xref: /illumos-gate/usr/src/uts/common/sys/fs/pc_dir.h (revision 2d6eb4a5)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*
22  * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 #ifndef	_SYS_FS_PC_DIR_H
27 #define	_SYS_FS_PC_DIR_H
28 
29 #include <sys/dirent.h>
30 
31 #ifdef	__cplusplus
32 extern "C" {
33 #endif
34 
35 #define	PCFNAMESIZE	8
36 #define	PCFEXTSIZE	3
37 #define	PCMAXNAMLEN	255
38 #define	PCMAXNAM_UTF16	(256 * sizeof (uint16_t))	/* for UTF-16 */
39 #define	PCLFNCHUNKSIZE	13
40 
41 struct pctime {
42 	ushort_t pct_time;		/* hh:mm:ss (little endian) */
43 	ushort_t pct_date;		/* yy:mm:dd (little endian) */
44 };
45 
46 /*
47  * Shifts and masks for time and date fields, in host byte order.
48  */
49 #define	SECSHIFT	0
50 #define	SECMASK		0x1F
51 #define	MINSHIFT	5
52 #define	MINMASK		0x3F
53 #define	HOURSHIFT	11
54 #define	HOURMASK	0x1F
55 
56 #define	DAYSHIFT	0
57 #define	DAYMASK		0x1F
58 #define	MONSHIFT	5
59 #define	MONMASK		0x0F
60 #define	YEARSHIFT	9
61 #define	YEARMASK	0x7F
62 
63 struct pcdir {
64 	char pcd_filename[PCFNAMESIZE];	/* file name */
65 	char pcd_ext[PCFEXTSIZE];	/* file extension */
66 	uchar_t	pcd_attr;		/* file attributes */
67 	uchar_t	pcd_ntattr;		/* reserved for NT attributes */
68 	uchar_t	pcd_crtime_msec;	/* milliseconds after the minute */
69 	struct pctime pcd_crtime;	/* creation time/date */
70 	ushort_t pcd_ladate;		/* last-access date */
71 	union {
72 		uint16_t pcd_eattr;	/* OS/2 extended attribute */
73 		pc_cluster16_t pcd_scluster_hi;
74 	} un;
75 	struct pctime pcd_mtime;	/* last modified time/date */
76 	pc_cluster16_t pcd_scluster_lo;	/* starting cluster (little endian) */
77 	uint_t	pcd_size;		/* file size (little endian) */
78 };
79 
80 #ifdef	__cplusplus
81 }
82 #endif
83 
84 #include <sys/fs/pc_node.h>
85 
86 #ifdef	__cplusplus
87 extern "C" {
88 #endif
89 
90 /*
91  * Long filename support (introduced by Windows 95) is an interesting
92  * exercise in compatibility. Basically, it is now no longer the case
93  * that an entry in a directory (as described by the 'pcdir' structure above)
94  * contains the entire name of a file. Now, a directory entry can consist of
95  * a long filename component (a series of 'pcdir'-like structures) followed
96  * by a short filename (the old form). Each long filename component is
97  * identified by having it's Read-Only, Hidden, System, and Volume Label
98  * attributes set.  Each can store 13 Unicode characters (16-bits, of
99  * which we only look at the lower 8 for now), broken into (gak) three
100  * sections of the entry (since that's the way the available bits fall out).
101  * In addition, each long filename entry has a sequence number (starting
102  * from 1). The first entry has bit 7 (0x40) set in the sequence number, and
103  * has the maximum value in the sequence. This may seem a bit backwards, and
104  * it gets worse: the first entry stores the last component of
105  * the name. So the directory entry for a file named
106  * "This is a very long filename indeed" might look like this:
107  *
108  * Offset  Sequence      Component         Attributes    Cluster    Size
109  *    0      0x43       "me indeed"          RSHV           0         0
110  *   32      0x02       "y long filena"      RSHV           0         0
111  *   64      0x01       "This is a ver"      RSHV           0         0
112  *   96      ----       "THISIS~1.TXT"       <whatever>  2122       110
113  *
114  * The last entry is for the short filename, which stores actual information
115  * about the file (like type, permissions, cluster, and size). The short name
116  * is also used by non-long-filename aware applications (like Windows 3.X and
117  * DOS). This name is generated by Windows 95 (and now Solaris) from the
118  * long name, and must (of course) be unique within the directory.
119  * Solaris continues to this entry to actually identify the file and its
120  * attributes (filenames only really matter when names are used, like at
121  * lookup/readdir/remove/create/rename time - for general access to the file
122  * they aren't used).
123  *
124  * Long filenames can also be broken by applications that don't
125  * understand them (for example, a Solaris 2.5.1 user could rename
126  * "THISIS~1.TXT" to "test.exe"). This can be detected because each long
127  * filename component has a checksum which is based on the short filename.
128  * After reading the long filename entry, if the checksum doesn't match the
129  * short name that follows, we simply ignore it and use the short name.
130  *
131  * One subtle thing - though long file names are case-sensitive,
132  * searches for them are not.
133  *
134  * Another _very_ subtle thing. The number of characters in the
135  * last long filename chunk (the first entry, with the 0x40 bit set) is
136  * either all the characters (if there is no null, '\0'), or all the
137  * characters up to the null. _However_, if the remaining characters are
138  * null, Norton Disk Doctor and Microsoft ScanDisk will claim
139  * that the filename entry is damaged. The remaining bytes must actually
140  * contain 0xff (discovered with Disk Doctor).
141  *
142  * Some information about long filename support can be found in the
143  * book "Inside Windows 95" by Adrian King.
144  */
145 
146 /*
147  * The number of bytes in each section of the name in a long filename
148  * entry. This is _bytes_, not characters: each character is actually
149  * 16 bits.
150  */
151 #define	PCLF_FIRSTNAMESIZE	10
152 #define	PCLF_SECONDNAMESIZE	12
153 #define	PCLF_THIRDNAMESIZE	4
154 
155 /*
156  * A long filename entry. It must match the 'pcdir' structure in size,
157  * and pcdl_attr must overlap pcd_attr.
158  */
159 struct pcdir_lfn {
160 	uchar_t pcdl_ordinal;	/* lfn order. First is 01, next 02, */
161 				/* last has bit 7 (0x40) set */
162 	uchar_t pcdl_firstfilename[PCLF_FIRSTNAMESIZE];
163 	uchar_t pcdl_attr;
164 	uchar_t pcdl_type;	/* type - always contains 0 for an LFN entry */
165 	uchar_t pcdl_checksum;	/* checksum to validate the LFN entry - */
166 				/* based on the short name */
167 	uchar_t pcdl_secondfilename[PCLF_SECONDNAMESIZE];
168 	pc_cluster16_t pcd_scluster;	/* (not used, always 0) */
169 	uchar_t pcdl_thirdfilename[PCLF_THIRDNAMESIZE];
170 };
171 
172 /*
173  * FAT LFN entries are consecutively numbered downwards, and the last
174  * entry of a LFN chain will have the 0x40 'termination' marker logically
175  * or'ed in. The entry immediately preceeding the short name has number 1,
176  * consecutively increasing. Since the filename length limit on FAT is
177  * 255 unicode characters and every LFN entry contributes 13 characters,
178  * the maximum sequence number is 255/13 + 1 == 20.
179  */
180 #define	PCDL_IS_LAST_LFN(x) ((x->pcdl_ordinal) & 0x40)
181 #define	PCDL_LFN_BITS (PCA_RDONLY | PCA_HIDDEN | PCA_SYSTEM | PCA_LABEL)
182 #define	PCDL_LFN_MASK (PCDL_LFN_BITS | PCA_DIR | PCA_ARCH)
183 #define	PCDL_LFN_VALID_ORD(x)						\
184 	(((((struct pcdir_lfn *)(x))->pcdl_ordinal & ~0x40) > 0) &&	\
185 	((((struct pcdir_lfn *)(x))->pcdl_ordinal & ~0x40) <= 20))
186 #define	PCDL_IS_LFN(x)							\
187 	(enable_long_filenames &&					\
188 	(((x)->pcd_attr & PCDL_LFN_MASK) == PCDL_LFN_BITS) &&		\
189 	PCDL_LFN_VALID_ORD((x)))
190 
191 /*
192  * The first char of the file name has special meaning as follows:
193  */
194 #define	PCD_UNUSED	((char)0x00)	/* entry has never been used */
195 #define	PCD_ERASED	((char)0xE5)	/* entry was erased */
196 
197 /*
198  * File attributes.
199  */
200 #define	PCA_RDONLY	0x01	/* file is read only */
201 #define	PCA_HIDDEN	0x02	/* file is hidden */
202 #define	PCA_SYSTEM	0x04	/* system file */
203 #define	PCA_LABEL	0x08	/* entry contains the volume label */
204 #define	PCA_DIR		0x10	/* subdirectory */
205 #define	PCA_ARCH	0x20	/* file has been modified since last backup */
206 
207 /*
208  * Avoid hidden files unless the private variable is set.
209  * Always avoid the label.
210  */
211 #define	PCA_IS_HIDDEN(fsp, attr) \
212 	((((attr) & PCA_LABEL) == PCA_LABEL) || \
213 	((((fsp)->pcfs_flags & PCFS_HIDDEN) == 0) && \
214 	    ((attr) & (PCA_HIDDEN | PCA_SYSTEM))))
215 
216 #define	PC_NAME_IS_DOT(namep) \
217 	(((namep)[0] == '.') && ((namep)[1] == '\0'))
218 #define	PC_NAME_IS_DOTDOT(namep) \
219 	(((namep)[0] == '.') && ((namep)[1] == '.') && ((namep)[2] == '\0'))
220 #define	PC_SHORTNAME_IS_DOT(namep) \
221 	(((namep)[0] == '.') && ((namep)[1] == ' '))
222 #define	PC_SHORTNAME_IS_DOTDOT(namep) \
223 	(((namep)[0] == '.') && ((namep)[1] == '.') && ((namep)[2] == ' '))
224 /*
225  * slot structure is used by the directory search routine to return
226  * the results of the search.  If the search is successful sl_blkno and
227  * sl_offset reflect the disk address of the entry and sl_ep points to
228  * the actual entry data in buffer sl_bp. sl_flags is set to whether the
229  * entry is dot or dotdot. If the search is unsuccessful sl_blkno and
230  * sl_offset points to an empty directory slot if there are any. Otherwise
231  * it is set to -1.
232  */
233 struct pcslot {
234 	enum {SL_NONE, SL_FOUND, SL_EXTEND} sl_status;	/* slot status */
235 	daddr_t		sl_blkno;	/* disk block number which has entry */
236 	int		sl_offset;	/* offset of entry within block */
237 	struct buf	*sl_bp;		/* buffer containing entry data */
238 	struct pcdir	*sl_ep;		/* pointer to entry data */
239 	int		sl_flags;	/* flags (see below) */
240 };
241 #define	SL_DOT		1	/* entry point to self */
242 #define	SL_DOTDOT	2	/* entry points to parent */
243 
244 /*
245  * A pcfs directory entry. Directory entries are actually variable
246  * length, but this is the maximum size.
247  *
248  * This _must_ match a dirent64 structure in format.
249  * d_name is 512 bytes long to accomodate 256 UTF-16 characters.
250  */
251 struct pc_dirent {
252 	ino64_t		d_ino;		/* "inode number" of entry */
253 	off64_t		d_off;		/* offset of disk directory entry */
254 	unsigned short	d_reclen;	/* length of this record */
255 	char		d_name[PCMAXNAM_UTF16];
256 };
257 
258 /*
259  * Check FAT 8.3 filename characters for validity.
260  * Lacking a kernel iconv, codepage support for short filenames
261  * is not provided.
262  * Short names must be uppercase ASCII (no support for MSDOS
263  * codepages right now, sorry) and may not contain any of
264  * *+=|\[];:",<>.?/ which are explicitly forbidden by the
265  * FAT specifications.
266  */
267 #define	pc_invalchar(c)						\
268 	(((c) >= 'a' && (c) <= 'z') ||				\
269 	(c) == '"' || (c) == '*' || (c) == '+' || (c) == ',' || \
270 	(c) == '.' || (c) == '/' || (c) == ':' || (c) == ';' || \
271 	(c) == '<' || (c) == '=' || (c) == '>' || (c) == '?' || \
272 	(c) == '[' || (c) == '|' || (c) == ']' || (c) == '\\')
273 
274 #define	pc_validchar(c)	(((c) >= ' ' && !((c) & ~0177)) && !pc_invalchar(c))
275 
276 
277 #ifdef _KERNEL
278 
279 /*
280  * macros for converting ASCII to/from upper or lower case.
281  * users may give and get names in lower case, but they are stored on the
282  * disk in upper case to be PCDOS compatible.
283  * These would better come from some shared source in <sys/...> but
284  * there is no such place yet.
285  */
286 #define	toupper(C)	(((C) >= 'a' && (C) <= 'z') ? (C) - 'a' + 'A' : (C))
287 #define	tolower(C)	(((C) >= 'A' && (C) <= 'Z') ? (C) - 'A' + 'a' : (C))
288 
289 extern int pc_tvtopct(timestruc_t *, struct pctime *);	/* timeval to pctime */
290 extern void pc_pcttotv(struct pctime *, int64_t *);	/* pctime to timeval */
291 extern int pc_valid_lfn_char(char);		/* valid long filename ch */
292 
293 extern int pc_read_long_fn(struct vnode *, struct uio *,
294     struct pc_dirent *, struct pcdir **, offset_t *, struct buf **);
295 extern int pc_read_short_fn(struct vnode *, struct uio *,
296     struct pc_dirent *, struct pcdir **, offset_t *, struct buf **);
297 extern int pc_match_long_fn(struct pcnode *, char *, struct pcdir **,
298     struct pcslot *, offset_t *);
299 extern int pc_match_short_fn(struct pcnode *, char *,
300     struct pcdir **, struct pcslot *, offset_t *);
301 extern uchar_t pc_checksum_long_fn(char *, char *);
302 extern void set_long_fn_chunk(struct pcdir_lfn *, char *, int);
303 extern int pc_valid_long_fn(char *, int);
304 extern int pc_extract_long_fn(struct pcnode *, char *,
305     struct pcdir **, offset_t *offset, struct buf **);
306 extern int pc_fname_ext_to_name(char *, char *, char *, int);
307 
308 extern pc_cluster32_t pc_getstartcluster(struct pcfs *, struct pcdir *);
309 extern void pc_setstartcluster(struct pcfs *, struct pcdir *, pc_cluster32_t);
310 
311 /*
312  * Private tunables
313  */
314 
315 /*
316  * Use long filenames (Windows 95). Disabling this causes pcfs
317  * to not recognize long filenames at all, which may cause it to
318  * break associations between the short and long names. This is likely
319  * to leave unused long filename entries  in directories (which may make
320  * apparently empty directories unremovable), and would require a fsck_pcfs
321  * to find and fix (or a Windows utility like Norton Disk Doctor or
322  * Microsoft ScanDisk).
323  */
324 extern int enable_long_filenames;	/* default: on */
325 
326 #endif
327 
328 #ifdef	__cplusplus
329 }
330 #endif
331 
332 #endif	/* _SYS_FS_PC_DIR_H */
333