1 // SPDX-License-Identifier: GPL-2.0-or-later
4 * Copyright (c) International Business Machines Corp., 2000,2009
8 #include <linux/slab.h>
9 #include "cifs_fs_sb.h"
10 #include "cifs_unicode.h"
11 #include "cifs_uniupr.h"
14 #include "cifs_debug.h"
16 int cifs_remap(struct cifs_sb_info *cifs_sb)
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;
25 map_type = NO_MAP_UNI_RSVD;
30 /* Convert character using the SFU - "Services for Unix" remapping range */
32 convert_sfu_char(const __u16 src_char, char *target)
35 * BB: Cannot handle remapping UNI_SLASH until all the calls to
36 * build_path_from_dentry are modified, as they use slash as
64 /* Convert character using the SFM - "Services for Mac" remapping range */
66 convert_sfm_char(const __u16 src_char, char *target)
68 if (src_char >= 0xF001 && src_char <= 0xF01F) {
69 *target = src_char - 0xF000;
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?
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).
119 cifs_mapchar(char *target, const __u16 *from, const struct nls_table *cp,
127 if ((maptype == SFM_MAP_UNI_RSVD) && convert_sfm_char(src_char, target))
129 else if ((maptype == SFU_MAP_UNI_RSVD) &&
130 convert_sfu_char(src_char, target))
133 /* if character not one of seven in special remap set */
134 len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
141 /* convert SURROGATE_PAIR and IVS */
142 if (strcmp(cp->charset, "utf8"))
144 len = utf16s_to_utf8s(from, 3, UTF16_LITTLE_ENDIAN, target, 6);
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?
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
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.
178 cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
179 const struct nls_table *codepage, int map_type)
181 int i, charlen, safelen;
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 */
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.
194 safelen = tolen - (NLS_MAX_CHARSET_SIZE + nullsize);
196 for (i = 0; i < fromwords; i++) {
197 ftmp[0] = get_unaligned_le16(&from[i]);
200 if (i + 1 < fromwords)
201 ftmp[1] = get_unaligned_le16(&from[i + 1]);
204 if (i + 2 < fromwords)
205 ftmp[2] = get_unaligned_le16(&from[i + 2]);
210 * check to see if converting this character might make the
211 * conversion bleed into the null terminator
213 if (outlen >= safelen) {
214 charlen = cifs_mapchar(tmp, ftmp, codepage, map_type);
215 if ((outlen + charlen) > (tolen - nullsize))
219 /* put converted char into 'to' buffer */
220 charlen = cifs_mapchar(&to[outlen], ftmp, codepage, map_type);
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) */
230 else if (charlen >= 5)
235 /* properly null-terminate string */
236 for (i = 0; i < nullsize; i++)
243 * NAME: cifs_strtoUTF16()
245 * FUNCTION: Convert character string to unicode string
249 cifs_strtoUTF16(__le16 *to, const char *from, int len,
250 const struct nls_table *codepage)
254 wchar_t wchar_to; /* needed to quiet sparse */
256 /* special case for utf8 to handle no plane0 chars */
257 if (!strcmp(codepage->charset, "utf8")) {
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)
263 i = utf8s_to_utf16s(from, len, UTF16_LITTLE_ENDIAN,
264 (wchar_t *) to, len);
266 /* if success terminate and exit */
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
277 for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
278 charlen = codepage->char2uni(from, len, &wchar_to);
280 cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n",
282 /* A question mark */
286 put_unaligned_le16(wchar_to, &to[i]);
290 put_unaligned_le16(0, &to[i]);
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
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.
305 cifs_utf16_bytes(const __le16 *from, int maxbytes,
306 const struct nls_table *codepage)
309 int charlen, outlen = 0;
310 int maxwords = maxbytes / 2;
311 char tmp[NLS_MAX_CHARSET_SIZE];
314 for (i = 0; i < maxwords; i++) {
315 ftmp[0] = get_unaligned_le16(&from[i]);
318 if (i + 1 < maxwords)
319 ftmp[1] = get_unaligned_le16(&from[i + 1]);
322 if (i + 2 < maxwords)
323 ftmp[2] = get_unaligned_le16(&from[i + 2]);
327 charlen = cifs_mapchar(tmp, ftmp, codepage, NO_MAP_UNI_RSVD);
335 * cifs_strndup_from_utf16 - copy a string from wire format to the local
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
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
347 cifs_strndup_from_utf16(const char *src, const int maxlen,
348 const bool is_unicode, const struct nls_table *codepage)
354 len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage);
355 len += nls_nullsize(codepage);
356 dst = kmalloc(len, GFP_KERNEL);
359 cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
362 dst = kstrndup(src, maxlen, GFP_KERNEL);
368 static __le16 convert_to_sfu_char(char src_char)
374 dest_char = cpu_to_le16(UNI_COLON);
377 dest_char = cpu_to_le16(UNI_ASTERISK);
380 dest_char = cpu_to_le16(UNI_QUESTION);
383 dest_char = cpu_to_le16(UNI_LESSTHAN);
386 dest_char = cpu_to_le16(UNI_GRTRTHAN);
389 dest_char = cpu_to_le16(UNI_PIPE);
398 static __le16 convert_to_sfm_char(char src_char, bool end_of_string)
402 if (src_char >= 0x01 && src_char <= 0x1F) {
403 dest_char = cpu_to_le16(src_char + 0xF000);
408 dest_char = cpu_to_le16(SFM_COLON);
411 dest_char = cpu_to_le16(SFM_DOUBLEQUOTE);
414 dest_char = cpu_to_le16(SFM_ASTERISK);
417 dest_char = cpu_to_le16(SFM_QUESTION);
420 dest_char = cpu_to_le16(SFM_LESSTHAN);
423 dest_char = cpu_to_le16(SFM_GRTRTHAN);
426 dest_char = cpu_to_le16(SFM_PIPE);
430 dest_char = cpu_to_le16(SFM_PERIOD);
436 dest_char = cpu_to_le16(SFM_SPACE);
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
454 cifsConvertToUTF16(__le16 *target, const char *source, int srclen,
455 const struct nls_table *cp, int map_chars)
462 wchar_t *wchar_to; /* UTF-16 */
466 if (map_chars == NO_MAP_UNI_RSVD)
467 return cifs_strtoUTF16(target, source, PATH_MAX, cp);
469 wchar_to = kzalloc(6, GFP_KERNEL);
471 for (i = 0; i < srclen; j++) {
472 src_char = source[i];
475 /* check if end of string */
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) {
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().
491 if ((i == srclen - 1) || (source[i+1] == '\\'))
492 end_of_string = true;
494 end_of_string = false;
496 dst_char = convert_to_sfm_char(src_char, end_of_string);
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.
505 charlen = cp->char2uni(source + i, srclen - i, &tmp);
506 dst_char = cpu_to_le16(tmp);
509 * if no match, use question mark, which at least in
510 * some cases serves as wild card
515 /* convert SURROGATE_PAIR */
516 if (strcmp(cp->charset, "utf8") || !wchar_to)
518 if (*(source + i) & 0x80) {
519 charlen = utf8_to_utf32(source + i, 6, &u);
524 ret = utf8s_to_utf16s(source + i, charlen,
531 dst_char = cpu_to_le16(*wchar_to);
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));
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));
548 put_unaligned(dst_char, &target[j]);
549 dst_char = cpu_to_le16(*(wchar_to + 2));
551 put_unaligned(dst_char, &target[j]);
556 dst_char = cpu_to_le16(0x003f);
562 * character may take more than one byte in the source string,
563 * but will take exactly two bytes in the target string
566 put_unaligned(dst_char, &target[j]);
570 put_unaligned(0, &target[j]); /* Null terminate target unicode string */
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
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.
587 cifs_local_to_utf16_bytes(const char *from, int len,
588 const struct nls_table *codepage)
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 */
600 return 2 * i; /* UTF16 characters are two bytes */
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
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
616 cifs_strndup_to_utf16(const char *src, const int maxlen, int *utf16_len,
617 const struct nls_table *cp, int remap)
622 len = cifs_local_to_utf16_bytes(src, maxlen, cp);
624 dst = kmalloc(len, GFP_KERNEL);
629 cifsConvertToUTF16(dst, src, strlen(src), cp, remap);