1 /*
2  * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
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/mman.h>
18 #include <sys/stat.h>
19 
20 #include <netdb.h>
21 
22 #include <assert.h>
23 #include <errno.h>
24 #include <fcntl.h>
25 #include <string.h>
26 
27 #include "gmid.h"
28 
29 int connected_clients;
30 
31 int
32 check_path(struct client *c, const char *path, int *fd)
33 {
34 	struct stat sb;
35 
36 	assert(path != NULL);
37 	if ((*fd = openat(c->host->dirfd, *path ? path : ".",
38 	    O_RDONLY | O_NOFOLLOW)) == -1) {
39 		return FILE_MISSING;
40 	}
41 
42 	if (fstat(*fd, &sb) == -1) {
43 		LOGN(c, "failed stat for %s: %s", path, strerror(errno));
44 		return FILE_MISSING;
45 	}
46 
47 	if (S_ISDIR(sb.st_mode))
48 		return FILE_DIRECTORY;
49 
50 	if (sb.st_mode & S_IXUSR)
51 		return FILE_EXECUTABLE;
52 
53 	return FILE_EXISTS;
54 }
55 
56 int
57 open_file(char *fpath, char *query, struct pollfd *fds, struct client *c)
58 {
59 	switch (check_path(c, fpath, &c->fd)) {
60 	case FILE_EXECUTABLE:
61 		if (c->host->cgi != NULL && starts_with(fpath, c->host->cgi))
62 			return start_cgi(fpath, "", query, fds, c);
63 
64 		/* fallthrough */
65 
66 	case FILE_EXISTS:
67 		if ((c->len = filesize(c->fd)) == -1) {
68 			LOGE(c, "failed to get file size for %s", fpath);
69 			goodbye(fds, c);
70 			return 0;
71 		}
72 
73 		if ((c->buf = mmap(NULL, c->len, PROT_READ, MAP_PRIVATE,
74 			    c->fd, 0)) == MAP_FAILED) {
75 			LOGW(c, "mmap: %s: %s", fpath, strerror(errno));
76 			goodbye(fds, c);
77 			return 0;
78 		}
79 		c->i = c->buf;
80 		return start_reply(fds, c, SUCCESS, mime(c->host, fpath));
81 
82 	case FILE_DIRECTORY:
83 		LOGD(c, "%s is a directory, trying %s/index.gmi", fpath, fpath);
84 		close(c->fd);
85 		c->fd = -1;
86 		send_dir(fpath, fds, c);
87 		return 0;
88 
89 	case FILE_MISSING:
90 		if (c->host->cgi != NULL && starts_with(fpath, c->host->cgi))
91 			return check_for_cgi(fpath, query, fds, c);
92 
93 		if (!start_reply(fds, c, NOT_FOUND, "not found"))
94 			return 0;
95 		goodbye(fds, c);
96 		return 0;
97 
98 	default:
99 		/* unreachable */
100 		abort();
101 	}
102 }
103 
104 
105 /*
106  * the inverse of this algorithm, i.e. starting from the start of the
107  * path + strlen(cgi), and checking if each component, should be
108  * faster.  But it's tedious to write.  This does the opposite: starts
109  * from the end and strip one component at a time, until either an
110  * executable is found or we emptied the path.
111  */
112 int
113 check_for_cgi(char *path, char *query, struct pollfd *fds, struct client *c)
114 {
115 	char *end;
116 	end = strchr(path, '\0');
117 
118 	/* NB: assume CGI is enabled and path matches cgi */
119 
120 	while (end > path) {
121 		/* go up one level.  UNIX paths are simple and POSIX
122 		 * dirname, with its ambiguities on if the given path
123 		 * is changed or not, gives me headaches. */
124 		while (*end != '/')
125 			end--;
126 		*end = '\0';
127 
128 		switch (check_path(c, path, &c->fd)) {
129 		case FILE_EXECUTABLE:
130 			return start_cgi(path, end+1, query, fds,c);
131 		case FILE_MISSING:
132 			break;
133 		default:
134 			goto err;
135 		}
136 
137 		*end = '/';
138 		end--;
139 	}
140 
141 err:
142 	if (!start_reply(fds, c, NOT_FOUND, "not found"))
143 		return 0;
144 	goodbye(fds, c);
145 	return 0;
146 }
147 
148 void
149 mark_nonblock(int fd)
150 {
151 	int flags;
152 
153 	if ((flags = fcntl(fd, F_GETFL)) == -1)
154 		fatal("fcntl(F_GETFL): %s", strerror(errno));
155 	if (fcntl(fd, F_SETFL, flags | O_NONBLOCK) == -1)
156 		fatal("fcntl(F_SETFL): %s", strerror(errno));
157 }
158 
159 void
160 handle_handshake(struct pollfd *fds, struct client *c)
161 {
162 	struct vhost *h;
163 	const char *servname;
164 
165 	switch (tls_handshake(c->ctx)) {
166 	case 0:  /* success */
167 	case -1: /* already handshaked */
168 		break;
169 	case TLS_WANT_POLLIN:
170 		fds->events = POLLIN;
171 		return;
172 	case TLS_WANT_POLLOUT:
173 		fds->events = POLLOUT;
174 		return;
175 	default:
176 		/* unreachable */
177 		abort();
178 	}
179 
180 	servname = tls_conn_servername(c->ctx);
181 	if (servname == NULL)
182 		goto hostnotfound;
183 
184 	for (h = hosts; h->domain != NULL; ++h) {
185 		if (!strcmp(h->domain, servname) || !strcmp(h->domain, "*"))
186 			break;
187 	}
188 
189 	if (h->domain != NULL) {
190 		c->state = S_OPEN;
191 		c->host = h;
192 		handle_open_conn(fds, c);
193 		return;
194 	}
195 
196 hostnotfound:
197 	/* XXX: check the correct response */
198 	if (!start_reply(fds, c, BAD_REQUEST, "Wrong host or missing SNI"))
199 		return;
200 	goodbye(fds, c);
201 }
202 
203 void
204 handle_open_conn(struct pollfd *fds, struct client *c)
205 {
206 	char buf[GEMINI_URL_LEN];
207 	const char *parse_err = "invalid request";
208 	struct iri iri;
209 
210 	bzero(buf, sizeof(buf));
211 
212 	switch (tls_read(c->ctx, buf, sizeof(buf)-1)) {
213 	case -1:
214 		LOGE(c, "tls_read: %s", tls_error(c->ctx));
215 		goodbye(fds, c);
216 		return;
217 
218 	case TLS_WANT_POLLIN:
219 		fds->events = POLLIN;
220 		return;
221 
222 	case TLS_WANT_POLLOUT:
223 		fds->events = POLLOUT;
224 		return;
225 	}
226 
227 	if (!trim_req_iri(buf) || !parse_iri(buf, &iri, &parse_err)) {
228 		if (!start_reply(fds, c, BAD_REQUEST, parse_err))
229 			return;
230 		goodbye(fds, c);
231 		return;
232 	}
233 
234 	if (strcmp(iri.schema, "gemini") || iri.port_no != conf.port) {
235 		if (!start_reply(fds, c, PROXY_REFUSED, "won't proxy request"))
236 			return;
237 		goodbye(fds, c);
238 		return;
239 	}
240 
241 	LOGI(c, "GET %s%s%s",
242 	    *iri.path ? iri.path : "/",
243 	    *iri.query ? "?" : "",
244 	    *iri.query ? iri.query : "");
245 
246 	send_file(iri.path, iri.query, fds, c);
247 }
248 
249 int
250 start_reply(struct pollfd *pfd, struct client *c, int code, const char *meta)
251 {
252 	char buf[1030];		/* status + ' ' + max reply len + \r\n\0 */
253 	int len;
254 
255 	c->code = code;
256 	c->meta = meta;
257 	c->state = S_INITIALIZING;
258 
259 	snprintf(buf, sizeof(buf), "%d ", code);
260 	strlcat(buf, meta, sizeof(buf));
261 	if (!strcmp(meta, "text/gemini") && c->host->lang != NULL) {
262 		strlcat(buf, "; lang=", sizeof(buf));
263 		strlcat(buf, c->host->lang, sizeof(buf));
264 	}
265 
266 	len = strlcat(buf, "\r\n", sizeof(buf));
267 	assert(len < (int)sizeof(buf));
268 
269 	switch (tls_write(c->ctx, buf, len)) {
270 	case TLS_WANT_POLLIN:
271 		pfd->events = POLLIN;
272 		return 0;
273 	case TLS_WANT_POLLOUT:
274 		pfd->events = POLLOUT;
275 		return 0;
276 	default:
277 		return 1;
278 	}
279 }
280 
281 int
282 start_cgi(const char *spath, const char *relpath, const char *query,
283     struct pollfd *fds, struct client *c)
284 {
285 	char addr[NI_MAXHOST];
286 	const char *ruser, *cissuer, *chash;
287 	int e;
288 
289 	e = getnameinfo((struct sockaddr*)&c->addr, sizeof(c->addr),
290 	    addr, sizeof(addr),
291 	    NULL, 0,
292 	    NI_NUMERICHOST);
293 	if (e != 0)
294 		goto err;
295 
296 	if (tls_peer_cert_provided(c->ctx)) {
297 		ruser = tls_peer_cert_subject(c->ctx);
298 		cissuer = tls_peer_cert_issuer(c->ctx);
299 		chash = tls_peer_cert_hash(c->ctx);
300 	} else {
301 		ruser = NULL;
302 		cissuer = NULL;
303 		chash = NULL;
304 	}
305 
306 	if (!send_string(exfd, spath)
307 	    || !send_string(exfd, relpath)
308 	    || !send_string(exfd, query)
309 	    || !send_string(exfd, addr)
310 	    || !send_string(exfd, ruser)
311 	    || !send_string(exfd, cissuer)
312 	    || !send_string(exfd, chash)
313 	    || !send_vhost(exfd, c->host))
314 		goto err;
315 
316 	close(c->fd);
317 	if ((c->fd = recv_fd(exfd)) == -1) {
318 		if (!start_reply(fds, c, TEMP_FAILURE, "internal server error"))
319 			return 0;
320 		goodbye(fds, c);
321 		return 0;
322 	}
323 	c->child = 1;
324 	c->state = S_SENDING;
325 	cgi_poll_on_child(fds, c);
326 	/* handle_cgi(fds, c); */
327 	return 0;
328 
329 err:
330 	/* fatal("cannot talk to the executor process: %s", strerror(errno)); */
331 	fatal("cannot talk to the executor process");
332 }
333 
334 void
335 send_file(char *path, char *query, struct pollfd *fds, struct client *c)
336 {
337 	ssize_t ret, len;
338 
339 	if (c->fd == -1) {
340 		if (!open_file(path, query, fds, c))
341 			return;
342 		c->state = S_SENDING;
343 	}
344 
345 	len = (c->buf + c->len) - c->i;
346 
347 	while (len > 0) {
348 		switch (ret = tls_write(c->ctx, c->i, len)) {
349 		case -1:
350 			LOGE(c, "tls_write: %s", tls_error(c->ctx));
351 			goodbye(fds, c);
352 			return;
353 
354 		case TLS_WANT_POLLIN:
355 			fds->events = POLLIN;
356 			return;
357 
358 		case TLS_WANT_POLLOUT:
359 			fds->events = POLLOUT;
360 			return;
361 
362 		default:
363 			c->i += ret;
364 			len -= ret;
365 			break;
366 		}
367 	}
368 
369 	goodbye(fds, c);
370 }
371 
372 void
373 send_dir(char *path, struct pollfd *fds, struct client *client)
374 {
375 	char fpath[PATHBUF];
376 	size_t len;
377 
378 	bzero(fpath, PATHBUF);
379 
380 	if (path[0] != '.')
381 		fpath[0] = '.';
382 
383 	/* this cannot fail since sizeof(fpath) > maxlen of path */
384 	strlcat(fpath, path, PATHBUF);
385 	len = strlen(fpath);
386 
387 	/* add a trailing / in case. */
388 	if (fpath[len-1] != '/') {
389 		fpath[len] = '/';
390 	}
391 
392 	strlcat(fpath, "index.gmi", sizeof(fpath));
393 
394 	send_file(fpath, NULL, fds, client);
395 }
396 
397 void
398 cgi_poll_on_child(struct pollfd *fds, struct client *c)
399 {
400 	int fd;
401 
402 	if (c->waiting_on_child)
403 		return;
404 	c->waiting_on_child = 1;
405 
406 	fds->events = POLLIN;
407 
408 	fd = fds->fd;
409 	fds->fd = c->fd;
410 	c->fd = fd;
411 }
412 
413 void
414 cgi_poll_on_client(struct pollfd *fds, struct client *c)
415 {
416 	int fd;
417 
418 	if (!c->waiting_on_child)
419 		return;
420 	c->waiting_on_child = 0;
421 
422 	fd = fds->fd;
423 	fds->fd = c->fd;
424 	c->fd = fd;
425 }
426 
427 void
428 handle_cgi(struct pollfd *fds, struct client *c)
429 {
430 	ssize_t r;
431 
432 	/* ensure c->fd is the child and fds->fd the client */
433 	cgi_poll_on_client(fds, c);
434 
435 	while (1) {
436 		if (c->len == 0) {
437 			if ((r = read(c->fd, c->sbuf, sizeof(c->sbuf))) == 0)
438 				goto end;
439 			if (r == -1) {
440 				if (errno == EAGAIN || errno == EWOULDBLOCK) {
441 					cgi_poll_on_child(fds, c);
442 					return;
443 				}
444                                 goto end;
445 			}
446 			c->len = r;
447 			c->off = 0;
448 		}
449 
450 		while (c->len > 0) {
451 			switch (r = tls_write(c->ctx, c->sbuf + c->off, c->len)) {
452 			case -1:
453 				goto end;
454 
455 			case TLS_WANT_POLLOUT:
456 				fds->events = POLLOUT;
457 				return;
458 
459 			case TLS_WANT_POLLIN:
460 				fds->events = POLLIN;
461 				return;
462 
463 			default:
464                                 c->off += r;
465 				c->len -= r;
466 				break;
467 			}
468 		}
469 	}
470 
471 end:
472 	goodbye(fds, c);
473 }
474 
475 void
476 goodbye(struct pollfd *pfd, struct client *c)
477 {
478 	c->state = S_CLOSING;
479 
480 	switch (tls_close(c->ctx)) {
481 	case TLS_WANT_POLLIN:
482 		pfd->events = POLLIN;
483 		return;
484 	case TLS_WANT_POLLOUT:
485 		pfd->events = POLLOUT;
486 		return;
487 	}
488 
489 	connected_clients--;
490 
491 	tls_free(c->ctx);
492 	c->ctx = NULL;
493 
494 	if (c->buf != MAP_FAILED)
495 		munmap(c->buf, c->len);
496 
497 	if (c->fd != -1)
498 		close(c->fd);
499 
500 	close(pfd->fd);
501 	pfd->fd = -1;
502 }
503 
504 void
505 do_accept(int sock, struct tls *ctx, struct pollfd *fds, struct client *clients)
506 {
507 	int i, fd;
508 	struct sockaddr_storage addr;
509 	socklen_t len;
510 
511 	len = sizeof(addr);
512 	if ((fd = accept(sock, (struct sockaddr*)&addr, &len)) == -1) {
513 		if (errno == EWOULDBLOCK)
514 			return;
515 		fatal("accept: %s", strerror(errno));
516 	}
517 
518 	mark_nonblock(fd);
519 
520 	for (i = 0; i < MAX_USERS; ++i) {
521 		if (fds[i].fd == -1) {
522 			bzero(&clients[i], sizeof(struct client));
523 			if (tls_accept_socket(ctx, &clients[i].ctx, fd) == -1)
524 				break; /* goodbye fd! */
525 
526 			fds[i].fd = fd;
527 			fds[i].events = POLLIN;
528 
529 			clients[i].state = S_HANDSHAKE;
530 			clients[i].fd = -1;
531 			clients[i].child = 0;
532 			clients[i].waiting_on_child = 0;
533 			clients[i].buf = MAP_FAILED;
534 			clients[i].af = AF_INET;
535 			clients[i].addr = addr;
536 
537 			connected_clients++;
538 			return;
539 		}
540 	}
541 
542 	close(fd);
543 }
544 
545 void
546 handle(struct pollfd *fds, struct client *client)
547 {
548 	switch (client->state) {
549 	case S_HANDSHAKE:
550 		handle_handshake(fds, client);
551 		break;
552 
553 	case S_OPEN:
554                 handle_open_conn(fds, client);
555 		break;
556 
557 	case S_INITIALIZING:
558 		if (!start_reply(fds, client, client->code, client->meta))
559 			return;
560 
561 		if (client->code != SUCCESS) {
562 			/* we don't need a body */
563 			goodbye(fds, client);
564 			return;
565 		}
566 
567 		client->state = S_SENDING;
568 
569 		/* fallthrough */
570 
571 	case S_SENDING:
572 		if (client->child)
573 			handle_cgi(fds, client);
574 		else
575 			send_file(NULL, NULL, fds, client);
576 		break;
577 
578 	case S_CLOSING:
579 		goodbye(fds, client);
580 		break;
581 
582 	default:
583 		/* unreachable */
584 		abort();
585 	}
586 }
587 
588 void
589 loop(struct tls *ctx, int sock4, int sock6)
590 {
591 	int i;
592 	struct client clients[MAX_USERS];
593 	struct pollfd fds[MAX_USERS];
594 
595 	connected_clients = 0;
596 
597 	for (i = 0; i < MAX_USERS; ++i) {
598 		fds[i].fd = -1;
599 		fds[i].events = POLLIN;
600 		bzero(&clients[i], sizeof(struct client));
601 	}
602 
603 	fds[0].fd = sock4;
604 	fds[1].fd = sock6;
605 
606 	for (;;) {
607 		if (poll(fds, MAX_USERS, INFTIM) == -1) {
608 			if (errno == EINTR) {
609                                 fprintf(stderr, "connected clients: %d\n",
610 				    connected_clients);
611 				continue;
612 			}
613 			fatal("poll: %s", strerror(errno));
614 		}
615 
616 		for (i = 0; i < MAX_USERS; i++) {
617 			if (fds[i].revents == 0)
618 				continue;
619 
620 			if (fds[i].revents & (POLLERR|POLLNVAL))
621 				fatal("bad fd %d: %s", fds[i].fd,
622 				    strerror(errno));
623 
624 			if (fds[i].revents & POLLHUP) {
625 				/* fds[i] may be the fd of the stdin
626 				 * of a cgi script that has exited. */
627 				if (!clients[i].waiting_on_child) {
628 					goodbye(&fds[i], &clients[i]);
629 					continue;
630 				}
631 			}
632 
633 			if (fds[i].fd == sock4)
634 				do_accept(sock4, ctx, fds, clients);
635 			else if (fds[i].fd == sock6)
636 				do_accept(sock6, ctx, fds, clients);
637 			else
638 				handle(&fds[i], &clients[i]);
639 		}
640 	}
641 }