1 /*-
2  * Copyright (c) 2007 Kai Wang
3  * Copyright (c) 2007 Tim Kientzle
4  * All rights reserved.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer
11  *    in this position and unchanged.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
17  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
20  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26  */
27 
28 #include "archive_platform.h"
29 __FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_ar.c 201108 2009-12-28 03:28:21Z kientzle $");
30 
31 #ifdef HAVE_ERRNO_H
32 #include <errno.h>
33 #endif
34 #ifdef HAVE_STDLIB_H
35 #include <stdlib.h>
36 #endif
37 #ifdef HAVE_STRING_H
38 #include <string.h>
39 #endif
40 
41 #include "archive.h"
42 #include "archive_entry.h"
43 #include "archive_private.h"
44 #include "archive_write_private.h"
45 #include "archive_write_set_format_private.h"
46 
47 struct ar_w {
48 	uint64_t	 entry_bytes_remaining;
49 	uint64_t	 entry_padding;
50 	int		 is_strtab;
51 	int		 has_strtab;
52 	char		 wrote_global_header;
53 	char		*strtab;
54 };
55 
56 /*
57  * Define structure of the "ar" header.
58  */
59 #define AR_name_offset 0
60 #define AR_name_size 16
61 #define AR_date_offset 16
62 #define AR_date_size 12
63 #define AR_uid_offset 28
64 #define AR_uid_size 6
65 #define AR_gid_offset 34
66 #define AR_gid_size 6
67 #define AR_mode_offset 40
68 #define AR_mode_size 8
69 #define AR_size_offset 48
70 #define AR_size_size 10
71 #define AR_fmag_offset 58
72 #define AR_fmag_size 2
73 
74 static int		 archive_write_set_format_ar(struct archive_write *);
75 static int		 archive_write_ar_header(struct archive_write *,
76 			     struct archive_entry *);
77 static ssize_t		 archive_write_ar_data(struct archive_write *,
78 			     const void *buff, size_t s);
79 static int		 archive_write_ar_free(struct archive_write *);
80 static int		 archive_write_ar_close(struct archive_write *);
81 static int		 archive_write_ar_finish_entry(struct archive_write *);
82 static const char	*ar_basename(const char *path);
83 static int		 format_octal(int64_t v, char *p, int s);
84 static int		 format_decimal(int64_t v, char *p, int s);
85 
86 int
archive_write_set_format_ar_bsd(struct archive * _a)87 archive_write_set_format_ar_bsd(struct archive *_a)
88 {
89 	struct archive_write *a = (struct archive_write *)_a;
90 	int r;
91 
92 	archive_check_magic(_a, ARCHIVE_WRITE_MAGIC,
93 	    ARCHIVE_STATE_NEW, "archive_write_set_format_ar_bsd");
94 	r = archive_write_set_format_ar(a);
95 	if (r == ARCHIVE_OK) {
96 		a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD;
97 		a->archive.archive_format_name = "ar (BSD)";
98 	}
99 	return (r);
100 }
101 
102 int
archive_write_set_format_ar_svr4(struct archive * _a)103 archive_write_set_format_ar_svr4(struct archive *_a)
104 {
105 	struct archive_write *a = (struct archive_write *)_a;
106 	int r;
107 
108 	archive_check_magic(_a, ARCHIVE_WRITE_MAGIC,
109 	    ARCHIVE_STATE_NEW, "archive_write_set_format_ar_svr4");
110 	r = archive_write_set_format_ar(a);
111 	if (r == ARCHIVE_OK) {
112 		a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU;
113 		a->archive.archive_format_name = "ar (GNU/SVR4)";
114 	}
115 	return (r);
116 }
117 
118 /*
119  * Generic initialization.
120  */
121 static int
archive_write_set_format_ar(struct archive_write * a)122 archive_write_set_format_ar(struct archive_write *a)
123 {
124 	struct ar_w *ar;
125 
126 	/* If someone else was already registered, unregister them. */
127 	if (a->format_free != NULL)
128 		(a->format_free)(a);
129 
130 	ar = (struct ar_w *)calloc(1, sizeof(*ar));
131 	if (ar == NULL) {
132 		archive_set_error(&a->archive, ENOMEM, "Can't allocate ar data");
133 		return (ARCHIVE_FATAL);
134 	}
135 	a->format_data = ar;
136 
137 	a->format_name = "ar";
138 	a->format_write_header = archive_write_ar_header;
139 	a->format_write_data = archive_write_ar_data;
140 	a->format_close = archive_write_ar_close;
141 	a->format_free = archive_write_ar_free;
142 	a->format_finish_entry = archive_write_ar_finish_entry;
143 	return (ARCHIVE_OK);
144 }
145 
146 static int
archive_write_ar_header(struct archive_write * a,struct archive_entry * entry)147 archive_write_ar_header(struct archive_write *a, struct archive_entry *entry)
148 {
149 	int ret, append_fn;
150 	char buff[60];
151 	char *ss, *se;
152 	struct ar_w *ar;
153 	const char *pathname;
154 	const char *filename;
155 	int64_t size;
156 
157 	append_fn = 0;
158 	ar = (struct ar_w *)a->format_data;
159 	ar->is_strtab = 0;
160 	filename = NULL;
161 	size = archive_entry_size(entry);
162 
163 
164 	/*
165 	 * Reject files with empty name.
166 	 */
167 	pathname = archive_entry_pathname(entry);
168 	if (pathname == NULL || *pathname == '\0') {
169 		archive_set_error(&a->archive, EINVAL,
170 		    "Invalid filename");
171 		return (ARCHIVE_WARN);
172 	}
173 
174 	/*
175 	 * If we are now at the beginning of the archive,
176 	 * we need first write the ar global header.
177 	 */
178 	if (!ar->wrote_global_header) {
179 		__archive_write_output(a, "!<arch>\n", 8);
180 		ar->wrote_global_header = 1;
181 	}
182 
183 	memset(buff, ' ', 60);
184 	memcpy(&buff[AR_fmag_offset], "`\n", 2);
185 
186 	if (strcmp(pathname, "/") == 0 ) {
187 		/* Entry is archive symbol table in GNU format */
188 		buff[AR_name_offset] = '/';
189 		goto stat;
190 	}
191 	if (strcmp(pathname, "/SYM64/") == 0) {
192 		/* Entry is archive symbol table in GNU 64-bit format */
193 		memcpy(buff + AR_name_offset, "/SYM64/", 7);
194 		goto stat;
195 	}
196 	if (strcmp(pathname, "__.SYMDEF") == 0) {
197 		/* Entry is archive symbol table in BSD format */
198 		memcpy(buff + AR_name_offset, "__.SYMDEF", 9);
199 		goto stat;
200 	}
201 	if (strcmp(pathname, "//") == 0) {
202 		/*
203 		 * Entry is archive filename table, inform that we should
204 		 * collect strtab in next _data call.
205 		 */
206 		ar->is_strtab = 1;
207 		buff[AR_name_offset] = buff[AR_name_offset + 1] = '/';
208 		/*
209 		 * For archive string table, only ar_size field should
210 		 * be set.
211 		 */
212 		goto size;
213 	}
214 
215 	/*
216 	 * Otherwise, entry is a normal archive member.
217 	 * Strip leading paths from filenames, if any.
218 	 */
219 	if ((filename = ar_basename(pathname)) == NULL) {
220 		/* Reject filenames with trailing "/" */
221 		archive_set_error(&a->archive, EINVAL,
222 		    "Invalid filename");
223 		return (ARCHIVE_WARN);
224 	}
225 
226 	if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) {
227 		/*
228 		 * SVR4/GNU variant use a "/" to mark then end of the filename,
229 		 * make it possible to have embedded spaces in the filename.
230 		 * So, the longest filename here (without extension) is
231 		 * actually 15 bytes.
232 		 */
233 		if (strlen(filename) <= 15) {
234 			memcpy(&buff[AR_name_offset],
235 			    filename, strlen(filename));
236 			buff[AR_name_offset + strlen(filename)] = '/';
237 		} else {
238 			/*
239 			 * For filename longer than 15 bytes, GNU variant
240 			 * makes use of a string table and instead stores the
241 			 * offset of the real filename to in the ar_name field.
242 			 * The string table should have been written before.
243 			 */
244 			if (ar->has_strtab <= 0) {
245 				archive_set_error(&a->archive, EINVAL,
246 				    "Can't find string table");
247 				return (ARCHIVE_WARN);
248 			}
249 
250 			se = (char *)malloc(strlen(filename) + 3);
251 			if (se == NULL) {
252 				archive_set_error(&a->archive, ENOMEM,
253 				    "Can't allocate filename buffer");
254 				return (ARCHIVE_FATAL);
255 			}
256 
257 			memcpy(se, filename, strlen(filename));
258 			strcpy(se + strlen(filename), "/\n");
259 
260 			ss = strstr(ar->strtab, se);
261 			free(se);
262 
263 			if (ss == NULL) {
264 				archive_set_error(&a->archive, EINVAL,
265 				    "Invalid string table");
266 				return (ARCHIVE_WARN);
267 			}
268 
269 			/*
270 			 * GNU variant puts "/" followed by digits into
271 			 * ar_name field. These digits indicates the real
272 			 * filename string's offset to the string table.
273 			 */
274 			buff[AR_name_offset] = '/';
275 			if (format_decimal(ss - ar->strtab,
276 			    buff + AR_name_offset + 1,
277 			    AR_name_size - 1)) {
278 				archive_set_error(&a->archive, ERANGE,
279 				    "string table offset too large");
280 				return (ARCHIVE_WARN);
281 			}
282 		}
283 	} else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) {
284 		/*
285 		 * BSD variant: for any file name which is more than
286 		 * 16 chars or contains one or more embedded space(s), the
287 		 * string "#1/" followed by the ASCII length of the name is
288 		 * put into the ar_name field. The file size (stored in the
289 		 * ar_size field) is incremented by the length of the name.
290 		 * The name is then written immediately following the
291 		 * archive header.
292 		 */
293 		if (strlen(filename) <= 16 && strchr(filename, ' ') == NULL) {
294 			memcpy(&buff[AR_name_offset], filename, strlen(filename));
295 			buff[AR_name_offset + strlen(filename)] = ' ';
296 		}
297 		else {
298 			memcpy(buff + AR_name_offset, "#1/", 3);
299 			if (format_decimal(strlen(filename),
300 			    buff + AR_name_offset + 3,
301 			    AR_name_size - 3)) {
302 				archive_set_error(&a->archive, ERANGE,
303 				    "File name too long");
304 				return (ARCHIVE_WARN);
305 			}
306 			append_fn = 1;
307 			size += strlen(filename);
308 		}
309 	}
310 
311 stat:
312 	if (format_decimal(archive_entry_mtime(entry), buff + AR_date_offset, AR_date_size)) {
313 		archive_set_error(&a->archive, ERANGE,
314 		    "File modification time too large");
315 		return (ARCHIVE_WARN);
316 	}
317 	if (format_decimal(archive_entry_uid(entry), buff + AR_uid_offset, AR_uid_size)) {
318 		archive_set_error(&a->archive, ERANGE,
319 		    "Numeric user ID too large");
320 		return (ARCHIVE_WARN);
321 	}
322 	if (format_decimal(archive_entry_gid(entry), buff + AR_gid_offset, AR_gid_size)) {
323 		archive_set_error(&a->archive, ERANGE,
324 		    "Numeric group ID too large");
325 		return (ARCHIVE_WARN);
326 	}
327 	if (format_octal(archive_entry_mode(entry), buff + AR_mode_offset, AR_mode_size)) {
328 		archive_set_error(&a->archive, ERANGE,
329 		    "Numeric mode too large");
330 		return (ARCHIVE_WARN);
331 	}
332 	/*
333 	 * Sanity Check: A non-pseudo archive member should always be
334 	 * a regular file.
335 	 */
336 	if (filename != NULL && archive_entry_filetype(entry) != AE_IFREG) {
337 		archive_set_error(&a->archive, EINVAL,
338 		    "Regular file required for non-pseudo member");
339 		return (ARCHIVE_WARN);
340 	}
341 
342 size:
343 	if (format_decimal(size, buff + AR_size_offset, AR_size_size)) {
344 		archive_set_error(&a->archive, ERANGE,
345 		    "File size out of range");
346 		return (ARCHIVE_WARN);
347 	}
348 
349 	ret = __archive_write_output(a, buff, 60);
350 	if (ret != ARCHIVE_OK)
351 		return (ret);
352 
353 	ar->entry_bytes_remaining = size;
354 	ar->entry_padding = ar->entry_bytes_remaining % 2;
355 
356 	if (append_fn > 0) {
357 		ret = __archive_write_output(a, filename, strlen(filename));
358 		if (ret != ARCHIVE_OK)
359 			return (ret);
360 		ar->entry_bytes_remaining -= strlen(filename);
361 	}
362 
363 	return (ARCHIVE_OK);
364 }
365 
366 static ssize_t
archive_write_ar_data(struct archive_write * a,const void * buff,size_t s)367 archive_write_ar_data(struct archive_write *a, const void *buff, size_t s)
368 {
369 	struct ar_w *ar;
370 	int ret;
371 
372 	ar = (struct ar_w *)a->format_data;
373 	if (s > ar->entry_bytes_remaining)
374 		s = (size_t)ar->entry_bytes_remaining;
375 
376 	if (ar->is_strtab > 0) {
377 		if (ar->has_strtab > 0) {
378 			archive_set_error(&a->archive, EINVAL,
379 			    "More than one string tables exist");
380 			return (ARCHIVE_WARN);
381 		}
382 
383 		ar->strtab = (char *)malloc(s + 1);
384 		if (ar->strtab == NULL) {
385 			archive_set_error(&a->archive, ENOMEM,
386 			    "Can't allocate strtab buffer");
387 			return (ARCHIVE_FATAL);
388 		}
389 		memcpy(ar->strtab, buff, s);
390 		ar->strtab[s] = '\0';
391 		ar->has_strtab = 1;
392 	}
393 
394 	ret = __archive_write_output(a, buff, s);
395 	if (ret != ARCHIVE_OK)
396 		return (ret);
397 
398 	ar->entry_bytes_remaining -= s;
399 	return (s);
400 }
401 
402 static int
archive_write_ar_free(struct archive_write * a)403 archive_write_ar_free(struct archive_write *a)
404 {
405 	struct ar_w *ar;
406 
407 	ar = (struct ar_w *)a->format_data;
408 
409 	if (ar == NULL)
410 		return (ARCHIVE_OK);
411 
412 	if (ar->has_strtab > 0) {
413 		free(ar->strtab);
414 		ar->strtab = NULL;
415 	}
416 
417 	free(ar);
418 	a->format_data = NULL;
419 	return (ARCHIVE_OK);
420 }
421 
422 static int
archive_write_ar_close(struct archive_write * a)423 archive_write_ar_close(struct archive_write *a)
424 {
425 	struct ar_w *ar;
426 	int ret;
427 
428 	/*
429 	 * If we haven't written anything yet, we need to write
430 	 * the ar global header now to make it a valid ar archive.
431 	 */
432 	ar = (struct ar_w *)a->format_data;
433 	if (!ar->wrote_global_header) {
434 		ar->wrote_global_header = 1;
435 		ret = __archive_write_output(a, "!<arch>\n", 8);
436 		return (ret);
437 	}
438 
439 	return (ARCHIVE_OK);
440 }
441 
442 static int
archive_write_ar_finish_entry(struct archive_write * a)443 archive_write_ar_finish_entry(struct archive_write *a)
444 {
445 	struct ar_w *ar;
446 	int ret;
447 
448 	ar = (struct ar_w *)a->format_data;
449 
450 	if (ar->entry_bytes_remaining != 0) {
451 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
452 		    "Entry remaining bytes larger than 0");
453 		return (ARCHIVE_WARN);
454 	}
455 
456 	if (ar->entry_padding == 0) {
457 		return (ARCHIVE_OK);
458 	}
459 
460 	if (ar->entry_padding != 1) {
461 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
462 		    "Padding wrong size: %ju should be 1 or 0",
463 		    (uintmax_t)ar->entry_padding);
464 		return (ARCHIVE_WARN);
465 	}
466 
467 	ret = __archive_write_output(a, "\n", 1);
468 	return (ret);
469 }
470 
471 /*
472  * Format a number into the specified field using base-8.
473  * NB: This version is slightly different from the one in
474  * _ustar.c
475  */
476 static int
format_octal(int64_t v,char * p,int s)477 format_octal(int64_t v, char *p, int s)
478 {
479 	int len;
480 	char *h;
481 
482 	len = s;
483 	h = p;
484 
485 	/* Octal values can't be negative, so use 0. */
486 	if (v < 0) {
487 		while (len-- > 0)
488 			*p++ = '0';
489 		return (-1);
490 	}
491 
492 	p += s;		/* Start at the end and work backwards. */
493 	do {
494 		*--p = (char)('0' + (v & 7));
495 		v >>= 3;
496 	} while (--s > 0 && v > 0);
497 
498 	if (v == 0) {
499 		memmove(h, p, len - s);
500 		p = h + len - s;
501 		while (s-- > 0)
502 			*p++ = ' ';
503 		return (0);
504 	}
505 	/* If it overflowed, fill field with max value. */
506 	while (len-- > 0)
507 		*p++ = '7';
508 
509 	return (-1);
510 }
511 
512 /*
513  * Format a number into the specified field using base-10.
514  */
515 static int
format_decimal(int64_t v,char * p,int s)516 format_decimal(int64_t v, char *p, int s)
517 {
518 	int len;
519 	char *h;
520 
521 	len = s;
522 	h = p;
523 
524 	/* Negative values in ar header are meaningless, so use 0. */
525 	if (v < 0) {
526 		while (len-- > 0)
527 			*p++ = '0';
528 		return (-1);
529 	}
530 
531 	p += s;
532 	do {
533 		*--p = (char)('0' + (v % 10));
534 		v /= 10;
535 	} while (--s > 0 && v > 0);
536 
537 	if (v == 0) {
538 		memmove(h, p, len - s);
539 		p = h + len - s;
540 		while (s-- > 0)
541 			*p++ = ' ';
542 		return (0);
543 	}
544 	/* If it overflowed, fill field with max value. */
545 	while (len-- > 0)
546 		*p++ = '9';
547 
548 	return (-1);
549 }
550 
551 static const char *
ar_basename(const char * path)552 ar_basename(const char *path)
553 {
554 	const char *endp, *startp;
555 
556 	endp = path + strlen(path) - 1;
557 	/*
558 	 * For filename with trailing slash(es), we return
559 	 * NULL indicating an error.
560 	 */
561 	if (*endp == '/')
562 		return (NULL);
563 
564 	/* Find the start of the base */
565 	startp = endp;
566 	while (startp > path && *(startp - 1) != '/')
567 		startp--;
568 
569 	return (startp);
570 }
571