blob: 0227b45ef00a372e4c201519710a123f63d44f35 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * fs/cifs/cifs_unicode.c
3 *
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)
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
Steve French221601c2007-06-05 20:35:06 +00009 * the Free Software Foundation; either version 2 of the License, or
Linus Torvalds1da177e2005-04-16 15:20:36 -070010 * (at your option) any later version.
Steve French221601c2007-06-05 20:35:06 +000011 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070012 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
15 * the GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
Steve French221601c2007-06-05 20:35:06 +000018 * along with this program; if not, write to the Free Software
Linus Torvalds1da177e2005-04-16 15:20:36 -070019 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 */
21#include <linux/fs.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090022#include <linux/slab.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070023#include "cifs_unicode.h"
24#include "cifs_uniupr.h"
25#include "cifspdu.h"
Steve French39798772006-05-31 22:40:51 +000026#include "cifsglob.h"
Linus Torvalds1da177e2005-04-16 15:20:36 -070027#include "cifs_debug.h"
28
29/*
Steve Frenchacbbb762012-01-18 22:32:33 -060030 * cifs_utf16_bytes - how long will a string be after conversion?
31 * @utf16 - pointer to input string
Jeff Layton69f801f2009-04-30 06:46:32 -040032 * @maxbytes - don't go past this many bytes of input string
33 * @codepage - destination codepage
34 *
Steve Frenchacbbb762012-01-18 22:32:33 -060035 * Walk a utf16le string and return the number of bytes that the string will
Jeff Layton69f801f2009-04-30 06:46:32 -040036 * be after being converted to the given charset, not including any null
37 * termination required. Don't walk past maxbytes in the source buffer.
38 */
39int
Steve Frenchacbbb762012-01-18 22:32:33 -060040cifs_utf16_bytes(const __le16 *from, int maxbytes,
Jeff Layton69f801f2009-04-30 06:46:32 -040041 const struct nls_table *codepage)
42{
43 int i;
44 int charlen, outlen = 0;
45 int maxwords = maxbytes / 2;
46 char tmp[NLS_MAX_CHARSET_SIZE];
Jeff Laytonba2dbf32011-01-20 13:36:51 -050047 __u16 ftmp;
Jeff Layton69f801f2009-04-30 06:46:32 -040048
Jeff Laytonba2dbf32011-01-20 13:36:51 -050049 for (i = 0; i < maxwords; i++) {
50 ftmp = get_unaligned_le16(&from[i]);
51 if (ftmp == 0)
52 break;
53
54 charlen = codepage->uni2char(ftmp, tmp, NLS_MAX_CHARSET_SIZE);
Jeff Layton69f801f2009-04-30 06:46:32 -040055 if (charlen > 0)
56 outlen += charlen;
57 else
58 outlen++;
59 }
60
61 return outlen;
62}
63
64/*
Jeff Laytonba2dbf32011-01-20 13:36:51 -050065 * cifs_mapchar - convert a host-endian char to proper char in codepage
Jeff Layton7fabf0c2009-04-30 06:46:15 -040066 * @target - where converted character should be copied
Jeff Laytonba2dbf32011-01-20 13:36:51 -050067 * @src_char - 2 byte host-endian source character
Jeff Layton7fabf0c2009-04-30 06:46:15 -040068 * @cp - codepage to which character should be converted
69 * @mapchar - should character be mapped according to mapchars mount option?
70 *
71 * This function handles the conversion of a single character. It is the
72 * responsibility of the caller to ensure that the target buffer is large
73 * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE).
74 */
75static int
Jeff Laytonba2dbf32011-01-20 13:36:51 -050076cifs_mapchar(char *target, const __u16 src_char, const struct nls_table *cp,
Jeff Layton7fabf0c2009-04-30 06:46:15 -040077 bool mapchar)
78{
79 int len = 1;
80
81 if (!mapchar)
82 goto cp_convert;
83
84 /*
85 * BB: Cannot handle remapping UNI_SLASH until all the calls to
86 * build_path_from_dentry are modified, as they use slash as
87 * separator.
88 */
Jeff Laytonba2dbf32011-01-20 13:36:51 -050089 switch (src_char) {
Jeff Layton7fabf0c2009-04-30 06:46:15 -040090 case UNI_COLON:
91 *target = ':';
92 break;
Jeff Layton581ade42011-04-05 15:02:37 -040093 case UNI_ASTERISK:
Jeff Layton7fabf0c2009-04-30 06:46:15 -040094 *target = '*';
95 break;
96 case UNI_QUESTION:
97 *target = '?';
98 break;
99 case UNI_PIPE:
100 *target = '|';
101 break;
102 case UNI_GRTRTHAN:
103 *target = '>';
104 break;
105 case UNI_LESSTHAN:
106 *target = '<';
107 break;
108 default:
109 goto cp_convert;
110 }
111
112out:
113 return len;
114
115cp_convert:
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500116 len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400117 if (len <= 0) {
118 *target = '?';
119 len = 1;
120 }
121 goto out;
122}
123
124/*
Steve Frenchacbbb762012-01-18 22:32:33 -0600125 * cifs_from_utf16 - convert utf16le string to local charset
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400126 * @to - destination buffer
127 * @from - source buffer
128 * @tolen - destination buffer size (in bytes)
129 * @fromlen - source buffer size (in bytes)
130 * @codepage - codepage to which characters should be converted
131 * @mapchar - should characters be remapped according to the mapchars option?
132 *
Steve Frenchacbbb762012-01-18 22:32:33 -0600133 * Convert a little-endian utf16le string (as sent by the server) to a string
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400134 * in the provided codepage. The tolen and fromlen parameters are to ensure
135 * that the code doesn't walk off of the end of the buffer (which is always
136 * a danger if the alignment of the source buffer is off). The destination
137 * string is always properly null terminated and fits in the destination
138 * buffer. Returns the length of the destination string in bytes (including
139 * null terminator).
140 *
141 * Note that some windows versions actually send multiword UTF-16 characters
Steve Frenchacbbb762012-01-18 22:32:33 -0600142 * instead of straight UTF16-2. The linux nls routines however aren't able to
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400143 * deal with those characters properly. In the event that we get some of
144 * those characters, they won't be translated properly.
145 */
146int
Steve Frenchacbbb762012-01-18 22:32:33 -0600147cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400148 const struct nls_table *codepage, bool mapchar)
149{
150 int i, charlen, safelen;
151 int outlen = 0;
152 int nullsize = nls_nullsize(codepage);
153 int fromwords = fromlen / 2;
154 char tmp[NLS_MAX_CHARSET_SIZE];
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500155 __u16 ftmp;
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400156
157 /*
158 * because the chars can be of varying widths, we need to take care
159 * not to overflow the destination buffer when we get close to the
160 * end of it. Until we get to this offset, we don't need to check
161 * for overflow however.
162 */
163 safelen = tolen - (NLS_MAX_CHARSET_SIZE + nullsize);
164
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500165 for (i = 0; i < fromwords; i++) {
166 ftmp = get_unaligned_le16(&from[i]);
167 if (ftmp == 0)
168 break;
169
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400170 /*
171 * check to see if converting this character might make the
172 * conversion bleed into the null terminator
173 */
174 if (outlen >= safelen) {
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500175 charlen = cifs_mapchar(tmp, ftmp, codepage, mapchar);
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400176 if ((outlen + charlen) > (tolen - nullsize))
177 break;
178 }
179
180 /* put converted char into 'to' buffer */
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500181 charlen = cifs_mapchar(&to[outlen], ftmp, codepage, mapchar);
Jeff Layton7fabf0c2009-04-30 06:46:15 -0400182 outlen += charlen;
183 }
184
185 /* properly null-terminate string */
186 for (i = 0; i < nullsize; i++)
187 to[outlen++] = 0;
188
189 return outlen;
190}
191
192/*
Steve Frenchacbbb762012-01-18 22:32:33 -0600193 * NAME: cifs_strtoUTF16()
Linus Torvalds1da177e2005-04-16 15:20:36 -0700194 *
195 * FUNCTION: Convert character string to unicode string
196 *
197 */
198int
Steve Frenchacbbb762012-01-18 22:32:33 -0600199cifs_strtoUTF16(__le16 *to, const char *from, int len,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700200 const struct nls_table *codepage)
201{
202 int charlen;
203 int i;
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500204 wchar_t wchar_to; /* needed to quiet sparse */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700205
Frediano Zigliofd3ba422012-08-07 04:33:03 -0500206 /* special case for utf8 to handle no plane0 chars */
207 if (!strcmp(codepage->charset, "utf8")) {
208 /*
209 * convert utf8 -> utf16, we assume we have enough space
210 * as caller should have assumed conversion does not overflow
211 * in destination len is length in wchar_t units (16bits)
212 */
213 i = utf8s_to_utf16s(from, len, UTF16_LITTLE_ENDIAN,
214 (wchar_t *) to, len);
215
216 /* if success terminate and exit */
217 if (i >= 0)
218 goto success;
219 /*
220 * if fails fall back to UCS encoding as this
221 * function should not return negative values
222 * currently can fail only if source contains
223 * invalid encoded characters
224 */
225 }
226
Linus Torvalds1da177e2005-04-16 15:20:36 -0700227 for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500228 charlen = codepage->char2uni(from, len, &wchar_to);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700229 if (charlen < 1) {
Joe Perchesf96637b2013-05-04 22:12:25 -0500230 cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n",
231 *from, charlen);
Steve French69114082005-11-10 19:28:44 -0800232 /* A question mark */
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500233 wchar_to = 0x003f;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234 charlen = 1;
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500235 }
236 put_unaligned_le16(wchar_to, &to[i]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237 }
238
Frediano Zigliofd3ba422012-08-07 04:33:03 -0500239success:
Jeff Laytonba2dbf32011-01-20 13:36:51 -0500240 put_unaligned_le16(0, &to[i]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700241 return i;
242}
243
Jeff Layton066ce682009-04-30 07:16:14 -0400244/*
Steve Frenchacbbb762012-01-18 22:32:33 -0600245 * cifs_strndup_from_utf16 - copy a string from wire format to the local
246 * codepage
Jeff Layton066ce682009-04-30 07:16:14 -0400247 * @src - source string
248 * @maxlen - don't walk past this many bytes in the source string
249 * @is_unicode - is this a unicode string?
250 * @codepage - destination codepage
251 *
252 * Take a string given by the server, convert it to the local codepage and
253 * put it in a new buffer. Returns a pointer to the new string or NULL on
254 * error.
255 */
256char *
Steve Frenchacbbb762012-01-18 22:32:33 -0600257cifs_strndup_from_utf16(const char *src, const int maxlen,
258 const bool is_unicode, const struct nls_table *codepage)
Jeff Layton066ce682009-04-30 07:16:14 -0400259{
260 int len;
261 char *dst;
262
263 if (is_unicode) {
Steve Frenchacbbb762012-01-18 22:32:33 -0600264 len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage);
Jeff Layton066ce682009-04-30 07:16:14 -0400265 len += nls_nullsize(codepage);
266 dst = kmalloc(len, GFP_KERNEL);
267 if (!dst)
268 return NULL;
Steve Frenchacbbb762012-01-18 22:32:33 -0600269 cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
Jeff Layton066ce682009-04-30 07:16:14 -0400270 false);
271 } else {
272 len = strnlen(src, maxlen);
273 len++;
274 dst = kmalloc(len, GFP_KERNEL);
275 if (!dst)
276 return NULL;
277 strlcpy(dst, src, len);
278 }
279
280 return dst;
281}
282
Jeff Layton84cdf742011-01-20 13:36:51 -0500283/*
284 * Convert 16 bit Unicode pathname to wire format from string in current code
285 * page. Conversion may involve remapping up the six characters that are
286 * only legal in POSIX-like OS (if they are present in the string). Path
287 * names are little endian 16 bit Unicode on the wire
288 */
289int
Steve Frenchacbbb762012-01-18 22:32:33 -0600290cifsConvertToUTF16(__le16 *target, const char *source, int srclen,
Jeff Layton84cdf742011-01-20 13:36:51 -0500291 const struct nls_table *cp, int mapChars)
292{
293 int i, j, charlen;
Jeff Layton84cdf742011-01-20 13:36:51 -0500294 char src_char;
Jeff Layton581ade42011-04-05 15:02:37 -0400295 __le16 dst_char;
296 wchar_t tmp;
Jeff Layton84cdf742011-01-20 13:36:51 -0500297
298 if (!mapChars)
Steve Frenchacbbb762012-01-18 22:32:33 -0600299 return cifs_strtoUTF16(target, source, PATH_MAX, cp);
Jeff Layton84cdf742011-01-20 13:36:51 -0500300
Jeff Layton581ade42011-04-05 15:02:37 -0400301 for (i = 0, j = 0; i < srclen; j++) {
Jeff Layton84cdf742011-01-20 13:36:51 -0500302 src_char = source[i];
Jeff Layton11379b52011-05-17 15:28:21 -0400303 charlen = 1;
Jeff Layton84cdf742011-01-20 13:36:51 -0500304 switch (src_char) {
305 case 0:
Jeff Layton581ade42011-04-05 15:02:37 -0400306 put_unaligned(0, &target[j]);
Steve Frenchacbbb762012-01-18 22:32:33 -0600307 goto ctoUTF16_out;
Jeff Layton84cdf742011-01-20 13:36:51 -0500308 case ':':
Jeff Layton581ade42011-04-05 15:02:37 -0400309 dst_char = cpu_to_le16(UNI_COLON);
Jeff Layton84cdf742011-01-20 13:36:51 -0500310 break;
311 case '*':
Jeff Layton581ade42011-04-05 15:02:37 -0400312 dst_char = cpu_to_le16(UNI_ASTERISK);
Jeff Layton84cdf742011-01-20 13:36:51 -0500313 break;
314 case '?':
Jeff Layton581ade42011-04-05 15:02:37 -0400315 dst_char = cpu_to_le16(UNI_QUESTION);
Jeff Layton84cdf742011-01-20 13:36:51 -0500316 break;
317 case '<':
Jeff Layton581ade42011-04-05 15:02:37 -0400318 dst_char = cpu_to_le16(UNI_LESSTHAN);
Jeff Layton84cdf742011-01-20 13:36:51 -0500319 break;
320 case '>':
Jeff Layton581ade42011-04-05 15:02:37 -0400321 dst_char = cpu_to_le16(UNI_GRTRTHAN);
Jeff Layton84cdf742011-01-20 13:36:51 -0500322 break;
323 case '|':
Jeff Layton581ade42011-04-05 15:02:37 -0400324 dst_char = cpu_to_le16(UNI_PIPE);
Jeff Layton84cdf742011-01-20 13:36:51 -0500325 break;
326 /*
327 * FIXME: We can not handle remapping backslash (UNI_SLASH)
328 * until all the calls to build_path_from_dentry are modified,
329 * as they use backslash as separator.
330 */
331 default:
Jeff Layton581ade42011-04-05 15:02:37 -0400332 charlen = cp->char2uni(source + i, srclen - i, &tmp);
333 dst_char = cpu_to_le16(tmp);
334
Jeff Layton84cdf742011-01-20 13:36:51 -0500335 /*
336 * if no match, use question mark, which at least in
337 * some cases serves as wild card
338 */
339 if (charlen < 1) {
Jeff Layton581ade42011-04-05 15:02:37 -0400340 dst_char = cpu_to_le16(0x003f);
Jeff Layton84cdf742011-01-20 13:36:51 -0500341 charlen = 1;
342 }
Jeff Layton84cdf742011-01-20 13:36:51 -0500343 }
Jeff Layton11379b52011-05-17 15:28:21 -0400344 /*
345 * character may take more than one byte in the source string,
346 * but will take exactly two bytes in the target string
347 */
348 i += charlen;
Jeff Layton581ade42011-04-05 15:02:37 -0400349 put_unaligned(dst_char, &target[j]);
Jeff Layton84cdf742011-01-20 13:36:51 -0500350 }
351
Steve Frenchacbbb762012-01-18 22:32:33 -0600352ctoUTF16_out:
Jeff Laytonc73f6932012-09-18 14:21:01 -0400353 return j;
Jeff Layton84cdf742011-01-20 13:36:51 -0500354}
Pavel Shilovsky2503a0d2011-12-26 22:58:46 +0400355
356#ifdef CONFIG_CIFS_SMB2
357/*
358 * cifs_local_to_utf16_bytes - how long will a string be after conversion?
359 * @from - pointer to input string
360 * @maxbytes - don't go past this many bytes of input string
361 * @codepage - source codepage
362 *
363 * Walk a string and return the number of bytes that the string will
364 * be after being converted to the given charset, not including any null
365 * termination required. Don't walk past maxbytes in the source buffer.
366 */
367
368static int
369cifs_local_to_utf16_bytes(const char *from, int len,
370 const struct nls_table *codepage)
371{
372 int charlen;
373 int i;
374 wchar_t wchar_to;
375
376 for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
377 charlen = codepage->char2uni(from, len, &wchar_to);
378 /* Failed conversion defaults to a question mark */
379 if (charlen < 1)
380 charlen = 1;
381 }
382 return 2 * i; /* UTF16 characters are two bytes */
383}
384
385/*
386 * cifs_strndup_to_utf16 - copy a string to wire format from the local codepage
387 * @src - source string
388 * @maxlen - don't walk past this many bytes in the source string
389 * @utf16_len - the length of the allocated string in bytes (including null)
390 * @cp - source codepage
391 * @remap - map special chars
392 *
393 * Take a string convert it from the local codepage to UTF16 and
394 * put it in a new buffer. Returns a pointer to the new string or NULL on
395 * error.
396 */
397__le16 *
398cifs_strndup_to_utf16(const char *src, const int maxlen, int *utf16_len,
399 const struct nls_table *cp, int remap)
400{
401 int len;
402 __le16 *dst;
403
404 len = cifs_local_to_utf16_bytes(src, maxlen, cp);
405 len += 2; /* NULL */
406 dst = kmalloc(len, GFP_KERNEL);
407 if (!dst) {
408 *utf16_len = 0;
409 return NULL;
410 }
411 cifsConvertToUTF16(dst, src, strlen(src), cp, remap);
412 *utf16_len = len;
413 return dst;
414}
415#endif /* CONFIG_CIFS_SMB2 */