op public repos

Blob

Date:: Wed Oct 14 08:09:06 2020 UTC
Message:: switching to mmap() based file handling
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2020 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/socket.h>
19 #include <sys/stat.h>
20 
21 #include <netinet/in.h>
22 
23 #include <assert.h>
24 #include <err.h>
25 #include <errno.h>
26 #include <fcntl.h>
27 #include <poll.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <tls.h>
32 #include <unistd.h>
33 
34 #ifndef __OpenBSD__
35 # define pledge(a, b) 0
36 # define unveil(a, b) 0
37 #endif /* __OpenBSD__ */
38 
39 #ifndef INFTIM
40 # define INFTIM -1
41 #endif /* INFTIM */
42 
43 #define GEMINI_URL_LEN (1024+3)	/* URL max len + \r\n + \0 */
44 
45 /* large enough to hold a copy of a gemini URL and still have extra room */
46 #define PATHBUF		2048
47 
48 #define SUCCESS		20
49 #define NOT_FOUND	51
50 #define BAD_REQUEST	59
51 
52 #ifndef MAX_USERS
53 #define MAX_USERS	64
54 #endif
55 
56 enum {
57 	S_OPEN,
58 	S_INITIALIZING,
59 	S_SENDING,
60 	S_CLOSING,
61 };
62 
63 struct client {
64 	struct tls	*ctx;
65 	int		 state;
66 	int		 code;
67 	const char	*meta;
68 	int		 fd;
69 	void		*buf, *i;
70 	ssize_t		 len, off;
71 };
72 
73 struct etm {			/* file extension to mime */
74 	const char	*mime;
75 	const char	*ext;
76 } filetypes[] = {
77 	{"application/pdf",	"pdf"},
78 
79 	{"image/gif",		"gif"},
80 	{"image/jpeg",		"jpg"},
81 	{"image/jpeg",		"jpeg"},
82 	{"image/png",		"png"},
83 	{"image/svg+xml",	"svg"},
84 
85 	{"text/gemini",		"gemini"},
86 	{"text/gemini",		"gmi"},
87 	{"text/markdown",	"markdown"},
88 	{"text/markdown",	"md"},
89 	{"text/plain",		"txt"},
90 
91 	{NULL, NULL}
92 };
93 
94 int dirfd;
95 
96 char		*url_after_proto(char*);
97 char		*url_start_of_request(char*);
98 int		 url_trim(char*);
99 void		 adjust_path(char*);
100 int		 path_isdir(char*);
101 ssize_t		 filesize(int);
102 
103 int		 start_reply(struct pollfd*, struct client*, int, const char*);
104 int		 isdir(int);
105 const char	*path_ext(const char*);
106 const char	*mime(const char*);
107 int		 open_file(char*, struct pollfd*, struct client*);
108 void		 send_file(char*, struct pollfd*, struct client*);
109 void		 send_dir(char*, struct pollfd*, struct client*);
110 void		 handle(struct pollfd*, struct client*);
111 
112 void		 mark_nonblock(int);
113 int		 make_soket(int);
114 void		 do_accept(int, struct tls*, struct pollfd*, struct client*);
115 void		 goodbye(struct pollfd*, struct client*);
116 void		 loop(struct tls*, int);
117 
118 void		 usage(const char*);
119 
120 char *
121 url_after_proto(char *url)
122 {
123 	char *s;
124 	const char *proto = "gemini";
125 	const char *marker = "://";
126 
127 	if ((s = strstr(url, marker)) == NULL)
128 		return url;
129 
130 	/* not a gemini:// URL */
131 
132 	if (s - strlen(proto) < url)
133 		return NULL;
134 	/* TODO: */
135 	/* if (strcmp(s - strlen(proto), proto)) */
136 	/* 	return NULL; */
137 
138 	/* a valid gemini:// URL */
139 	return s + strlen(marker);
140 }
141 
142 char *
143 url_start_of_request(char *url)
144 {
145 	char *s, *t;
146 
147 	if ((s = url_after_proto(url)) == NULL)
148 		return NULL;
149 
150 	if ((t = strstr(s, "/")) == NULL)
151 		return s + strlen(s);
152 	return t;
153 }
154 
155 int
156 url_trim(char *url)
157 {
158 	const char *e = "\r\n";
159 	char *s;
160 
161 	if ((s = strstr(url, e)) == NULL)
162 		return 0;
163 	s[0] = '\0';
164 	s[1] = '\0';
165 
166 	if (s[2] != '\0') {
167 		fprintf(stderr, "the request was longer than 1024 bytes\n");
168 		return 0;
169 	}
170 
171 	return 1;
172 }
173 
174 void
175 adjust_path(char *path)
176 {
177 	char *s;
178 	size_t len;
179 
180 	/* /.. -> / */
181 	len = strlen(path);
182 	if (len >= 3) {
183 		if (!strcmp(&path[len-3], "/..")) {
184 			path[len-2] = '\0';
185 		}
186 	}
187 
188 	/* if the path is only `..` trim out and exit */
189 	if (!strcmp(path, "..")) {
190 		path[0] = '\0';
191 		return;
192 	}
193 
194 	/* remove every ../ in the path */
195 	while (1) {
196 		if ((s = strstr(path, "../")) == NULL)
197 			return;
198 		memmove(s, s+3, strlen(s)+1);	/* copy also the \0 */
199 	}
200 }
201 
202 int
203 path_isdir(char *path)
204 {
205 	if (*path == '\0')
206 		return 1;
207 	return path[strlen(path)-1] == '/';
208 }
209 
210 int
211 start_reply(struct pollfd *pfd, struct client *client, int code, const char *reason)
212 {
213 	char buf[1030] = {0}; 	/* status + ' ' + max reply len + \r\n\0 */
214 	int len;
215 	int ret;
216 
217 	client->code = code;
218 	client->meta = reason;
219 	client->state = S_INITIALIZING;
220 
221 	len = snprintf(buf, sizeof(buf), "%d %s\r\n", code, reason);
222 	assert(len < (int)sizeof(buf));
223 	ret = tls_write(client->ctx, buf, len);
224 	if (ret == TLS_WANT_POLLIN) {
225 		pfd->events = POLLIN;
226 		return 0;
227 	}
228 
229 	if (ret == TLS_WANT_POLLOUT) {
230 		pfd->events = POLLOUT;
231 		return 0;
232 	}
233 
234 	return 1;
235 }
236 
237 int
238 isdir(int fd)
239 {
240 	struct stat sb;
241 
242 	if (fstat(fd, &sb) == -1) {
243 		warn("fstat");
244 		return 1; 	/* we'll probably fail later on anyway */
245 	}
246 
247 	return S_ISDIR(sb.st_mode);
248 }
249 
250 ssize_t
251 filesize(int fd)
252 {
253 	ssize_t len;
254 
255 	if ((len = lseek(fd, 0, SEEK_END)) == -1)
256 		return -1;
257 	if (lseek(fd, 0, SEEK_SET) == -1)
258 		return -1;
259 	return len;
260 }
261 
262 const char *
263 path_ext(const char *path)
264 {
265 	const char *end;
266 
267 	end = path + strlen(path)-1; /* the last byte before the NUL */
268 	for (; end != path; --end) {
269 		if (*end == '.')
270 			return end+1;
271 		if (*end == '/')
272 			break;
273 	}
274 
275 	return NULL;
276 }
277 
278 const char *
279 mime(const char *path)
280 {
281 	const char *ext, *def = "application/octet-stream";
282 	struct etm *t;
283 
284 	if ((ext = path_ext(path)) == NULL)
285 		return def;
286 
287 	for (t = filetypes; t->mime != NULL; ++t)
288 		if (!strcmp(ext, t->ext))
289 			return t->mime;
290 
291 	return def;
292 }
293 
294 int
295 open_file(char *path, struct pollfd *fds, struct client *c)
296 {
297 	char fpath[PATHBUF];
298 
299 	assert(path != NULL);
300 
301 	bzero(fpath, sizeof(fpath));
302 
303 	if (*path != '.')
304 		fpath[0] = '.';
305 	strlcat(fpath, path, PATHBUF);
306 
307 	if ((c->fd = openat(dirfd, fpath, O_RDONLY | O_NOFOLLOW)) == -1) {
308 		warn("open: %s", fpath);
309 		if (!start_reply(fds, c, NOT_FOUND, "not found"))
310 			return 0;
311 		goodbye(fds, c);
312 		return 0;
313 	}
314 
315 	if (isdir(c->fd)) {
316 		warnx("%s is a directory, trying %s/index.gmi", fpath, fpath);
317 		close(c->fd);
318 		c->fd = -1;
319 		send_dir(fpath, fds, c);
320 		return 0;
321 	}
322 
323 	if ((c->len = filesize(c->fd)) == -1) {
324 		warn("filesize: %s", fpath);
325 		goodbye(fds, c);
326 		return 0;
327 	}
328 
329 	if ((c->buf = mmap(NULL, c->len, PROT_READ, MAP_PRIVATE,
330 		    c->fd, 0)) == MAP_FAILED) {
331 		warn("mmap: %s", fpath);
332 		goodbye(fds, c);
333 		return 0;
334 	}
335 	c->i = c->buf;
336 
337 	return start_reply(fds, c, SUCCESS, mime(fpath));
338 }
339 
340 void
341 send_file(char *path, struct pollfd *fds, struct client *c)
342 {
343 	ssize_t ret, len;
344 
345 	if (c->fd == -1) {
346 		if (!open_file(path, fds, c))
347 			return;
348 		c->state = S_SENDING;
349 	}
350 
351 	len = (c->buf + c->len) - c->i;
352 
353 	while (len > 0) {
354 		switch (ret = tls_write(c->ctx, c->i, len)) {
355 		case -1:
356 			warnx("tls_write: %s", tls_error(c->ctx));
357 			goodbye(fds, c);
358 			return;
359 
360 		case TLS_WANT_POLLIN:
361 			fds->events = POLLIN;
362 			return;
363 
364 		case TLS_WANT_POLLOUT:
365 			fds->events = POLLOUT;
366 			return;
367 
368 		default:
369 			c->i += ret;
370 			len -= ret;
371 			break;
372 		}
373 	}
374 
375 	goodbye(fds, c);
376 }
377 
378 void
379 send_dir(char *path, struct pollfd *fds, struct client *client)
380 {
381 	char fpath[PATHBUF];
382 	size_t len;
383 
384 	bzero(fpath, PATHBUF);
385 
386 	if (path[0] != '.')
387 		fpath[0] = '.';
388 
389 	/* this cannot fail since sizeof(fpath) > maxlen of path */
390 	strlcat(fpath, path, PATHBUF);
391 	len = strlen(fpath);
392 
393 	/* add a trailing / in case. */
394 	if (fpath[len-1] != '/') {
395 		fpath[len] = '/';
396 	}
397 
398 	strlcat(fpath, "index.gmi", sizeof(fpath));
399 
400 	send_file(fpath, fds, client);
401 }
402 
403 void
404 handle(struct pollfd *fds, struct client *client)
405 {
406 	char buf[GEMINI_URL_LEN];
407 	char *path;
408 
409 	switch (client->state) {
410 	case S_OPEN:
411 		bzero(buf, GEMINI_URL_LEN);
412 		switch (tls_read(client->ctx, buf, sizeof(buf)-1)) {
413 		case -1:
414 			warnx("tls_read: %s", tls_error(client->ctx));
415 			goodbye(fds, client);
416 			return;
417 
418 		case TLS_WANT_POLLIN:
419 			fds->events = POLLIN;
420 			return;
421 
422 		case TLS_WANT_POLLOUT:
423 			fds->events = POLLOUT;
424 			return;
425 		}
426 
427 		if (!url_trim(buf)) {
428 			if (!start_reply(fds, client, BAD_REQUEST, "bad request"))
429 				return;
430 			goodbye(fds, client);
431 			return;
432 		}
433 
434 		if ((path = url_start_of_request(buf)) == NULL) {
435 			if (!start_reply(fds, client, BAD_REQUEST, "bad request"))
436 				return;
437 			goodbye(fds, client);
438 			return;
439 		}
440 
441 		adjust_path(path);
442 		fprintf(stderr, "requested path: %s\n", path);
443 
444 		if (path_isdir(path))
445 			send_dir(path, fds, client);
446 		else
447 			send_file(path, fds, client);
448 		break;
449 
450 	case S_INITIALIZING:
451 		if (!start_reply(fds, client, client->code, client->meta))
452 			return;
453 
454 		if (client->code != SUCCESS) {
455 			/* we don't need a body */
456 			goodbye(fds, client);
457 			return;
458 		}
459 
460 		client->state = S_SENDING;
461 
462 		/* fallthrough */
463 
464 	case S_SENDING:
465 		send_file(NULL, fds, client);
466 		break;
467 
468 	case S_CLOSING:
469 		goodbye(fds, client);
470 		break;
471 
472 	default:
473 		/* unreachable */
474 		abort();
475 	}
476 }
477 
478 void
479 mark_nonblock(int fd)
480 {
481 	int flags;
482 
483 	if ((flags = fcntl(fd, F_GETFL)) == -1)
484 		err(1, "fcntl(F_GETFL)");
485 	if (fcntl(fd, F_SETFL, flags | O_NONBLOCK) == -1)
486 		err(1, "fcntl(F_SETFL)");
487 }
488 
489 int
490 make_socket(int port)
491 {
492 	int sock, v;
493 	struct sockaddr_in addr;
494 
495 	if ((sock = socket(AF_INET, SOCK_STREAM, 0)) == -1)
496 		err(1, "socket");
497 
498 	v = 1;
499 	if (setsockopt(sock, SOL_SOCKET, SO_REUSEADDR, &v, sizeof(v)) == -1)
500 		err(1, "setsockopt(SO_REUSEADDR)");
501 
502 	v = 1;
503 	if (setsockopt(sock, SOL_SOCKET, SO_REUSEPORT, &v, sizeof(v)) == -1)
504 		err(1, "setsockopt(SO_REUSEPORT)");
505 
506 	mark_nonblock(sock);
507 
508 	bzero(&addr, sizeof(addr));
509 	addr.sin_family = AF_INET;
510 	addr.sin_port = htons(port);
511 	addr.sin_addr.s_addr = INADDR_ANY;
512 
513 	if (bind(sock, (struct sockaddr*)&addr, sizeof(addr)) == -1)
514 		err(1, "bind");
515 
516 	if (listen(sock, 16) == -1)
517 		err(1, "listen");
518 
519 	return sock;
520 }
521 
522 void
523 do_accept(int sock, struct tls *ctx, struct pollfd *fds, struct client *clients)
524 {
525 	int i, fd;
526 	struct sockaddr_in addr;
527 	socklen_t len;
528 
529 	len = sizeof(addr);
530 	if ((fd = accept(sock, (struct sockaddr*)&addr, &len)) == -1) {
531 		if (errno == EWOULDBLOCK)
532 			return;
533 		err(1, "accept");
534 	}
535 
536 	mark_nonblock(fd);
537 
538 	for (i = 0; i < MAX_USERS; ++i) {
539 		if (fds[i].fd == -1) {
540 			bzero(&clients[i], sizeof(struct client));
541 			if (tls_accept_socket(ctx, &clients[i].ctx, fd) == -1)
542 				break; /* goodbye fd! */
543 
544 			fds[i].fd = fd;
545 			fds[i].events = POLLIN;
546 
547 			clients[i].state = S_OPEN;
548 			clients[i].fd = -1;
549 			clients[i].buf = MAP_FAILED;
550 
551 			return;
552 		}
553 	}
554 
555 	close(fd);
556 }
557 
558 void
559 goodbye(struct pollfd *pfd, struct client *c)
560 {
561 	ssize_t ret;
562 
563 	c->state = S_CLOSING;
564 
565 	ret = tls_close(c->ctx);
566 	if (ret == TLS_WANT_POLLIN) {
567 		pfd->events = POLLIN;
568 		return;
569 	}
570 	if (ret == TLS_WANT_POLLOUT) {
571 		pfd->events = POLLOUT;
572 		return;
573 	}
574 
575 	tls_free(c->ctx);
576 	c->ctx = NULL;
577 
578 	if (c->buf != MAP_FAILED)
579 		munmap(c->buf, c->len);
580 
581 	if (c->fd != -1)
582 		close(c->fd);
583 
584 	close(pfd->fd);
585 	pfd->fd = -1;
586 }
587 
588 void
589 loop(struct tls *ctx, int sock)
590 {
591 	int i, todo;
592 	struct client clients[MAX_USERS];
593 	struct pollfd fds[MAX_USERS];
594 
595 	for (i = 0; i < MAX_USERS; ++i) {
596 		fds[i].fd = -1;
597 		fds[i].events = POLLIN;
598 		bzero(&clients[i], sizeof(struct client));
599 	}
600 
601 	fds[0].fd = sock;
602 
603 	for (;;) {
604 		if ((todo = poll(fds, MAX_USERS, INFTIM)) == -1)
605 			err(1, "poll");
606 
607 		for (i = 0; i < MAX_USERS; i++) {
608 			assert(i < MAX_USERS);
609 
610 			if (fds[i].revents == 0)
611 				continue;
612 
613 			if (fds[i].revents & (POLLERR|POLLNVAL))
614 				err(1, "bad fd %d", fds[i].fd);
615 
616 			if (fds[i].revents & POLLHUP) {
617 				goodbye(&fds[i], &clients[i]);
618 				continue;
619 			}
620 
621 			todo--;
622 
623 			if (i == 0) { /* new client */
624 				do_accept(sock, ctx, fds, clients);
625 				continue;
626 			}
627 
628 			handle(&fds[i], &clients[i]);
629 		}
630 	}
631 }
632 
633 void
634 usage(const char *me)
635 {
636 	fprintf(stderr,
637 	    "USAGE: %s [-h] [-c cert.pem] [-d docs] [-k key.pem]\n",
638 	    me);
639 }
640 
641 int
642 main(int argc, char **argv)
643 {
644 	const char *cert = "cert.pem", *key = "key.pem", *dir = "docs";
645 	struct tls *ctx = NULL;
646 	struct tls_config *conf;
647 	int sock, ch;
648 
649 	while ((ch = getopt(argc, argv, "c:d:hk:")) != -1) {
650 		switch (ch) {
651 		case 'c':
652 			cert = optarg;
653 			break;
654 
655 		case 'd':
656 			dir = optarg;
657 			break;
658 
659 		case 'h':
660 			usage(*argv);
661 			return 0;
662 
663 		case 'k':
664 			key = optarg;
665 			break;
666 
667 		default:
668 			usage(*argv);
669 			return 1;
670 		}
671 	}
672 
673 	if ((conf = tls_config_new()) == NULL)
674 		err(1, "tls_config_new");
675 
676 	if (tls_config_set_protocols(conf,
677 	    TLS_PROTOCOL_TLSv1_2 | TLS_PROTOCOL_TLSv1_3) == -1)
678 		err(1, "tls_config_set_protocols");
679 
680 	if (tls_config_set_cert_file(conf, cert) == -1)
681 		err(1, "tls_config_set_cert_file: %s", cert);
682 
683 	if (tls_config_set_key_file(conf, key) == -1)
684 		err(1, "tls_config_set_key_file: %s", key);
685 
686 	if ((ctx = tls_server()) == NULL)
687 		err(1, "tls_server");
688 
689 	if (tls_configure(ctx, conf) == -1)
690 		errx(1, "tls_configure: %s", tls_error(ctx));
691 
692 	sock = make_socket(1965);
693 
694 	if ((dirfd = open(dir, O_RDONLY | O_DIRECTORY)) == -1)
695 		err(1, "open: %s", dir);
696 
697 	if (unveil(dir, "r") == -1)
698 		err(1, "unveil");
699 
700 	if (pledge("stdio rpath inet", "") == -1)
701 		err(1, "pledge");
702 
703 	loop(ctx, sock);
704 
705 	close(sock);
706 	tls_free(ctx);
707 	tls_config_free(conf);
708 }