOT_ERR_BAD_PACKFILE, "short packfile");
356 
357 	n = 0;
358 	SHA1Init(&ctx);
359 	while (n < sz - 20) {
360 		nr = sizeof(buf);
361 		if (sz - n - 20 < sizeof(buf))
362 			nr = sz - n - 20;
363 		r = read(fd, buf, nr);
364 		if (r == -1)
365 			return got_error_from_errno("read");
366 		if (r != nr)
367 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
368 			    "short pack file");
369 		SHA1Update(&ctx, buf, nr);
370 		n += r;
371 	}
372 	SHA1Final(hcomp, &ctx);
373 
374 	r = read(fd, hexpect, sizeof(hexpect));
375 	if (r == -1)
376 		return got_error_from_errno("read");
377 	if (r != sizeof(hexpect))
378 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
379 		    "short pack file");
380 
381 	if (memcmp(hcomp, hexpect, SHA1_DIGEST_LENGTH) != 0)
382 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
383 		    "packfile checksum mismatch");
384 
385 	return NULL;
386 }
387 
388 const struct got_error*
389 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
390     struct got_pathlist_head *symrefs, int fetchfd, struct got_repository *repo,
391     got_fetch_progress_cb progress_cb, void *progress_arg)
392 {
393 	int imsg_fetchfds[2], imsg_idxfds[2];
394 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
395 	int tmpfds[3], i;
396 	int fetchstatus, idxstatus, done = 0;
397 	const struct got_error *err;
398 	struct imsgbuf fetchibuf, idxibuf;
399 	pid_t fetchpid, idxpid;
400 	char *tmppackpath = NULL, *tmpidxpath = NULL;
401 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
402 	const char *repo_path = got_repo_get_path(repo);
403 	struct got_pathlist_head have_refs;
404 	struct got_pathlist_entry *pe;
405 	off_t packfile_size = 0;
406 	char *path;
407 
408 	*pack_hash = NULL;
409 	for (i = 0; i < nitems(tmpfds); i++)
410 		tmpfds[i] = -1;
411 
412 	TAILQ_INIT(&have_refs);
413 
414 	if (asprintf(&path, "%s/%s/fetching.pack",
415 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
416 		err = got_error_from_errno("asprintf");
417 		goto done;
418 	}
419 	err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
420 	free(path);
421 	if (err)
422 		goto done;
423 	npackfd = dup(packfd);
424 	if (npackfd == -1) {
425 		err = got_error_from_errno("dup");
426 		goto done;
427 	}
428 	if (asprintf(&path, "%s/%s/fetching.idx",
429 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
430 		err = got_error_from_errno("asprintf");
431 		goto done;
432 	}
433 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
434 	free(path);
435 	if (err)
436 		goto done;
437 	nidxfd = dup(idxfd);
438 	if (nidxfd == -1) {
439 		err = got_error_from_errno("dup");
440 		goto done;
441 	}
442 
443 	for (i = 0; i < nitems(tmpfds); i++) {
444 		tmpfds[i] = got_opentempfd();
445 		if (tmpfds[i] == -1) {
446 			err = got_error_from_errno("got_opentempfd");
447 			goto done;
448 		}
449 	}
450 
451 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
452 		err = got_error_from_errno("socketpair");
453 		goto done;
454 	}
455 
456 	fetchpid = fork();
457 	if (fetchpid == -1) {
458 		err = got_error_from_errno("fork");
459 		goto done;
460 	} else if (fetchpid == 0){
461 		got_privsep_exec_child(imsg_fetchfds,
462 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
463 	}
464 
465 	if (close(imsg_fetchfds[1]) != 0) {
466 		err = got_error_from_errno("close");
467 		goto done;
468 	}
469 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
470 	nfetchfd = dup(fetchfd);
471 	if (nfetchfd == -1) {
472 		err = got_error_from_errno("dup");
473 		goto done;
474 	}
475 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs);
476 	if (err != NULL)
477 		goto done;
478 	nfetchfd = -1;
479 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
480 	if (err != NULL)
481 		goto done;
482 	npackfd = dup(packfd);
483 	if (npackfd == -1) {
484 		err = got_error_from_errno("dup");
485 		goto done;
486 	}
487 
488 	packfile_size = 0;
489 	while (!done) {
490 		struct got_object_id *id = NULL;
491 		char *refname = NULL;
492 		char *server_progress = NULL;
493 		off_t packfile_size_cur;
494 
495 		err = got_privsep_recv_fetch_progress(&done,
496 		    &id, &refname, symrefs, &server_progress,
497 		    &packfile_size_cur, &fetchibuf);
498 		if (err != NULL)
499 			goto done;
500 		if (done)
501 			*pack_hash = id;
502 		else if (refname && id) {
503 			err = got_pathlist_append(refs, refname, id);
504 			if (err)
505 				goto done;
506 		} else if (server_progress) {
507 			char *s, *s0 = server_progress;
508 			while ((s = strsep(&s0, "\r")) != NULL) {
509 				if (*s == '\0')
510 					continue;
511 				err = progress_cb(progress_arg, s,
512 				    packfile_size_cur, 0, 0, 0, 0);
513 				if (err)
514 					break;
515 			}
516 			free(server_progress);
517 			if (err)
518 				goto done;
519 		} else if (packfile_size_cur != packfile_size) {
520 			err = progress_cb(progress_arg, NULL,
521 			    packfile_size_cur, 0, 0, 0, 0);
522 			if (err)
523 				break;
524 			packfile_size = packfile_size_cur;
525 		}
526 	}
527 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
528 		err = got_error_from_errno("waitpid");
529 		goto done;
530 	}
531 
532 	if (lseek(packfd, 0, SEEK_SET) == -1) {
533 		err = got_error_from_errno("lseek");
534 		goto done;
535 	}
536 	err = check_pack_hash(packfd, packfile_size, (*pack_hash)->sha1);
537 	if (err)
538 		goto done;
539  
540 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
541 		err = got_error_from_errno("socketpair");
542 		goto done;
543 	}
544 	idxpid = fork();
545 	if (idxpid == -1) {
546 		err= got_error_from_errno("fork");
547 		goto done;
548 	} else if (idxpid == 0)
549 		got_privsep_exec_child(imsg_idxfds,
550 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
551 	if (close(imsg_idxfds[1]) != 0) {
552 		err = got_error_from_errno("close");
553 		goto done;
554 	}
555 	imsg_init(&idxibuf, imsg_idxfds[0]);
556 
557 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
558 	    npackfd);
559 	if (err != NULL)
560 		goto done;
561 	npackfd = -1;
562 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
563 	if (err != NULL)
564 		goto done;
565 	nidxfd = -1;
566 	for (i = 0; i < nitems(tmpfds); i++) {
567 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
568 		if (err != NULL)
569 			goto done;
570 		tmpfds[i] = -1;
571 	}
572 	done = 0;
573 	while (!done) {
574 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
575 
576 		err = got_privsep_recv_index_progress(&done, &nobj_total,
577 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
578 		    &idxibuf);
579 		if (err != NULL)
580 			goto done;
581 		if (nobj_indexed != 0) {
582 			err = progress_cb(progress_arg, NULL,
583 			    packfile_size, nobj_total,
584 			    nobj_indexed, nobj_loose, nobj_resolved);
585 			if (err)
586 				break;
587 		}
588 		imsg_clear(&idxibuf);
589 	}
590 	if (close(imsg_idxfds[0]) == -1) {
591 		err = got_error_from_errno("close");
592 		goto done;
593 	}
594 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
595 		err = got_error_from_errno("waitpid");
596 		goto done;
597 	}
598 
599 	err = got_object_id_str(&id_str, *pack_hash);
600 	if (err)
601 		goto done;
602 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
603 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
604 		err = got_error_from_errno("asprintf");
605 		goto done;
606 	}
607 
608 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
609 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
610 		err = got_error_from_errno("asprintf");
611 		goto done;
612 	}
613 
614 	if (rename(tmppackpath, packpath) == -1) {
615 		err = got_error_from_errno3("rename", tmppackpath, packpath);
616 		goto done;
617 	}
618 	if (rename(tmpidxpath, idxpath) == -1) {
619 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
620 		goto done;
621 	}
622 
623 done:
624 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
625 		err = got_error_from_errno("close");
626 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
627 		err = got_error_from_errno("close");
628 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
629 		err = got_error_from_errno("close");
630 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
631 		err = got_error_from_errno("close");
632 	for (i = 0; i < nitems(tmpfds); i++) {
633 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
634 			err = got_error_from_errno("close");
635 	}
636 	free(tmppackpath);
637 	free(tmpidxpath);
638 	free(idxpath);
639 	free(packpath);
640 
641 	if (err) {
642 		free(*pack_hash);
643 		*pack_hash = NULL;
644 		TAILQ_FOREACH(pe, refs, entry) {
645 			free((void *)pe->path);
646 			free(pe->data);
647 		}
648 		got_pathlist_free(refs);
649 		TAILQ_FOREACH(pe, symrefs, entry) {
650 			free((void *)pe->path);
651 			free(pe->data);
652 		}
653 		got_pathlist_free(symrefs);
654 	}
655 	return err;
656 }