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, hasport;
233 
234 	*proto = *host = *port = *server_path = *repo_name = NULL;
235 
236 	p = strstr(uri, "://");
237 	if (!p) {
238 		return got_error(GOT_ERR_PARSE_URI);
239 	}
240 	*proto = strndup(uri, p - uri);
241 	if (proto == NULL) {
242 		err = got_error_from_errno("strndup");
243 		goto done;
244 	}
245 
246 	hasport = (strcmp(*proto, "git") == 0 ||
247 	    strstr(*proto, "http") == *proto);
248 	s = p + 3;
249 	p = NULL;
250 	if (!hasport) {
251 		p = strstr(s, ":");
252 		if (p != NULL)
253 			p++;
254 	}
255 	if (p == NULL)
256 		p = strstr(s, "/");
257 	if (p == NULL || strlen(p) == 1) {
258 		err = got_error(GOT_ERR_PARSE_URI);
259 		goto done;
260 	}
261 
262 	q = memchr(s, ':', p - s);
263 	if (q) {
264 		*host = strndup(s, q - s);
265 		if (*host == NULL) {
266 			err = got_error_from_errno("strndup");
267 			goto done;
268 		}
269 		*port = strndup(q + 1, p - (q + 1));
270 		if (*port == NULL) {
271 			err = got_error_from_errno("strndup");
272 			goto done;
273 		}
274 	} else {
275 		*host = strndup(s, p - s);
276 		if (*host == NULL) {
277 			err = got_error_from_errno("strndup");
278 			goto done;
279 		}
280 		if (asprintf(port, "%u", GOT_DEFAULT_GIT_PORT) == -1) {
281 			err = got_error_from_errno("asprintf");
282 			goto done;
283 		}
284 	}
285 
286 	*server_path = strdup(p);
287 	if (*server_path == NULL) {
288 		err = got_error_from_errno("strdup");
289 		goto done;
290 	}
291 
292 	p = strrchr(p, '/') + 1;
293 	if (!p || strlen(p) == 0) {
294 		//werrstr("missing repository in uri");
295 		err = got_error(GOT_ERR_PARSE_URI);
296 		goto done;
297 	}
298 	n = strlen(p);
299 	if (hassuffix(p, ".git"))
300 		n -= 4;
301 	*repo_name = strndup(p, (p + n) - p);
302 	if (*repo_name == NULL) {
303 		err = got_error_from_errno("strndup");
304 		goto done;
305 	}
306 done:
307 	if (err) {
308 		free(*proto);
309 		*proto = NULL;
310 		free(*host);
311 		*host = NULL;
312 		free(*port);
313 		*port = NULL;
314 		free(*server_path);
315 		*server_path = NULL;
316 		free(*repo_name);
317 		*repo_name = NULL;
318 	}
319 	return err;
320 }
321 
322 static const struct got_error *
323 check_pack_hash(int fd, size_t sz, uint8_t *hcomp)
324 {
325 	SHA1_CTX ctx;
326 	uint8_t hexpect[SHA1_DIGEST_LENGTH];
327 	uint8_t buf[32 * 1024];
328 	ssize_t n, r, nr;
329 
330 	if (sz < sizeof(struct got_packfile_hdr) + SHA1_DIGEST_LENGTH)
331 		return got_error_msg(GOT_ERR_BAD_PACKFILE, "short packfile");
332 
333 	n = 0;
334 	SHA1Init(&ctx);
335 	while (n < sz - 20) {
336 		nr = sizeof(buf);
337 		if (sz - n - 20 < sizeof(buf))
338 			nr = sz - n - 20;
339 		r = read(fd, buf, nr);
340 		if (r == -1)
341 			return got_error_from_errno("read");
342 		if (r != nr)
343 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
344 			    "short pack file");
345 		SHA1Update(&ctx, buf, nr);
346 		n += r;
347 	}
348 	SHA1Final(hcomp, &ctx);
349 
350 	r = read(fd, hexpect, sizeof(hexpect));
351 	if (r == -1)
352 		return got_error_from_errno("read");
353 	if (r != sizeof(hexpect))
354 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
355 		    "short pack file");
356 
357 	if (memcmp(hcomp, hexpect, SHA1_DIGEST_LENGTH) != 0)
358 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
359 		    "packfile checksum mismatch");
360 
361 	return NULL;
362 }
363 
364 const struct got_error*
365 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
366     struct got_pathlist_head *symrefs, int fetchfd, struct got_repository *repo,
367     got_fetch_progress_cb progress_cb, void *progress_arg)
368 {
369 	int imsg_fetchfds[2], imsg_idxfds[2];
370 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
371 	int tmpfds[3], i;
372 	int fetchstatus, idxstatus, done = 0;
373 	const struct got_error *err;
374 	struct imsgbuf fetchibuf, idxibuf;
375 	pid_t fetchpid, idxpid;
376 	char *tmppackpath = NULL, *tmpidxpath = NULL;
377 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
378 	const char *repo_path = got_repo_get_path(repo);
379 	struct got_pathlist_head have_refs;
380 	struct got_pathlist_entry *pe;
381 	off_t packfile_size = 0;
382 	char *path;
383 
384 	*pack_hash = NULL;
385 	for (i = 0; i < nitems(tmpfds); i++)
386 		tmpfds[i] = -1;
387 
388 	TAILQ_INIT(&have_refs);
389 
390 	if (asprintf(&path, "%s/%s/fetching.pack",
391 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
392 		err = got_error_from_errno("asprintf");
393 		goto done;
394 	}
395 	err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
396 	free(path);
397 	if (err)
398 		goto done;
399 	npackfd = dup(packfd);
400 	if (npackfd == -1) {
401 		err = got_error_from_errno("dup");
402 		goto done;
403 	}
404 	if (asprintf(&path, "%s/%s/fetching.idx",
405 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
406 		err = got_error_from_errno("asprintf");
407 		goto done;
408 	}
409 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
410 	free(path);
411 	if (err)
412 		goto done;
413 	nidxfd = dup(idxfd);
414 	if (nidxfd == -1) {
415 		err = got_error_from_errno("dup");
416 		goto done;
417 	}
418 
419 	for (i = 0; i < nitems(tmpfds); i++) {
420 		tmpfds[i] = got_opentempfd();
421 		if (tmpfds[i] == -1) {
422 			err = got_error_from_errno("got_opentempfd");
423 			goto done;
424 		}
425 	}
426 
427 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
428 		err = got_error_from_errno("socketpair");
429 		goto done;
430 	}
431 
432 	fetchpid = fork();
433 	if (fetchpid == -1) {
434 		err = got_error_from_errno("fork");
435 		goto done;
436 	} else if (fetchpid == 0){
437 		got_privsep_exec_child(imsg_fetchfds,
438 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
439 	}
440 
441 	if (close(imsg_fetchfds[1]) != 0) {
442 		err = got_error_from_errno("close");
443 		goto done;
444 	}
445 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
446 	nfetchfd = dup(fetchfd);
447 	if (nfetchfd == -1) {
448 		err = got_error_from_errno("dup");
449 		goto done;
450 	}
451 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs);
452 	if (err != NULL)
453 		goto done;
454 	nfetchfd = -1;
455 	err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
456 	if (err != NULL)
457 		goto done;
458 	npackfd = dup(packfd);
459 	if (npackfd == -1) {
460 		err = got_error_from_errno("dup");
461 		goto done;
462 	}
463 
464 	packfile_size = 0;
465 	while (!done) {
466 		struct got_object_id *id = NULL;
467 		char *refname = NULL;
468 		char *server_progress = NULL;
469 		off_t packfile_size_cur;
470 
471 		err = got_privsep_recv_fetch_progress(&done,
472 		    &id, &refname, symrefs, &server_progress,
473 		    &packfile_size_cur, &fetchibuf);
474 		if (err != NULL)
475 			goto done;
476 		if (done)
477 			*pack_hash = id;
478 		else if (refname && id) {
479 			err = got_pathlist_append(refs, refname, id);
480 			if (err)
481 				goto done;
482 		} else if (server_progress) {
483 			char *s, *s0 = server_progress;
484 			while ((s = strsep(&s0, "\r")) != NULL) {
485 				if (*s == '\0')
486 					continue;
487 				err = progress_cb(progress_arg, s,
488 				    packfile_size_cur, 0, 0, 0, 0);
489 				if (err)
490 					break;
491 			}
492 			free(server_progress);
493 			if (err)
494 				goto done;
495 		} else if (packfile_size_cur != packfile_size) {
496 			err = progress_cb(progress_arg, NULL,
497 			    packfile_size_cur, 0, 0, 0, 0);
498 			if (err)
499 				break;
500 			packfile_size = packfile_size_cur;
501 		}
502 	}
503 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
504 		err = got_error_from_errno("waitpid");
505 		goto done;
506 	}
507 
508 	if (lseek(packfd, 0, SEEK_SET) == -1) {
509 		err = got_error_from_errno("lseek");
510 		goto done;
511 	}
512 	err = check_pack_hash(packfd, packfile_size, (*pack_hash)->sha1);
513 	if (err)
514 		goto done;
515  
516 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
517 		err = got_error_from_errno("socketpair");
518 		goto done;
519 	}
520 	idxpid = fork();
521 	if (idxpid == -1) {
522 		err= got_error_from_errno("fork");
523 		goto done;
524 	} else if (idxpid == 0)
525 		got_privsep_exec_child(imsg_idxfds,
526 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
527 	if (close(imsg_idxfds[1]) != 0) {
528 		err = got_error_from_errno("close");
529 		goto done;
530 	}
531 	imsg_init(&idxibuf, imsg_idxfds[0]);
532 
533 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
534 	    npackfd);
535 	if (err != NULL)
536 		goto done;
537 	npackfd = -1;
538 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
539 	if (err != NULL)
540 		goto done;
541 	nidxfd = -1;
542 	for (i = 0; i < nitems(tmpfds); i++) {
543 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
544 		if (err != NULL)
545 			goto done;
546 		tmpfds[i] = -1;
547 	}
548 	done = 0;
549 	while (!done) {
550 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
551 
552 		err = got_privsep_recv_index_progress(&done, &nobj_total,
553 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
554 		    &idxibuf);
555 		if (err != NULL)
556 			goto done;
557 		if (nobj_indexed != 0) {
558 			err = progress_cb(progress_arg, NULL,
559 			    packfile_size, nobj_total,
560 			    nobj_indexed, nobj_loose, nobj_resolved);
561 			if (err)
562 				break;
563 		}
564 		imsg_clear(&idxibuf);
565 	}
566 	if (close(imsg_idxfds[0]) == -1) {
567 		err = got_error_from_errno("close");
568 		goto done;
569 	}
570 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
571 		err = got_error_from_errno("waitpid");
572 		goto done;
573 	}
574 
575 	err = got_object_id_str(&id_str, *pack_hash);
576 	if (err)
577 		goto done;
578 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
579 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
580 		err = got_error_from_errno("asprintf");
581 		goto done;
582 	}
583 
584 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
585 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
586 		err = got_error_from_errno("asprintf");
587 		goto done;
588 	}
589 
590 	if (rename(tmppackpath, packpath) == -1) {
591 		err = got_error_from_errno3("rename", tmppackpath, packpath);
592 		goto done;
593 	}
594 	if (rename(tmpidxpath, idxpath) == -1) {
595 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
596 		goto done;
597 	}
598 
599 done:
600 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
601 		err = got_error_from_errno("close");
602 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
603 		err = got_error_from_errno("close");
604 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
605 		err = got_error_from_errno("close");
606 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
607 		err = got_error_from_errno("close");
608 	for (i = 0; i < nitems(tmpfds); i++) {
609 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
610 			err = got_error_from_errno("close");
611 	}
612 	free(tmppackpath);
613 	free(tmpidxpath);
614 	free(idxpath);
615 	free(packpath);
616 
617 	if (err) {
618 		free(*pack_hash);
619 		*pack_hash = NULL;
620 		TAILQ_FOREACH(pe, refs, entry) {
621 			free((void *)pe->path);
622 			free(pe->data);
623 		}
624 		got_pathlist_free(refs);
625 		TAILQ_FOREACH(pe, symrefs, entry) {
626 			free((void *)pe->path);
627 			free(pe->data);
628 		}
629 		got_pathlist_free(symrefs);
630 	}
631 	return err;
632 }