1/*
2 * Copyright (c) 2000-2008,2010-2012 Apple Inc. All rights reserved.
3 *
4 * @APPLE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. Please obtain a copy of the License at
10 * http://www.opensource.apple.com/apsl/ and read it before using this
11 * file.
12 *
13 * The Original Code and all software distributed under the License are
14 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
15 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
16 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
18 * Please see the License for the specific language governing rights and
19 * limitations under the License.
20 *
21 * @APPLE_LICENSE_HEADER_END@
22 */
23/* $FreeBSD: src/sys/msdosfs/msdosfs_conv.c,v 1.29 1999/08/28 00:48:08 peter Exp $ */
24/*	$NetBSD: msdosfs_conv.c,v 1.25 1997/11/17 15:36:40 ws Exp $	*/
25
26/*-
27 * Copyright (C) 1995, 1997 Wolfgang Solfrank.
28 * Copyright (C) 1995, 1997 TooLs GmbH.
29 * All rights reserved.
30 * Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
31 *
32 * Redistribution and use in source and binary forms, with or without
33 * modification, are permitted provided that the following conditions
34 * are met:
35 * 1. Redistributions of source code must retain the above copyright
36 *    notice, this list of conditions and the following disclaimer.
37 * 2. Redistributions in binary form must reproduce the above copyright
38 *    notice, this list of conditions and the following disclaimer in the
39 *    documentation and/or other materials provided with the distribution.
40 * 3. All advertising materials mentioning features or use of this software
41 *    must display the following acknowledgement:
42 *	This product includes software developed by TooLs GmbH.
43 * 4. The name of TooLs GmbH may not be used to endorse or promote products
44 *    derived from this software without specific prior written permission.
45 *
46 * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
47 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
48 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
49 * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
50 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
51 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
52 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
53 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
54 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
55 * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
56 */
57/*
58 * Written by Paul Popelka (paulp@uts.amdahl.com)
59 *
60 * You can do anything you want with this software, just don't say you wrote
61 * it, and don't remove this notice.
62 *
63 * This software is provided "as is".
64 *
65 * The author supplies this software to be publicly redistributed on the
66 * understanding that the author is not responsible for the correct
67 * functioning of this software in any circumstances and is not liable for
68 * any damages caused by this software.
69 *
70 * October 1992
71 */
72
73/*
74 * System include files.
75 */
76#include <sys/param.h>
77#include <sys/time.h>
78#include <sys/systm.h>
79#include <sys/dirent.h>
80
81/*
82 * MSDOSFS include files.
83 */
84#include "direntry.h"
85
86/*
87 * Total number of days that have passed for each month in a regular year.
88 */
89static uint16_t regyear[] = {
90	31, 59, 90, 120, 151, 181,
91	212, 243, 273, 304, 334, 365
92};
93
94/*
95 * Total number of days that have passed for each month in a leap year.
96 */
97static uint16_t leapyear[] = {
98	31, 60, 91, 121, 152, 182,
99	213, 244, 274, 305, 335, 366
100};
101
102/*
103 * Variables used to remember parts of the last time conversion.  Maybe we
104 * can avoid a full conversion.
105 */
106static uint32_t  lasttime;
107static uint32_t  lastday;
108static uint16_t lastddate;
109static uint16_t lastdtime;
110
111/*
112 * This variable contains the number of seconds that local time is west of GMT
113 * It is updated every time an msdosfs volume is mounted.  This value is
114 * essentially tz_minuteswest * 60 - (tz_dsttime ? 3600 : 0) based on the
115 * timezone returned by gettimeofday() in the mount_msdosfs tool.
116 *
117 * This has a problem with daylight savings time.  If the current daylight
118 * savings time setting does not match the date being converted, then it
119 * will be off by one hour.  The only way to properly fix this is to know
120 * (inside the kernel) what dates should be adjusted for daylight savings,
121 * and which should not.  That would let us return correct GMT for dates
122 * throughout the year.  Of course, the GUI would have to do the opposite
123 * conversion when converting to local time for display to the user.
124 */
125__private_extern__ int32_t msdos_secondsWest = 0;
126
127/*
128 * Convert the unix version of time to dos's idea of time to be used in
129 * file timestamps. The passed in unix time is assumed to be in GMT.
130 */
131void msdosfs_unix2dostime(struct timespec *tsp, u_int16_t *ddp, u_int16_t *dtp, u_int8_t *dhp)
132{
133	uint32_t t;
134	uint32_t days;
135	uint32_t inc;
136	uint32_t year;
137	uint32_t month;
138	uint16_t *months;
139
140	/*
141	 * If the time from the last conversion is the same as now, then
142	 * skip the computations and use the saved result.
143	 */
144	t = (uint32_t)(tsp->tv_sec - msdos_secondsWest);
145	t &= ~1;	/* Round down to multiple of 2 seconds */
146	if (lasttime != t) {
147		lasttime = t;
148		lastdtime = (((t / 2) % 30) << DT_2SECONDS_SHIFT)
149		    + (((t / 60) % 60) << DT_MINUTES_SHIFT)
150		    + (((t / 3600) % 24) << DT_HOURS_SHIFT);
151
152		/*
153		 * If the number of days since 1970 is the same as the last
154		 * time we did the computation then skip all this leap year
155		 * and month stuff.
156		 */
157		days = t / (24 * 60 * 60);
158		if (days != lastday) {
159			lastday = days;
160			for (year = 1970;; year++) {
161				inc = year & 0x03 ? 365 : 366;
162				if (days < inc)
163					break;
164				days -= inc;
165			}
166			months = year & 0x03 ? regyear : leapyear;
167			for (month = 0; days >= months[month]; month++)
168				;
169			if (month > 0)
170				days -= months[month - 1];
171			lastddate = ((days + 1) << DD_DAY_SHIFT)
172			    + ((month + 1) << DD_MONTH_SHIFT);
173			/*
174			 * Remember dos's idea of time is relative to 1980.
175			 * unix's is relative to 1970.  If somehow we get a
176			 * time before 1980 then don't give totally crazy
177			 * results.
178			 */
179			if (year > 1980)
180				lastddate += (year - 1980) << DD_YEAR_SHIFT;
181		}
182	}
183	if (dtp)
184		*dtp = lastdtime;
185	if (dhp)
186		*dhp = (tsp->tv_sec & 1) * 100 + tsp->tv_nsec / 10000000;
187
188	*ddp = lastddate;
189}
190
191/*
192 * The number of seconds between Jan 1, 1970 and Jan 1, 1980. In that
193 * interval there were 8 regular years and 2 leap years.
194 */
195#define	SECONDSTO1980	(((8 * 365) + (2 * 366)) * (24 * 60 * 60))
196
197static uint16_t lastdosdate;
198static uint32_t  lastseconds;
199
200/*
201 * Convert from dos' idea of time to unix'. This will probably only be
202 * called from the stat(), and fstat() system calls and so probably need
203 * not be too efficient.
204 */
205void msdosfs_dos2unixtime(u_int dd, u_int dt, u_int dh, struct timespec *tsp)
206{
207	uint32_t seconds;
208	uint32_t month;
209	uint32_t year;
210	uint32_t days;
211	uint16_t *months;
212
213	if (dd == 0) {
214		/*
215		 * Uninitialized field, return the epoch.
216		 */
217		tsp->tv_sec = 0;
218		tsp->tv_nsec = 0;
219		return;
220	}
221	seconds = (((dt & DT_2SECONDS_MASK) >> DT_2SECONDS_SHIFT) << 1)
222	    + ((dt & DT_MINUTES_MASK) >> DT_MINUTES_SHIFT) * 60
223	    + ((dt & DT_HOURS_MASK) >> DT_HOURS_SHIFT) * 3600
224	    + dh / 100;
225	/*
226	 * If the year, month, and day from the last conversion are the
227	 * same then use the saved value.
228	 */
229	if (lastdosdate != dd) {
230		lastdosdate = dd;
231		year = (dd & DD_YEAR_MASK) >> DD_YEAR_SHIFT;
232		days = year * 365;
233		days += year / 4 + 1;	/* add in leap days */
234		if ((year & 0x03) == 0)
235			days--;		/* if year is a leap year */
236		months = year & 0x03 ? regyear : leapyear;
237		month = (dd & DD_MONTH_MASK) >> DD_MONTH_SHIFT;
238		if (month < 1 || month > 12) {
239			month = 1;
240		}
241		if (month > 1)
242			days += months[month - 2];
243		days += ((dd & DD_DAY_MASK) >> DD_DAY_SHIFT) - 1;
244		lastseconds = (days * 24 * 60 * 60) + SECONDSTO1980;
245	}
246	tsp->tv_sec = seconds + lastseconds + msdos_secondsWest;
247	tsp->tv_nsec = (dh % 100) * 10000000;
248}
249
250/*
251 * Unicode (LSB) to Win Latin1 (ANSI CodePage 1252).
252 * Also converts ASCII to upper case.
253 *
254 * 0 - character disallowed in long file name.
255 * 1 - character should be replaced by '_' in DOS file name,
256 *     and generation number inserted.
257 * 2 - character ('.' and ' ') should be skipped in DOS file name,
258 *     and generation number inserted.
259 */
260static u_char
261unilsb2dos[256] = {
262	0,    1,    1,    1,    1,    1,    1,    1,	/* 00-07 */
263	1,    1,    1,    1,    1,    1,    1,    1,	/* 08-0f */
264	1,    1,    1,    1,    1,    1,    1,    1,	/* 10-17 */
265	1,    1,    1,    1,    1,    1,    1,    1,	/* 18-1f */
266	2,    0x21, 1,    0x23, 0x24, 0x25, 0x26, 0x27,	/* 20-27 */
267	0x28, 0x29, 1,    1,    1,    0x2d, 2,    0,	/* 28-2f */
268	0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,	/* 30-37 */
269	0x38, 0x39, 1,    1,    1,    1,    1,    1,	/* 38-3f */
270	0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,	/* 40-47 */
271	0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f,	/* 48-4f */
272	0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,	/* 50-57 */
273	0x58, 0x59, 0x5a, 1,    1,    1,    0x5e, 0x5f,	/* 58-5f */
274	0x60, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,	/* 60-67 */
275	0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f,	/* 68-6f */
276	0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,	/* 70-77 */
277	0x58, 0x59, 0x5a, 0x7b, 1,    0x7d, 0x7e, 0x7f,	/* 78-7f */
278	0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,	/* 80-87 */
279	0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f,	/* 88-8f */
280	0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,	/* 90-97 */
281	0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,	/* 98-9f */
282	0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,	/* a0-a7 */
283	0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,	/* a8-af */
284	0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,	/* b0-b7 */
285	0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,	/* b8-bf */
286	0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,	/* c0-c7 */
287	0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,	/* c8-cf */
288	0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,	/* d0-d7 */
289	0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,	/* d8-df */
290	0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,	/* e0-e7 */
291	0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef,	/* e8-ef */
292	0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,	/* f0-f7 */
293	0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,	/* f8-ff */
294};
295
296/* Unicode punctuation marks to Win Latin1 (ANSI CodePage 1252) */
297static u_char
298unipunct2dos[48] = {
299	1,    1,    1,    0x96, 0x97, 1,    1,    1,  /* 2010-2017 */
300	0x91, 0x92, 0x82, 1,    0x93, 0x94, 0x84, 1,  /* 2018-201F */
301	0x86, 0x87, 0x95, 1,    1,    1,    0x85, 1,  /* 2020-2027 */
302	1,    1,    1,    1,    1,    1,    1,    1,  /* 2028-202F */
303	0x89, 1,    1,    1,    1,    1,    1,    1,  /* 2030-2037 */
304	1,    0x8B, 0x9B, 1,    1,    1,    1,    1   /* 2038-203F */
305};
306
307/* Win Latin1 (ANSI CodePage 1252) to Unicode */
308__private_extern__ u_int16_t
309dos2unicode[32] = {
310  0x20AC, 0x003f, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021, /* 80-87 */
311  0x02C6, 0x2030, 0x0160, 0x2039, 0x0152, 0x003f, 0x017D, 0x003f, /* 88-8F */
312  0x003f, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014, /* 90-97 */
313  0x02DC, 0x2122, 0x0161, 0x203A, 0x0153, 0x003f, 0x017E, 0x0178, /* 98-9F */
314};
315
316
317/*
318 * Case folding table for Latin-1.
319 *
320 * Maps 'A'..'Z' to 'a'..'z' and 0xc0..0xd6 to 0xe0..0xf6 and
321 * 0xd8..0xde to 0xf8..0xfe.
322 */
323__private_extern__ u_char
324l2u[256] = {
325	0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, /* 00-07 */
326	0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, /* 08-0f */
327	0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, /* 10-17 */
328	0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, /* 18-1f */
329	0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, /* 20-27 */
330	0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, /* 28-2f */
331	0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, /* 30-37 */
332	0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, /* 38-3f */
333	0x40, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, /* 40-47 */
334	0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, /* 48-4f */
335	0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, /* 50-57 */
336	0x78, 0x79, 0x7a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, /* 58-5f */
337	0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, /* 60-67 */
338	0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, /* 68-6f */
339	0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, /* 70-77 */
340	0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, /* 78-7f */
341	0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, /* 80-87 */
342	0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f, /* 88-8f */
343	0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, /* 90-97 */
344	0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, /* 98-9f */
345	0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, /* a0-a7 */
346	0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, /* a8-af */
347	0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, /* b0-b7 */
348	0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, /* b8-bf */
349	0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, /* c0-c7 */
350	0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, /* c8-cf */
351	0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xd7, /* d0-d7 */
352	0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xdf, /* d8-df */
353	0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, /* e0-e7 */
354	0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, /* e8-ef */
355	0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, /* f0-f7 */
356	0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff, /* f8-ff */
357};
358
359/*
360 *	Look-up table to determine whether a given ASCII character is
361 *	considered upper case, lower case, or neither.  Also indicates
362 *	which characters should require generation of a long name.
363 *
364 *	Values are bit masks composed of the following constants:
365 */
366enum {
367	CASE_LOWER	= 1,
368	CASE_UPPER	= 2,
369	CASE_LONG	= 4,	/* A long name should be generated */
370};
371
372static u_char
373ascii_case[128] = {
374	0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04,	/* 00-07 */
375	0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04,	/* 08-0F */
376	0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04,	/* 10-17 */
377	0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04,	/* 18-1F */
378	0x04, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00,	/* 20-27 */
379	0x00, 0x00, 0x04, 0x04, 0x04, 0x00, 0x04, 0x04,	/* 28-2F */
380	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,	/* 30-37 */
381	0x00, 0x00, 0x04, 0x04, 0x04, 0x04, 0x04, 0x04,	/* 38-3F */
382	0x00, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02,	/* 40-47 */
383	0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02,	/* 48-4F */
384	0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02,	/* 50-57 */
385	0x02, 0x02, 0x02, 0x04, 0x04, 0x04, 0x00, 0x00,	/* 58-5F */
386	0x00, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,	/* 60-67 */
387	0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,	/* 68-6F */
388	0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,	/* 70-77 */
389	0x01, 0x01, 0x01, 0x00, 0x04, 0x00, 0x00, 0x00,	/* 78-7F */
390};
391
392#if 0
393/*
394 * Macintosh Unicode (LSB) to Microsoft Services for Macintosh (SFM) Unicode
395 */
396static u_int16_t
397mac2sfm[128] = {
398	0x0,    0xf001, 0xf002, 0xf003, 0xf004, 0xf005, 0xf006, 0xf007,	/* 00-07 */
399	0xf008, 0xf009, 0xf00a, 0xf00b, 0xf00c, 0xf00d, 0xf00e, 0xf00f,	/* 08-0f */
400	0xf010, 0xf011, 0xf012, 0xf013, 0xf014, 0xf015, 0xf016, 0xf017,	/* 10-17 */
401	0xf018, 0xf019, 0xf01a, 0xf01b, 0xf01c, 0xf01d, 0xf01e, 0xf01f,	/* 18-1f */
402	0x20,   0x21,   0xf020, 0x23,   0x24,   0x25,   0x26,   0x27,	/* 20-27 */
403	0x28,   0x29,   0xf021, 0x2b,   0x2c,   0x2d,   0x2e,   0x2f,  	/* 28-2f */
404	0x30,   0x31,   0x32,   0x33,   0x34,   0x35,   0x36,   0x37,	/* 30-37 */
405	0x38,   0x39,   0xf022, 0x3b,   0xf023, 0x3d,   0xf024, 0xf025, /* 38-3f */
406	0x40,   0x41,   0x42,   0x43,   0x44,   0x45,   0x46,   0x47,	/* 40-47 */
407	0x48,   0x49,   0x4a,   0x4b,   0x4c,   0x4d,   0x4e,   0x4f,	/* 48-4f */
408	0x50,   0x51,   0x52,   0x53,   0x54,   0x55,   0x56,   0x57,	/* 50-57 */
409	0x58,   0x59,   0x5a,   0x5b,   0xf026, 0x5d,   0x5e,   0x5f,	/* 58-5f */
410	0x60,   0x61,   0x62,   0x63,   0x64,   0x65,   0x66,   0x67,	/* 60-67 */
411	0x68,   0x69,   0x6a,   0x6b,   0x6c,   0x6d,   0x6e,   0x6f,	/* 68-6f */
412	0x70,   0x71,   0x72,   0x73,   0x74,   0x75,   0x76,   0x77,	/* 70-77 */
413	0x78,   0x79,   0x7a,   0x7b,   0xf027, 0x7d,   0x7e,   0x7f,   /* 78-7f */
414};
415
416#define MAX_MAC2SFM			0x80
417#define MAX_SFM2MAC			0x29
418#define SFMCODE_PREFIX_MASK	0xf000
419/*
420 * SFM Unicode (LSB) to Macintosh Unicode (LSB)
421 */
422static u_char
423sfm2mac[42] = {
424	0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,	/* 00-07 */
425	0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,	/* 08-0F */
426	0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,	/* 10-17 */
427	0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,	/* 18-1F */
428	0x22, 0x2a, 0x3a, 0x3c, 0x3e, 0x3f, 0x5c, 0x7c,	/* 20-27 */
429	0x20, 0x2e, 	 							 	/* 28-29 */
430};
431#endif
432
433/* map a Unicode char into a DOS char */
434u_char msdosfs_unicode2dos(u_int16_t uc)
435{
436	if (uc < 0x100)
437		return (unilsb2dos[uc]);
438
439	if (uc > 0x2122)
440		return (1);
441
442	if (uc >= 0x2010 && uc <= 0x203F)
443		return (unipunct2dos[uc - 0x2010]);
444
445	if (uc >= 0x0152 && uc <= 0x02DC)
446		switch (uc) {
447		case 0x0152:
448		    return (0x8C);  /* LATIN CAPITAL LIGATURE OE */
449		case 0x0153:
450		    return (0x9C);  /* LATIN SMALL LIGATURE OE */
451		case 0x0160:
452		    return (0x8A);  /* CAPITAL LETTER S WITH CARON */
453		case 0x0161:
454		    return (0x9A);  /* SMALL LETTER S WITH CARON */
455		case 0x0178:
456		    return (0x9F);  /* CAPITAL LETTER Y WITH DIAERESIS */
457		case 0x017D:
458		    return (0x8E);  /* CAPITAL LETTER Z WITH CARON */
459		case 0x017E:
460		    return (0x9E);  /* SMALL LETTER Z WITH CARON */
461		case 0x0192:
462		    return (0x83);  /* SMALL LETTER F WITH HOOK */
463		case 0x02C6:
464		    return (0x88);  /* MODIFIER LETTER CIRCUMFLEX ACCENT */
465		case 0x02DC:
466		    return (0x98);  /* SMALL TILDE */
467		default:
468			return (1);
469		}
470
471	if (uc == 0x20AC)
472		return (0x80);  /* EURO SIGN */
473	if (uc == 0x2122)
474		return (0x99);  /* TRADE MARK SIGN */
475
476	return (1);
477}
478
479/*
480 * DOS filenames are made of 2 parts, the name part and the extension part.
481 * The name part is 8 characters long and the extension part is 3
482 * characters long.  They may contain trailing blanks if the name or
483 * extension are not long enough to fill their respective fields.
484 */
485
486/*
487 * Convert a DOS filename to a Unicode filename. And, return the number of
488 * characters in the resulting unix filename.
489 */
490size_t msdosfs_dos2unicodefn(u_char dn[SHORT_NAME_LEN], u_int16_t *un, int lower)
491{
492	int i;
493	u_char dc;
494	int unichars = 0;
495
496	/*
497	 * Copy the base name portion into the Unicode string.
498	 */
499	for (i = 0; i < 8; i++) {
500		dc = *dn++;
501
502		/*
503		 * If first char of the filename is SLOT_E5 (0x05), then
504		 * the real first char of the filename should be 0xe5.
505		 * But, they couldn't just have a 0xe5 mean 0xe5 because
506		 * that is used to mean a freed directory slot.
507		 */
508		if (i == 0 && dc == SLOT_E5)
509			dc = 0xe5;
510
511		/*
512		 * If the base name was supposed to be lower case,
513		 * then convert it.
514		 */
515		if (lower & LCASE_BASE)
516			dc = l2u[dc];	/* Map to lower case equivalent */
517
518		un[unichars++] = (dc < 0x80 || dc > 0x9F ? (u_int16_t)dc : dos2unicode[dc - 0x80]);
519	}
520
521	/*
522	 * Get rid of trailing space padding in the base name.
523	 */
524	while (unichars > 0 && un[unichars-1]==' ')
525	{
526		-- unichars;
527	}
528
529	/*
530	 * Copy the extension portion (with dot) into the Unicode string.
531	 */
532	un[unichars++] = '.';
533	for (i = 0; i < 3; i++) {
534		dc = *dn++;
535
536		/*
537		 * If the extension was supposed to be all lower case,
538		 * then convert it.
539		 */
540		if (lower & LCASE_EXT)
541			dc = l2u[dc];	/* Map to lower case equivalent */
542
543		un[unichars++] = (dc < 0x80 || dc > 0x9F ? (u_int16_t)dc : dos2unicode[dc - 0x80]);
544	}
545
546	/*
547	 * Get rid of trailing space padding in the extension (and the dot
548	 * if there was no extension).
549	 */
550	for (i=0; i<3 && un[unichars-1]==' '; ++i)
551	{
552		--unichars;
553	}
554	if (i==3)			/* Was the extension entirely spaces? */
555		--unichars;		/* Yes, so remove the dot, too. */
556
557	return unichars;
558}
559
560
561/*
562 * Convert a Unicode filename to an 8.3-style short name.
563 *
564 * This function implements the Basis-Name Generation Algorithm
565 * from the FAT32 specification.  It does NOT implement the numeric
566 * tail generation; that part is implemented elsewhere.
567 *
568 * In order to support setting the LCASE_BASE and LCASE_EXT flags, this
569 * routine returns the corresponding value for those flags when the name can
570 * be converted (when the function result is 1).
571 *
572 * The test is whether the name (base or extension) contains at least one
573 * lower case letter, no upper case, and zero or more case-neutral characters
574 * allowed in an 8.3 name (such as digits, dollar sign, ampersand, etc.).
575 *
576 * By experimentation with Windows XP, it seems to always create long names
577 * for names that contain extended characters, such as letter e with acute
578 * (regardless of upper/lower case).  This is true even on a U.S. system
579 * with characters that are representable in code page 437 (which is in
580 * fact what you see in the short name).  I'm guessing it uses pure short
581 * names only if all characters are pure ASCII.
582 *
583 * Inputs:
584 *	unicode			Points to a buffer of UTF-16 (native endian)
585 *					containing the original name.
586 *	unicode_length	The length of the Unicode name in UTF-16 code
587 *					points.
588 *
589 * Outputs:
590 *	short_name		The corresponding 8.3-style short name is returned
591 *					in this buffer.  The short name is upper case.
592 *	lower_case		A pointer to returned lower case flags.  The pointer
593 *					must not be NULL.
594 *
595 * Returns:
596 *	0	if name couldn't be converted (has disallowed characters).
597 *	1	if the converted name is the same as the original
598 *		(no long filename entry necessary for Win95).
599 *	2	if conversion was successful, and long name entries should be created,
600 *		but a generation number is not necessary (for example, the name had a
601 *		mix of upper and lower case, contained non-ASCII characters, contained
602 *		an embedded space, or contained a dot in the base name).
603 *	3	if conversion was successful, long name entries should be created, and
604 *		one or more characters of the original name were removed (thus, a
605 *		generation number should also be added).
606 */
607int msdosfs_unicode_to_dos_name(const uint16_t *unicode,
608								size_t unicode_length,
609								uint8_t short_name[SHORT_NAME_LEN],
610								u_int8_t *lower_case)
611{
612	int i;					/* An index into the Unicode name. */
613	int j;					/* An index into the short name. */
614	int conv = 1;			/* The function's result. */
615	const uint16_t *cp;		/* A pointer for looping over the Unicode name. */
616	const uint16_t *dp;		/* A pointer to the first char of extension (or NULL). */
617	const uint16_t *dp1;	/* A pointer after last seen dot character. */
618	uint16_t c;				/* The current character being examined. */
619	int case_flags;			/* Accumulates upper/low/neither case information */
620
621	*lower_case = 0;	/* default to all upper case, and clear undefined bits */
622
623	/*
624	 * Fill the dos filename string with blanks. These are DOS's pad
625	 * characters.
626	 */
627	memset(short_name, ' ', SHORT_NAME_LEN);
628
629	/*
630	 * The filenames "." and ".." are handled specially, since they
631	 * don't follow dos filename rules.
632	 */
633	if (unicode_length == 1 && unicode[0] == '.') {
634		short_name[0] = '.';
635		return 1;
636	}
637	if (unicode_length == 2 && unicode[0] == '.' && unicode[1] == '.') {
638		short_name[0] = '.';
639		short_name[1] = '.';
640		return 1;
641	}
642
643	/*
644	 * Filenames with some characters are not allowed!
645	 *
646	 * NOTE: For a name passed into the POSIX APIs, most of the disallowed
647	 * characters will have already been replaced with alternate characters
648	 * according to the Services For Macintosh conversions.
649	 */
650	for (cp = unicode, i = (int)unicode_length; --i >= 0; cp++)
651		if (msdosfs_unicode2dos(*cp) == 0)
652			return 0;
653
654	/*
655	 * Find the extension (if any).
656	 *
657	 * Note: A dot as the first or last character do not count as
658	 * an extension.  Trailing blanks are supposed to be ignored,
659	 * but we don't (and didn't previously) do that.
660	 */
661	dp = dp1 = NULL;
662	for (cp = unicode + 1, i = (int)unicode_length - 1; --i >= 0;) {
663		switch (*cp++) {
664			case '.':
665				if (!dp1)
666					dp1 = cp;
667				break;
668			default:
669				if (dp1)
670					dp = dp1;
671				dp1 = NULL;
672				break;
673		}
674	}
675
676	/*
677	 * Convert the extension (if any).
678	 */
679	if (dp) {
680		int l;		/* The length of the extension (in UTF-16 code points). */
681
682		if (dp1)
683			l = (int)(dp1 - dp);            /* Ignore trailing dots. */
684		else
685			l = (int)(unicode_length - (dp - unicode));
686
687		/*
688		 * Convert up to 3 characters of the extension.
689		 */
690		for (case_flags = i = 0, j = 8; i < l && j < SHORT_NAME_LEN; i++, j++) {
691			c = dp[i];
692			if (c < 0x80)
693				case_flags |= ascii_case[c];
694			else
695				case_flags |= CASE_LONG;	/* Non-ASCII always requires a long name */
696			if (c < 0x100)
697   				c = l2u[c];
698			c = msdosfs_unicode2dos(c);
699			short_name[j] = c;
700			if (c == 1) {
701				conv = 3;					/* Character is not allowed in short names */
702				short_name[j] = '_';		/* and must be replaced with underscore */
703			}
704			if (c == 2) {
705				conv = 3;					/* Character is not allowed in short names */
706				short_name[j--] = ' ';		/* and is not substituted */
707			}
708		}
709
710		/*
711		 * Check for other conditions which might require a long name.
712		 */
713		if ((case_flags & CASE_LONG) != 0 && conv != 3)
714			conv = 2;			/* Force a long name for things like embedded spaces or non-ASCII */
715		if (conv == 1) {
716			if ((case_flags & (CASE_LOWER | CASE_UPPER)) == (CASE_LOWER | CASE_UPPER))
717				conv = 2;		/* Force a long name for names with mixed case */
718			else if (case_flags & CASE_LOWER)
719				*lower_case |= LCASE_EXT;	/* Extension has lower case */
720		}
721		if (i < l)
722			conv = 3;			/* Extension was longer than 3 characters */
723
724		dp--;					/* dp points at the dot at the start of the extension. */
725	} else {
726		dp = cp;				/* dp points past the end of the Unicode name. */
727	}
728
729	/*
730	 * Now convert the base name
731	 *
732	 * When we get here, dp points just past the last character of the base name.
733	 */
734	for (case_flags = i = j = 0; unicode < dp && j < 8; i++, j++, unicode++) {
735        c = *unicode;
736		if (c < 0x80)
737			case_flags |= ascii_case[c];
738		else
739			case_flags |= CASE_LONG;	/* Non-ASCII always requires a long name */
740        if (c < 0x100)
741            c = l2u[c];
742        c = msdosfs_unicode2dos(c);
743        short_name[j] = c;
744		if (c == 1) {
745			conv = 3;		/* Character is not allowed in short names */
746			short_name[j] = '_';	/* and must be replaced with underscore */
747		}
748		if (c == 2) {
749			conv = 3;		/* Character is not allowed in short names */
750			short_name[j--] = ' ';	/* and is not substituted */
751		}
752	}
753	if ((case_flags & CASE_LONG) != 0 && conv != 3)
754		conv = 2;	/* Force a long name for things like embedded spaces */
755	if (conv == 1) {
756		if ((case_flags & (CASE_LOWER | CASE_UPPER)) == (CASE_LOWER | CASE_UPPER))
757			conv = 2;	/* Force a long name for names with mixed case */
758		else if (case_flags & CASE_LOWER)
759			*lower_case |= LCASE_BASE;	/* Base name has lower case */
760	}
761
762	if (unicode < dp)
763		conv = 3;	/* Base name was longer than 8 characters */
764
765	/*
766	 * If the resulting base name was empty, generate a default
767	 */
768	if (!j)
769		short_name[0] = '_';
770
771	/*
772	 * The first character cannot be E5, because that means a deleted entry
773	 */
774	if (short_name[0] == 0xe5)
775		short_name[0] = SLOT_E5;
776
777	/*
778	 * If the name couldn't be represented as a short name, make sure the
779	 * lower case flags are clear (in case the base or extension was all
780	 * lower case, but the other was not, in which case we left one of the
781	 * bits set above).
782	 *
783	 * That is, the lower case flags are only set if the name is a valid
784	 * 8.3-style short name, where any letters in the base and extension
785	 * are all upper or all lower case.  (The case of the base may differ
786	 * from the case of the extension.)  Non-letter ASCII characters do
787	 * not affect the lower case flags.
788	 */
789	if (conv != 1)
790		*lower_case = 0;
791
792	return conv;
793}
794
795
796/*
797 * Add or update a generation number, and preceding tilde (~), to a short name.
798 *
799 * The base name is up to 8 characters.  We want to keep at least the first
800 * character.  We need to insert a tilde before the generation number.  That
801 * leaves us with room for up to 6 characters for the generation number.
802 *
803 * Parameters:
804 *	short_name		Modified in place.  On input, a short name with no
805 *					generation number, or a generation number less than
806 *					or equal to the "generation" parameter.  On output,
807 *					the short name with generation number.
808 *	generation		The generation number to add or update.
809 *
810 * Result:
811 *	0				The generation number was successfully added.
812 *	ENAMETOOLONG	The generation number was too large to fit in the short name.
813 */
814int msdosfs_apply_generation_to_short_name(uint8_t short_name[SHORT_NAME_LEN],
815										   unsigned generation)
816{
817	uint8_t	generation_text[6];			/* Note: this is stored in reverse order. */
818	unsigned generation_text_length = 0;
819	unsigned base_name_index;			/* Index into base name where generation string
820										   should be stored. */
821	/* Convert the generation number to ASCII. */
822	do {
823		generation_text[generation_text_length++] = '0' + (generation % 10);
824		generation /= 10;
825	} while (generation > 0 && generation_text_length < 6);
826
827	/* Check for a too-large generation number. */
828	if (generation != 0)
829		return ENAMETOOLONG;
830
831	/*
832	 * Add or replace the generation string.
833	 *
834	 * We start by assuming the base name is the maximum 8 characters, then scan left
835	 * to find a non-space character.  (FAT uses trailing space padding in short names.
836	 * This also handles the case where the short name had an embedded space, and we'd
837	 * like to avoid putting the generation string after an embedded space.)
838	 */
839	base_name_index = 8 - (generation_text_length + 1);		/* +1 is for the tilde. */
840	while (short_name[base_name_index] == ' ' && base_name_index > 0)
841	{
842		--base_name_index;		/* Skip over trailing or embedded spaces. */
843	}
844	short_name[base_name_index++] = '~';
845	while (generation_text_length > 0)
846	{
847		short_name[base_name_index++] = generation_text[--generation_text_length];
848	}
849
850	return 0;
851}
852
853
854/*
855 * Create a Win95 long name directory entry
856 * Note: assumes that the filename is valid,
857 *	 i.e. doesn't consist solely of blanks and dots
858 */
859int msdosfs_unicode2winfn(const u_int16_t *un, int unlen, struct winentry *wep, int cnt, int chksum)
860{
861	u_int8_t *wcp;
862	int i;
863	u_int16_t code;
864
865	un += (cnt - 1) * WIN_CHARS;
866	unlen -= (cnt - 1) * WIN_CHARS;
867
868	/*
869	 * Initialize winentry to some useful default
870	 */
871	for (wcp = (u_int8_t *)wep, i = sizeof(*wep); --i >= 0; *wcp++ = 0xff);
872	wep->weCnt = cnt;
873	wep->weAttributes = ATTR_WIN95;
874	wep->weReserved1 = 0;
875	wep->weChksum = chksum;
876	wep->weReserved2 = 0;
877
878	/*
879	 * Now convert the filename parts
880	 */
881	for (wcp = wep->wePart1, i = sizeof(wep->wePart1)/2; --i >= 0;) {
882		if (--unlen < 0)
883			goto done;
884		code = *un++;
885		*wcp++ = code & 0x00ff;
886		*wcp++ = code >> 8;
887	}
888	for (wcp = wep->wePart2, i = sizeof(wep->wePart2)/2; --i >= 0;) {
889		if (--unlen < 0)
890			goto done;
891		code = *un++;
892		*wcp++ = code & 0x00ff;
893		*wcp++ = code >> 8;
894	}
895	for (wcp = wep->wePart3, i = sizeof(wep->wePart3)/2; --i >= 0;) {
896		if (--unlen < 0)
897			goto done;
898		code = *un++;
899		*wcp++ = code & 0x00ff;
900		*wcp++ = code >> 8;
901	}
902	if (!unlen)
903		wep->weCnt |= WIN_LAST;
904	return unlen;
905
906done:
907	*wcp++ = 0;
908	*wcp   = 0;
909	wep->weCnt |= WIN_LAST;
910	return 0;
911}
912
913
914
915/*
916 * Convert a Unicode character to a single known case.  Upper and lower case
917 * variants of the same character produce the same result.
918 *
919 * As an implementation detail, we convert the character to lower case.
920 *
921 * Note: this currently only handles case folding of ASCII characters.  The
922 * Unicode standard defines case equivalence for other characters (such as
923 * precomposed characters), but I don't know whether Windows considers them
924 * case equivalents.
925 */
926static inline u_int16_t case_fold(u_int16_t ch)
927{
928	if (ch < 0x100)
929		return l2u[ch];
930	else
931		return ch;
932}
933
934
935/*
936 * Compare a single long name entry to the corresponding portion of a
937 * filename.  If the name matches, we return the on-disk name and a
938 * flag indicating whether the name needed to be case folded.
939 *
940 * We also make sure all long name entries have the same checksum
941 * value; if not, the long name is invalid and is not a match.
942 *
943 * Inputs:
944 *	un[]		The name being looked up (the search name).
945 *	ucslen		The number of code points in the name (length of un[] in elements).
946 *	wep			The current long name entry being compared.
947 *	chksum		The checksum field from previous long name entries.
948 *
949 * Outputs:
950 *	found_name	The actual on-disk name that matched.  OPTIONAL.  MAY BE NULL.
951 *	case_folded	Set to true if the on-disk name is a case variant of un[] (that is
952 *				NOT a case-sensitive match), or if the on-disk name is not a match
953 *				to un[].  Unchanged if the on-disk name is exactly equal to un[].
954 *
955 * Function result:
956 *	-1			The name or checksums did not match
957 * <checksum>	The name matched.  <checksum> is the checksum stored in all
958 *				of the long name entries.
959 */
960int msdosfs_winChkName(const u_int16_t *un, int ucslen, struct winentry *wep, int chksum,
961					   u_int16_t *found_name, boolean_t *case_folded)
962{
963	u_int8_t *cp;
964	int i;
965	u_int16_t code;
966
967	/*
968	 * First compare checksums
969	 */
970	if (wep->weCnt&WIN_LAST)
971		chksum = wep->weChksum;
972	else if (chksum != wep->weChksum)
973		chksum = -1;
974	if (chksum == -1)
975		return -1;
976
977	/*
978	 * Offset of this entry
979	 */
980	i = ((wep->weCnt&WIN_CNT) - 1) * WIN_CHARS;
981	un += i;
982	if (found_name)
983		found_name += i;
984
985	if ((ucslen -= i) < 0)	/* Was "<=".  See below. */
986		return -1;	/* More long name entries than the name would need */
987	if ((wep->weCnt&WIN_LAST) && ucslen > WIN_CHARS)
988		return -1;	/* Too few long name entries to hold the name */
989
990        /*
991         * [2865792] Some FAT implementations have a bug when the long
992         * is an exact multiple of WIN_CHARS long.  They make an extra
993         * long name entry containing only a terminating 0x0000 and
994         * the 0xFFFF pad characters.  While this is out-of-spec
995         * (i.e. corrupt), we can be graceful and handle it anyway,
996         * like Windows does.
997         *
998         * We handle this case by falling through here with ucslen == 0.
999         * We then expect to return during the first iteration of the
1000         * following for() loop where --ucslen goes negative, and
1001         * "cp" points to two zero bytes.
1002         */
1003
1004	/*
1005	 * Compare the name parts
1006	 */
1007	for (cp = wep->wePart1, i = sizeof(wep->wePart1)/2; --i >= 0;) {
1008		code = (cp[1] << 8) | cp[0];
1009		if (--ucslen < 0) {
1010			/* Got to end of input name.  Are we also at end of on-disk name? */
1011			if (code == 0)
1012				return chksum;
1013			else
1014				return -1;
1015		}
1016		if (found_name)
1017			*found_name++ = code;
1018		if (code != *un)
1019		{
1020			/* Not an exact match.  Try case-insensitive match. */
1021			*case_folded = TRUE;
1022			if (case_fold(code) != case_fold(*un))
1023				return (-1);
1024		}
1025		cp += 2;
1026		un++;
1027	}
1028	for (cp = wep->wePart2, i = sizeof(wep->wePart2)/2; --i >= 0;) {
1029		code = (cp[1] << 8) | cp[0];
1030		if (--ucslen < 0) {
1031			/* Got to end of input name.  Are we also at end of on-disk name? */
1032			if (code == 0)
1033				return chksum;
1034			else
1035				return -1;
1036		}
1037		if (found_name)
1038			*found_name++ = code;
1039		if (code != *un)
1040		{
1041			/* Not an exact match.  Try case-insensitive match. */
1042			*case_folded = TRUE;
1043			if (case_fold(code) != case_fold(*un))
1044				return (-1);
1045		}
1046		cp += 2;
1047		un++;
1048	}
1049	for (cp = wep->wePart3, i = sizeof(wep->wePart3)/2; --i >= 0;) {
1050		code = (cp[1] << 8) | cp[0];
1051		if (--ucslen < 0) {
1052			/* Got to end of input name.  Are we also at end of on-disk name? */
1053			if (code == 0)
1054				return chksum;
1055			else
1056				return -1;
1057		}
1058		if (found_name)
1059			*found_name++ = code;
1060		if (code != *un)
1061		{
1062			/* Not an exact match.  Try case-insensitive match. */
1063			*case_folded = TRUE;
1064			if (case_fold(code) != case_fold(*un))
1065				return (-1);
1066		}
1067		cp += 2;
1068		un++;
1069	}
1070	return chksum;
1071}
1072
1073
1074/*
1075 * Collect Win95 filename Unicode chars into ucfn.  Stores the number
1076 * of characters in *unichars.  Returns the checksum, or -1 if impossible.
1077 */
1078int msdosfs_getunicodefn(struct winentry *wep, u_int16_t ucfn[WIN_MAXLEN], u_int16_t *unichars, int chksum)
1079{
1080	u_int8_t *cp;
1081	u_int16_t *np;
1082	u_int16_t *ep = &ucfn[WIN_MAXLEN];
1083	u_int16_t code;
1084	int i;
1085
1086	if ((wep->weCnt&WIN_CNT) > howmany(WIN_MAXLEN, WIN_CHARS)
1087	    || !(wep->weCnt&WIN_CNT))
1088	{
1089		return -1;
1090	}
1091
1092	/*
1093	 * First compare checksums
1094	 */
1095	if (wep->weCnt&WIN_LAST)
1096	{
1097		/*
1098		 * The "last" entry is the one we encounter first in the directory,
1099		 * so save off the checksum to compare against the other entries.
1100		 *
1101		 * The length we return here assumes the name is an exact multiple
1102		 * of WIN_CHARS long, and therefore has no terminator in the "last"
1103		 * entry.  If we in fact find a terminator, we'll adjust the length.
1104		 */
1105		chksum = wep->weChksum;
1106		*unichars = (wep->weCnt&WIN_CNT) * WIN_CHARS;
1107	}
1108	else if (chksum != wep->weChksum)
1109	{
1110		chksum = -1;
1111	}
1112	if (chksum == -1)
1113		return -1;
1114
1115	/*
1116	 * Find the offset within ucfn where the first character of this
1117	 * long name entry should get stored.
1118	 */
1119	np = &ucfn[((wep->weCnt&WIN_CNT) - 1) * WIN_CHARS];
1120
1121	/*
1122	 * Extract the characters from the three discontiguous parts.
1123	 *
1124	 * A maximum name (255 characters) will occupy 19 full entries
1125	 * with 13 characters per entry, and a partial (20th) entry
1126	 * containing 8 characters.  In that case, we'd better see the
1127	 * terminating 0-character in wePart2.  We only need to check
1128	 * for maximum name length while checking wePart2.
1129	 */
1130	for (cp = wep->wePart1, i = sizeof(wep->wePart1)/2; --i >= 0;)
1131	{
1132		code = (cp[1] << 8) | cp[0];
1133		switch (code) {
1134		case 0:
1135			*unichars = np - ucfn;
1136			return chksum;
1137		case '/':
1138			return -1;
1139		default:
1140			*np++ = code;
1141			break;
1142		}
1143		cp += 2;
1144	}
1145	for (cp = wep->wePart2, i = sizeof(wep->wePart2)/2; --i >= 0;)
1146	{
1147		code = (cp[1] << 8) | cp[0];
1148		switch (code) {
1149		case 0:
1150			*unichars = np - ucfn;
1151			return chksum;
1152		case '/':
1153			return -1;
1154		default:
1155			if (np >= ep)
1156			{
1157				/* The name is too long.  Return error. */
1158				return -1;
1159			}
1160			*np++ = code;
1161			break;
1162		}
1163		cp += 2;
1164	}
1165	for (cp = wep->wePart3, i = sizeof(wep->wePart3)/2; --i >= 0;)
1166	{
1167		code = (cp[1] << 8) | cp[0];
1168		switch (code) {
1169		case 0:
1170			*unichars = np - ucfn;
1171			return chksum;
1172		case '/':
1173			return -1;
1174		default:
1175			*np++ = code;
1176			break;
1177		}
1178		cp += 2;
1179	}
1180	return chksum;
1181}
1182
1183
1184/*
1185 * Compute the checksum of a DOS filename for Win95 use
1186 */
1187u_int8_t msdosfs_winChksum(u_int8_t *name)
1188{
1189	int i;
1190	u_int8_t s;
1191
1192	for (s = 0, i = SHORT_NAME_LEN; --i >= 0; s += *name++)
1193		s = (s << 7)|(s >> 1);
1194	return s;
1195}
1196
1197/*
1198 * Determine the number of slots necessary for Win95 names
1199 */
1200int msdosfs_winSlotCnt(const u_int16_t *un, int unlen)
1201{
1202#pragma unused (un)
1203	if (unlen > WIN_MAXLEN)
1204		return 0;
1205	return howmany(unlen, WIN_CHARS);
1206}
1207