Skip to content

Commit 484a10f

Browse files
agabbasovjankara
agabbasov
authored and
jankara
committed
udf: Merge linux specific translation into CS0 conversion function
Current implementation of udf_translate_to_linux function does not support multi-bytes characters at all: it counts bytes while calculating extension length, when inserting CRC inside the name it doesn't take into account inter-character boundaries and can break into the middle of the character. The most efficient way to properly support multi-bytes characters is merging of translation operations directly into conversion function. This can help to avoid extra passes along the string or parsing the multi-bytes character back into unicode to find out it's length. Signed-off-by: Andrew Gabbasov <andrew_gabbasov@mentor.com> Signed-off-by: Jan Kara <jack@suse.cz>
1 parent 9293fcf commit 484a10f

File tree

1 file changed

+152
-128
lines changed

1 file changed

+152
-128
lines changed

‎fs/udf/unicode.c

Lines changed: 152 additions & 128 deletions
Original file line numberDiff line numberDiff line change
@@ -28,9 +28,6 @@
2828

2929
#include "udf_sb.h"
3030

31-
static int udf_translate_to_linux(uint8_t *, int, const uint8_t *, int,
32-
const uint8_t *, int);
33-
3431
static int udf_uni2char_utf8(wchar_t uni,
3532
unsigned char *out,
3633
int boundlen)
@@ -114,13 +111,83 @@ static int udf_char2uni_utf8(const unsigned char *in,
114111
return u_len;
115112
}
116113

114+
#define ILLEGAL_CHAR_MARK '_'
115+
#define EXT_MARK '.'
116+
#define CRC_MARK '#'
117+
#define EXT_SIZE 5
118+
/* Number of chars we need to store generated CRC to make filename unique */
119+
#define CRC_LEN 5
120+
121+
static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
122+
int *str_o_idx,
123+
const uint8_t *str_i, int str_i_max_len,
124+
int *str_i_idx,
125+
int u_ch, int *needsCRC,
126+
int (*conv_f)(wchar_t, unsigned char *, int),
127+
int translate)
128+
{
129+
uint32_t c;
130+
int illChar = 0;
131+
int len, gotch = 0;
132+
133+
for (; (!gotch) && (*str_i_idx < str_i_max_len); *str_i_idx += u_ch) {
134+
if (*str_o_idx >= str_o_max_len) {
135+
*needsCRC = 1;
136+
return gotch;
137+
}
138+
139+
/* Expand OSTA compressed Unicode to Unicode */
140+
c = str_i[*str_i_idx];
141+
if (u_ch > 1)
142+
c = (c << 8) | str_i[*str_i_idx + 1];
143+
144+
if (translate && (c == '/' || c == 0))
145+
illChar = 1;
146+
else if (illChar)
147+
break;
148+
else
149+
gotch = 1;
150+
}
151+
if (illChar) {
152+
*needsCRC = 1;
153+
c = ILLEGAL_CHAR_MARK;
154+
gotch = 1;
155+
}
156+
if (gotch) {
157+
len = conv_f(c, &str_o[*str_o_idx], str_o_max_len - *str_o_idx);
158+
/* Valid character? */
159+
if (len >= 0)
160+
*str_o_idx += len;
161+
else if (len == -ENAMETOOLONG) {
162+
*needsCRC = 1;
163+
gotch = 0;
164+
} else {
165+
str_o[(*str_o_idx)++] = '?';
166+
*needsCRC = 1;
167+
}
168+
}
169+
return gotch;
170+
}
171+
117172
static int udf_name_from_CS0(uint8_t *str_o, int str_max_len,
118173
const uint8_t *ocu, int ocu_len,
119-
int (*conv_f)(wchar_t, unsigned char *, int))
174+
int (*conv_f)(wchar_t, unsigned char *, int),
175+
int translate)
120176
{
177+
uint32_t c;
121178
uint8_t cmp_id;
122-
int i, len;
123-
int str_o_len = 0;
179+
int idx, len;
180+
int u_ch;
181+
int needsCRC = 0;
182+
int ext_i_len, ext_max_len;
183+
int str_o_len = 0; /* Length of resulting output */
184+
int ext_o_len = 0; /* Extension output length */
185+
int ext_crc_len = 0; /* Extension output length if used with CRC */
186+
int i_ext = -1; /* Extension position in input buffer */
187+
int o_crc = 0; /* Rightmost possible output pos for CRC+ext */
188+
unsigned short valueCRC;
189+
uint8_t ext[EXT_SIZE * NLS_MAX_CHARSET_SIZE + 1];
190+
uint8_t crc[CRC_LEN];
124191

125192
if (str_max_len <= 0)
126193
return 0;
@@ -133,24 +200,88 @@ static int udf_name_from_CS0(uint8_t *str_o, int str_max_len,
133200
cmp_id = ocu[0];
134201
if (cmp_id != 8 && cmp_id != 16) {
135202
memset(str_o, 0, str_max_len);
136-
pr_err("unknown compression code (%d) stri=%s\n", cmp_id, ocu);
203+
pr_err("unknown compression code (%d)\n", cmp_id);
137204
return -EINVAL;
138205
}
206+
u_ch = cmp_id >> 3;
139207

140-
for (i = 1; (i < ocu_len) && (str_o_len < str_max_len);) {
141-
/* Expand OSTA compressed Unicode to Unicode */
142-
uint32_t c = ocu[i++];
143-
if (cmp_id == 16)
144-
c = (c << 8) | ocu[i++];
208+
ocu++;
209+
ocu_len--;
145210

146-
len = conv_f(c, &str_o[str_o_len], str_max_len - str_o_len);
147-
/* Valid character? */
148-
if (len >= 0)
149-
str_o_len += len;
150-
else if (len == -ENAMETOOLONG)
211+
if (ocu_len % u_ch) {
212+
pr_err("incorrect filename length (%d)\n", ocu_len + 1);
213+
return -EINVAL;
214+
}
215+
216+
if (translate) {
217+
/* Look for extension */
218+
for (idx = ocu_len - u_ch, ext_i_len = 0;
219+
(idx >= 0) && (ext_i_len < EXT_SIZE);
220+
idx -= u_ch, ext_i_len++) {
221+
c = ocu[idx];
222+
if (u_ch > 1)
223+
c = (c << 8) | ocu[idx + 1];
224+
225+
if (c == EXT_MARK) {
226+
if (ext_i_len)
227+
i_ext = idx;
228+
break;
229+
}
230+
}
231+
if (i_ext >= 0) {
232+
/* Convert extension */
233+
ext_max_len = min_t(int, sizeof(ext), str_max_len);
234+
ext[ext_o_len++] = EXT_MARK;
235+
idx = i_ext + u_ch;
236+
while (udf_name_conv_char(ext, ext_max_len, &ext_o_len,
237+
ocu, ocu_len, &idx,
238+
u_ch, &needsCRC,
239+
conv_f, translate)) {
240+
if ((ext_o_len + CRC_LEN) < str_max_len)
241+
ext_crc_len = ext_o_len;
242+
}
243+
}
244+
}
245+
246+
idx = 0;
247+
while (1) {
248+
if (translate && (idx == i_ext)) {
249+
if (str_o_len > (str_max_len - ext_o_len))
250+
needsCRC = 1;
151251
break;
152-
else
153-
str_o[str_o_len++] = '?';
252+
}
253+
254+
if (!udf_name_conv_char(str_o, str_max_len, &str_o_len,
255+
ocu, ocu_len, &idx,
256+
u_ch, &needsCRC, conv_f, translate))
257+
break;
258+
259+
if (translate &&
260+
(str_o_len <= (str_max_len - ext_o_len - CRC_LEN)))
261+
o_crc = str_o_len;
262+
}
263+
264+
if (translate) {
265+
if (str_o_len <= 2 && str_o[0] == '.' &&
266+
(str_o_len == 1 || str_o[1] == '.'))
267+
needsCRC = 1;
268+
if (needsCRC) {
269+
str_o_len = o_crc;
270+
valueCRC = crc_itu_t(0, ocu, ocu_len);
271+
crc[0] = CRC_MARK;
272+
crc[1] = hex_asc_upper_hi(valueCRC >> 8);
273+
crc[2] = hex_asc_upper_lo(valueCRC >> 8);
274+
crc[3] = hex_asc_upper_hi(valueCRC);
275+
crc[4] = hex_asc_upper_lo(valueCRC);
276+
len = min_t(int, CRC_LEN, str_max_len - str_o_len);
277+
memcpy(&str_o[str_o_len], crc, len);
278+
str_o_len += len;
279+
ext_o_len = ext_crc_len;
280+
}
281+
if (ext_o_len > 0) {
282+
memcpy(&str_o[str_o_len], ext, ext_o_len);
283+
str_o_len += ext_o_len;
284+
}
154285
}
155286

156287
return str_o_len;
@@ -207,13 +338,12 @@ static int udf_name_to_CS0(uint8_t *ocu, int ocu_max_len,
207338
int udf_CS0toUTF8(uint8_t *utf_o, int o_len, const uint8_t *ocu_i, int i_len)
208339
{
209340
return udf_name_from_CS0(utf_o, o_len, ocu_i, i_len,
210-
udf_uni2char_utf8);
341+
udf_uni2char_utf8, 0);
211342
}
212343

213344
int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
214345
uint8_t *dname, int dlen)
215346
{
216-
uint8_t *filename;
217347
int (*conv_f)(wchar_t, unsigned char *, int);
218348
int ret;
219349

@@ -223,30 +353,17 @@ int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
223353
if (dlen <= 0)
224354
return 0;
225355

226-
filename = kmalloc(dlen, GFP_NOFS);
227-
if (!filename)
228-
return -ENOMEM;
229-
230356
if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
231357
conv_f = udf_uni2char_utf8;
232358
} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
233359
conv_f = UDF_SB(sb)->s_nls_map->uni2char;
234360
} else
235361
BUG();
236362

237-
ret = udf_name_from_CS0(filename, dlen, sname, slen, conv_f);
238-
if (ret < 0) {
239-
udf_debug("Failed in udf_get_filename: sname = %s\n", sname);
240-
goto out2;
241-
}
242-
243-
ret = udf_translate_to_linux(dname, dlen, filename, dlen,
244-
sname + 1, slen - 1);
363+
ret = udf_name_from_CS0(dname, dlen, sname, slen, conv_f, 1);
245364
/* Zero length filename isn't valid... */
246365
if (ret == 0)
247366
ret = -EINVAL;
248-
out2:
249-
kfree(filename);
250367
return ret;
251368
}
252369

@@ -265,96 +382,3 @@ int udf_put_filename(struct super_block *sb, const uint8_t *sname, int slen,
265382
return udf_name_to_CS0(dname, dlen, sname, slen, conv_f);
266383
}
267384

268-
#define ILLEGAL_CHAR_MARK '_'
269-
#define EXT_MARK '.'
270-
#define CRC_MARK '#'
271-
#define EXT_SIZE 5
272-
/* Number of chars we need to store generated CRC to make filename unique */
273-
#define CRC_LEN 5
274-
275-
static int udf_translate_to_linux(uint8_t *newName, int newLen,
276-
const uint8_t *udfName, int udfLen,
277-
const uint8_t *fidName, int fidNameLen)
278-
{
279-
int index, newIndex = 0, needsCRC = 0;
280-
int extIndex = 0, newExtIndex = 0, hasExt = 0;
281-
unsigned short valueCRC;
282-
uint8_t curr;
283-
284-
if (udfName[0] == '.' &&
285-
(udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
286-
needsCRC = 1;
287-
newIndex = udfLen;
288-
memcpy(newName, udfName, udfLen);
289-
} else {
290-
for (index = 0; index < udfLen; index++) {
291-
curr = udfName[index];
292-
if (curr == '/' || curr == 0) {
293-
needsCRC = 1;
294-
curr = ILLEGAL_CHAR_MARK;
295-
while (index + 1 < udfLen &&
296-
(udfName[index + 1] == '/' ||
297-
udfName[index + 1] == 0))
298-
index++;
299-
}
300-
if (curr == EXT_MARK &&
301-
(udfLen - index - 1) <= EXT_SIZE) {
302-
if (udfLen == index + 1)
303-
hasExt = 0;
304-
else {
305-
hasExt = 1;
306-
extIndex = index;
307-
newExtIndex = newIndex;
308-
}
309-
}
310-
if (newIndex < newLen)
311-
newName[newIndex++] = curr;
312-
else
313-
needsCRC = 1;
314-
}
315-
}
316-
if (needsCRC) {
317-
uint8_t ext[EXT_SIZE];
318-
int localExtIndex = 0;
319-
320-
if (hasExt) {
321-
int maxFilenameLen;
322-
for (index = 0;
323-
index < EXT_SIZE && extIndex + index + 1 < udfLen;
324-
index++) {
325-
curr = udfName[extIndex + index + 1];
326-
327-
if (curr == '/' || curr == 0) {
328-
needsCRC = 1;
329-
curr = ILLEGAL_CHAR_MARK;
330-
while (extIndex + index + 2 < udfLen &&
331-
(index + 1 < EXT_SIZE &&
332-
(udfName[extIndex + index + 2] == '/' ||
333-
udfName[extIndex + index + 2] == 0)))
334-
index++;
335-
}
336-
ext[localExtIndex++] = curr;
337-
}
338-
maxFilenameLen = newLen - CRC_LEN - localExtIndex;
339-
if (newIndex > maxFilenameLen)
340-
newIndex = maxFilenameLen;
341-
else
342-
newIndex = newExtIndex;
343-
} else if (newIndex > newLen - CRC_LEN)
344-
newIndex = newLen - CRC_LEN;
345-
newName[newIndex++] = CRC_MARK;
346-
valueCRC = crc_itu_t(0, fidName, fidNameLen);
347-
newName[newIndex++] = hex_asc_upper_hi(valueCRC >> 8);
348-
newName[newIndex++] = hex_asc_upper_lo(valueCRC >> 8);
349-
newName[newIndex++] = hex_asc_upper_hi(valueCRC);
350-
newName[newIndex++] = hex_asc_upper_lo(valueCRC);
351-
352-
if (hasExt) {
353-
newName[newIndex++] = EXT_MARK;
354-
for (index = 0; index < localExtIndex; index++)
355-
newName[newIndex++] = ext[index];
356-
}
357-
}
358-
359-
return newIndex;
360-
}

0 commit comments

Comments
 (0)