blob: e7582dd791794cb3aa31b499785063a45eef8fa8 [file] [log] [blame]
Thomas Gleixner1a59d1b82019-05-27 08:55:05 +02001// SPDX-License-Identifier: GPL-2.0-or-later
Linus Torvalds1da177e2005-04-16 15:20:36 -07002/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07003 *
Steve Frenchd185cda2009-04-30 17:45:10 +00004 * Copyright (c) International Business Machines Corp., 2000,2009
Linus Torvalds1da177e2005-04-16 15:20:36 -07005 * Modified by Steve French (sfrench@us.ibm.com)
Linus Torvalds1da177e2005-04-16 15:20:36 -07006 */
7#include <linux/fs.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +09008#include <linux/slab.h>
Steve French2baa2682014-09-27 02:19:01 -05009#include "cifs_fs_sb.h"
Linus Torvalds1da177e2005-04-16 15:20:36 -070010#include "cifs_unicode.h"
11#include "cifs_uniupr.h"
12#include "cifspdu.h"
Steve French39798772006-05-31 22:40:51 +000013#include "cifsglob.h"
Linus Torvalds1da177e2005-04-16 15:20:36 -070014#include "cifs_debug.h"
15
Steve French2baa2682014-09-27 02:19:01 -050016int 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
Steve Frenchb6938552014-09-25 13:20:05 -050030/* Convert character using the SFU - "Services for Unix" remapping range */
31static bool
32convert_sfu_char(const __u16 src_char, char *target)
Jeff Layton7fabf0c2009-04-30 06:46:15 -040033{
Jeff Layton7fabf0c2009-04-30 06:46:15 -040034 /*
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 */
Jeff Laytonba2dbf32011-01-20 13:36:51 -050039 switch (src_char) {
Jeff Layton7fabf0c2009-04-30 06:46:15 -040040 case UNI_COLON:
41 *target = ':';
42 break;
Jeff Layton581ade42011-04-05 15:02:37 -040043 case UNI_ASTERISK:
Jeff Layton7fabf0c2009-04-30 06:46:15 -040044 *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:
Steve Frenchb6938552014-09-25 13:20:05 -050059 return false;
Jeff Layton7fabf0c2009-04-30 06:46:15 -040060 }
Steve Frenchb6938552014-09-25 13:20:05 -050061 return true;
62}
Jeff Layton7fabf0c2009-04-30 06:46:15 -040063
Steve Frenchb6938552014-09-25 13:20:05 -050064/* Convert character using the SFM - "Services for Mac" remapping range */
65static bool
66convert_sfm_char(const __u16 src_char, char *target)
67{
Björn JACKE7e46f092017-06-01 11:00:06 +020068 if (src_char >= 0xF001 && src_char <= 0xF01F) {
69 *target = src_char - 0xF000;
70 return true;
71 }
Steve Frenchb6938552014-09-25 13:20:05 -050072 switch (src_char) {
73 case SFM_COLON:
74 *target = ':';
75 break;
Björn Jacke85435d72017-05-05 04:36:16 +020076 case SFM_DOUBLEQUOTE:
77 *target = '"';
78 break;
Steve Frenchb6938552014-09-25 13:20:05 -050079 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;
Steve French45e8a252016-06-22 21:07:32 -050094 case SFM_SPACE:
95 *target = ' ';
96 break;
97 case SFM_PERIOD:
98 *target = '.';
99 break;
Steve Frenchb6938552014-09-25 13:20:05 -0500100 default:
101 return false;
102 }
103 return true;
104}
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400105
Steve Frenchb6938552014-09-25 13:20:05 -0500106
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 */
118static int
Nakajima Akirab2910302015-04-09 17:27:39 +0900119cifs_mapchar(char *target, const __u16 *from, const struct nls_table *cp,
Steve Frenchb6938552014-09-25 13:20:05 -0500120 int maptype)
121{
122 int len = 1;
Nakajima Akirab2910302015-04-09 17:27:39 +0900123 __u16 src_char;
124
125 src_char = *from;
Steve Frenchb6938552014-09-25 13:20:05 -0500126
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 */
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500134 len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
Nakajima Akirab2910302015-04-09 17:27:39 +0900135 if (len <= 0)
136 goto surrogate_pair;
137
138 return len;
139
140surrogate_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
149unknown:
150 *target = '?';
151 len = 1;
Steve Frenchb6938552014-09-25 13:20:05 -0500152 return len;
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400153}
154
155/*
Steve Frenchacbbb762012-01-18 22:32:33 -0600156 * cifs_from_utf16 - convert utf16le string to local charset
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400157 * @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 *
Steve Frenchacbbb762012-01-18 22:32:33 -0600164 * Convert a little-endian utf16le string (as sent by the server) to a string
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400165 * 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
Steve Frenchacbbb762012-01-18 22:32:33 -0600173 * instead of straight UTF16-2. The linux nls routines however aren't able to
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400174 * deal with those characters properly. In the event that we get some of
175 * those characters, they won't be translated properly.
176 */
177int
Steve Frenchacbbb762012-01-18 22:32:33 -0600178cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
Steve Frenchb6938552014-09-25 13:20:05 -0500179 const struct nls_table *codepage, int map_type)
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400180{
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];
Nakajima Akirab2910302015-04-09 17:27:39 +0900186 __u16 ftmp[3]; /* ftmp[3] = 3array x 2bytes = 6bytes UTF-16 */
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400187
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
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500196 for (i = 0; i < fromwords; i++) {
Nakajima Akirab2910302015-04-09 17:27:39 +0900197 ftmp[0] = get_unaligned_le16(&from[i]);
198 if (ftmp[0] == 0)
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500199 break;
Nakajima Akirab2910302015-04-09 17:27:39 +0900200 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;
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500208
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400209 /*
210 * check to see if converting this character might make the
211 * conversion bleed into the null terminator
212 */
213 if (outlen >= safelen) {
Steve Frenchb6938552014-09-25 13:20:05 -0500214 charlen = cifs_mapchar(tmp, ftmp, codepage, map_type);
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400215 if ((outlen + charlen) > (tolen - nullsize))
216 break;
217 }
218
219 /* put converted char into 'to' buffer */
Steve Frenchb6938552014-09-25 13:20:05 -0500220 charlen = cifs_mapchar(&to[outlen], ftmp, codepage, map_type);
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400221 outlen += charlen;
Nakajima Akirab2910302015-04-09 17:27:39 +0900222
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;
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400233 }
234
235 /* properly null-terminate string */
236 for (i = 0; i < nullsize; i++)
237 to[outlen++] = 0;
238
239 return outlen;
240}
241
242/*
Steve Frenchacbbb762012-01-18 22:32:33 -0600243 * NAME: cifs_strtoUTF16()
Linus Torvalds1da177e2005-04-16 15:20:36 -0700244 *
245 * FUNCTION: Convert character string to unicode string
246 *
247 */
248int
Steve Frenchacbbb762012-01-18 22:32:33 -0600249cifs_strtoUTF16(__le16 *to, const char *from, int len,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250 const struct nls_table *codepage)
251{
252 int charlen;
253 int i;
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500254 wchar_t wchar_to; /* needed to quiet sparse */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700255
Frediano Zigliofd3ba422012-08-07 04:33:03 -0500256 /* 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700277 for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500278 charlen = codepage->char2uni(from, len, &wchar_to);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700279 if (charlen < 1) {
Joe Perchesf96637b2013-05-04 22:12:25 -0500280 cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n",
281 *from, charlen);
Steve French69114082005-11-10 19:28:44 -0800282 /* A question mark */
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500283 wchar_to = 0x003f;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284 charlen = 1;
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500285 }
286 put_unaligned_le16(wchar_to, &to[i]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700287 }
288
Frediano Zigliofd3ba422012-08-07 04:33:03 -0500289success:
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500290 put_unaligned_le16(0, &to[i]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291 return i;
292}
293
Jeff Layton066ce682009-04-30 07:16:14 -0400294/*
Nakajima Akirab2910302015-04-09 17:27:39 +0900295 * 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 */
304int
305cifs_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/*
Steve Frenchacbbb762012-01-18 22:32:33 -0600335 * cifs_strndup_from_utf16 - copy a string from wire format to the local
336 * codepage
Jeff Layton066ce682009-04-30 07:16:14 -0400337 * @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 */
346char *
Steve Frenchacbbb762012-01-18 22:32:33 -0600347cifs_strndup_from_utf16(const char *src, const int maxlen,
348 const bool is_unicode, const struct nls_table *codepage)
Jeff Layton066ce682009-04-30 07:16:14 -0400349{
350 int len;
351 char *dst;
352
353 if (is_unicode) {
Steve Frenchacbbb762012-01-18 22:32:33 -0600354 len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage);
Jeff Layton066ce682009-04-30 07:16:14 -0400355 len += nls_nullsize(codepage);
356 dst = kmalloc(len, GFP_KERNEL);
357 if (!dst)
358 return NULL;
Steve Frenchacbbb762012-01-18 22:32:33 -0600359 cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
Len Bakerf980d052021-08-17 12:27:09 +0200360 NO_MAP_UNI_RSVD);
Jeff Layton066ce682009-04-30 07:16:14 -0400361 } else {
Len Bakerf980d052021-08-17 12:27:09 +0200362 dst = kstrndup(src, maxlen, GFP_KERNEL);
Jeff Layton066ce682009-04-30 07:16:14 -0400363 }
364
365 return dst;
366}
367
Steve Frencha4153cb2014-09-25 14:01:34 -0500368static __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
Steve French45e8a252016-06-22 21:07:32 -0500398static __le16 convert_to_sfm_char(char src_char, bool end_of_string)
Steve Frencha4153cb2014-09-25 14:01:34 -0500399{
400 __le16 dest_char;
401
Björn JACKE7e46f092017-06-01 11:00:06 +0200402 if (src_char >= 0x01 && src_char <= 0x1F) {
403 dest_char = cpu_to_le16(src_char + 0xF000);
404 return dest_char;
405 }
Steve Frencha4153cb2014-09-25 14:01:34 -0500406 switch (src_char) {
407 case ':':
408 dest_char = cpu_to_le16(SFM_COLON);
409 break;
Björn Jacke85435d72017-05-05 04:36:16 +0200410 case '"':
411 dest_char = cpu_to_le16(SFM_DOUBLEQUOTE);
412 break;
Steve Frencha4153cb2014-09-25 14:01:34 -0500413 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;
Steve French45e8a252016-06-22 21:07:32 -0500428 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;
Steve Frencha4153cb2014-09-25 14:01:34 -0500440 default:
441 dest_char = 0;
442 }
443
444 return dest_char;
445}
446
Jeff Layton84cdf742011-01-20 13:36:51 -0500447/*
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 */
453int
Steve Frenchacbbb762012-01-18 22:32:33 -0600454cifsConvertToUTF16(__le16 *target, const char *source, int srclen,
Steve Frencha4153cb2014-09-25 14:01:34 -0500455 const struct nls_table *cp, int map_chars)
Jeff Layton84cdf742011-01-20 13:36:51 -0500456{
Steve Frenchce36d9a2014-06-22 20:38:49 -0500457 int i, charlen;
458 int j = 0;
Jeff Layton84cdf742011-01-20 13:36:51 -0500459 char src_char;
Jeff Layton581ade42011-04-05 15:02:37 -0400460 __le16 dst_char;
461 wchar_t tmp;
Nakajima Akirab2910302015-04-09 17:27:39 +0900462 wchar_t *wchar_to; /* UTF-16 */
463 int ret;
464 unicode_t u;
Jeff Layton84cdf742011-01-20 13:36:51 -0500465
Steve Frencha4153cb2014-09-25 14:01:34 -0500466 if (map_chars == NO_MAP_UNI_RSVD)
Steve Frenchacbbb762012-01-18 22:32:33 -0600467 return cifs_strtoUTF16(target, source, PATH_MAX, cp);
Jeff Layton84cdf742011-01-20 13:36:51 -0500468
Nakajima Akirab2910302015-04-09 17:27:39 +0900469 wchar_to = kzalloc(6, GFP_KERNEL);
470
Steve Frenchce36d9a2014-06-22 20:38:49 -0500471 for (i = 0; i < srclen; j++) {
Jeff Layton84cdf742011-01-20 13:36:51 -0500472 src_char = source[i];
Jeff Layton11379b52011-05-17 15:28:21 -0400473 charlen = 1;
Steve Frencha4153cb2014-09-25 14:01:34 -0500474
475 /* check if end of string */
476 if (src_char == 0)
Steve Frenchacbbb762012-01-18 22:32:33 -0600477 goto ctoUTF16_out;
Steve Frencha4153cb2014-09-25 14:01:34 -0500478
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);
Steve French45e8a252016-06-22 21:07:32 -0500482 else if (map_chars == SFM_MAP_UNI_RSVD) {
483 bool end_of_string;
484
Boris Protopopov57c17602020-09-24 00:36:38 +0000485 /**
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] == '\\'))
Steve French45e8a252016-06-22 21:07:32 -0500492 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
Steve Frencha4153cb2014-09-25 14:01:34 -0500498 dst_char = 0;
Jeff Layton84cdf742011-01-20 13:36:51 -0500499 /*
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 */
Steve Frencha4153cb2014-09-25 14:01:34 -0500504 if (dst_char == 0) {
Jeff Layton581ade42011-04-05 15:02:37 -0400505 charlen = cp->char2uni(source + i, srclen - i, &tmp);
506 dst_char = cpu_to_le16(tmp);
507
Jeff Layton84cdf742011-01-20 13:36:51 -0500508 /*
509 * if no match, use question mark, which at least in
510 * some cases serves as wild card
511 */
Nakajima Akirab2910302015-04-09 17:27:39 +0900512 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]);
Jeff Layton84cdf742011-01-20 13:36:51 -0500552 }
Nakajima Akirab2910302015-04-09 17:27:39 +0900553 continue;
554
555unknown:
556 dst_char = cpu_to_le16(0x003f);
557 charlen = 1;
Jeff Layton84cdf742011-01-20 13:36:51 -0500558 }
Nakajima Akirab2910302015-04-09 17:27:39 +0900559
560ctoUTF16:
Jeff Layton11379b52011-05-17 15:28:21 -0400561 /*
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;
Jeff Layton581ade42011-04-05 15:02:37 -0400566 put_unaligned(dst_char, &target[j]);
Jeff Layton84cdf742011-01-20 13:36:51 -0500567 }
568
Steve Frenchacbbb762012-01-18 22:32:33 -0600569ctoUTF16_out:
Steve Frenchce36d9a2014-06-22 20:38:49 -0500570 put_unaligned(0, &target[j]); /* Null terminate target unicode string */
Nakajima Akirab2910302015-04-09 17:27:39 +0900571 kfree(wchar_to);
Jeff Laytonc73f6932012-09-18 14:21:01 -0400572 return j;
Jeff Layton84cdf742011-01-20 13:36:51 -0500573}
Pavel Shilovsky2503a0d2011-12-26 22:58:46 +0400574
Pavel Shilovsky2503a0d2011-12-26 22:58:46 +0400575/*
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
586static int
587cifs_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 *
616cifs_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}