aboutsummaryrefslogtreecommitdiffstats
path: root/fs/udf/unicode.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/udf/unicode.c')
-rw-r--r--fs/udf/unicode.c260
1 files changed, 126 insertions, 134 deletions
diff --git a/fs/udf/unicode.c b/fs/udf/unicode.c
index 16a8ad21b77e..45234791fec2 100644
--- a/fs/udf/unicode.c
+++ b/fs/udf/unicode.c
@@ -28,101 +28,64 @@
28 28
29#include "udf_sb.h" 29#include "udf_sb.h"
30 30
31#define PLANE_SIZE 0x10000
32#define UNICODE_MAX 0x10ffff
31#define SURROGATE_MASK 0xfffff800 33#define SURROGATE_MASK 0xfffff800
32#define SURROGATE_PAIR 0x0000d800 34#define SURROGATE_PAIR 0x0000d800
35#define SURROGATE_LOW 0x00000400
36#define SURROGATE_CHAR_BITS 10
37#define SURROGATE_CHAR_MASK ((1 << SURROGATE_CHAR_BITS) - 1)
33 38
34static int udf_uni2char_utf8(wchar_t uni, 39#define ILLEGAL_CHAR_MARK '_'
35 unsigned char *out, 40#define EXT_MARK '.'
36 int boundlen) 41#define CRC_MARK '#'
37{ 42#define EXT_SIZE 5
38 int u_len = 0; 43/* Number of chars we need to store generated CRC to make filename unique */
39 44#define CRC_LEN 5
40 if (boundlen <= 0)
41 return -ENAMETOOLONG;
42 45
43 if ((uni & SURROGATE_MASK) == SURROGATE_PAIR) 46static unicode_t get_utf16_char(const uint8_t *str_i, int str_i_max_len,
44 return -EINVAL; 47 int str_i_idx, int u_ch, unicode_t *ret)
48{
49 unicode_t c;
50 int start_idx = str_i_idx;
51
52 /* Expand OSTA compressed Unicode to Unicode */
53 c = str_i[str_i_idx++];
54 if (u_ch > 1)
55 c = (c << 8) | str_i[str_i_idx++];
56 if ((c & SURROGATE_MASK) == SURROGATE_PAIR) {
57 unicode_t next;
58
59 /* Trailing surrogate char */
60 if (str_i_idx >= str_i_max_len) {
61 c = UNICODE_MAX + 1;
62 goto out;
63 }
45 64
46 if (uni < 0x80) { 65 /* Low surrogate must follow the high one... */
47 out[u_len++] = (unsigned char)uni; 66 if (c & SURROGATE_LOW) {
48 } else if (uni < 0x800) { 67 c = UNICODE_MAX + 1;
49 if (boundlen < 2) 68 goto out;
50 return -ENAMETOOLONG; 69 }
51 out[u_len++] = (unsigned char)(0xc0 | (uni >> 6));
52 out[u_len++] = (unsigned char)(0x80 | (uni & 0x3f));
53 } else {
54 if (boundlen < 3)
55 return -ENAMETOOLONG;
56 out[u_len++] = (unsigned char)(0xe0 | (uni >> 12));
57 out[u_len++] = (unsigned char)(0x80 | ((uni >> 6) & 0x3f));
58 out[u_len++] = (unsigned char)(0x80 | (uni & 0x3f));
59 }
60 return u_len;
61}
62 70
63static int udf_char2uni_utf8(const unsigned char *in, 71 WARN_ON_ONCE(u_ch != 2);
64 int boundlen, 72 next = str_i[str_i_idx++] << 8;
65 wchar_t *uni) 73 next |= str_i[str_i_idx++];
66{ 74 if ((next & SURROGATE_MASK) != SURROGATE_PAIR ||
67 unsigned int utf_char; 75 !(next & SURROGATE_LOW)) {
68 unsigned char c; 76 c = UNICODE_MAX + 1;
69 int utf_cnt, u_len; 77 goto out;
70
71 utf_char = 0;
72 utf_cnt = 0;
73 for (u_len = 0; u_len < boundlen;) {
74 c = in[u_len++];
75
76 /* Complete a multi-byte UTF-8 character */
77 if (utf_cnt) {
78 utf_char = (utf_char << 6) | (c & 0x3f);
79 if (--utf_cnt)
80 continue;
81 } else {
82 /* Check for a multi-byte UTF-8 character */
83 if (c & 0x80) {
84 /* Start a multi-byte UTF-8 character */
85 if ((c & 0xe0) == 0xc0) {
86 utf_char = c & 0x1f;
87 utf_cnt = 1;
88 } else if ((c & 0xf0) == 0xe0) {
89 utf_char = c & 0x0f;
90 utf_cnt = 2;
91 } else if ((c & 0xf8) == 0xf0) {
92 utf_char = c & 0x07;
93 utf_cnt = 3;
94 } else if ((c & 0xfc) == 0xf8) {
95 utf_char = c & 0x03;
96 utf_cnt = 4;
97 } else if ((c & 0xfe) == 0xfc) {
98 utf_char = c & 0x01;
99 utf_cnt = 5;
100 } else {
101 utf_cnt = -1;
102 break;
103 }
104 continue;
105 } else {
106 /* Single byte UTF-8 character (most common) */
107 utf_char = c;
108 }
109 } 78 }
110 *uni = utf_char; 79
111 break; 80 c = PLANE_SIZE +
112 } 81 ((c & SURROGATE_CHAR_MASK) << SURROGATE_CHAR_BITS) +
113 if (utf_cnt) { 82 (next & SURROGATE_CHAR_MASK);
114 *uni = '?';
115 return -EINVAL;
116 } 83 }
117 return u_len; 84out:
85 *ret = c;
86 return str_i_idx - start_idx;
118} 87}
119 88
120#define ILLEGAL_CHAR_MARK '_'
121#define EXT_MARK '.'
122#define CRC_MARK '#'
123#define EXT_SIZE 5
124/* Number of chars we need to store generated CRC to make filename unique */
125#define CRC_LEN 5
126 89
127static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len, 90static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
128 int *str_o_idx, 91 int *str_o_idx,
@@ -132,27 +95,29 @@ static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
132 int (*conv_f)(wchar_t, unsigned char *, int), 95 int (*conv_f)(wchar_t, unsigned char *, int),
133 int translate) 96 int translate)
134{ 97{
135 uint32_t c; 98 unicode_t c;
136 int illChar = 0; 99 int illChar = 0;
137 int len, gotch = 0; 100 int len, gotch = 0;
138 101
139 for (; (!gotch) && (*str_i_idx < str_i_max_len); *str_i_idx += u_ch) { 102 while (!gotch && *str_i_idx < str_i_max_len) {
140 if (*str_o_idx >= str_o_max_len) { 103 if (*str_o_idx >= str_o_max_len) {
141 *needsCRC = 1; 104 *needsCRC = 1;
142 return gotch; 105 return gotch;
143 } 106 }
144 107
145 /* Expand OSTA compressed Unicode to Unicode */ 108 len = get_utf16_char(str_i, str_i_max_len, *str_i_idx, u_ch,
146 c = str_i[*str_i_idx]; 109 &c);
147 if (u_ch > 1) 110 /* These chars cannot be converted. Replace them. */
148 c = (c << 8) | str_i[*str_i_idx + 1]; 111 if (c == 0 || c > UNICODE_MAX || (conv_f && c > MAX_WCHAR_T) ||
149 112 (translate && c == '/')) {
150 if (translate && (c == '/' || c == 0))
151 illChar = 1; 113 illChar = 1;
152 else if (illChar) 114 if (!translate)
115 gotch = 1;
116 } else if (illChar)
153 break; 117 break;
154 else 118 else
155 gotch = 1; 119 gotch = 1;
120 *str_i_idx += len;
156 } 121 }
157 if (illChar) { 122 if (illChar) {
158 *needsCRC = 1; 123 *needsCRC = 1;
@@ -160,7 +125,15 @@ static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
160 gotch = 1; 125 gotch = 1;
161 } 126 }
162 if (gotch) { 127 if (gotch) {
163 len = conv_f(c, &str_o[*str_o_idx], str_o_max_len - *str_o_idx); 128 if (conv_f) {
129 len = conv_f(c, &str_o[*str_o_idx],
130 str_o_max_len - *str_o_idx);
131 } else {
132 len = utf32_to_utf8(c, &str_o[*str_o_idx],
133 str_o_max_len - *str_o_idx);
134 if (len < 0)
135 len = -ENAMETOOLONG;
136 }
164 /* Valid character? */ 137 /* Valid character? */
165 if (len >= 0) 138 if (len >= 0)
166 *str_o_idx += len; 139 *str_o_idx += len;
@@ -168,16 +141,16 @@ static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
168 *needsCRC = 1; 141 *needsCRC = 1;
169 gotch = 0; 142 gotch = 0;
170 } else { 143 } else {
171 str_o[(*str_o_idx)++] = '?'; 144 str_o[(*str_o_idx)++] = ILLEGAL_CHAR_MARK;
172 *needsCRC = 1; 145 *needsCRC = 1;
173 } 146 }
174 } 147 }
175 return gotch; 148 return gotch;
176} 149}
177 150
178static int udf_name_from_CS0(uint8_t *str_o, int str_max_len, 151static int udf_name_from_CS0(struct super_block *sb,
152 uint8_t *str_o, int str_max_len,
179 const uint8_t *ocu, int ocu_len, 153 const uint8_t *ocu, int ocu_len,
180 int (*conv_f)(wchar_t, unsigned char *, int),
181 int translate) 154 int translate)
182{ 155{
183 uint32_t c; 156 uint32_t c;
@@ -194,6 +167,7 @@ static int udf_name_from_CS0(uint8_t *str_o, int str_max_len,
194 unsigned short valueCRC; 167 unsigned short valueCRC;
195 uint8_t ext[EXT_SIZE * NLS_MAX_CHARSET_SIZE + 1]; 168 uint8_t ext[EXT_SIZE * NLS_MAX_CHARSET_SIZE + 1];
196 uint8_t crc[CRC_LEN]; 169 uint8_t crc[CRC_LEN];
170 int (*conv_f)(wchar_t, unsigned char *, int);
197 171
198 if (str_max_len <= 0) 172 if (str_max_len <= 0)
199 return 0; 173 return 0;
@@ -203,6 +177,11 @@ static int udf_name_from_CS0(uint8_t *str_o, int str_max_len,
203 return 0; 177 return 0;
204 } 178 }
205 179
180 if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP))
181 conv_f = UDF_SB(sb)->s_nls_map->uni2char;
182 else
183 conv_f = NULL;
184
206 cmp_id = ocu[0]; 185 cmp_id = ocu[0];
207 if (cmp_id != 8 && cmp_id != 16) { 186 if (cmp_id != 8 && cmp_id != 16) {
208 memset(str_o, 0, str_max_len); 187 memset(str_o, 0, str_max_len);
@@ -293,18 +272,24 @@ static int udf_name_from_CS0(uint8_t *str_o, int str_max_len,
293 return str_o_len; 272 return str_o_len;
294} 273}
295 274
296static int udf_name_to_CS0(uint8_t *ocu, int ocu_max_len, 275static int udf_name_to_CS0(struct super_block *sb,
297 const uint8_t *str_i, int str_len, 276 uint8_t *ocu, int ocu_max_len,
298 int (*conv_f)(const unsigned char *, int, wchar_t *)) 277 const uint8_t *str_i, int str_len)
299{ 278{
300 int i, len; 279 int i, len;
301 unsigned int max_val; 280 unsigned int max_val;
302 wchar_t uni_char;
303 int u_len, u_ch; 281 int u_len, u_ch;
282 unicode_t uni_char;
283 int (*conv_f)(const unsigned char *, int, wchar_t *);
304 284
305 if (ocu_max_len <= 0) 285 if (ocu_max_len <= 0)
306 return 0; 286 return 0;
307 287
288 if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP))
289 conv_f = UDF_SB(sb)->s_nls_map->char2uni;
290 else
291 conv_f = NULL;
292
308 memset(ocu, 0, ocu_max_len); 293 memset(ocu, 0, ocu_max_len);
309 ocu[0] = 8; 294 ocu[0] = 8;
310 max_val = 0xff; 295 max_val = 0xff;
@@ -312,36 +297,61 @@ static int udf_name_to_CS0(uint8_t *ocu, int ocu_max_len,
312 297
313try_again: 298try_again:
314 u_len = 1; 299 u_len = 1;
315 for (i = 0; i < str_len; i++) { 300 for (i = 0; i < str_len; i += len) {
316 /* Name didn't fit? */ 301 /* Name didn't fit? */
317 if (u_len + u_ch > ocu_max_len) 302 if (u_len + u_ch > ocu_max_len)
318 return 0; 303 return 0;
319 len = conv_f(&str_i[i], str_len - i, &uni_char); 304 if (conv_f) {
320 if (!len) 305 wchar_t wchar;
321 continue; 306
307 len = conv_f(&str_i[i], str_len - i, &wchar);
308 if (len > 0)
309 uni_char = wchar;
310 } else {
311 len = utf8_to_utf32(&str_i[i], str_len - i,
312 &uni_char);
313 }
322 /* Invalid character, deal with it */ 314 /* Invalid character, deal with it */
323 if (len < 0) { 315 if (len <= 0 || uni_char > UNICODE_MAX) {
324 len = 1; 316 len = 1;
325 uni_char = '?'; 317 uni_char = '?';
326 } 318 }
327 319
328 if (uni_char > max_val) { 320 if (uni_char > max_val) {
329 max_val = 0xffff; 321 unicode_t c;
330 ocu[0] = 0x10; 322
331 u_ch = 2; 323 if (max_val == 0xff) {
332 goto try_again; 324 max_val = 0xffff;
325 ocu[0] = 0x10;
326 u_ch = 2;
327 goto try_again;
328 }
329 /*
330 * Use UTF-16 encoding for chars outside we
331 * cannot encode directly.
332 */
333 if (u_len + 2 * u_ch > ocu_max_len)
334 return 0;
335
336 uni_char -= PLANE_SIZE;
337 c = SURROGATE_PAIR |
338 ((uni_char >> SURROGATE_CHAR_BITS) &
339 SURROGATE_CHAR_MASK);
340 ocu[u_len++] = (uint8_t)(c >> 8);
341 ocu[u_len++] = (uint8_t)(c & 0xff);
342 uni_char = SURROGATE_PAIR | SURROGATE_LOW |
343 (uni_char & SURROGATE_CHAR_MASK);
333 } 344 }
334 345
335 if (max_val == 0xffff) 346 if (max_val == 0xffff)
336 ocu[u_len++] = (uint8_t)(uni_char >> 8); 347 ocu[u_len++] = (uint8_t)(uni_char >> 8);
337 ocu[u_len++] = (uint8_t)(uni_char & 0xff); 348 ocu[u_len++] = (uint8_t)(uni_char & 0xff);
338 i += len - 1;
339 } 349 }
340 350
341 return u_len; 351 return u_len;
342} 352}
343 353
344int udf_dstrCS0toUTF8(uint8_t *utf_o, int o_len, 354int udf_dstrCS0toChar(struct super_block *sb, uint8_t *utf_o, int o_len,
345 const uint8_t *ocu_i, int i_len) 355 const uint8_t *ocu_i, int i_len)
346{ 356{
347 int s_len = 0; 357 int s_len = 0;
@@ -355,14 +365,12 @@ int udf_dstrCS0toUTF8(uint8_t *utf_o, int o_len,
355 } 365 }
356 } 366 }
357 367
358 return udf_name_from_CS0(utf_o, o_len, ocu_i, s_len, 368 return udf_name_from_CS0(sb, utf_o, o_len, ocu_i, s_len, 0);
359 udf_uni2char_utf8, 0);
360} 369}
361 370
362int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen, 371int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
363 uint8_t *dname, int dlen) 372 uint8_t *dname, int dlen)
364{ 373{
365 int (*conv_f)(wchar_t, unsigned char *, int);
366 int ret; 374 int ret;
367 375
368 if (!slen) 376 if (!slen)
@@ -371,14 +379,7 @@ int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
371 if (dlen <= 0) 379 if (dlen <= 0)
372 return 0; 380 return 0;
373 381
374 if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) { 382 ret = udf_name_from_CS0(sb, dname, dlen, sname, slen, 1);
375 conv_f = udf_uni2char_utf8;
376 } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
377 conv_f = UDF_SB(sb)->s_nls_map->uni2char;
378 } else
379 BUG();
380
381 ret = udf_name_from_CS0(dname, dlen, sname, slen, conv_f, 1);
382 /* Zero length filename isn't valid... */ 383 /* Zero length filename isn't valid... */
383 if (ret == 0) 384 if (ret == 0)
384 ret = -EINVAL; 385 ret = -EINVAL;
@@ -388,15 +389,6 @@ int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
388int udf_put_filename(struct super_block *sb, const uint8_t *sname, int slen, 389int udf_put_filename(struct super_block *sb, const uint8_t *sname, int slen,
389 uint8_t *dname, int dlen) 390 uint8_t *dname, int dlen)
390{ 391{
391 int (*conv_f)(const unsigned char *, int, wchar_t *); 392 return udf_name_to_CS0(sb, dname, dlen, sname, slen);
392
393 if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
394 conv_f = udf_char2uni_utf8;
395 } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
396 conv_f = UDF_SB(sb)->s_nls_map->char2uni;
397 } else
398 BUG();
399
400 return udf_name_to_CS0(dname, dlen, sname, slen, conv_f);
401} 393}
402 394
generated by cgit v1.2.2 (git 2.25.0) at 2025-09-04 16:42:36 -0400