1 // SPDX-License-Identifier: GPL-2.0+
3 * EFI Unicode collation protocol
5 * Copyright (c) 2018 Heinrich Schuchardt <xypron.glpk@gmx.de>
12 #include <efi_loader.h>
14 /* Characters that may not be used in FAT 8.3 file names */
15 static const char illegal[] = "+,<=>:;\"/\\|?*[]\x7f";
18 * EDK2 assumes codepage 1250 when creating FAT 8.3 file names.
19 * Linux defaults to codepage 437 for FAT 8.3 file names.
21 #if CONFIG_FAT_DEFAULT_CODEPAGE == 1250
22 /* Unicode code points for code page 1250 characters 0x80 - 0xff */
23 static const u16 codepage[] = CP1250;
25 /* Unicode code points for code page 437 characters 0x80 - 0xff */
26 static const u16 codepage[] = CP437;
29 /* GUID of the EFI_UNICODE_COLLATION_PROTOCOL2 */
30 const efi_guid_t efi_guid_unicode_collation_protocol2 =
31 EFI_UNICODE_COLLATION_PROTOCOL2_GUID;
34 * efi_stri_coll() - compare utf-16 strings case-insenitively
36 * @this: unicode collation protocol instance
40 * This function implements the StriColl() service of the
41 * EFI_UNICODE_COLLATION_PROTOCOL.
43 * See the Unified Extensible Firmware Interface (UEFI) specification for
47 * The implementation does not follow the Unicode collation algorithm.
48 * For ASCII characters it results in the same sort order as EDK2.
49 * We could use table UNICODE_CAPITALIZATION_TABLE for better results.
51 * Return: 0: s1 == s2, > 0: s1 > s2, < 0: s1 < s2
53 static efi_intn_t EFIAPI efi_stri_coll(
54 struct efi_unicode_collation_protocol *this, u16 *s1, u16 *s2)
59 EFI_ENTRY("%p, %ls, %ls", this, s1, s2);
60 for (; *s1 | *s2; ++s1, ++s2) {
61 c1 = utf_to_upper(*s1);
62 c2 = utf_to_upper(*s2);
72 EFI_EXIT(EFI_SUCCESS);
77 * next_lower() - get next codepoint converted to lower case
79 * @string: pointer to u16 string, on return advanced by one codepoint
80 * Return: first codepoint of string converted to lower case
82 static s32 next_lower(const u16 **string)
84 return utf_to_lower(utf16_get(string));
88 * metai_match() - compare utf-16 string with a pattern string case-insenitively
90 * @string: string to compare
91 * @pattern: pattern string
93 * The pattern string may use these:
94 * - * matches >= 0 characters
95 * - ? matches 1 character
96 * - [<char1><char2>...<charN>] match any character in the set
97 * - [<char1>-<char2>] matches any character in the range
99 * This function is called my efi_metai_match().
101 * For '*' pattern searches this function calls itself recursively.
102 * Performance-wise this is suboptimal, especially for multiple '*' wildcards.
103 * But it results in simple code.
105 * Return: true if the string is matched.
107 static bool metai_match(const u16 *string, const u16 *pattern)
111 for (; *string && *pattern;) {
112 const u16 *string_old = string;
114 s = next_lower(&string);
115 p = next_lower(&pattern);
119 /* Match 0 or more characters */
120 for (;; s = next_lower(&string)) {
121 if (metai_match(string_old, pattern))
128 /* Match any one character */
131 /* Match any character in the set */
132 p = next_lower(&pattern);
137 p = next_lower(&pattern);
140 p = next_lower(&pattern);
141 if (s < first || s > p)
143 p = next_lower(&pattern);
152 for (; p && p != ']';
153 p = next_lower(&pattern)) {
157 if (!hit || p != ']')
162 /* Match one character */
167 if (!*pattern && !*string)
173 * efi_metai_match() - compare utf-16 string with a pattern string
176 * @this: unicode collation protocol instance
177 * @s: string to compare
180 * The pattern string may use these:
181 * - * matches >= 0 characters
182 * - ? matches 1 character
183 * - [<char1><char2>...<charN>] match any character in the set
184 * - [<char1>-<char2>] matches any character in the range
186 * This function implements the MetaMatch() service of the
187 * EFI_UNICODE_COLLATION_PROTOCOL.
189 * Return: true if the string is matched.
191 static bool EFIAPI efi_metai_match(struct efi_unicode_collation_protocol *this,
192 const u16 *string, const u16 *pattern)
196 EFI_ENTRY("%p, %ls, %ls", this, string, pattern);
197 ret = metai_match(string, pattern);
198 EFI_EXIT(EFI_SUCCESS);
203 * efi_str_lwr() - convert to lower case
205 * @this: unicode collation protocol instance
206 * @string: string to convert
209 * The conversion is done in place. As long as upper and lower letters use the
210 * same number of words this does not pose a problem.
212 * This function implements the StrLwr() service of the
213 * EFI_UNICODE_COLLATION_PROTOCOL.
215 static void EFIAPI efi_str_lwr(struct efi_unicode_collation_protocol *this,
218 EFI_ENTRY("%p, %ls", this, string);
219 for (; *string; ++string)
220 *string = utf_to_lower(*string);
221 EFI_EXIT(EFI_SUCCESS);
225 * efi_str_upr() - convert to upper case
227 * @this: unicode collation protocol instance
228 * @string: string to convert
231 * The conversion is done in place. As long as upper and lower letters use the
232 * same number of words this does not pose a problem.
234 * This function implements the StrUpr() service of the
235 * EFI_UNICODE_COLLATION_PROTOCOL.
237 static void EFIAPI efi_str_upr(struct efi_unicode_collation_protocol *this,
240 EFI_ENTRY("%p, %ls", this, string);
241 for (; *string; ++string)
242 *string = utf_to_upper(*string);
243 EFI_EXIT(EFI_SUCCESS);
247 * efi_fat_to_str() - convert an 8.3 file name from an OEM codepage to Unicode
249 * @this: unicode collation protocol instance
250 * @fat_size: size of the string to convert
251 * @fat: string to convert
252 * @string: converted string
254 * This function implements the FatToStr() service of the
255 * EFI_UNICODE_COLLATION_PROTOCOL.
257 static void EFIAPI efi_fat_to_str(struct efi_unicode_collation_protocol *this,
258 efi_uintn_t fat_size, char *fat, u16 *string)
263 EFI_ENTRY("%p, %zu, %s, %p", this, fat_size, fat, string);
264 for (i = 0; i < fat_size; ++i) {
265 c = (unsigned char)fat[i];
267 c = codepage[i - 0x80];
273 EFI_EXIT(EFI_SUCCESS);
277 * efi_fat_to_str() - convert a utf-16 string to legal characters for a FAT
278 * file name in an OEM code page
280 * @this: unicode collation protocol instance
281 * @string: Unicode string to convert
282 * @fat_size: size of the target buffer
283 * @fat: converted string
285 * This function implements the StrToFat() service of the
286 * EFI_UNICODE_COLLATION_PROTOCOL.
288 * Return: true if an illegal character was substituted by '_'.
290 static bool EFIAPI efi_str_to_fat(struct efi_unicode_collation_protocol *this,
291 const u16 *string, efi_uintn_t fat_size,
298 EFI_ENTRY("%p, %ls, %zu, %p", this, string, fat_size, fat);
299 for (i = 0; i < fat_size;) {
300 c = utf16_get(&string);
302 /* Ignore period and space */
313 /* Look for codepage translation */
314 for (j = 0; j < 0x80; ++j) {
315 if (c == codepage[j]) {
324 } else if (c && (c < 0x20 || strchr(illegal, c))) {
334 EFI_EXIT(EFI_SUCCESS);
338 const struct efi_unicode_collation_protocol efi_unicode_collation_protocol2 = {
339 .stri_coll = efi_stri_coll,
340 .metai_match = efi_metai_match,
341 .str_lwr = efi_str_lwr,
342 .str_upr = efi_str_upr,
343 .fat_to_str = efi_fat_to_str,
344 .str_to_fat = efi_str_to_fat,
345 .supported_languages = "en",
349 * In EFI 1.10 a version of the Unicode collation protocol using ISO 639-2
350 * language codes existed. This protocol is not part of the UEFI specification
351 * any longer. Unfortunately it is required to run the UEFI Self Certification
352 * Test (SCT) II, version 2.6, 2017. So we implement it here for the sole
353 * purpose of running the SCT. It can be removed when a compliant SCT is
356 #if CONFIG_IS_ENABLED(EFI_UNICODE_COLLATION_PROTOCOL)
358 /* GUID of the EFI_UNICODE_COLLATION_PROTOCOL */
359 const efi_guid_t efi_guid_unicode_collation_protocol =
360 EFI_UNICODE_COLLATION_PROTOCOL_GUID;
362 const struct efi_unicode_collation_protocol efi_unicode_collation_protocol = {
363 .stri_coll = efi_stri_coll,
364 .metai_match = efi_metai_match,
365 .str_lwr = efi_str_lwr,
366 .str_upr = efi_str_upr,
367 .fat_to_str = efi_fat_to_str,
368 .str_to_fat = efi_str_to_fat,
369 /* ISO 639-2 language code */
370 .supported_languages = "eng",