xref: /linux/include/linux/unicode.h (revision 1dbb7422053fa0f4fb5f9acc3931617fc0091939)
1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_UNICODE_H
3 #define _LINUX_UNICODE_H
4 
5 #include <linux/init.h>
6 #include <linux/dcache.h>
7 
8 struct utf8data;
9 struct utf8data_table;
10 
11 #define UNICODE_MAJ_SHIFT		16
12 #define UNICODE_MIN_SHIFT		8
13 
14 #define UNICODE_AGE(MAJ, MIN, REV)			\
15 	(((unsigned int)(MAJ) << UNICODE_MAJ_SHIFT) |	\
16 	 ((unsigned int)(MIN) << UNICODE_MIN_SHIFT) |	\
17 	 ((unsigned int)(REV)))
18 
19 #define UTF8_LATEST        UNICODE_AGE(12, 1, 0)
20 
unicode_major(unsigned int age)21 static inline u8 unicode_major(unsigned int age)
22 {
23 	return (age >> UNICODE_MAJ_SHIFT) & 0xff;
24 }
25 
unicode_minor(unsigned int age)26 static inline u8 unicode_minor(unsigned int age)
27 {
28 	return (age >> UNICODE_MIN_SHIFT) & 0xff;
29 }
30 
unicode_rev(unsigned int age)31 static inline u8 unicode_rev(unsigned int age)
32 {
33 	return age & 0xff;
34 }
35 
36 /*
37  * Two normalization forms are supported:
38  * 1) NFDI
39  *   - Apply unicode normalization form NFD.
40  *   - Remove any Default_Ignorable_Code_Point.
41  * 2) NFDICF
42  *   - Apply unicode normalization form NFD.
43  *   - Remove any Default_Ignorable_Code_Point.
44  *   - Apply a full casefold (C + F).
45  */
46 enum utf8_normalization {
47 	UTF8_NFDI = 0,
48 	UTF8_NFDICF,
49 	UTF8_NMAX,
50 };
51 
52 struct unicode_map {
53 	unsigned int version;
54 	const struct utf8data *ntab[UTF8_NMAX];
55 	const struct utf8data_table *tables;
56 	struct rcu_head rcu;
57 };
58 
59 int utf8_validate(const struct unicode_map *um, const struct qstr *str);
60 
61 int utf8_strncmp(const struct unicode_map *um,
62 		 const struct qstr *s1, const struct qstr *s2);
63 
64 int utf8_strncasecmp(const struct unicode_map *um,
65 		 const struct qstr *s1, const struct qstr *s2);
66 int utf8_strncasecmp_folded(const struct unicode_map *um,
67 			    const struct qstr *cf,
68 			    const struct qstr *s1);
69 
70 int utf8_casefold(const struct unicode_map *um, const struct qstr *str,
71 		  unsigned char *dest, size_t dlen);
72 
73 int utf8_casefold_hash(const struct unicode_map *um, const void *salt,
74 		       struct qstr *str);
75 
76 struct unicode_map *utf8_load(unsigned int version);
77 void utf8_unload(struct unicode_map *um);
78 
79 int utf8_parse_version(char *version);
80 
81 #endif /* _LINUX_UNICODE_H */
82