xref: /freebsd/bin/cat/cat.c (revision 2e3f49888ec8851bafb22011533217487764fdb0)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1989, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Kevin Fall.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 
35 #include <sys/capsicum.h>
36 #include <sys/param.h>
37 #include <sys/stat.h>
38 #ifndef NO_UDOM_SUPPORT
39 #include <sys/socket.h>
40 #include <sys/un.h>
41 #include <netdb.h>
42 #endif
43 
44 #include <capsicum_helpers.h>
45 #include <ctype.h>
46 #include <err.h>
47 #include <errno.h>
48 #include <fcntl.h>
49 #include <locale.h>
50 #include <stdio.h>
51 #include <stdlib.h>
52 #include <string.h>
53 #include <unistd.h>
54 #include <wchar.h>
55 #include <wctype.h>
56 
57 #include <libcasper.h>
58 #include <casper/cap_fileargs.h>
59 #include <casper/cap_net.h>
60 
61 static int bflag, eflag, lflag, nflag, sflag, tflag, vflag;
62 static int rval;
63 static const char *filename;
64 static fileargs_t *fa;
65 
66 static void usage(void) __dead2;
67 static void scanfiles(char *argv[], int cooked);
68 #ifndef BOOTSTRAP_CAT
69 static void cook_cat(FILE *);
70 static ssize_t in_kernel_copy(int);
71 #endif
72 static void raw_cat(int);
73 
74 #ifndef NO_UDOM_SUPPORT
75 static cap_channel_t *capnet;
76 
77 static int udom_open(const char *path, int flags);
78 #endif
79 
80 /*
81  * Memory strategy threshold, in pages: if physmem is larger than this,
82  * use a large buffer.
83  */
84 #define	PHYSPAGES_THRESHOLD (32 * 1024)
85 
86 /* Maximum buffer size in bytes - do not allow it to grow larger than this. */
87 #define	BUFSIZE_MAX (2 * 1024 * 1024)
88 
89 /*
90  * Small (default) buffer size in bytes. It's inefficient for this to be
91  * smaller than MAXPHYS.
92  */
93 #define	BUFSIZE_SMALL (MAXPHYS)
94 
95 
96 /*
97  * For the bootstrapped cat binary (needed for locked appending to METALOG), we
98  * disable all flags except -l and -u to avoid non-portable function calls.
99  * In the future we may instead want to write a small portable bootstrap tool
100  * that locks the output file before writing to it. However, for now
101  * bootstrapping cat without multibyte support is the simpler solution.
102  */
103 #ifdef BOOTSTRAP_CAT
104 #define SUPPORTED_FLAGS "lu"
105 #else
106 #define SUPPORTED_FLAGS "belnstuv"
107 #endif
108 
109 #ifndef NO_UDOM_SUPPORT
110 static void
111 init_casper_net(cap_channel_t *casper)
112 {
113 	cap_net_limit_t *limit;
114 	int familylimit;
115 
116 	capnet = cap_service_open(casper, "system.net");
117 	if (capnet == NULL)
118 		err(EXIT_FAILURE, "unable to create network service");
119 
120 	limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR |
121 	    CAPNET_CONNECTDNS);
122 	if (limit == NULL)
123 		err(EXIT_FAILURE, "unable to create limits");
124 
125 	familylimit = AF_LOCAL;
126 	cap_net_limit_name2addr_family(limit, &familylimit, 1);
127 
128 	if (cap_net_limit(limit) < 0)
129 		err(EXIT_FAILURE, "unable to apply limits");
130 }
131 #endif
132 
133 static void
134 init_casper(int argc, char *argv[])
135 {
136 	cap_channel_t *casper;
137 	cap_rights_t rights;
138 
139 	casper = cap_init();
140 	if (casper == NULL)
141 		err(EXIT_FAILURE, "unable to create Casper");
142 
143 	fa = fileargs_cinit(casper, argc, argv, O_RDONLY, 0,
144 	    cap_rights_init(&rights, CAP_READ | CAP_FSTAT | CAP_FCNTL | CAP_SEEK),
145 	    FA_OPEN | FA_REALPATH);
146 	if (fa == NULL)
147 		err(EXIT_FAILURE, "unable to create fileargs");
148 
149 #ifndef NO_UDOM_SUPPORT
150 	init_casper_net(casper);
151 #endif
152 
153 	cap_close(casper);
154 }
155 
156 int
157 main(int argc, char *argv[])
158 {
159 	int ch;
160 	struct flock stdout_lock;
161 
162 	setlocale(LC_CTYPE, "");
163 
164 	while ((ch = getopt(argc, argv, SUPPORTED_FLAGS)) != -1)
165 		switch (ch) {
166 		case 'b':
167 			bflag = nflag = 1;	/* -b implies -n */
168 			break;
169 		case 'e':
170 			eflag = vflag = 1;	/* -e implies -v */
171 			break;
172 		case 'l':
173 			lflag = 1;
174 			break;
175 		case 'n':
176 			nflag = 1;
177 			break;
178 		case 's':
179 			sflag = 1;
180 			break;
181 		case 't':
182 			tflag = vflag = 1;	/* -t implies -v */
183 			break;
184 		case 'u':
185 			setbuf(stdout, NULL);
186 			break;
187 		case 'v':
188 			vflag = 1;
189 			break;
190 		default:
191 			usage();
192 		}
193 	argv += optind;
194 	argc -= optind;
195 
196 	if (lflag) {
197 		stdout_lock.l_len = 0;
198 		stdout_lock.l_start = 0;
199 		stdout_lock.l_type = F_WRLCK;
200 		stdout_lock.l_whence = SEEK_SET;
201 		if (fcntl(STDOUT_FILENO, F_SETLKW, &stdout_lock) == -1)
202 			err(EXIT_FAILURE, "stdout");
203 	}
204 
205 	init_casper(argc, argv);
206 
207 	caph_cache_catpages();
208 
209 	if (caph_enter_casper() < 0)
210 		err(EXIT_FAILURE, "capsicum");
211 
212 	if (bflag || eflag || nflag || sflag || tflag || vflag)
213 		scanfiles(argv, 1);
214 	else
215 		scanfiles(argv, 0);
216 	if (fclose(stdout))
217 		err(1, "stdout");
218 	exit(rval);
219 	/* NOTREACHED */
220 }
221 
222 static void
223 usage(void)
224 {
225 
226 	fprintf(stderr, "usage: cat [-" SUPPORTED_FLAGS "] [file ...]\n");
227 	exit(1);
228 	/* NOTREACHED */
229 }
230 
231 static void
232 scanfiles(char *argv[], int cooked __unused)
233 {
234 	int fd, i;
235 	char *path;
236 #ifndef BOOTSTRAP_CAT
237 	FILE *fp;
238 #endif
239 
240 	i = 0;
241 	fd = -1;
242 	while ((path = argv[i]) != NULL || i == 0) {
243 		if (path == NULL || strcmp(path, "-") == 0) {
244 			filename = "stdin";
245 			fd = STDIN_FILENO;
246 		} else {
247 			filename = path;
248 			fd = fileargs_open(fa, path);
249 #ifndef NO_UDOM_SUPPORT
250 			if (fd < 0 && errno == EOPNOTSUPP)
251 				fd = udom_open(path, O_RDONLY);
252 #endif
253 		}
254 		if (fd < 0) {
255 			warn("%s", path);
256 			rval = 1;
257 #ifndef BOOTSTRAP_CAT
258 		} else if (cooked) {
259 			if (fd == STDIN_FILENO)
260 				cook_cat(stdin);
261 			else {
262 				fp = fdopen(fd, "r");
263 				cook_cat(fp);
264 				fclose(fp);
265 			}
266 #endif
267 		} else {
268 #ifndef BOOTSTRAP_CAT
269 			if (in_kernel_copy(fd) == -1) {
270 				if (errno == EINVAL || errno == EBADF ||
271 				    errno == EISDIR)
272 					raw_cat(fd);
273 				else
274 					err(1, "stdout");
275 			}
276 #else
277 			raw_cat(fd);
278 #endif
279 			if (fd != STDIN_FILENO)
280 				close(fd);
281 		}
282 		if (path == NULL)
283 			break;
284 		++i;
285 	}
286 }
287 
288 #ifndef BOOTSTRAP_CAT
289 static void
290 cook_cat(FILE *fp)
291 {
292 	int ch, gobble, line, prev;
293 	wint_t wch;
294 
295 	/* Reset EOF condition on stdin. */
296 	if (fp == stdin && feof(stdin))
297 		clearerr(stdin);
298 
299 	line = gobble = 0;
300 	for (prev = '\n'; (ch = getc(fp)) != EOF; prev = ch) {
301 		if (prev == '\n') {
302 			if (sflag) {
303 				if (ch == '\n') {
304 					if (gobble)
305 						continue;
306 					gobble = 1;
307 				} else
308 					gobble = 0;
309 			}
310 			if (nflag) {
311 				if (!bflag || ch != '\n') {
312 					(void)fprintf(stdout, "%6d\t", ++line);
313 					if (ferror(stdout))
314 						break;
315 				} else if (eflag) {
316 					(void)fprintf(stdout, "%6s\t", "");
317 					if (ferror(stdout))
318 						break;
319 				}
320 			}
321 		}
322 		if (ch == '\n') {
323 			if (eflag && putchar('$') == EOF)
324 				break;
325 		} else if (ch == '\t') {
326 			if (tflag) {
327 				if (putchar('^') == EOF || putchar('I') == EOF)
328 					break;
329 				continue;
330 			}
331 		} else if (vflag) {
332 			(void)ungetc(ch, fp);
333 			/*
334 			 * Our getwc(3) doesn't change file position
335 			 * on error.
336 			 */
337 			if ((wch = getwc(fp)) == WEOF) {
338 				if (ferror(fp) && errno == EILSEQ) {
339 					clearerr(fp);
340 					/* Resync attempt. */
341 					memset(&fp->_mbstate, 0, sizeof(mbstate_t));
342 					if ((ch = getc(fp)) == EOF)
343 						break;
344 					wch = ch;
345 					goto ilseq;
346 				} else
347 					break;
348 			}
349 			if (!iswascii(wch) && !iswprint(wch)) {
350 ilseq:
351 				if (putchar('M') == EOF || putchar('-') == EOF)
352 					break;
353 				wch = toascii(wch);
354 			}
355 			if (iswcntrl(wch)) {
356 				ch = toascii(wch);
357 				ch = (ch == '\177') ? '?' : (ch | 0100);
358 				if (putchar('^') == EOF || putchar(ch) == EOF)
359 					break;
360 				continue;
361 			}
362 			if (putwchar(wch) == WEOF)
363 				break;
364 			ch = -1;
365 			continue;
366 		}
367 		if (putchar(ch) == EOF)
368 			break;
369 	}
370 	if (ferror(fp)) {
371 		warn("%s", filename);
372 		rval = 1;
373 		clearerr(fp);
374 	}
375 	if (ferror(stdout))
376 		err(1, "stdout");
377 }
378 
379 static ssize_t
380 in_kernel_copy(int rfd)
381 {
382 	int wfd;
383 	ssize_t ret;
384 
385 	wfd = fileno(stdout);
386 	ret = 1;
387 
388 	while (ret > 0)
389 		ret = copy_file_range(rfd, NULL, wfd, NULL, SSIZE_MAX, 0);
390 
391 	return (ret);
392 }
393 #endif /* BOOTSTRAP_CAT */
394 
395 static void
396 raw_cat(int rfd)
397 {
398 	long pagesize;
399 	int off, wfd;
400 	ssize_t nr, nw;
401 	static size_t bsize;
402 	static char *buf = NULL;
403 	struct stat sbuf;
404 
405 	wfd = fileno(stdout);
406 	if (buf == NULL) {
407 		if (fstat(wfd, &sbuf))
408 			err(1, "stdout");
409 		if (S_ISREG(sbuf.st_mode)) {
410 			/* If there's plenty of RAM, use a large copy buffer */
411 			if (sysconf(_SC_PHYS_PAGES) > PHYSPAGES_THRESHOLD)
412 				bsize = MIN(BUFSIZE_MAX, MAXPHYS * 8);
413 			else
414 				bsize = BUFSIZE_SMALL;
415 		} else {
416 			bsize = sbuf.st_blksize;
417 			pagesize = sysconf(_SC_PAGESIZE);
418 			if (pagesize > 0)
419 				bsize = MAX(bsize, (size_t)pagesize);
420 		}
421 		if ((buf = malloc(bsize)) == NULL)
422 			err(1, "malloc() failure of IO buffer");
423 	}
424 	while ((nr = read(rfd, buf, bsize)) > 0)
425 		for (off = 0; nr; nr -= nw, off += nw)
426 			if ((nw = write(wfd, buf + off, (size_t)nr)) < 0)
427 				err(1, "stdout");
428 	if (nr < 0) {
429 		warn("%s", filename);
430 		rval = 1;
431 	}
432 }
433 
434 #ifndef NO_UDOM_SUPPORT
435 
436 static int
437 udom_open(const char *path, int flags)
438 {
439 	struct addrinfo hints, *res, *res0;
440 	char rpath[PATH_MAX];
441 	int error, fd, serrno;
442 	cap_rights_t rights;
443 
444 	/*
445 	 * Construct the unix domain socket address and attempt to connect.
446 	 */
447 	bzero(&hints, sizeof(hints));
448 	hints.ai_family = AF_LOCAL;
449 
450 	if (fileargs_realpath(fa, path, rpath) == NULL)
451 		return (-1);
452 
453 	error = cap_getaddrinfo(capnet, rpath, NULL, &hints, &res0);
454 	if (error) {
455 		warn("%s", gai_strerror(error));
456 		errno = EINVAL;
457 		return (-1);
458 	}
459 	cap_rights_init(&rights, CAP_CONNECT, CAP_READ, CAP_WRITE,
460 	    CAP_SHUTDOWN, CAP_FSTAT, CAP_FCNTL);
461 
462 	/* Default error if something goes wrong. */
463 	serrno = EINVAL;
464 
465 	for (res = res0; res != NULL; res = res->ai_next) {
466 		fd = socket(res->ai_family, res->ai_socktype,
467 		    res->ai_protocol);
468 		if (fd < 0) {
469 			serrno = errno;
470 			freeaddrinfo(res0);
471 			errno = serrno;
472 			return (-1);
473 		}
474 		if (caph_rights_limit(fd, &rights) < 0) {
475 			serrno = errno;
476 			close(fd);
477 			freeaddrinfo(res0);
478 			errno = serrno;
479 			return (-1);
480 		}
481 		error = cap_connect(capnet, fd, res->ai_addr, res->ai_addrlen);
482 		if (error == 0)
483 			break;
484 		else {
485 			serrno = errno;
486 			close(fd);
487 		}
488 	}
489 	freeaddrinfo(res0);
490 
491 	if (res == NULL) {
492 		errno = serrno;
493 		return (-1);
494 	}
495 
496 	/*
497 	 * handle the open flags by shutting down appropriate directions
498 	 */
499 
500 	switch (flags & O_ACCMODE) {
501 	case O_RDONLY:
502 		cap_rights_clear(&rights, CAP_WRITE);
503 		if (shutdown(fd, SHUT_WR) == -1)
504 			warn(NULL);
505 		break;
506 	case O_WRONLY:
507 		cap_rights_clear(&rights, CAP_READ);
508 		if (shutdown(fd, SHUT_RD) == -1)
509 			warn(NULL);
510 		break;
511 	default:
512 		break;
513 	}
514 
515 	cap_rights_clear(&rights, CAP_CONNECT, CAP_SHUTDOWN);
516 	if (caph_rights_limit(fd, &rights) < 0) {
517 		serrno = errno;
518 		close(fd);
519 		errno = serrno;
520 		return (-1);
521 	}
522 	return (fd);
523 }
524 
525 #endif
526