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 #define GOT_PROTOMAX	64
66 #define GOT_HOSTMAX	256
67 #define GOT_PATHMAX	512
68 #define GOT_REPOMAX	256
69 #define GOT_PORTMAX	16
70 #define GOT_URIMAX	1024
71 
72 static int
73 hassuffix(char *base, char *suf)
74 {
75 	int nb, ns;
76 
77 	nb = strlen(base);
78 	ns = strlen(suf);
79 	if (ns <= nb && strcmp(base + (nb - ns), suf) == 0)
80 		return 1;
81 	return 0;
82 }
83 
84 static const struct got_error *
85 dial_ssh(int *fetchfd, const char *host, const char *port, const char *path,
86     const char *direction)
87 {
88 	const struct got_error *error = NULL;
89 	int pid, pfd[2];
90 	char cmd[64];
91 
92 	*fetchfd = -1;
93 
94 	if (pipe(pfd) == -1)
95 		return got_error_from_errno("pipe");
96 
97 	pid = fork();
98 	if (pid == -1) {
99 		error = got_error_from_errno("fork");
100 		close(pfd[0]);
101 		close(pfd[1]);
102 		return error;
103 	} else if (pid == 0) {
104 		int n;
105 		close(pfd[1]);
106 		dup2(pfd[0], 0);
107 		dup2(pfd[0], 1);
108 		n = snprintf(cmd, sizeof(cmd), "git-%s-pack", direction);
109 		if (n < 0 || n >= sizeof(cmd))
110 			err(1, "snprintf");
111 		if (execlp("ssh", "ssh", host, cmd, path, NULL) == -1)
112 			err(1, "execlp");
113 		abort(); /* not reached */
114 	} else {
115 		close(pfd[0]);
116 		*fetchfd = pfd[1];
117 		return NULL;
118 	}
119 }
120 
121 static const struct got_error *
122 dial_git(int *fetchfd, const char *host, const char *port, const char *path,
123     const char *direction)
124 {
125 	const struct got_error *err = NULL;
126 	struct addrinfo hints, *servinfo, *p;
127 	char *cmd = NULL, *pkt = NULL;
128 	int fd = -1, totlen, r, eaicode;
129 
130 	*fetchfd = -1;
131 
132 	memset(&hints, 0, sizeof hints);
133 	hints.ai_family = AF_UNSPEC;
134 	hints.ai_socktype = SOCK_STREAM;
135 	eaicode = getaddrinfo(host, port, &hints, &servinfo);
136 	if (eaicode)
137 		return got_error_msg(GOT_ERR_ADDRINFO, gai_strerror(eaicode));
138 
139 	for (p = servinfo; p != NULL; p = p->ai_next) {
140 		if ((fd = socket(p->ai_family, p->ai_socktype,
141 		    p->ai_protocol)) == -1)
142 			continue;
143 		if (connect(fd, p->ai_addr, p->ai_addrlen) == 0)
144 			break;
145 		err = got_error_from_errno("connect");
146 		close(fd);
147 	}
148 	if (p == NULL)
149 		goto done;
150 
151 	if (asprintf(&cmd, "git-%s-pack %s", direction, path) == -1) {
152 		err = got_error_from_errno("asprintf");
153 		goto done;
154 	}
155 	totlen = 4 + strlen(cmd) + 1 + strlen("host=") + strlen(host) + 1;
156 	if (asprintf(&pkt, "%04x%s", totlen, cmd) == -1) {
157 		err = got_error_from_errno("asprintf");
158 		goto done;
159 	}
160 	r = write(fd, pkt, strlen(pkt) + 1);
161 	if (r == -1) {
162 		err = got_error_from_errno("write");
163 		goto done;
164 	}
165 	if (asprintf(&pkt, "host=%s", host) == -1) {
166 		err = got_error_from_errno("asprintf");
167 		goto done;
168 	}
169 	r = write(fd, pkt, strlen(pkt) + 1);
170 	if (r == -1) {
171 		err = got_error_from_errno("write");
172 		goto done;
173 	}
174 done:
175 	free(cmd);
176 	free(pkt);
177 	if (err) {
178 		if (fd != -1)
179 			close(fd);
180 	} else
181 		*fetchfd = fd;
182 	return err;
183 }
184 
185 const struct got_error *
186 got_fetch_connect(int *fetchfd, const char *proto, const char *host,
187     const char *port, const char *server_path)
188 {
189 	const struct got_error *err = NULL;
190 
191 	*fetchfd = -1;
192 
193 	if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
194 		err = dial_ssh(fetchfd, host, port, server_path, "upload");
195 	else if (strcmp(proto, "git") == 0)
196 		err = dial_git(fetchfd, host, port, server_path, "upload");
197 	else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
198 		err = got_error_path(proto, GOT_ERR_NOT_IMPL);
199 	else
200 		err = got_error_path(proto, GOT_ERR_BAD_PROTO);
201 	return err;
202 }
203 
204 const struct got_error *
205 got_fetch_parse_uri(char **proto, char **host, char **port,
206     char **server_path, char **repo_name, const char *uri)
207 {
208 	const struct got_error *err = NULL;
209 	char *s, *p, *q;
210 	int n, hasport;
211 
212 	*proto = *host = *port = *server_path = *repo_name = NULL;
213 
214 	p = strstr(uri, "://");
215 	if (!p) {
216 		return got_error(GOT_ERR_PARSE_URI);
217 	}
218 	*proto = strndup(uri, p - uri);
219 	if (proto == NULL) {
220 		err = got_error_from_errno("strndup");
221 		goto done;
222 	}
223 
224 	hasport = (strcmp(*proto, "git") == 0 ||
225 	    strstr(*proto, "http") == *proto);
226 	s = p + 3;
227 	p = NULL;
228 	if (!hasport) {
229 		p = strstr(s, ":");
230 		if (p != NULL)
231 			p++;
232 	}
233 	if (p == NULL)
234 		p = strstr(s, "/");
235 	if (p == NULL || strlen(p) == 1) {
236 		err = got_error(GOT_ERR_PARSE_URI);
237 		goto done;
238 	}
239 
240 	q = memchr(s, ':', p - s);
241 	if (q) {
242 		*host = strndup(s, q - s);
243 		if (*host == NULL) {
244 			err = got_error_from_errno("strndup");
245 			goto done;
246 		}
247 		*port = strndup(q + 1, p - (q + 1));
248 		if (*port == NULL) {
249 			err = got_error_from_errno("strndup");
250 			goto done;
251 		}
252 	} else {
253 		*host = strndup(s, p - s);
254 		if (*host == NULL) {
255 			err = got_error_from_errno("strndup");
256 			goto done;
257 		}
258 		if (asprintf(port, "%u", GOT_DEFAULT_GIT_PORT) == -1) {
259 			err = got_error_from_errno("asprintf");
260 			goto done;
261 		}
262 	}
263 
264 	*server_path = strdup(p);
265 	if (*server_path == NULL) {
266 		err = got_error_from_errno("strdup");
267 		goto done;
268 	}
269 
270 	p = strrchr(p, '/') + 1;
271 	if (!p || strlen(p) == 0) {
272 		//werrstr("missing repository in uri");
273 		err = got_error(GOT_ERR_PARSE_URI);
274 		goto done;
275 	}
276 	n = strlen(p);
277 	if (hassuffix(p, ".git"))
278 		n -= 4;
279 	*repo_name = strndup(p, (p + n) - p);
280 	if (*repo_name == NULL) {
281 		err = got_error_from_errno("strndup");
282 		goto done;
283 	}
284 done:
285 	if (err) {
286 		free(*proto);
287 		*proto = NULL;
288 		free(*host);
289 		*host = NULL;
290 		free(*port);
291 		*port = NULL;
292 		free(*server_path);
293 		*server_path = NULL;
294 		free(*repo_name);
295 		*repo_name = NULL;
296 	}
297 	return err;
298 }
299 
300 static const struct got_error *
301 check_pack_hash(int fd, size_t sz, uint8_t *hcomp)
302 {
303 	SHA1_CTX ctx;
304 	uint8_t hexpect[SHA1_DIGEST_LENGTH];
305 	uint8_t buf[32 * 1024];
306 	ssize_t n, r, nr;
307 
308 	if (sz < sizeof(struct got_packfile_hdr) + SHA1_DIGEST_LENGTH)
309 		return got_error_msg(GOT_ERR_BAD_PACKFILE, "short packfile");
310 
311 	n = 0;
312 	SHA1Init(&ctx);
313 	while (n < sz - 20) {
314 		nr = sizeof(buf);
315 		if (sz - n - 20 < sizeof(buf))
316 			nr = sz - n - 20;
317 		r = read(fd, buf, nr);
318 		if (r == -1)
319 			return got_error_from_errno("read");
320 		if (r != nr)
321 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
322 			    "short pack file");
323 		SHA1Update(&ctx, buf, nr);
324 		n += r;
325 	}
326 	SHA1Final(hcomp, &ctx);
327 
328 	r = read(fd, hexpect, sizeof(hexpect));
329 	if (r == -1)
330 		return got_error_from_errno("read");
331 	if (r != sizeof(hexpect))
332 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
333 		    "short pack file");
334 
335 	if (memcmp(hcomp, hexpect, SHA1_DIGEST_LENGTH) != 0)
336 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
337 		    "packfile checksum mismatch");
338 
339 	return NULL;
340 }
341 
342 const struct got_error*
343 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
344     struct got_pathlist_head *symrefs, int fetchfd, struct got_repository *repo,
345     got_fetch_progress_cb progress_cb, void *progress_arg)
346 {
347 	int imsg_fetchfds[2], imsg_idxfds[2];
348 	int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
349 	int fetchstatus, idxstatus, done = 0;
350 	const struct got_error *err;
351 	struct imsgbuf fetchibuf, idxibuf;
352 	pid_t fetchpid, idxpid;
353 	char *tmppackpath = NULL, *tmpidxpath = NULL;
354 	char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
355 	const char *repo_path = got_repo_get_path(repo);
356 	struct got_pathlist_head have_refs;
357 	struct got_pathlist_entry *pe;
358 	off_t packfile_size = 0;
359 	char *path;
360 
361 	*pack_hash = NULL;
362 
363 	TAILQ_INIT(&have_refs);
364 
365 	if (asprintf(&path, "%s/%s/fetching.pack",
366 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
367 		err = got_error_from_errno("asprintf");
368 		goto done;
369 	}
370 	err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
371 	free(path);
372 	if (err)
373 		goto done;
374 	npackfd = dup(packfd);
375 	if (npackfd == -1) {
376 		err = got_error_from_errno("dup");
377 		goto done;
378 	}
379 	if (asprintf(&path, "%s/%s/fetching.idx",
380 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
381 		err = got_error_from_errno("asprintf");
382 		goto done;
383 	}
384 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
385 	free(path);
386 	if (err)
387 		goto done;
388 	nidxfd = dup(idxfd);
389 	if (nidxfd == -1) {
390 		err = got_error_from_errno("dup");
391 		goto done;
392 	}
393 
394 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
395 		err = got_error_from_errno("socketpair");
396 		goto done;
397 	}
398 
399 	fetchpid = fork();
400 	if (fetchpid == -1) {
401 		err = got_error_from_errno("fork");
402 		goto done;
403 	} else if (fetchpid == 0){
404 		got_privsep_exec_child(imsg_fetchfds,
405 		    GOT_PATH_PROG_FETCH_PACK, tmppackpath);
406 	}
407 
408 	if (close(imsg_fetchfds[1]) != 0) {
409 		err = got_error_from_errno("close");
410 		goto done;
411 	}
412 	imsg_init(&fetchibuf, imsg_fetchfds[0]);
413 	nfetchfd = dup(fetchfd);
414 	if (nfetchfd == -1) {
415 		err = got_error_from_errno("dup");
416 		goto done;
417 	}
418 	err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs);
419 	if (err != NULL)
420 		goto done;
421 	nfetchfd = -1;
422 	err = got_privsep_send_tmpfd(&fetchibuf, npackfd);
423 	if (err != NULL)
424 		goto done;
425 	npackfd = dup(packfd);
426 	if (npackfd == -1) {
427 		err = got_error_from_errno("dup");
428 		goto done;
429 	}
430 
431 	packfile_size = 0;
432 	while (!done) {
433 		struct got_object_id *id = NULL;
434 		char *refname = NULL;
435 		char *server_progress = NULL;
436 		off_t packfile_size_cur;
437 
438 		err = got_privsep_recv_fetch_progress(&done,
439 		    &id, &refname, symrefs, &server_progress,
440 		    &packfile_size_cur, &fetchibuf);
441 		if (err != NULL)
442 			goto done;
443 		if (done)
444 			*pack_hash = id;
445 		else if (refname && id) {
446 			err = got_pathlist_append(refs, refname, id);
447 			if (err)
448 				goto done;
449 		} else if (server_progress) {
450 			char *s, *s0 = server_progress;
451 			while ((s = strsep(&s0, "\r")) != NULL) {
452 				if (*s == '\0')
453 					continue;
454 				err = progress_cb(progress_arg, s,
455 				    packfile_size_cur, 0, 0, 0, 0);
456 				if (err)
457 					break;
458 			}
459 			free(server_progress);
460 			if (err)
461 				goto done;
462 		} else if (packfile_size_cur != packfile_size) {
463 			err = progress_cb(progress_arg, NULL,
464 			    packfile_size_cur, 0, 0, 0, 0);
465 			if (err)
466 				break;
467 			packfile_size = packfile_size_cur;
468 		}
469 	}
470 	if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
471 		err = got_error_from_errno("waitpid");
472 		goto done;
473 	}
474 
475 	if (lseek(packfd, 0, SEEK_SET) == -1) {
476 		err = got_error_from_errno("lseek");
477 		goto done;
478 	}
479 	err = check_pack_hash(packfd, packfile_size, (*pack_hash)->sha1);
480 	if (err)
481 		goto done;
482  
483 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
484 		err = got_error_from_errno("socketpair");
485 		goto done;
486 	}
487 	idxpid = fork();
488 	if (idxpid == -1) {
489 		err= got_error_from_errno("fork");
490 		goto done;
491 	} else if (idxpid == 0)
492 		got_privsep_exec_child(imsg_idxfds,
493 		    GOT_PATH_PROG_INDEX_PACK, tmppackpath);
494 	if (close(imsg_idxfds[1]) != 0) {
495 		err = got_error_from_errno("close");
496 		goto done;
497 	}
498 	imsg_init(&idxibuf, imsg_idxfds[0]);
499 
500 	err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
501 	    npackfd);
502 	if (err != NULL)
503 		goto done;
504 	npackfd = -1;
505 	err = got_privsep_send_tmpfd(&idxibuf, nidxfd);
506 	if (err != NULL)
507 		goto done;
508 	nidxfd = -1;
509 	done = 0;
510 	while (!done) {
511 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
512 
513 		err = got_privsep_recv_index_progress(&done, &nobj_total,
514 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
515 		    &idxibuf);
516 		if (err != NULL)
517 			goto done;
518 		if (nobj_indexed != 0) {
519 			err = progress_cb(progress_arg, NULL,
520 			    packfile_size, nobj_total,
521 			    nobj_indexed, nobj_loose, nobj_resolved);
522 			if (err)
523 				break;
524 		}
525 		imsg_clear(&idxibuf);
526 	}
527 	if (close(imsg_idxfds[0]) == -1) {
528 		err = got_error_from_errno("close");
529 		goto done;
530 	}
531 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
532 		err = got_error_from_errno("waitpid");
533 		goto done;
534 	}
535 
536 	err = got_object_id_str(&id_str, *pack_hash);
537 	if (err)
538 		goto done;
539 	if (asprintf(&packpath, "%s/%s/pack-%s.pack",
540 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
541 		err = got_error_from_errno("asprintf");
542 		goto done;
543 	}
544 
545 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
546 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
547 		err = got_error_from_errno("asprintf");
548 		goto done;
549 	}
550 
551 	if (rename(tmppackpath, packpath) == -1) {
552 		err = got_error_from_errno3("rename", tmppackpath, packpath);
553 		goto done;
554 	}
555 	if (rename(tmpidxpath, idxpath) == -1) {
556 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
557 		goto done;
558 	}
559 
560 done:
561 	if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
562 		err = got_error_from_errno("close");
563 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
564 		err = got_error_from_errno("close");
565 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
566 		err = got_error_from_errno("close");
567 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
568 		err = got_error_from_errno("close");
569 	free(tmppackpath);
570 	free(tmpidxpath);
571 	free(idxpath);
572 	free(packpath);
573 
574 	if (err) {
575 		free(*pack_hash);
576 		*pack_hash = NULL;
577 		TAILQ_FOREACH(pe, refs, entry) {
578 			free((void *)pe->path);
579 			free(pe->data);
580 		}
581 		got_pathlist_free(refs);
582 		TAILQ_FOREACH(pe, symrefs, entry) {
583 			free((void *)pe->path);
584 			free(pe->data);
585 		}
586 		got_pathlist_free(symrefs);
587 	}
588 	return err;
589 }