xref: /freebsd/bin/dd/dd.c (revision 6a66acb565dde6d5b0ee52eef2a631fbb09df153)
1 /*-
2  * Copyright (c) 1991, 1993, 1994
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * This code is derived from software contributed to Berkeley by
6  * Keith Muller of the University of California, San Diego and Lance
7  * Visser of Convex Computer Corporation.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by the University of
20  *	California, Berkeley and its contributors.
21  * 4. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  */
37 
38 #ifndef lint
39 static char const copyright[] =
40 "@(#) Copyright (c) 1991, 1993, 1994\n\
41 	The Regents of the University of California.  All rights reserved.\n";
42 #endif /* not lint */
43 
44 #ifndef lint
45 #if 0
46 static char sccsid[] = "@(#)dd.c	8.5 (Berkeley) 4/2/94";
47 #endif
48 #endif /* not lint */
49 #include <sys/cdefs.h>
50 __FBSDID("$FreeBSD$");
51 
52 #include <sys/param.h>
53 #include <sys/stat.h>
54 #include <sys/conf.h>
55 #include <sys/disklabel.h>
56 #include <sys/filio.h>
57 #include <sys/time.h>
58 
59 #include <ctype.h>
60 #include <err.h>
61 #include <errno.h>
62 #include <fcntl.h>
63 #include <inttypes.h>
64 #include <locale.h>
65 #include <stdio.h>
66 #include <stdlib.h>
67 #include <string.h>
68 #include <unistd.h>
69 
70 #include "dd.h"
71 #include "extern.h"
72 
73 static void dd_close(void);
74 static void dd_in(void);
75 static void getfdtype(IO *);
76 static void setup(void);
77 
78 IO	in, out;		/* input/output state */
79 STAT	st;			/* statistics */
80 void	(*cfunc)(void);		/* conversion function */
81 uintmax_t cpy_cnt;		/* # of blocks to copy */
82 static off_t	pending = 0;	/* pending seek if sparse */
83 u_int	ddflags = 0;		/* conversion options */
84 size_t	cbsz;			/* conversion block size */
85 uintmax_t files_cnt = 1;	/* # of files to copy */
86 const	u_char *ctab;		/* conversion table */
87 
88 int
89 main(int argc __unused, char *argv[])
90 {
91 	(void)setlocale(LC_CTYPE, "");
92 	jcl(argv);
93 	setup();
94 
95 	(void)signal(SIGINFO, summaryx);
96 	(void)signal(SIGINT, terminate);
97 
98 	atexit(summary);
99 
100 	while (files_cnt--)
101 		dd_in();
102 
103 	dd_close();
104 	exit(0);
105 }
106 
107 static void
108 setup(void)
109 {
110 	u_int cnt;
111 	struct timeval tv;
112 
113 	if (in.name == NULL) {
114 		in.name = "stdin";
115 		in.fd = STDIN_FILENO;
116 	} else {
117 		in.fd = open(in.name, O_RDONLY, 0);
118 		if (in.fd == -1)
119 			err(1, "%s", in.name);
120 	}
121 
122 	getfdtype(&in);
123 
124 	if (files_cnt > 1 && !(in.flags & ISTAPE))
125 		errx(1, "files is not supported for non-tape devices");
126 
127 	if (out.name == NULL) {
128 		/* No way to check for read access here. */
129 		out.fd = STDOUT_FILENO;
130 		out.name = "stdout";
131 	} else {
132 #define	OFLAGS \
133     (O_CREAT | (ddflags & (C_SEEK | C_NOTRUNC) ? 0 : O_TRUNC))
134 		out.fd = open(out.name, O_RDWR | OFLAGS, DEFFILEMODE);
135 		/*
136 		 * May not have read access, so try again with write only.
137 		 * Without read we may have a problem if output also does
138 		 * not support seeks.
139 		 */
140 		if (out.fd == -1) {
141 			out.fd = open(out.name, O_WRONLY | OFLAGS, DEFFILEMODE);
142 			out.flags |= NOREAD;
143 		}
144 		if (out.fd == -1)
145 			err(1, "%s", out.name);
146 	}
147 
148 	getfdtype(&out);
149 
150 	/*
151 	 * Allocate space for the input and output buffers.  If not doing
152 	 * record oriented I/O, only need a single buffer.
153 	 */
154 	if (!(ddflags & (C_BLOCK | C_UNBLOCK))) {
155 		if ((in.db = malloc(out.dbsz + in.dbsz - 1)) == NULL)
156 			err(1, "input buffer");
157 		out.db = in.db;
158 	} else if ((in.db = malloc(MAX(in.dbsz, cbsz) + cbsz)) == NULL ||
159 	    (out.db = malloc(out.dbsz + cbsz)) == NULL)
160 		err(1, "output buffer");
161 	in.dbp = in.db;
162 	out.dbp = out.db;
163 
164 	/* Position the input/output streams. */
165 	if (in.offset)
166 		pos_in();
167 	if (out.offset)
168 		pos_out();
169 
170 	/*
171 	 * Truncate the output file.  If it fails on a type of output file
172 	 * that it should _not_ fail on, error out.
173 	 */
174 	if ((ddflags & (C_OF | C_SEEK | C_NOTRUNC)) == (C_OF | C_SEEK) &&
175 	    out.flags & ISTRUNC)
176 		if (ftruncate(out.fd, out.offset * out.dbsz) == -1)
177 			err(1, "truncating %s", out.name);
178 
179 	/*
180 	 * If converting case at the same time as another conversion, build a
181 	 * table that does both at once.  If just converting case, use the
182 	 * built-in tables.
183 	 */
184 	if (ddflags & (C_LCASE | C_UCASE)) {
185 		if (ddflags & (C_ASCII | C_EBCDIC)) {
186 			if (ddflags & C_LCASE) {
187 				for (cnt = 0; cnt <= 0377; ++cnt)
188 					casetab[cnt] = tolower(ctab[cnt]);
189 			} else {
190 				for (cnt = 0; cnt <= 0377; ++cnt)
191 					casetab[cnt] = toupper(ctab[cnt]);
192 			}
193 		} else {
194 			if (ddflags & C_LCASE) {
195 				for (cnt = 0; cnt <= 0377; ++cnt)
196 					casetab[cnt] = tolower((int)cnt);
197 			} else {
198 				for (cnt = 0; cnt <= 0377; ++cnt)
199 					casetab[cnt] = toupper((int)cnt);
200 			}
201 		}
202 		ctab = casetab;
203 	}
204 
205 	(void)gettimeofday(&tv, (struct timezone *)NULL);
206 	st.start = tv.tv_sec + tv.tv_usec * 1e-6;
207 }
208 
209 static void
210 getfdtype(IO *io)
211 {
212 	struct stat sb;
213 	int type;
214 
215 	if (fstat(io->fd, &sb) == -1)
216 		err(1, "%s", io->name);
217 	if (S_ISREG(sb.st_mode))
218 		io->flags |= ISTRUNC;
219 	if (S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode)) {
220 		if (ioctl(io->fd, FIODTYPE, &type) == -1) {
221 			err(1, "%s", io->name);
222 		} else {
223 			if (type & D_TAPE)
224 				io->flags |= ISTAPE;
225 			else if (type & (D_DISK | D_MEM))
226 				io->flags |= ISSEEK;
227 			if (S_ISCHR(sb.st_mode) && (type & D_TAPE) == 0)
228 				io->flags |= ISCHR;
229 		}
230 		return;
231 	}
232 	errno = 0;
233 	if (lseek(io->fd, (off_t)0, SEEK_CUR) == -1 && errno == ESPIPE)
234 		io->flags |= ISPIPE;
235 	else
236 		io->flags |= ISSEEK;
237 }
238 
239 static void
240 dd_in(void)
241 {
242 	ssize_t n;
243 
244 	for (;;) {
245 		switch (cpy_cnt) {
246 		case -1:			/* count=0 was specified */
247 			return;
248 		case 0:
249 			break;
250 		default:
251 			if (st.in_full + st.in_part >= (uintmax_t)cpy_cnt)
252 				return;
253 			break;
254 		}
255 
256 		/*
257 		 * Zero the buffer first if sync; if doing block operations,
258 		 * use spaces.
259 		 */
260 		if (ddflags & C_SYNC) {
261 			if (ddflags & (C_BLOCK | C_UNBLOCK))
262 				memset(in.dbp, ' ', in.dbsz);
263 			else
264 				memset(in.dbp, 0, in.dbsz);
265 		}
266 
267 		n = read(in.fd, in.dbp, in.dbsz);
268 		if (n == 0) {
269 			in.dbrcnt = 0;
270 			return;
271 		}
272 
273 		/* Read error. */
274 		if (n == -1) {
275 			/*
276 			 * If noerror not specified, die.  POSIX requires that
277 			 * the warning message be followed by an I/O display.
278 			 */
279 			if (!(ddflags & C_NOERROR))
280 				err(1, "%s", in.name);
281 			warn("%s", in.name);
282 			summary();
283 
284 			/*
285 			 * If it's a seekable file descriptor, seek past the
286 			 * error.  If your OS doesn't do the right thing for
287 			 * raw disks this section should be modified to re-read
288 			 * in sector size chunks.
289 			 */
290 			if (in.flags & ISSEEK &&
291 			    lseek(in.fd, (off_t)in.dbsz, SEEK_CUR))
292 				warn("%s", in.name);
293 
294 			/* If sync not specified, omit block and continue. */
295 			if (!(ddflags & C_SYNC))
296 				continue;
297 
298 			/* Read errors count as full blocks. */
299 			in.dbcnt += in.dbrcnt = in.dbsz;
300 			++st.in_full;
301 
302 		/* Handle full input blocks. */
303 		} else if ((size_t)n == in.dbsz) {
304 			in.dbcnt += in.dbrcnt = n;
305 			++st.in_full;
306 
307 		/* Handle partial input blocks. */
308 		} else {
309 			/* If sync, use the entire block. */
310 			if (ddflags & C_SYNC)
311 				in.dbcnt += in.dbrcnt = in.dbsz;
312 			else
313 				in.dbcnt += in.dbrcnt = n;
314 			++st.in_part;
315 		}
316 
317 		/*
318 		 * POSIX states that if bs is set and no other conversions
319 		 * than noerror, notrunc or sync are specified, the block
320 		 * is output without buffering as it is read.
321 		 */
322 		if (ddflags & C_BS) {
323 			out.dbcnt = in.dbcnt;
324 			dd_out(1);
325 			in.dbcnt = 0;
326 			continue;
327 		}
328 
329 		if (ddflags & C_SWAB) {
330 			if ((n = in.dbrcnt) & 1) {
331 				++st.swab;
332 				--n;
333 			}
334 			swab(in.dbp, in.dbp, (size_t)n);
335 		}
336 
337 		in.dbp += in.dbrcnt;
338 		(*cfunc)();
339 	}
340 }
341 
342 /*
343  * Clean up any remaining I/O and flush output.  If necessary, the output file
344  * is truncated.
345  */
346 static void
347 dd_close(void)
348 {
349 	if (cfunc == def)
350 		def_close();
351 	else if (cfunc == block)
352 		block_close();
353 	else if (cfunc == unblock)
354 		unblock_close();
355 	if (ddflags & C_OSYNC && out.dbcnt && out.dbcnt < out.dbsz) {
356 		if (ddflags & (C_BLOCK | C_UNBLOCK))
357 			memset(out.dbp, ' ', out.dbsz - out.dbcnt);
358 		else
359 			memset(out.dbp, 0, out.dbsz - out.dbcnt);
360 		out.dbcnt = out.dbsz;
361 	}
362 	if (out.dbcnt || pending)
363 		dd_out(1);
364 }
365 
366 void
367 dd_out(int force)
368 {
369 	u_char *outp;
370 	size_t cnt, i, n;
371 	ssize_t nw;
372 	static int warned;
373 	int sparse;
374 
375 	/*
376 	 * Write one or more blocks out.  The common case is writing a full
377 	 * output block in a single write; increment the full block stats.
378 	 * Otherwise, we're into partial block writes.  If a partial write,
379 	 * and it's a character device, just warn.  If a tape device, quit.
380 	 *
381 	 * The partial writes represent two cases.  1: Where the input block
382 	 * was less than expected so the output block was less than expected.
383 	 * 2: Where the input block was the right size but we were forced to
384 	 * write the block in multiple chunks.  The original versions of dd(1)
385 	 * never wrote a block in more than a single write, so the latter case
386 	 * never happened.
387 	 *
388 	 * One special case is if we're forced to do the write -- in that case
389 	 * we play games with the buffer size, and it's usually a partial write.
390 	 */
391 	outp = out.db;
392 	for (n = force ? out.dbcnt : out.dbsz;; n = out.dbsz) {
393 		for (cnt = n;; cnt -= nw) {
394 			sparse = 0;
395 			if (ddflags & C_SPARSE) {
396 				sparse = 1;	/* Is buffer sparse? */
397 				for (i = 0; i < cnt; i++)
398 					if (outp[i] != 0) {
399 						sparse = 0;
400 						break;
401 					}
402 			}
403 			if (sparse && !force) {
404 				pending += cnt;
405 				nw = cnt;
406 			} else {
407 				if (pending != 0) {
408 					if (force)
409 						pending--;
410 					if (lseek(out.fd, pending, SEEK_CUR) ==
411 					    -1)
412 						err(2, "%s: seek error creating sparse file",
413 						    out.name);
414 					if (force)
415 						write(out.fd, outp, 1);
416 					pending = 0;
417 				}
418 				if (cnt)
419 					nw = write(out.fd, outp, cnt);
420 				else
421 					return;
422 			}
423 
424 			if (nw <= 0) {
425 				if (nw == 0)
426 					errx(1, "%s: end of device", out.name);
427 				if (errno != EINTR)
428 					err(1, "%s", out.name);
429 				nw = 0;
430 			}
431 			outp += nw;
432 			st.bytes += nw;
433 			if ((size_t)nw == n) {
434 				if (n != out.dbsz)
435 					++st.out_part;
436 				else
437 					++st.out_full;
438 				break;
439 			}
440 			++st.out_part;
441 			if ((size_t)nw == cnt)
442 				break;
443 			if (out.flags & ISTAPE)
444 				errx(1, "%s: short write on tape device",
445 				    out.name);
446 			if (out.flags & ISCHR && !warned) {
447 				warned = 1;
448 				warnx("%s: short write on character device",
449 				    out.name);
450 			}
451 		}
452 		if ((out.dbcnt -= n) < out.dbsz)
453 			break;
454 	}
455 
456 	/* Reassemble the output block. */
457 	if (out.dbcnt)
458 		(void)memmove(out.db, out.dbp - out.dbcnt, out.dbcnt);
459 	out.dbp = out.db + out.dbcnt;
460 }
461