xref: /freebsd/bin/cat/cat.c (revision 401ab69cff8fa2320a9f8ea4baa114a6da6c952b)
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 #if 0
36 #ifndef lint
37 static char const copyright[] =
38 "@(#) Copyright (c) 1989, 1993\n\
39 	The Regents of the University of California.  All rights reserved.\n";
40 #endif /* not lint */
41 #endif
42 
43 #ifndef lint
44 #if 0
45 static char sccsid[] = "@(#)cat.c	8.2 (Berkeley) 4/27/95";
46 #endif
47 #endif /* not lint */
48 #include <sys/cdefs.h>
49 #include <sys/capsicum.h>
50 #include <sys/param.h>
51 #include <sys/stat.h>
52 #ifndef NO_UDOM_SUPPORT
53 #include <sys/socket.h>
54 #include <sys/un.h>
55 #include <netdb.h>
56 #endif
57 
58 #include <capsicum_helpers.h>
59 #include <ctype.h>
60 #include <err.h>
61 #include <errno.h>
62 #include <fcntl.h>
63 #include <locale.h>
64 #include <stdio.h>
65 #include <stdlib.h>
66 #include <string.h>
67 #include <unistd.h>
68 #include <wchar.h>
69 #include <wctype.h>
70 
71 #include <libcasper.h>
72 #include <casper/cap_fileargs.h>
73 #include <casper/cap_net.h>
74 
75 static int bflag, eflag, lflag, nflag, sflag, tflag, vflag;
76 static int rval;
77 static const char *filename;
78 static fileargs_t *fa;
79 
80 static void usage(void) __dead2;
81 static void scanfiles(char *argv[], int cooked);
82 #ifndef BOOTSTRAP_CAT
83 static void cook_cat(FILE *);
84 static ssize_t in_kernel_copy(int);
85 #endif
86 static void raw_cat(int);
87 
88 #ifndef NO_UDOM_SUPPORT
89 static cap_channel_t *capnet;
90 
91 static int udom_open(const char *path, int flags);
92 #endif
93 
94 /*
95  * Memory strategy threshold, in pages: if physmem is larger than this,
96  * use a large buffer.
97  */
98 #define	PHYSPAGES_THRESHOLD (32 * 1024)
99 
100 /* Maximum buffer size in bytes - do not allow it to grow larger than this. */
101 #define	BUFSIZE_MAX (2 * 1024 * 1024)
102 
103 /*
104  * Small (default) buffer size in bytes. It's inefficient for this to be
105  * smaller than MAXPHYS.
106  */
107 #define	BUFSIZE_SMALL (MAXPHYS)
108 
109 
110 /*
111  * For the bootstrapped cat binary (needed for locked appending to METALOG), we
112  * disable all flags except -l and -u to avoid non-portable function calls.
113  * In the future we may instead want to write a small portable bootstrap tool
114  * that locks the output file before writing to it. However, for now
115  * bootstrapping cat without multibyte support is the simpler solution.
116  */
117 #ifdef BOOTSTRAP_CAT
118 #define SUPPORTED_FLAGS "lu"
119 #else
120 #define SUPPORTED_FLAGS "belnstuv"
121 #endif
122 
123 #ifndef NO_UDOM_SUPPORT
124 static void
125 init_casper_net(cap_channel_t *casper)
126 {
127 	cap_net_limit_t *limit;
128 	int familylimit;
129 
130 	capnet = cap_service_open(casper, "system.net");
131 	if (capnet == NULL)
132 		err(EXIT_FAILURE, "unable to create network service");
133 
134 	limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR |
135 	    CAPNET_CONNECTDNS);
136 	if (limit == NULL)
137 		err(EXIT_FAILURE, "unable to create limits");
138 
139 	familylimit = AF_LOCAL;
140 	cap_net_limit_name2addr_family(limit, &familylimit, 1);
141 
142 	if (cap_net_limit(limit) < 0)
143 		err(EXIT_FAILURE, "unable to apply limits");
144 }
145 #endif
146 
147 static void
148 init_casper(int argc, char *argv[])
149 {
150 	cap_channel_t *casper;
151 	cap_rights_t rights;
152 
153 	casper = cap_init();
154 	if (casper == NULL)
155 		err(EXIT_FAILURE, "unable to create Casper");
156 
157 	fa = fileargs_cinit(casper, argc, argv, O_RDONLY, 0,
158 	    cap_rights_init(&rights, CAP_READ | CAP_FSTAT | CAP_FCNTL),
159 	    FA_OPEN | FA_REALPATH);
160 	if (fa == NULL)
161 		err(EXIT_FAILURE, "unable to create fileargs");
162 
163 #ifndef NO_UDOM_SUPPORT
164 	init_casper_net(casper);
165 #endif
166 
167 	cap_close(casper);
168 }
169 
170 int
171 main(int argc, char *argv[])
172 {
173 	int ch;
174 	struct flock stdout_lock;
175 
176 	setlocale(LC_CTYPE, "");
177 
178 	while ((ch = getopt(argc, argv, SUPPORTED_FLAGS)) != -1)
179 		switch (ch) {
180 		case 'b':
181 			bflag = nflag = 1;	/* -b implies -n */
182 			break;
183 		case 'e':
184 			eflag = vflag = 1;	/* -e implies -v */
185 			break;
186 		case 'l':
187 			lflag = 1;
188 			break;
189 		case 'n':
190 			nflag = 1;
191 			break;
192 		case 's':
193 			sflag = 1;
194 			break;
195 		case 't':
196 			tflag = vflag = 1;	/* -t implies -v */
197 			break;
198 		case 'u':
199 			setbuf(stdout, NULL);
200 			break;
201 		case 'v':
202 			vflag = 1;
203 			break;
204 		default:
205 			usage();
206 		}
207 	argv += optind;
208 	argc -= optind;
209 
210 	if (lflag) {
211 		stdout_lock.l_len = 0;
212 		stdout_lock.l_start = 0;
213 		stdout_lock.l_type = F_WRLCK;
214 		stdout_lock.l_whence = SEEK_SET;
215 		if (fcntl(STDOUT_FILENO, F_SETLKW, &stdout_lock) == -1)
216 			err(EXIT_FAILURE, "stdout");
217 	}
218 
219 	init_casper(argc, argv);
220 
221 	caph_cache_catpages();
222 
223 	if (caph_enter_casper() < 0)
224 		err(EXIT_FAILURE, "capsicum");
225 
226 	if (bflag || eflag || nflag || sflag || tflag || vflag)
227 		scanfiles(argv, 1);
228 	else
229 		scanfiles(argv, 0);
230 	if (fclose(stdout))
231 		err(1, "stdout");
232 	exit(rval);
233 	/* NOTREACHED */
234 }
235 
236 static void
237 usage(void)
238 {
239 
240 	fprintf(stderr, "usage: cat [-" SUPPORTED_FLAGS "] [file ...]\n");
241 	exit(1);
242 	/* NOTREACHED */
243 }
244 
245 static void
246 scanfiles(char *argv[], int cooked __unused)
247 {
248 	int fd, i;
249 	char *path;
250 #ifndef BOOTSTRAP_CAT
251 	FILE *fp;
252 #endif
253 
254 	i = 0;
255 	fd = -1;
256 	while ((path = argv[i]) != NULL || i == 0) {
257 		if (path == NULL || strcmp(path, "-") == 0) {
258 			filename = "stdin";
259 			fd = STDIN_FILENO;
260 		} else {
261 			filename = path;
262 			fd = fileargs_open(fa, path);
263 #ifndef NO_UDOM_SUPPORT
264 			if (fd < 0 && errno == EOPNOTSUPP)
265 				fd = udom_open(path, O_RDONLY);
266 #endif
267 		}
268 		if (fd < 0) {
269 			warn("%s", path);
270 			rval = 1;
271 #ifndef BOOTSTRAP_CAT
272 		} else if (cooked) {
273 			if (fd == STDIN_FILENO)
274 				cook_cat(stdin);
275 			else {
276 				fp = fdopen(fd, "r");
277 				cook_cat(fp);
278 				fclose(fp);
279 			}
280 #endif
281 		} else {
282 #ifndef BOOTSTRAP_CAT
283 			if (in_kernel_copy(fd) == -1) {
284 				if (errno == EINVAL || errno == EBADF)
285 					raw_cat(fd);
286 				else
287 					err(1, "stdout");
288 			}
289 #else
290 			raw_cat(fd);
291 #endif
292 			if (fd != STDIN_FILENO)
293 				close(fd);
294 		}
295 		if (path == NULL)
296 			break;
297 		++i;
298 	}
299 }
300 
301 #ifndef BOOTSTRAP_CAT
302 static void
303 cook_cat(FILE *fp)
304 {
305 	int ch, gobble, line, prev;
306 	wint_t wch;
307 
308 	/* Reset EOF condition on stdin. */
309 	if (fp == stdin && feof(stdin))
310 		clearerr(stdin);
311 
312 	line = gobble = 0;
313 	for (prev = '\n'; (ch = getc(fp)) != EOF; prev = ch) {
314 		if (prev == '\n') {
315 			if (sflag) {
316 				if (ch == '\n') {
317 					if (gobble)
318 						continue;
319 					gobble = 1;
320 				} else
321 					gobble = 0;
322 			}
323 			if (nflag) {
324 				if (!bflag || ch != '\n') {
325 					(void)fprintf(stdout, "%6d\t", ++line);
326 					if (ferror(stdout))
327 						break;
328 				} else if (eflag) {
329 					(void)fprintf(stdout, "%6s\t", "");
330 					if (ferror(stdout))
331 						break;
332 				}
333 			}
334 		}
335 		if (ch == '\n') {
336 			if (eflag && putchar('$') == EOF)
337 				break;
338 		} else if (ch == '\t') {
339 			if (tflag) {
340 				if (putchar('^') == EOF || putchar('I') == EOF)
341 					break;
342 				continue;
343 			}
344 		} else if (vflag) {
345 			(void)ungetc(ch, fp);
346 			/*
347 			 * Our getwc(3) doesn't change file position
348 			 * on error.
349 			 */
350 			if ((wch = getwc(fp)) == WEOF) {
351 				if (ferror(fp) && errno == EILSEQ) {
352 					clearerr(fp);
353 					/* Resync attempt. */
354 					memset(&fp->_mbstate, 0, sizeof(mbstate_t));
355 					if ((ch = getc(fp)) == EOF)
356 						break;
357 					wch = ch;
358 					goto ilseq;
359 				} else
360 					break;
361 			}
362 			if (!iswascii(wch) && !iswprint(wch)) {
363 ilseq:
364 				if (putchar('M') == EOF || putchar('-') == EOF)
365 					break;
366 				wch = toascii(wch);
367 			}
368 			if (iswcntrl(wch)) {
369 				ch = toascii(wch);
370 				ch = (ch == '\177') ? '?' : (ch | 0100);
371 				if (putchar('^') == EOF || putchar(ch) == EOF)
372 					break;
373 				continue;
374 			}
375 			if (putwchar(wch) == WEOF)
376 				break;
377 			ch = -1;
378 			continue;
379 		}
380 		if (putchar(ch) == EOF)
381 			break;
382 	}
383 	if (ferror(fp)) {
384 		warn("%s", filename);
385 		rval = 1;
386 		clearerr(fp);
387 	}
388 	if (ferror(stdout))
389 		err(1, "stdout");
390 }
391 
392 static ssize_t
393 in_kernel_copy(int rfd)
394 {
395 	int wfd;
396 	ssize_t ret;
397 
398 	wfd = fileno(stdout);
399 	ret = 1;
400 
401 	while (ret > 0)
402 		ret = copy_file_range(rfd, NULL, wfd, NULL, SSIZE_MAX, 0);
403 
404 	return (ret);
405 }
406 #endif /* BOOTSTRAP_CAT */
407 
408 static void
409 raw_cat(int rfd)
410 {
411 	long pagesize;
412 	int off, wfd;
413 	ssize_t nr, nw;
414 	static size_t bsize;
415 	static char *buf = NULL;
416 	struct stat sbuf;
417 
418 	wfd = fileno(stdout);
419 	if (buf == NULL) {
420 		if (fstat(wfd, &sbuf))
421 			err(1, "stdout");
422 		if (S_ISREG(sbuf.st_mode)) {
423 			/* If there's plenty of RAM, use a large copy buffer */
424 			if (sysconf(_SC_PHYS_PAGES) > PHYSPAGES_THRESHOLD)
425 				bsize = MIN(BUFSIZE_MAX, MAXPHYS * 8);
426 			else
427 				bsize = BUFSIZE_SMALL;
428 		} else {
429 			bsize = sbuf.st_blksize;
430 			pagesize = sysconf(_SC_PAGESIZE);
431 			if (pagesize > 0)
432 				bsize = MAX(bsize, (size_t)pagesize);
433 		}
434 		if ((buf = malloc(bsize)) == NULL)
435 			err(1, "malloc() failure of IO buffer");
436 	}
437 	while ((nr = read(rfd, buf, bsize)) > 0)
438 		for (off = 0; nr; nr -= nw, off += nw)
439 			if ((nw = write(wfd, buf + off, (size_t)nr)) < 0)
440 				err(1, "stdout");
441 	if (nr < 0) {
442 		warn("%s", filename);
443 		rval = 1;
444 	}
445 }
446 
447 #ifndef NO_UDOM_SUPPORT
448 
449 static int
450 udom_open(const char *path, int flags)
451 {
452 	struct addrinfo hints, *res, *res0;
453 	char rpath[PATH_MAX];
454 	int error, fd, serrno;
455 	cap_rights_t rights;
456 
457 	/*
458 	 * Construct the unix domain socket address and attempt to connect.
459 	 */
460 	bzero(&hints, sizeof(hints));
461 	hints.ai_family = AF_LOCAL;
462 
463 	if (fileargs_realpath(fa, path, rpath) == NULL)
464 		return (-1);
465 
466 	error = cap_getaddrinfo(capnet, rpath, NULL, &hints, &res0);
467 	if (error) {
468 		warn("%s", gai_strerror(error));
469 		errno = EINVAL;
470 		return (-1);
471 	}
472 	cap_rights_init(&rights, CAP_CONNECT, CAP_READ, CAP_WRITE,
473 	    CAP_SHUTDOWN, CAP_FSTAT, CAP_FCNTL);
474 
475 	/* Default error if something goes wrong. */
476 	serrno = EINVAL;
477 
478 	for (res = res0; res != NULL; res = res->ai_next) {
479 		fd = socket(res->ai_family, res->ai_socktype,
480 		    res->ai_protocol);
481 		if (fd < 0) {
482 			serrno = errno;
483 			freeaddrinfo(res0);
484 			errno = serrno;
485 			return (-1);
486 		}
487 		if (caph_rights_limit(fd, &rights) < 0) {
488 			serrno = errno;
489 			close(fd);
490 			freeaddrinfo(res0);
491 			errno = serrno;
492 			return (-1);
493 		}
494 		error = cap_connect(capnet, fd, res->ai_addr, res->ai_addrlen);
495 		if (error == 0)
496 			break;
497 		else {
498 			serrno = errno;
499 			close(fd);
500 		}
501 	}
502 	freeaddrinfo(res0);
503 
504 	if (res == NULL) {
505 		errno = serrno;
506 		return (-1);
507 	}
508 
509 	/*
510 	 * handle the open flags by shutting down appropriate directions
511 	 */
512 
513 	switch (flags & O_ACCMODE) {
514 	case O_RDONLY:
515 		cap_rights_clear(&rights, CAP_WRITE);
516 		if (shutdown(fd, SHUT_WR) == -1)
517 			warn(NULL);
518 		break;
519 	case O_WRONLY:
520 		cap_rights_clear(&rights, CAP_READ);
521 		if (shutdown(fd, SHUT_RD) == -1)
522 			warn(NULL);
523 		break;
524 	default:
525 		break;
526 	}
527 
528 	cap_rights_clear(&rights, CAP_CONNECT, CAP_SHUTDOWN);
529 	if (caph_rights_limit(fd, &rights) < 0) {
530 		serrno = errno;
531 		close(fd);
532 		errno = serrno;
533 		return (-1);
534 	}
535 	return (fd);
536 }
537 
538 #endif
539