xref: /linux/fs/smb/client/cifs_unicode.c (revision 4c630f307455c06f99bdeca7f7a1ab5318604fe0)
1 // SPDX-License-Identifier: GPL-2.0-or-later
2 /*
3  *
4  *   Copyright (c) International Business Machines  Corp., 2000,2009
5  *   Modified by Steve French (sfrench@us.ibm.com)
6  */
7 #include <linux/fs.h>
8 #include <linux/slab.h>
9 #include "cifs_fs_sb.h"
10 #include "cifs_unicode.h"
11 #include "cifs_uniupr.h"
12 #include "cifspdu.h"
13 #include "cifsglob.h"
14 #include "cifs_debug.h"
15 
16 int cifs_remap(struct cifs_sb_info *cifs_sb)
17 {
18 	int map_type;
19 
20 	if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SFM_CHR)
21 		map_type = SFM_MAP_UNI_RSVD;
22 	else if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SPECIAL_CHR)
23 		map_type = SFU_MAP_UNI_RSVD;
24 	else
25 		map_type = NO_MAP_UNI_RSVD;
26 
27 	return map_type;
28 }
29 
30 /* Convert character using the SFU - "Services for Unix" remapping range */
31 static bool
32 convert_sfu_char(const __u16 src_char, char *target)
33 {
34 	/*
35 	 * BB: Cannot handle remapping UNI_SLASH until all the calls to
36 	 *     build_path_from_dentry are modified, as they use slash as
37 	 *     separator.
38 	 */
39 	switch (src_char) {
40 	case UNI_COLON:
41 		*target = ':';
42 		break;
43 	case UNI_ASTERISK:
44 		*target = '*';
45 		break;
46 	case UNI_QUESTION:
47 		*target = '?';
48 		break;
49 	case UNI_PIPE:
50 		*target = '|';
51 		break;
52 	case UNI_GRTRTHAN:
53 		*target = '>';
54 		break;
55 	case UNI_LESSTHAN:
56 		*target = '<';
57 		break;
58 	default:
59 		return false;
60 	}
61 	return true;
62 }
63 
64 /* Convert character using the SFM - "Services for Mac" remapping range */
65 static bool
66 convert_sfm_char(const __u16 src_char, char *target)
67 {
68 	if (src_char >= 0xF001 && src_char <= 0xF01F) {
69 		*target = src_char - 0xF000;
70 		return true;
71 	}
72 	switch (src_char) {
73 	case SFM_COLON:
74 		*target = ':';
75 		break;
76 	case SFM_DOUBLEQUOTE:
77 		*target = '"';
78 		break;
79 	case SFM_ASTERISK:
80 		*target = '*';
81 		break;
82 	case SFM_QUESTION:
83 		*target = '?';
84 		break;
85 	case SFM_PIPE:
86 		*target = '|';
87 		break;
88 	case SFM_GRTRTHAN:
89 		*target = '>';
90 		break;
91 	case SFM_LESSTHAN:
92 		*target = '<';
93 		break;
94 	case SFM_SPACE:
95 		*target = ' ';
96 		break;
97 	case SFM_PERIOD:
98 		*target = '.';
99 		break;
100 	default:
101 		return false;
102 	}
103 	return true;
104 }
105 
106 
107 /*
108  * cifs_mapchar - convert a host-endian char to proper char in codepage
109  * @target - where converted character should be copied
110  * @src_char - 2 byte host-endian source character
111  * @cp - codepage to which character should be converted
112  * @map_type - How should the 7 NTFS/SMB reserved characters be mapped to UCS2?
113  *
114  * This function handles the conversion of a single character. It is the
115  * responsibility of the caller to ensure that the target buffer is large
116  * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE).
117  */
118 static int
119 cifs_mapchar(char *target, const __u16 *from, const struct nls_table *cp,
120 	     int maptype)
121 {
122 	int len = 1;
123 	__u16 src_char;
124 
125 	src_char = *from;
126 
127 	if ((maptype == SFM_MAP_UNI_RSVD) && convert_sfm_char(src_char, target))
128 		return len;
129 	else if ((maptype == SFU_MAP_UNI_RSVD) &&
130 		  convert_sfu_char(src_char, target))
131 		return len;
132 
133 	/* if character not one of seven in special remap set */
134 	len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
135 	if (len <= 0)
136 		goto surrogate_pair;
137 
138 	return len;
139 
140 surrogate_pair:
141 	/* convert SURROGATE_PAIR and IVS */
142 	if (strcmp(cp->charset, "utf8"))
143 		goto unknown;
144 	len = utf16s_to_utf8s(from, 3, UTF16_LITTLE_ENDIAN, target, 6);
145 	if (len <= 0)
146 		goto unknown;
147 	return len;
148 
149 unknown:
150 	*target = '?';
151 	len = 1;
152 	return len;
153 }
154 
155 /*
156  * cifs_from_utf16 - convert utf16le string to local charset
157  * @to - destination buffer
158  * @from - source buffer
159  * @tolen - destination buffer size (in bytes)
160  * @fromlen - source buffer size (in bytes)
161  * @codepage - codepage to which characters should be converted
162  * @mapchar - should characters be remapped according to the mapchars option?
163  *
164  * Convert a little-endian utf16le string (as sent by the server) to a string
165  * in the provided codepage. The tolen and fromlen parameters are to ensure
166  * that the code doesn't walk off of the end of the buffer (which is always
167  * a danger if the alignment of the source buffer is off). The destination
168  * string is always properly null terminated and fits in the destination
169  * buffer. Returns the length of the destination string in bytes (including
170  * null terminator).
171  *
172  * Note that some windows versions actually send multiword UTF-16 characters
173  * instead of straight UTF16-2. The linux nls routines however aren't able to
174  * deal with those characters properly. In the event that we get some of
175  * those characters, they won't be translated properly.
176  */
177 int
178 cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
179 		const struct nls_table *codepage, int map_type)
180 {
181 	int i, charlen, safelen;
182 	int outlen = 0;
183 	int nullsize = nls_nullsize(codepage);
184 	int fromwords = fromlen / 2;
185 	char tmp[NLS_MAX_CHARSET_SIZE];
186 	__u16 ftmp[3];		/* ftmp[3] = 3array x 2bytes = 6bytes UTF-16 */
187 
188 	/*
189 	 * because the chars can be of varying widths, we need to take care
190 	 * not to overflow the destination buffer when we get close to the
191 	 * end of it. Until we get to this offset, we don't need to check
192 	 * for overflow however.
193 	 */
194 	safelen = tolen - (NLS_MAX_CHARSET_SIZE + nullsize);
195 
196 	for (i = 0; i < fromwords; i++) {
197 		ftmp[0] = get_unaligned_le16(&from[i]);
198 		if (ftmp[0] == 0)
199 			break;
200 		if (i + 1 < fromwords)
201 			ftmp[1] = get_unaligned_le16(&from[i + 1]);
202 		else
203 			ftmp[1] = 0;
204 		if (i + 2 < fromwords)
205 			ftmp[2] = get_unaligned_le16(&from[i + 2]);
206 		else
207 			ftmp[2] = 0;
208 
209 		/*
210 		 * check to see if converting this character might make the
211 		 * conversion bleed into the null terminator
212 		 */
213 		if (outlen >= safelen) {
214 			charlen = cifs_mapchar(tmp, ftmp, codepage, map_type);
215 			if ((outlen + charlen) > (tolen - nullsize))
216 				break;
217 		}
218 
219 		/* put converted char into 'to' buffer */
220 		charlen = cifs_mapchar(&to[outlen], ftmp, codepage, map_type);
221 		outlen += charlen;
222 
223 		/* charlen (=bytes of UTF-8 for 1 character)
224 		 * 4bytes UTF-8(surrogate pair) is charlen=4
225 		 *   (4bytes UTF-16 code)
226 		 * 7-8bytes UTF-8(IVS) is charlen=3+4 or 4+4
227 		 *   (2 UTF-8 pairs divided to 2 UTF-16 pairs) */
228 		if (charlen == 4)
229 			i++;
230 		else if (charlen >= 5)
231 			/* 5-6bytes UTF-8 */
232 			i += 2;
233 	}
234 
235 	/* properly null-terminate string */
236 	for (i = 0; i < nullsize; i++)
237 		to[outlen++] = 0;
238 
239 	return outlen;
240 }
241 
242 /*
243  * NAME:	cifs_strtoUTF16()
244  *
245  * FUNCTION:	Convert character string to unicode string
246  *
247  */
248 int
249 cifs_strtoUTF16(__le16 *to, const char *from, int len,
250 	      const struct nls_table *codepage)
251 {
252 	int charlen;
253 	int i;
254 	wchar_t wchar_to; /* needed to quiet sparse */
255 
256 	/* special case for utf8 to handle no plane0 chars */
257 	if (!strcmp(codepage->charset, "utf8")) {
258 		/*
259 		 * convert utf8 -> utf16, we assume we have enough space
260 		 * as caller should have assumed conversion does not overflow
261 		 * in destination len is length in wchar_t units (16bits)
262 		 */
263 		i  = utf8s_to_utf16s(from, len, UTF16_LITTLE_ENDIAN,
264 				       (wchar_t *) to, len);
265 
266 		/* if success terminate and exit */
267 		if (i >= 0)
268 			goto success;
269 		/*
270 		 * if fails fall back to UCS encoding as this
271 		 * function should not return negative values
272 		 * currently can fail only if source contains
273 		 * invalid encoded characters
274 		 */
275 	}
276 
277 	for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
278 		charlen = codepage->char2uni(from, len, &wchar_to);
279 		if (charlen < 1) {
280 			cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n",
281 				 *from, charlen);
282 			/* A question mark */
283 			wchar_to = 0x003f;
284 			charlen = 1;
285 		}
286 		put_unaligned_le16(wchar_to, &to[i]);
287 	}
288 
289 success:
290 	put_unaligned_le16(0, &to[i]);
291 	return i;
292 }
293 
294 /*
295  * cifs_utf16_bytes - how long will a string be after conversion?
296  * @utf16 - pointer to input string
297  * @maxbytes - don't go past this many bytes of input string
298  * @codepage - destination codepage
299  *
300  * Walk a utf16le string and return the number of bytes that the string will
301  * be after being converted to the given charset, not including any null
302  * termination required. Don't walk past maxbytes in the source buffer.
303  */
304 int
305 cifs_utf16_bytes(const __le16 *from, int maxbytes,
306 		const struct nls_table *codepage)
307 {
308 	int i;
309 	int charlen, outlen = 0;
310 	int maxwords = maxbytes / 2;
311 	char tmp[NLS_MAX_CHARSET_SIZE];
312 	__u16 ftmp[3];
313 
314 	for (i = 0; i < maxwords; i++) {
315 		ftmp[0] = get_unaligned_le16(&from[i]);
316 		if (ftmp[0] == 0)
317 			break;
318 		if (i + 1 < maxwords)
319 			ftmp[1] = get_unaligned_le16(&from[i + 1]);
320 		else
321 			ftmp[1] = 0;
322 		if (i + 2 < maxwords)
323 			ftmp[2] = get_unaligned_le16(&from[i + 2]);
324 		else
325 			ftmp[2] = 0;
326 
327 		charlen = cifs_mapchar(tmp, ftmp, codepage, NO_MAP_UNI_RSVD);
328 		outlen += charlen;
329 	}
330 
331 	return outlen;
332 }
333 
334 /*
335  * cifs_strndup_from_utf16 - copy a string from wire format to the local
336  * codepage
337  * @src - source string
338  * @maxlen - don't walk past this many bytes in the source string
339  * @is_unicode - is this a unicode string?
340  * @codepage - destination codepage
341  *
342  * Take a string given by the server, convert it to the local codepage and
343  * put it in a new buffer. Returns a pointer to the new string or NULL on
344  * error.
345  */
346 char *
347 cifs_strndup_from_utf16(const char *src, const int maxlen,
348 			const bool is_unicode, const struct nls_table *codepage)
349 {
350 	int len;
351 	char *dst;
352 
353 	if (is_unicode) {
354 		len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage);
355 		len += nls_nullsize(codepage);
356 		dst = kmalloc(len, GFP_KERNEL);
357 		if (!dst)
358 			return NULL;
359 		cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
360 				NO_MAP_UNI_RSVD);
361 	} else {
362 		dst = kstrndup(src, maxlen, GFP_KERNEL);
363 	}
364 
365 	return dst;
366 }
367 
368 static __le16 convert_to_sfu_char(char src_char)
369 {
370 	__le16 dest_char;
371 
372 	switch (src_char) {
373 	case ':':
374 		dest_char = cpu_to_le16(UNI_COLON);
375 		break;
376 	case '*':
377 		dest_char = cpu_to_le16(UNI_ASTERISK);
378 		break;
379 	case '?':
380 		dest_char = cpu_to_le16(UNI_QUESTION);
381 		break;
382 	case '<':
383 		dest_char = cpu_to_le16(UNI_LESSTHAN);
384 		break;
385 	case '>':
386 		dest_char = cpu_to_le16(UNI_GRTRTHAN);
387 		break;
388 	case '|':
389 		dest_char = cpu_to_le16(UNI_PIPE);
390 		break;
391 	default:
392 		dest_char = 0;
393 	}
394 
395 	return dest_char;
396 }
397 
398 static __le16 convert_to_sfm_char(char src_char, bool end_of_string)
399 {
400 	__le16 dest_char;
401 
402 	if (src_char >= 0x01 && src_char <= 0x1F) {
403 		dest_char = cpu_to_le16(src_char + 0xF000);
404 		return dest_char;
405 	}
406 	switch (src_char) {
407 	case ':':
408 		dest_char = cpu_to_le16(SFM_COLON);
409 		break;
410 	case '"':
411 		dest_char = cpu_to_le16(SFM_DOUBLEQUOTE);
412 		break;
413 	case '*':
414 		dest_char = cpu_to_le16(SFM_ASTERISK);
415 		break;
416 	case '?':
417 		dest_char = cpu_to_le16(SFM_QUESTION);
418 		break;
419 	case '<':
420 		dest_char = cpu_to_le16(SFM_LESSTHAN);
421 		break;
422 	case '>':
423 		dest_char = cpu_to_le16(SFM_GRTRTHAN);
424 		break;
425 	case '|':
426 		dest_char = cpu_to_le16(SFM_PIPE);
427 		break;
428 	case '.':
429 		if (end_of_string)
430 			dest_char = cpu_to_le16(SFM_PERIOD);
431 		else
432 			dest_char = 0;
433 		break;
434 	case ' ':
435 		if (end_of_string)
436 			dest_char = cpu_to_le16(SFM_SPACE);
437 		else
438 			dest_char = 0;
439 		break;
440 	default:
441 		dest_char = 0;
442 	}
443 
444 	return dest_char;
445 }
446 
447 /*
448  * Convert 16 bit Unicode pathname to wire format from string in current code
449  * page. Conversion may involve remapping up the six characters that are
450  * only legal in POSIX-like OS (if they are present in the string). Path
451  * names are little endian 16 bit Unicode on the wire
452  */
453 int
454 cifsConvertToUTF16(__le16 *target, const char *source, int srclen,
455 		 const struct nls_table *cp, int map_chars)
456 {
457 	int i, charlen;
458 	int j = 0;
459 	char src_char;
460 	__le16 dst_char;
461 	wchar_t tmp;
462 	wchar_t *wchar_to;	/* UTF-16 */
463 	int ret;
464 	unicode_t u;
465 
466 	if (map_chars == NO_MAP_UNI_RSVD)
467 		return cifs_strtoUTF16(target, source, PATH_MAX, cp);
468 
469 	wchar_to = kzalloc(6, GFP_KERNEL);
470 
471 	for (i = 0; i < srclen; j++) {
472 		src_char = source[i];
473 		charlen = 1;
474 
475 		/* check if end of string */
476 		if (src_char == 0)
477 			goto ctoUTF16_out;
478 
479 		/* see if we must remap this char */
480 		if (map_chars == SFU_MAP_UNI_RSVD)
481 			dst_char = convert_to_sfu_char(src_char);
482 		else if (map_chars == SFM_MAP_UNI_RSVD) {
483 			bool end_of_string;
484 
485 			/**
486 			 * Remap spaces and periods found at the end of every
487 			 * component of the path. The special cases of '.' and
488 			 * '..' do not need to be dealt with explicitly because
489 			 * they are addressed in namei.c:link_path_walk().
490 			 **/
491 			if ((i == srclen - 1) || (source[i+1] == '\\'))
492 				end_of_string = true;
493 			else
494 				end_of_string = false;
495 
496 			dst_char = convert_to_sfm_char(src_char, end_of_string);
497 		} else
498 			dst_char = 0;
499 		/*
500 		 * FIXME: We can not handle remapping backslash (UNI_SLASH)
501 		 * until all the calls to build_path_from_dentry are modified,
502 		 * as they use backslash as separator.
503 		 */
504 		if (dst_char == 0) {
505 			charlen = cp->char2uni(source + i, srclen - i, &tmp);
506 			dst_char = cpu_to_le16(tmp);
507 
508 			/*
509 			 * if no match, use question mark, which at least in
510 			 * some cases serves as wild card
511 			 */
512 			if (charlen > 0)
513 				goto ctoUTF16;
514 
515 			/* convert SURROGATE_PAIR */
516 			if (strcmp(cp->charset, "utf8") || !wchar_to)
517 				goto unknown;
518 			if (*(source + i) & 0x80) {
519 				charlen = utf8_to_utf32(source + i, 6, &u);
520 				if (charlen < 0)
521 					goto unknown;
522 			} else
523 				goto unknown;
524 			ret  = utf8s_to_utf16s(source + i, charlen,
525 					       UTF16_LITTLE_ENDIAN,
526 					       wchar_to, 6);
527 			if (ret < 0)
528 				goto unknown;
529 
530 			i += charlen;
531 			dst_char = cpu_to_le16(*wchar_to);
532 			if (charlen <= 3)
533 				/* 1-3bytes UTF-8 to 2bytes UTF-16 */
534 				put_unaligned(dst_char, &target[j]);
535 			else if (charlen == 4) {
536 				/* 4bytes UTF-8(surrogate pair) to 4bytes UTF-16
537 				 * 7-8bytes UTF-8(IVS) divided to 2 UTF-16
538 				 *   (charlen=3+4 or 4+4) */
539 				put_unaligned(dst_char, &target[j]);
540 				dst_char = cpu_to_le16(*(wchar_to + 1));
541 				j++;
542 				put_unaligned(dst_char, &target[j]);
543 			} else if (charlen >= 5) {
544 				/* 5-6bytes UTF-8 to 6bytes UTF-16 */
545 				put_unaligned(dst_char, &target[j]);
546 				dst_char = cpu_to_le16(*(wchar_to + 1));
547 				j++;
548 				put_unaligned(dst_char, &target[j]);
549 				dst_char = cpu_to_le16(*(wchar_to + 2));
550 				j++;
551 				put_unaligned(dst_char, &target[j]);
552 			}
553 			continue;
554 
555 unknown:
556 			dst_char = cpu_to_le16(0x003f);
557 			charlen = 1;
558 		}
559 
560 ctoUTF16:
561 		/*
562 		 * character may take more than one byte in the source string,
563 		 * but will take exactly two bytes in the target string
564 		 */
565 		i += charlen;
566 		put_unaligned(dst_char, &target[j]);
567 	}
568 
569 ctoUTF16_out:
570 	put_unaligned(0, &target[j]); /* Null terminate target unicode string */
571 	kfree(wchar_to);
572 	return j;
573 }
574 
575 /*
576  * cifs_local_to_utf16_bytes - how long will a string be after conversion?
577  * @from - pointer to input string
578  * @maxbytes - don't go past this many bytes of input string
579  * @codepage - source codepage
580  *
581  * Walk a string and return the number of bytes that the string will
582  * be after being converted to the given charset, not including any null
583  * termination required. Don't walk past maxbytes in the source buffer.
584  */
585 
586 static int
587 cifs_local_to_utf16_bytes(const char *from, int len,
588 			  const struct nls_table *codepage)
589 {
590 	int charlen;
591 	int i;
592 	wchar_t wchar_to;
593 
594 	for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
595 		charlen = codepage->char2uni(from, len, &wchar_to);
596 		/* Failed conversion defaults to a question mark */
597 		if (charlen < 1)
598 			charlen = 1;
599 	}
600 	return 2 * i; /* UTF16 characters are two bytes */
601 }
602 
603 /*
604  * cifs_strndup_to_utf16 - copy a string to wire format from the local codepage
605  * @src - source string
606  * @maxlen - don't walk past this many bytes in the source string
607  * @utf16_len - the length of the allocated string in bytes (including null)
608  * @cp - source codepage
609  * @remap - map special chars
610  *
611  * Take a string convert it from the local codepage to UTF16 and
612  * put it in a new buffer. Returns a pointer to the new string or NULL on
613  * error.
614  */
615 __le16 *
616 cifs_strndup_to_utf16(const char *src, const int maxlen, int *utf16_len,
617 		      const struct nls_table *cp, int remap)
618 {
619 	int len;
620 	__le16 *dst;
621 
622 	len = cifs_local_to_utf16_bytes(src, maxlen, cp);
623 	len += 2; /* NULL */
624 	dst = kmalloc(len, GFP_KERNEL);
625 	if (!dst) {
626 		*utf16_len = 0;
627 		return NULL;
628 	}
629 	cifsConvertToUTF16(dst, src, strlen(src), cp, remap);
630 	*utf16_len = len;
631 	return dst;
632 }
633