/* $NetBSD: msdosfs_conv.c,v 1.17 2016/06/30 09:34:01 nonaka Exp $ */
/*-
* Copyright (C) 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*
*/
#if HAVE_NBTOOL_CONFIG_H
#include "nbtool_config.h"
#endif
#ifndef _KERNEL
#include <assert.h>
#define KASSERT(x) assert(x)
#endif
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: msdosfs_conv.c,v 1.17 2016/06/30 09:34:01 nonaka Exp $");
/*
* System include files.
*/
#include <sys/param.h>
#include <sys/time.h>
#include <sys/endian.h>
#ifdef _KERNEL
#include <sys/dirent.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/vnode.h>
#else
#include <stdio.h>
#include <string.h>
#include <dirent.h>
#include <sys/queue.h>
#endif
#include <dev/clock_subr.h>
/*
* MSDOSFS include files.
*/
#include <fs/msdosfs/direntry.h>
#include <fs/msdosfs/denode.h>
static int invalidname(const u_int16_t *, int);
static int ucs2utf8(const u_int16_t *, u_int8_t *, int);
static int utf8ucs2(const u_int8_t *, int, u_int16_t *);
static int ucs2utf8str(const u_int16_t *, int, u_int8_t *, int);
static int utf8ucs2str(const u_int8_t *, int, u_int16_t *, int);
static int ucs2char8str(const u_int16_t *, int, u_int8_t *, int);
static int char8ucs2str(const u_int8_t *, int, u_int16_t *, int);
static void ucs2pad(u_int16_t *, int, int);
static u_int16_t ucs2fold(u_int16_t);
static int ucs2match(u_int16_t *, u_int16_t *, int n);
static int char8match(u_int16_t *, u_int16_t *, int n);
/*
* The number of seconds between Jan 1, 1970 and Jan 1, 1980. In that
* interval there were 8 regular years and 2 leap years.
*/
#define DOSBIASYEAR 1980
#define SECONDSTO1980 (((8 * 365) + (2 * 366)) * (24 * 60 * 60))
/*
* msdos fs can not store dates beyound the year 2234
*/
#define DOSMAXYEAR ((DD_YEAR_MASK >> DD_YEAR_SHIFT) + DOSBIASYEAR)
/*
* Convert the unix version of time to dos's idea of time to be used in
* file timestamps. The passed in unix time is assumed to be in GMT.
*/
void
unix2dostime(const struct timespec *tsp, int gmtoff, u_int16_t *ddp, u_int16_t *dtp, u_int8_t *dhp)
{
u_long t;
struct clock_ymdhms ymd;
t = tsp->tv_sec + gmtoff; /* time zone correction */
/*
* DOS timestamps can not represent dates before 1980.
*/
if (t < SECONDSTO1980)
goto invalid_dos_date;
/*
* DOS granularity is 2 seconds
*/
t &= ~1;
/*
* Convert to year/month/day/.. format
*/
clock_secs_to_ymdhms(t, &ymd);
if (ymd.dt_year > DOSMAXYEAR)
goto invalid_dos_date;
/*
* Now transform to DOS format
*/
*ddp = (ymd.dt_day << DD_DAY_SHIFT)
+ (ymd.dt_mon << DD_MONTH_SHIFT)
+ ((ymd.dt_year - DOSBIASYEAR) << DD_YEAR_SHIFT);
if (dhp)
*dhp = (tsp->tv_sec & 1) * 100 + tsp->tv_nsec / 10000000;
if (dtp)
*dtp = (((t / 2) % 30) << DT_2SECONDS_SHIFT)
+ (((t / 60) % 60) << DT_MINUTES_SHIFT)
+ (((t / 3600) % 24) << DT_HOURS_SHIFT);
return;
invalid_dos_date:
*ddp = 0;
if (dtp)
*dtp = 0;
if (dhp)
*dhp = 0;
}
/*
* Convert from dos' idea of time to unix'. This will probably only be
* called from the stat(), and fstat() system calls and so probably need
* not be too efficient.
*/
void
dos2unixtime(u_int dd, u_int dt, u_int dh, int gmtoff, struct timespec *tsp)
{
time_t seconds;
struct clock_ymdhms ymd;
if (dd == 0) {
/*
* Uninitialized field, return the epoch.
*/
tsp->tv_sec = 0;
tsp->tv_nsec = 0;
return;
}
memset(&ymd, 0, sizeof(ymd));
ymd.dt_year = ((dd & DD_YEAR_MASK) >> DD_YEAR_SHIFT) + 1980 ;
ymd.dt_mon = ((dd & DD_MONTH_MASK) >> DD_MONTH_SHIFT);
ymd.dt_day = ((dd & DD_DAY_MASK) >> DD_DAY_SHIFT);
ymd.dt_hour = (dt & DT_HOURS_MASK) >> DT_HOURS_SHIFT;
ymd.dt_min = (dt & DT_MINUTES_MASK) >> DT_MINUTES_SHIFT;
ymd.dt_sec = ((dt & DT_2SECONDS_MASK) >> DT_2SECONDS_SHIFT) * 2;
seconds = clock_ymdhms_to_secs(&ymd);
tsp->tv_sec = seconds;
tsp->tv_sec -= gmtoff; /* time zone correction */
tsp->tv_nsec = (dh % 100) * 10000000;
}
static const u_char
unix2dos[256] = {
0, 0, 0, 0, 0, 0, 0, 0, /* 00-07 */
0, 0, 0, 0, 0, 0, 0, 0, /* 08-0f */
0, 0, 0, 0, 0, 0, 0, 0, /* 10-17 */
0, 0, 0, 0, 0, 0, 0, 0, /* 18-1f */
0, '!', 0, '#', '$', '%', '&', '\'', /* 20-27 */
'(', ')', 0, '+', 0, '-', 0, 0, /* 28-2f */
'0', '1', '2', '3', '4', '5', '6', '7', /* 30-37 */
'8', '9', 0, 0, 0, 0, 0, 0, /* 38-3f */
'@', 'A', 'B', 'C', 'D', 'E', 'F', 'G', /* 40-47 */
'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', /* 48-4f */
'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', /* 50-57 */
'X', 'Y', 'Z', 0, 0, 0, '^', '_', /* 58-5f */
'`', 'A', 'B', 'C', 'D', 'E', 'F', 'G', /* 60-67 */
'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', /* 68-6f */
'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', /* 70-77 */
'X', 'Y', 'Z', '{', 0, '}', '~', 0, /* 78-7f */
0, 0, 0, 0, 0, 0, 0, 0, /* 80-87 */
0, 0, 0, 0, 0, 0, 0, 0, /* 88-8f */
0, 0, 0, 0, 0, 0, 0, 0, /* 90-97 */
0, 0, 0, 0, 0, 0, 0, 0, /* 98-9f */
0, 0xad, 0xbd, 0x9c, 0xcf, 0xbe, 0xdd, 0xf5, /* a0-a7 */
0xf9, 0xb8, 0xa6, 0xae, 0xaa, 0xf0, 0xa9, 0xee, /* a8-af */
0xf8, 0xf1, 0xfd, 0xfc, 0xef, 0xe6, 0xf4, 0xfa, /* b0-b7 */
0xf7, 0xfb, 0xa7, 0xaf, 0xac, 0xab, 0xf3, 0xa8, /* b8-bf */
0xb7, 0xb5, 0xb6, 0xc7, 0x8e, 0x8f, 0x92, 0x80, /* c0-c7 */
0xd4, 0x90, 0xd2, 0xd3, 0xde, 0xd6, 0xd7, 0xd8, /* c8-cf */
0xd1, 0xa5, 0xe3, 0xe0, 0xe2, 0xe5, 0x99, 0x9e, /* d0-d7 */
0x9d, 0xeb, 0xe9, 0xea, 0x9a, 0xed, 0xe8, 0xe1, /* d8-df */
0xb7, 0xb5, 0xb6, 0xc7, 0x8e, 0x8f, 0x92, 0x80, /* e0-e7 */
0xd4, 0x90, 0xd2, 0xd3, 0xde, 0xd6, 0xd7, 0xd8, /* e8-ef */
0xd1, 0xa5, 0xe3, 0xe0, 0xe2, 0xe5, 0x99, 0xf6, /* f0-f7 */
0x9d, 0xeb, 0xe9, 0xea, 0x9a, 0xed, 0xe8, 0x98, /* f8-ff */
};
static const u_char
dos2unix[256] = {
'?', '?', '?', '?', '?', '?', '?', '?', /* 00-07 */
'?', '?', '?', '?', '?', '?', '?', '?', /* 08-0f */
'?', '?', '?', '?', '?', '?', '?', '?', /* 10-17 */
'?', '?', '?', '?', '?', '?', '?', '?', /* 18-1f */
' ', '!', '"', '#', '$', '%', '&', '\'', /* 20-27 */
'(', ')', '*', '+', ',', '-', '.', '/', /* 28-2f */
'0', '1', '2', '3', '4', '5', '6', '7', /* 30-37 */
'8', '9', ':', ';', '<', '=', '>', '?', /* 38-3f */
'@', 'A', 'B', 'C', 'D', 'E', 'F', 'G', /* 40-47 */
'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', /* 48-4f */
'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', /* 50-57 */
'X', 'Y', 'Z', '[', '\\', ']', '^', '_', /* 58-5f */
'`', 'a', 'b', 'c', 'd', 'e', 'f', 'g', /* 60-67 */
'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', /* 68-6f */
'p', 'q', 'r', 's', 't', 'u', 'v', 'w', /* 70-77 */
'x', 'y', 'z', '{', '|', '}', '~', 0x7f, /* 78-7f */
0xc7, 0xfc, 0xe9, 0xe2, 0xe4, 0xe0, 0xe5, 0xe7, /* 80-87 */
0xea, 0xeb, 0xe8, 0xef, 0xee, 0xec, 0xc4, 0xc5, /* 88-8f */
0xc9, 0xe6, 0xc6, 0xf4, 0xf6, 0xf2, 0xfb, 0xf9, /* 90-97 */
0xff, 0xd6, 0xdc, 0xf8, 0xa3, 0xd8, 0xd7, '?', /* 98-9f */
0xe1, 0xed, 0xf3, 0xfa, 0xf1, 0xd1, 0xaa, 0xba, /* a0-a7 */
0xbf, 0xae, 0xac, 0xbd, 0xbc, 0xa1, 0xab, 0xbb, /* a8-af */
'?', '?', '?', '?', '?', 0xc1, 0xc2, 0xc0, /* b0-b7 */
0xa9, '?', '?', '?', '?', 0xa2, 0xa5, '?', /* b8-bf */
'?', '?', '?', '?', '?', '?', 0xe3, 0xc3, /* c0-c7 */
'?', '?', '?', '?', '?', '?', '?', 0xa4, /* c8-cf */
0xf0, 0xd0, 0xca, 0xcb, 0xc8, '?', 0xcd, 0xce, /* d0-d7 */
0xcf, '?', '?', '?', '?', 0xa6, 0xcc, '?', /* d8-df */
0xd3, 0xdf, 0xd4, 0xd2, 0xf5, 0xd5, 0xb5, 0xfe, /* e0-e7 */
0xde, 0xda, 0xdb, 0xd9, 0xfd, 0xdd, 0xaf, 0x3f, /* e8-ef */
0xad, 0xb1, '?', 0xbe, 0xb6, 0xa7, 0xf7, 0xb8, /* f0-f7 */
0xb0, 0xa8, 0xb7, 0xb9, 0xb3, 0xb2, '?', '?', /* f8-ff */
};
static const u_char
u2l[256] = {
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, /* 00-07 */
0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, /* 08-0f */
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, /* 10-17 */
0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, /* 18-1f */
' ', '!', '"', '#', '$', '%', '&', '\'', /* 20-27 */
'(', ')', '*', '+', ',', '-', '.', '/', /* 28-2f */
'0', '1', '2', '3', '4', '5', '6', '7', /* 30-37 */
'8', '9', ':', ';', '<', '=', '>', '?', /* 38-3f */
'@', 'a', 'b', 'c', 'd', 'e', 'f', 'g', /* 40-47 */
'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', /* 48-4f */
'p', 'q', 'r', 's', 't', 'u', 'v', 'w', /* 50-57 */
'x', 'y', 'z', '[', '\\', ']', '^', '_', /* 58-5f */
'`', 'a', 'b', 'c', 'd', 'e', 'f', 'g', /* 60-67 */
'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', /* 68-6f */
'p', 'q', 'r', 's', 't', 'u', 'v', 'w', /* 70-77 */
'x', 'y', 'z', '{', '|', '}', '~', 0x7f, /* 78-7f */
0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, /* 80-87 */
0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f, /* 88-8f */
0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, /* 90-97 */
0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, /* 98-9f */
0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, /* a0-a7 */
0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, /* a8-af */
0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, /* b0-b7 */
0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, /* b8-bf */
0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, /* c0-c7 */
0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, /* c8-cf */
0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xd7, /* d0-d7 */
0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xdf, /* d8-df */
0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, /* e0-e7 */
0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, /* e8-ef */
0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, /* f0-f7 */
0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff, /* f8-ff */
};
/*
* DOS filenames are made of 2 parts, the name part and the extension part.
* The name part is 8 characters long and the extension part is 3
* characters long. They may contain trailing blanks if the name or
* extension are not long enough to fill their respective fields.
*/
/*
* Convert a DOS filename to a unix filename. And, return the number of
* characters in the resulting unix filename excluding the terminating
* null.
*/
int
dos2unixfn(u_char dn[11], u_char *un, int lower)
{
int i, j;
int thislong = 1;
u_char c;
/*
* If first char of the filename is SLOT_E5 (0x05), then the real
* first char of the filename should be 0xe5. But, they couldn't
* just have a 0xe5 mean 0xe5 because that is used to mean a freed
* directory slot. Another dos quirk.
*/
if (*dn == SLOT_E5)
c = dos2unix[0xe5];
else
c = dos2unix[*dn];
*un++ = lower ? u2l[c] : c;
/*
* Copy the rest into the unix filename string, ignoring
* trailing blanks.
*/
for (j=7; (j >= 0) && (dn[j] == ' '); j--)
;
for (i = 1; i <= j; i++) {
c = dos2unix[dn[i]];
*un++ = lower ? u2l[c] : c;
thislong++;
}
dn += 8;
/*
* Now, if there is an extension then put in a period and copy in
* the extension.
*/
if (*dn != ' ') {
*un++ = '.';
thislong++;
for (i = 0; i < 3 && *dn != ' '; i++) {
c = dos2unix[*dn++];
*un++ = lower ? u2l[c] : c;
thislong++;
}
}
*un++ = 0;
return (thislong);
}
/*
* Convert a unix filename to a DOS filename according to Win95 rules.
* If applicable and gen is not 0, it is inserted into the converted
* filename as a generation number.
* Returns
* 0 if name couldn't be converted
* 1 if the converted name is the same as the original
* (no long filename entry necessary for Win95)
* 2 if conversion was successful
* 3 if conversion was successful and generation number was inserted
*/
int
unix2dosfn(const u_char *un, u_char dn[12], int unlen, u_int gen)
{
int i, j, l;
int conv = 1;
const u_char *cp, *dp, *dp1;
u_char gentext[6], *wcp;
int shortlen;
/*
* Fill the dos filename string with blanks. These are DOS's pad
* characters.
*/
for (i = 0; i < 11; i++)
dn[i] = ' ';
dn[11] = 0;
/*
* The filenames "." and ".." are handled specially, since they
* don't follow dos filename rules.
*/
if (un[0] == '.' && unlen == 1) {
dn[0] = '.';
return gen <= 1;
}
if (un[0] == '.' && un[1] == '.' && unlen == 2) {
dn[0] = '.';
dn[1] = '.';
return gen <= 1;
}
/*
* Filenames with only blanks and dots are not allowed!
*/
for (cp = un, i = unlen; --i >= 0; cp++)
if (*cp != ' ' && *cp != '.')
break;
if (i < 0)
return 0;
/*
* Now find the extension
* Note: dot as first char doesn't start extension
* and trailing dots and blanks are ignored
*/
dp = dp1 = 0;
for (cp = un + 1, i = unlen - 1; --i >= 0;) {
switch (*cp++) {
case '.':
if (!dp1)
dp1 = cp;
break;
case ' ':
break;
default:
if (dp1)
dp = dp1;
dp1 = 0;
break;
}
}
/*
* Now convert it
*/
if (dp) {
if (dp1)
l = dp1 - dp;
else
l = unlen - (dp - un);
for (i = 0, j = 8; i < l && j < 11; i++, j++) {
if (dp[i] != (dn[j] = unix2dos[dp[i]])
&& conv != 3)
conv = 2;
if (!dn[j]) {
conv = 3;
dn[j--] = ' ';
}
}
if (i < l)
conv = 3;
dp--;
} else {
for (dp = cp; *--dp == ' ' || *dp == '.';);
dp++;
}
shortlen = (dp - un) <= 8;
/*
* Now convert the rest of the name
*/
for (i = j = 0; un < dp && j < 8; i++, j++, un++) {
if ((*un == ' ') && shortlen)
dn[j] = ' ';
else
dn[j] = unix2dos[*un];
if ((*un != dn[j])
&& conv != 3)
conv = 2;
if (!dn[j]) {
conv = 3;
dn[j--] = ' ';
}
}
if (un < dp)
conv = 3;
/*
* If we didn't have any chars in filename,
* generate a default
*/
if (!j)
dn[0] = '_';
/*
* The first character cannot be E5,
* because that means a deleted entry
*/
if (dn[0] == 0xe5)
dn[0] = SLOT_E5;
/*
* If there wasn't any char dropped,
* there is no place for generation numbers
*/
if (conv != 3) {
if (gen > 1)
return 0;
return conv;
}
/*
* Now insert the generation number into the filename part
*/
for (wcp = gentext + sizeof(gentext); wcp > gentext && gen; gen /= 10)
*--wcp = gen % 10 + '0';
if (gen)
return 0;
for (i = 8; dn[--i] == ' ';);
i++;
if (gentext + sizeof(gentext) - wcp + 1 > 8 - i)
i = 8 - (gentext + sizeof(gentext) - wcp + 1);
dn[i++] = '~';
while (wcp < gentext + sizeof(gentext))
dn[i++] = *wcp++;
return 3;
}
/*
* Create a Win95 long name directory entry
* Note: assumes that the filename is valid,
* i.e. doesn't consist solely of blanks and dots
*/
int
unix2winfn(const u_char *un, int unlen, struct winentry *wep, int cnt, int chksum, int utf8)
{
u_int16_t wn[WIN_MAXLEN], *p;
int i, len;
const u_char *cp;
/*
* Drop trailing blanks and dots
*/
for (cp = un + unlen; unlen > 0; unlen--)
if (*--cp != ' ' && *cp != '.')
break;
/*
* Offset of this entry
*/
i = (cnt - 1) * WIN_CHARS;
/*
* Translate UNIX name to ucs-2
*/
len = utf8 ? utf8ucs2str(un, unlen, wn, WIN_MAXLEN) : char8ucs2str(un, unlen, wn, WIN_MAXLEN);
ucs2pad(wn, len, WIN_MAXLEN);
/*
* Initialize winentry to some useful default
*/
memset(wep, 0xff, sizeof(*wep));
wep->weCnt = cnt;
wep->weAttributes = ATTR_WIN95;
wep->weReserved1 = 0;
wep->weChksum = chksum;
wep->weReserved2 = 0;
/*
* Store name segment into directory entry
*/
p = &wn[i];
memcpy(wep->wePart1, p, sizeof(wep->wePart1));
p += sizeof(wep->wePart1) / sizeof(*p);
memcpy(wep->wePart2, p, sizeof(wep->wePart2));
p += sizeof(wep->wePart2) / sizeof(*p);
memcpy(wep->wePart3, p, sizeof(wep->wePart3));
if (len > i + WIN_CHARS)
return 1;
wep->weCnt |= WIN_LAST;
return 0;
}
/*
* Compare our filename to the one in the Win95 entry
* Returns the checksum or -1 if no match
*/
int
winChkName(const u_char *un, int unlen, struct winentry *wep, int chksum, int utf8)
{
u_int16_t wn[WIN_MAXLEN], *p;
u_int16_t buf[WIN_CHARS];
int i, len;
/*
* First compare checksums
*/
if (wep->weCnt & WIN_LAST)
chksum = wep->weChksum;
else if (chksum != wep->weChksum)
chksum = -1;
if (chksum == -1)
return -1;
/*
* Offset of this entry
*/
i = ((wep->weCnt & WIN_CNT) - 1) * WIN_CHARS;
/*
* Translate UNIX name to ucs-2
*/
len = utf8 ? utf8ucs2str(un, unlen, wn, WIN_MAXLEN) : char8ucs2str(un, unlen, wn, WIN_MAXLEN);
ucs2pad(wn, len, WIN_MAXLEN);
if (i >= len + 1)
return -1;
if ((wep->weCnt & WIN_LAST) && (len - i > WIN_CHARS))
return -1;
/*
* Fetch name segment from directory entry
*/
p = &buf[0];
memcpy(p, wep->wePart1, sizeof(wep->wePart1));
p += sizeof(wep->wePart1) / sizeof(*p);
memcpy(p, wep->wePart2, sizeof(wep->wePart2));
p += sizeof(wep->wePart2) / sizeof(*p);
memcpy(p, wep->wePart3, sizeof(wep->wePart3));
/*
* And compare name segment
*/
if (! (utf8 ? ucs2match(&wn[i], buf, WIN_CHARS) : char8match(&wn[i], buf, WIN_CHARS)))
return -1;
return chksum;
}
/*
* Convert Win95 filename to dirbuf.
* Returns the checksum or -1 if impossible
*/
int
win2unixfn(struct winentry *wep, struct dirent *dp, int chksum,
uint16_t *namlen, int utf8)
{
u_int16_t wn[WIN_CHARS], *p;
u_int8_t buf[WIN_CHARS*3];
int len;
if ((wep->weCnt & WIN_CNT) > howmany(WIN_MAXLEN, WIN_CHARS)
|| !(wep->weCnt & WIN_CNT))
return -1;
/*
* First compare checksums
*/
if (wep->weCnt & WIN_LAST) {
chksum = wep->weChksum;
*namlen = 0;
} else if (chksum != wep->weChksum)
chksum = -1;
if (chksum == -1)
return -1;
/*
* Fetch name segment from directory entry
*/
p = &wn[0];
memcpy(p, wep->wePart1, sizeof(wep->wePart1));
p += sizeof(wep->wePart1) / sizeof(*p);
memcpy(p, wep->wePart2, sizeof(wep->wePart2));
p += sizeof(wep->wePart2) / sizeof(*p);
memcpy(p, wep->wePart3, sizeof(wep->wePart3));
/*
* Don't allow slashes in UNIX names. Discard that entry.
*/
if (invalidname(wn, WIN_CHARS))
return -1;
/*
* Translate ucs-2 to UNIX name
*/
len = utf8 ? ucs2utf8str(wn, WIN_CHARS, buf, sizeof(buf))
: ucs2char8str(wn, WIN_CHARS, buf, sizeof(buf));
KASSERT(len >= 0);
KASSERT((size_t)len <= MIN(sizeof(buf), sizeof(dp->d_name)-1));
/*
* Prepend name segment to directory entry
*
* This ignores the slot number from the windows entry but
* assumes that segments are read in reverse order.
*
* The UCS-2 name (up to 255 chars) can overflow the UNIX
* directory entry (up to 511 bytes). Trailing characters
* are silently discarded. This could also end in multiple
* files using the same (truncated) name.
*/
*namlen += len;
if (*namlen > sizeof(dp->d_name) - 1)
*namlen = sizeof(dp->d_name) - 1;
KASSERT(*namlen >= len);
memmove(&dp->d_name[len], &dp->d_name[0], *namlen - len);
memcpy(dp->d_name, buf, len);
return chksum;
}
/*
* Compute the checksum of a DOS filename for Win95 use
*/
u_int8_t
winChksum(u_int8_t *name)
{
int i;
u_int8_t s;
for (s = 0, i = 11; --i >= 0; s += *name++)
s = (s << 7) | (s >> 1);
return s;
}
/*
* Determine the number of slots necessary for Win95 names
*/
int
winSlotCnt(const u_char *un, int unlen, int utf8)
{
const u_char *cp;
int len;
/*
* Drop trailing blanks and dots
*/
for (cp = un + unlen; unlen > 0; unlen--)
if (*--cp != ' ' && *cp != '.')
break;
len = utf8 ? utf8ucs2str(un, unlen, NULL, WIN_MAXLEN) : unlen;
return howmany(len, WIN_CHARS);
}
/*
* Scan windows name for characters that must not
* appear in a UNIX filename
*/
static int
invalidname(const u_int16_t *in, int n)
{
while (n-- > 0) {
if (*in++ == '/')
return 1;
}
return 0;
}
/*
* Convert UCS-2 character into UTF-8
* return number of output bytes or 0 if output
* buffer is too short
*/
static int
ucs2utf8(const u_int16_t *in, u_int8_t *out, int n)
{
uint16_t inch = le16toh(in[0]);
if (inch <= 0x007f) {
if (n < 1) return 0;
if (out)
*out++ = inch;
return 1;
} else if (inch <= 0x07ff) {
if (n < 2) return 0;
if (out) {
*out++ = 0xc0 | (inch >> 6);
*out++ = 0x80 | (inch & 0x3f);
}
return 2;
} else {
if (n < 3) return 0;
if (out) {
*out++ = 0xe0 | (inch >> 12);
*out++ = 0x80 | ((inch >> 6) & 0x3f);
*out++ = 0x80 | (inch & 0x3f);
}
return 3;
}
}
/*
* Convert UTF-8 bytes into UCS-2 character
* return number of input bytes, 0 if input
* is too short and -1 if input is invalid
*/
static int
utf8ucs2(const u_int8_t *in, int n, u_int16_t *out)
{
uint16_t outch;
if (n < 1) return 0;
if (in[0] <= 0x7f) {
outch = in[0];
if (out)
*out = htole16(outch);
return 1;
} else if (in[0] <= 0xdf) {
if (n < 2) return 0;
outch = (in[0] & 0x1f) << 6 | (in[1] & 0x3f);
if (out)
*out = htole16(outch);
return 2;
} else if (in[0] <= 0xef) {
if (n < 3) return 0;
outch = (in[0] & 0x1f) << 12 | (in[1] & 0x3f) << 6 | (in[2] & 0x3f);
if (out)
*out = htole16(outch);
return 3;
}
return -1;
}
/*
* Convert UCS-2 string into UTF-8 string
* return total number of output bytes
*/
static int
ucs2utf8str(const u_int16_t *in, int n, u_int8_t *out, int m)
{
u_int8_t *p;
int outlen;
p = out;
while (n > 0 && *in != 0) {
outlen = ucs2utf8(in, out ? p : out, m);
if (outlen == 0)
break;
p += outlen;
m -= outlen;
in += 1;
n -= 1;
}
return p - out;
}
/*
* Convert UTF8 string into UCS-2 string
* return total number of output chacters
*/
static int
utf8ucs2str(const u_int8_t *in, int n, u_int16_t *out, int m)
{
u_int16_t *p;
int inlen;
p = out;
while (n > 0 && *in != 0) {
if (m < 1)
break;
inlen = utf8ucs2(in, n, out ? p : out);
if (inlen <= 0)
break;
in += inlen;
n -= inlen;
p += 1;
m -= 1;
}
return p - out;
}
/*
* Convert UCS-2 string into 8bit character string
* return total number of output bytes
*/
static int
ucs2char8str(const u_int16_t *in, int n, u_int8_t *out, int m)
{
u_int8_t *p;
u_int16_t inch;
p = out;
while (n > 0 && in[0] != 0) {
if (m < 1)
break;
inch = le16toh(in[0]);
if (inch > 255)
break;
if (p)
p[0] = inch;
p += 1;
m -= 1;
in += 1;
n -= 1;
}
return p - out;
}
/*
* Convert 8bit character string into UCS-2 string
* return total number of output chacters
*/
static int
char8ucs2str(const u_int8_t *in, int n, u_int16_t *out, int m)
{
u_int16_t *p;
p = out;
while (n > 0 && in[0] != 0) {
if (m < 1)
break;
if (p)
p[0] = htole16(in[0]);
p += 1;
m -= 1;
in += 1;
n -= 1;
}
return p - out;
}
static void
ucs2pad(u_int16_t *buf, int len, int size)
{
if (len < size-1)
buf[len++] = 0x0000;
while (len < size)
buf[len++] = 0xffff;
}
/*
* Fold UCS-2 character to uppercase
*/
static u_int16_t
ucs2fold(u_int16_t w)
{
int low,high,mid;
u_int16_t check;
extern const u_int16_t msdosfs_unicode_foldmap[];
extern size_t msdosfs_unicode_foldmap_entries;
w = le16toh(w);
low = 0;
high = msdosfs_unicode_foldmap_entries / 2;
while (low < high) {
mid = (low + high)/2;
check = msdosfs_unicode_foldmap[2*mid+0];
if (w == check) {
w = msdosfs_unicode_foldmap[2*mid+1];
break;
}
if (w < check)
high = mid;
else
low = mid+1;
}
w = le16toh(w);
return w;
}
/*
* Compare two UCS-2 strings case-insensitive
*
* uses the Unicode case folding table
*/
static int
ucs2match(u_int16_t *w1, u_int16_t *w2, int n)
{
u_int16_t u1, u2;
while (n > 0) {
if (*w1 == 0 || *w2 == 0)
return *w1 == *w2;
u1 = ucs2fold(*w1);
u2 = ucs2fold(*w2);
if (u1 != u2)
return 0;
++w1;
++w2;
--n;
}
return 1;
}
/*
* Compare two 8bit char conversions case-insensitive
*
* uses the DOS case folding table
*/
static int
char8match(u_int16_t *w1, u_int16_t *w2, int n)
{
u_int16_t u1, u2;
while (n > 0) {
u1 = le16toh(*w1);
u2 = le16toh(*w2);
if (u1 == 0 || u2 == 0)
return u1 == u2;
if (u1 > 255 || u2 > 255)
return 0;
u1 = u2l[u1 & 0xff];
u2 = u2l[u2 & 0xff];
if (u1 != u2)
return 0;
++w1;
++w2;
--n;
}
return 1;
}