| 1 | // SPDX-License-Identifier: GPL-2.0 |
| 2 | /* |
| 3 | * linux/fs/hfsplus/unicode.c |
| 4 | * |
| 5 | * Copyright (C) 2001 |
| 6 | * Brad Boyer (flar@allandria.com) |
| 7 | * (C) 2003 Ardis Technologies <roman@ardistech.com> |
| 8 | * |
| 9 | * Handler routines for unicode strings |
| 10 | */ |
| 11 | |
| 12 | #include <linux/types.h> |
| 13 | #include <linux/nls.h> |
| 14 | |
| 15 | #include <kunit/visibility.h> |
| 16 | |
| 17 | #include "hfsplus_fs.h" |
| 18 | #include "hfsplus_raw.h" |
| 19 | |
| 20 | /* Fold the case of a unicode char, given the 16 bit value */ |
| 21 | /* Returns folded char, or 0 if ignorable */ |
| 22 | static inline u16 case_fold(u16 c) |
| 23 | { |
| 24 | u16 tmp; |
| 25 | |
| 26 | tmp = hfsplus_case_fold_table[c >> 8]; |
| 27 | if (tmp) |
| 28 | tmp = hfsplus_case_fold_table[tmp + (c & 0xff)]; |
| 29 | else |
| 30 | tmp = c; |
| 31 | return tmp; |
| 32 | } |
| 33 | |
| 34 | /* Compare unicode strings, return values like normal strcmp */ |
| 35 | int hfsplus_strcasecmp(const struct hfsplus_unistr *s1, |
| 36 | const struct hfsplus_unistr *s2) |
| 37 | { |
| 38 | u16 len1, len2, c1, c2; |
| 39 | const hfsplus_unichr *p1, *p2; |
| 40 | |
| 41 | len1 = be16_to_cpu(s1->length); |
| 42 | len2 = be16_to_cpu(s2->length); |
| 43 | p1 = s1->unicode; |
| 44 | p2 = s2->unicode; |
| 45 | |
| 46 | if (len1 > HFSPLUS_MAX_STRLEN) { |
| 47 | len1 = HFSPLUS_MAX_STRLEN; |
| 48 | pr_err("invalid length %u has been corrected to %d\n" , |
| 49 | be16_to_cpu(s1->length), len1); |
| 50 | } |
| 51 | |
| 52 | if (len2 > HFSPLUS_MAX_STRLEN) { |
| 53 | len2 = HFSPLUS_MAX_STRLEN; |
| 54 | pr_err("invalid length %u has been corrected to %d\n" , |
| 55 | be16_to_cpu(s2->length), len2); |
| 56 | } |
| 57 | |
| 58 | while (1) { |
| 59 | c1 = c2 = 0; |
| 60 | |
| 61 | while (len1 && !c1) { |
| 62 | c1 = case_fold(be16_to_cpu(*p1)); |
| 63 | p1++; |
| 64 | len1--; |
| 65 | } |
| 66 | while (len2 && !c2) { |
| 67 | c2 = case_fold(be16_to_cpu(*p2)); |
| 68 | p2++; |
| 69 | len2--; |
| 70 | } |
| 71 | |
| 72 | if (c1 != c2) |
| 73 | return (c1 < c2) ? -1 : 1; |
| 74 | if (!c1 && !c2) |
| 75 | return 0; |
| 76 | } |
| 77 | } |
| 78 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_strcasecmp); |
| 79 | |
| 80 | /* Compare names as a sequence of 16-bit unsigned integers */ |
| 81 | int hfsplus_strcmp(const struct hfsplus_unistr *s1, |
| 82 | const struct hfsplus_unistr *s2) |
| 83 | { |
| 84 | u16 len1, len2, c1, c2; |
| 85 | const hfsplus_unichr *p1, *p2; |
| 86 | int len; |
| 87 | |
| 88 | len1 = be16_to_cpu(s1->length); |
| 89 | len2 = be16_to_cpu(s2->length); |
| 90 | p1 = s1->unicode; |
| 91 | p2 = s2->unicode; |
| 92 | |
| 93 | if (len1 > HFSPLUS_MAX_STRLEN) { |
| 94 | len1 = HFSPLUS_MAX_STRLEN; |
| 95 | pr_err("invalid length %u has been corrected to %d\n" , |
| 96 | be16_to_cpu(s1->length), len1); |
| 97 | } |
| 98 | |
| 99 | if (len2 > HFSPLUS_MAX_STRLEN) { |
| 100 | len2 = HFSPLUS_MAX_STRLEN; |
| 101 | pr_err("invalid length %u has been corrected to %d\n" , |
| 102 | be16_to_cpu(s2->length), len2); |
| 103 | } |
| 104 | |
| 105 | for (len = min(len1, len2); len > 0; len--) { |
| 106 | c1 = be16_to_cpu(*p1); |
| 107 | c2 = be16_to_cpu(*p2); |
| 108 | if (c1 != c2) |
| 109 | return c1 < c2 ? -1 : 1; |
| 110 | p1++; |
| 111 | p2++; |
| 112 | } |
| 113 | |
| 114 | return len1 < len2 ? -1 : |
| 115 | len1 > len2 ? 1 : 0; |
| 116 | } |
| 117 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_strcmp); |
| 118 | |
| 119 | #define Hangul_SBase 0xac00 |
| 120 | #define Hangul_LBase 0x1100 |
| 121 | #define Hangul_VBase 0x1161 |
| 122 | #define Hangul_TBase 0x11a7 |
| 123 | #define Hangul_SCount 11172 |
| 124 | #define Hangul_LCount 19 |
| 125 | #define Hangul_VCount 21 |
| 126 | #define Hangul_TCount 28 |
| 127 | #define Hangul_NCount (Hangul_VCount * Hangul_TCount) |
| 128 | |
| 129 | |
| 130 | static u16 *hfsplus_compose_lookup(u16 *p, u16 cc) |
| 131 | { |
| 132 | int i, s, e; |
| 133 | |
| 134 | s = 1; |
| 135 | e = p[1]; |
| 136 | if (!e || cc < p[s * 2] || cc > p[e * 2]) |
| 137 | return NULL; |
| 138 | do { |
| 139 | i = (s + e) / 2; |
| 140 | if (cc > p[i * 2]) |
| 141 | s = i + 1; |
| 142 | else if (cc < p[i * 2]) |
| 143 | e = i - 1; |
| 144 | else |
| 145 | return hfsplus_compose_table + p[i * 2 + 1]; |
| 146 | } while (s <= e); |
| 147 | return NULL; |
| 148 | } |
| 149 | |
| 150 | static int hfsplus_uni2asc(struct super_block *sb, |
| 151 | const struct hfsplus_unistr *ustr, |
| 152 | int max_len, char *astr, int *len_p) |
| 153 | { |
| 154 | const hfsplus_unichr *ip; |
| 155 | struct nls_table *nls = HFSPLUS_SB(sb)->nls; |
| 156 | u8 *op; |
| 157 | u16 cc, c0, c1; |
| 158 | u16 *ce1, *ce2; |
| 159 | int i, len, ustrlen, res, compose; |
| 160 | |
| 161 | op = astr; |
| 162 | ip = ustr->unicode; |
| 163 | |
| 164 | ustrlen = be16_to_cpu(ustr->length); |
| 165 | if (ustrlen > max_len) { |
| 166 | ustrlen = max_len; |
| 167 | pr_err("invalid length %u has been corrected to %d\n" , |
| 168 | be16_to_cpu(ustr->length), ustrlen); |
| 169 | } |
| 170 | |
| 171 | len = *len_p; |
| 172 | ce1 = NULL; |
| 173 | compose = !test_bit(HFSPLUS_SB_NODECOMPOSE, &HFSPLUS_SB(sb)->flags); |
| 174 | |
| 175 | while (ustrlen > 0) { |
| 176 | c0 = be16_to_cpu(*ip++); |
| 177 | ustrlen--; |
| 178 | /* search for single decomposed char */ |
| 179 | if (likely(compose)) |
| 180 | ce1 = hfsplus_compose_lookup(p: hfsplus_compose_table, cc: c0); |
| 181 | if (ce1) |
| 182 | cc = ce1[0]; |
| 183 | else |
| 184 | cc = 0; |
| 185 | if (cc) { |
| 186 | /* start of a possibly decomposed Hangul char */ |
| 187 | if (cc != 0xffff) |
| 188 | goto done; |
| 189 | if (!ustrlen) |
| 190 | goto same; |
| 191 | c1 = be16_to_cpu(*ip) - Hangul_VBase; |
| 192 | if (c1 < Hangul_VCount) { |
| 193 | /* compose the Hangul char */ |
| 194 | cc = (c0 - Hangul_LBase) * Hangul_VCount; |
| 195 | cc = (cc + c1) * Hangul_TCount; |
| 196 | cc += Hangul_SBase; |
| 197 | ip++; |
| 198 | ustrlen--; |
| 199 | if (!ustrlen) |
| 200 | goto done; |
| 201 | c1 = be16_to_cpu(*ip) - Hangul_TBase; |
| 202 | if (c1 > 0 && c1 < Hangul_TCount) { |
| 203 | cc += c1; |
| 204 | ip++; |
| 205 | ustrlen--; |
| 206 | } |
| 207 | goto done; |
| 208 | } |
| 209 | } |
| 210 | while (1) { |
| 211 | /* main loop for common case of not composed chars */ |
| 212 | if (!ustrlen) |
| 213 | goto same; |
| 214 | c1 = be16_to_cpu(*ip); |
| 215 | if (likely(compose)) |
| 216 | ce1 = hfsplus_compose_lookup( |
| 217 | p: hfsplus_compose_table, cc: c1); |
| 218 | if (ce1) |
| 219 | break; |
| 220 | switch (c0) { |
| 221 | case 0: |
| 222 | c0 = 0x2400; |
| 223 | break; |
| 224 | case '/': |
| 225 | c0 = ':'; |
| 226 | break; |
| 227 | } |
| 228 | res = nls->uni2char(c0, op, len); |
| 229 | if (res < 0) { |
| 230 | if (res == -ENAMETOOLONG) |
| 231 | goto out; |
| 232 | *op = '?'; |
| 233 | res = 1; |
| 234 | } |
| 235 | op += res; |
| 236 | len -= res; |
| 237 | c0 = c1; |
| 238 | ip++; |
| 239 | ustrlen--; |
| 240 | } |
| 241 | ce2 = hfsplus_compose_lookup(p: ce1, cc: c0); |
| 242 | if (ce2) { |
| 243 | i = 1; |
| 244 | while (i < ustrlen) { |
| 245 | ce1 = hfsplus_compose_lookup(p: ce2, |
| 246 | be16_to_cpu(ip[i])); |
| 247 | if (!ce1) |
| 248 | break; |
| 249 | i++; |
| 250 | ce2 = ce1; |
| 251 | } |
| 252 | cc = ce2[0]; |
| 253 | if (cc) { |
| 254 | ip += i; |
| 255 | ustrlen -= i; |
| 256 | goto done; |
| 257 | } |
| 258 | } |
| 259 | same: |
| 260 | switch (c0) { |
| 261 | case 0: |
| 262 | cc = 0x2400; |
| 263 | break; |
| 264 | case '/': |
| 265 | cc = ':'; |
| 266 | break; |
| 267 | default: |
| 268 | cc = c0; |
| 269 | } |
| 270 | done: |
| 271 | res = nls->uni2char(cc, op, len); |
| 272 | if (res < 0) { |
| 273 | if (res == -ENAMETOOLONG) |
| 274 | goto out; |
| 275 | *op = '?'; |
| 276 | res = 1; |
| 277 | } |
| 278 | op += res; |
| 279 | len -= res; |
| 280 | } |
| 281 | res = 0; |
| 282 | out: |
| 283 | *len_p = (char *)op - astr; |
| 284 | return res; |
| 285 | } |
| 286 | |
| 287 | inline int hfsplus_uni2asc_str(struct super_block *sb, |
| 288 | const struct hfsplus_unistr *ustr, char *astr, |
| 289 | int *len_p) |
| 290 | { |
| 291 | return hfsplus_uni2asc(sb, ustr, HFSPLUS_MAX_STRLEN, astr, len_p); |
| 292 | } |
| 293 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_uni2asc_str); |
| 294 | |
| 295 | inline int hfsplus_uni2asc_xattr_str(struct super_block *sb, |
| 296 | const struct hfsplus_attr_unistr *ustr, |
| 297 | char *astr, int *len_p) |
| 298 | { |
| 299 | return hfsplus_uni2asc(sb, ustr: (const struct hfsplus_unistr *)ustr, |
| 300 | HFSPLUS_ATTR_MAX_STRLEN, astr, len_p); |
| 301 | } |
| 302 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_uni2asc_xattr_str); |
| 303 | |
| 304 | /* |
| 305 | * Convert one or more ASCII characters into a single unicode character. |
| 306 | * Returns the number of ASCII characters corresponding to the unicode char. |
| 307 | */ |
| 308 | static inline int asc2unichar(struct super_block *sb, const char *astr, int len, |
| 309 | wchar_t *uc) |
| 310 | { |
| 311 | int size = HFSPLUS_SB(sb)->nls->char2uni(astr, len, uc); |
| 312 | if (size <= 0) { |
| 313 | *uc = '?'; |
| 314 | size = 1; |
| 315 | } |
| 316 | switch (*uc) { |
| 317 | case 0x2400: |
| 318 | *uc = 0; |
| 319 | break; |
| 320 | case ':': |
| 321 | *uc = '/'; |
| 322 | break; |
| 323 | } |
| 324 | return size; |
| 325 | } |
| 326 | |
| 327 | /* Decomposes a non-Hangul unicode character. */ |
| 328 | static u16 *hfsplus_decompose_nonhangul(wchar_t uc, int *size) |
| 329 | { |
| 330 | int off; |
| 331 | |
| 332 | off = hfsplus_decompose_table[(uc >> 12) & 0xf]; |
| 333 | if (off == 0 || off == 0xffff) |
| 334 | return NULL; |
| 335 | |
| 336 | off = hfsplus_decompose_table[off + ((uc >> 8) & 0xf)]; |
| 337 | if (!off) |
| 338 | return NULL; |
| 339 | |
| 340 | off = hfsplus_decompose_table[off + ((uc >> 4) & 0xf)]; |
| 341 | if (!off) |
| 342 | return NULL; |
| 343 | |
| 344 | off = hfsplus_decompose_table[off + (uc & 0xf)]; |
| 345 | *size = off & 3; |
| 346 | if (*size == 0) |
| 347 | return NULL; |
| 348 | return hfsplus_decompose_table + (off / 4); |
| 349 | } |
| 350 | |
| 351 | /* |
| 352 | * Try to decompose a unicode character as Hangul. Return 0 if @uc is not |
| 353 | * precomposed Hangul, otherwise return the length of the decomposition. |
| 354 | * |
| 355 | * This function was adapted from sample code from the Unicode Standard |
| 356 | * Annex #15: Unicode Normalization Forms, version 3.2.0. |
| 357 | * |
| 358 | * Copyright (C) 1991-2018 Unicode, Inc. All rights reserved. Distributed |
| 359 | * under the Terms of Use in http://www.unicode.org/copyright.html. |
| 360 | */ |
| 361 | static int hfsplus_try_decompose_hangul(wchar_t uc, u16 *result) |
| 362 | { |
| 363 | int index; |
| 364 | int l, v, t; |
| 365 | |
| 366 | index = uc - Hangul_SBase; |
| 367 | if (index < 0 || index >= Hangul_SCount) |
| 368 | return 0; |
| 369 | |
| 370 | l = Hangul_LBase + index / Hangul_NCount; |
| 371 | v = Hangul_VBase + (index % Hangul_NCount) / Hangul_TCount; |
| 372 | t = Hangul_TBase + index % Hangul_TCount; |
| 373 | |
| 374 | result[0] = l; |
| 375 | result[1] = v; |
| 376 | if (t != Hangul_TBase) { |
| 377 | result[2] = t; |
| 378 | return 3; |
| 379 | } |
| 380 | return 2; |
| 381 | } |
| 382 | |
| 383 | /* Decomposes a single unicode character. */ |
| 384 | static u16 *decompose_unichar(wchar_t uc, int *size, u16 *hangul_buffer) |
| 385 | { |
| 386 | u16 *result; |
| 387 | |
| 388 | /* Hangul is handled separately */ |
| 389 | result = hangul_buffer; |
| 390 | *size = hfsplus_try_decompose_hangul(uc, result); |
| 391 | if (*size == 0) |
| 392 | result = hfsplus_decompose_nonhangul(uc, size); |
| 393 | return result; |
| 394 | } |
| 395 | |
| 396 | int hfsplus_asc2uni(struct super_block *sb, |
| 397 | struct hfsplus_unistr *ustr, int max_unistr_len, |
| 398 | const char *astr, int len) |
| 399 | { |
| 400 | int size, dsize, decompose; |
| 401 | u16 *dstr, outlen = 0; |
| 402 | wchar_t c; |
| 403 | u16 dhangul[3]; |
| 404 | |
| 405 | decompose = !test_bit(HFSPLUS_SB_NODECOMPOSE, &HFSPLUS_SB(sb)->flags); |
| 406 | while (outlen < max_unistr_len && len > 0) { |
| 407 | size = asc2unichar(sb, astr, len, uc: &c); |
| 408 | |
| 409 | if (decompose) |
| 410 | dstr = decompose_unichar(uc: c, size: &dsize, hangul_buffer: dhangul); |
| 411 | else |
| 412 | dstr = NULL; |
| 413 | if (dstr) { |
| 414 | if (outlen + dsize > max_unistr_len) |
| 415 | break; |
| 416 | do { |
| 417 | ustr->unicode[outlen++] = cpu_to_be16(*dstr++); |
| 418 | } while (--dsize > 0); |
| 419 | } else |
| 420 | ustr->unicode[outlen++] = cpu_to_be16(c); |
| 421 | |
| 422 | astr += size; |
| 423 | len -= size; |
| 424 | } |
| 425 | ustr->length = cpu_to_be16(outlen); |
| 426 | if (len > 0) |
| 427 | return -ENAMETOOLONG; |
| 428 | return 0; |
| 429 | } |
| 430 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_asc2uni); |
| 431 | |
| 432 | /* |
| 433 | * Hash a string to an integer as appropriate for the HFS+ filesystem. |
| 434 | * Composed unicode characters are decomposed and case-folding is performed |
| 435 | * if the appropriate bits are (un)set on the superblock. |
| 436 | */ |
| 437 | int hfsplus_hash_dentry(const struct dentry *dentry, struct qstr *str) |
| 438 | { |
| 439 | struct super_block *sb = dentry->d_sb; |
| 440 | const char *astr; |
| 441 | const u16 *dstr; |
| 442 | int casefold, decompose, size, len; |
| 443 | unsigned long hash; |
| 444 | wchar_t c; |
| 445 | u16 c2; |
| 446 | u16 dhangul[3]; |
| 447 | |
| 448 | casefold = test_bit(HFSPLUS_SB_CASEFOLD, &HFSPLUS_SB(sb)->flags); |
| 449 | decompose = !test_bit(HFSPLUS_SB_NODECOMPOSE, &HFSPLUS_SB(sb)->flags); |
| 450 | hash = init_name_hash(dentry); |
| 451 | astr = str->name; |
| 452 | len = str->len; |
| 453 | while (len > 0) { |
| 454 | int dsize; |
| 455 | size = asc2unichar(sb, astr, len, uc: &c); |
| 456 | astr += size; |
| 457 | len -= size; |
| 458 | |
| 459 | if (decompose) |
| 460 | dstr = decompose_unichar(uc: c, size: &dsize, hangul_buffer: dhangul); |
| 461 | else |
| 462 | dstr = NULL; |
| 463 | if (dstr) { |
| 464 | do { |
| 465 | c2 = *dstr++; |
| 466 | if (casefold) |
| 467 | c2 = case_fold(c: c2); |
| 468 | if (!casefold || c2) |
| 469 | hash = partial_name_hash(c: c2, prevhash: hash); |
| 470 | } while (--dsize > 0); |
| 471 | } else { |
| 472 | c2 = c; |
| 473 | if (casefold) |
| 474 | c2 = case_fold(c: c2); |
| 475 | if (!casefold || c2) |
| 476 | hash = partial_name_hash(c: c2, prevhash: hash); |
| 477 | } |
| 478 | } |
| 479 | str->hash = end_name_hash(hash); |
| 480 | |
| 481 | return 0; |
| 482 | } |
| 483 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_hash_dentry); |
| 484 | |
| 485 | /* |
| 486 | * Compare strings with HFS+ filename ordering. |
| 487 | * Composed unicode characters are decomposed and case-folding is performed |
| 488 | * if the appropriate bits are (un)set on the superblock. |
| 489 | */ |
| 490 | int hfsplus_compare_dentry(const struct dentry *dentry, |
| 491 | unsigned int len, const char *str, const struct qstr *name) |
| 492 | { |
| 493 | struct super_block *sb = dentry->d_sb; |
| 494 | int casefold, decompose, size; |
| 495 | int dsize1, dsize2, len1, len2; |
| 496 | const u16 *dstr1, *dstr2; |
| 497 | const char *astr1, *astr2; |
| 498 | u16 c1, c2; |
| 499 | wchar_t c; |
| 500 | u16 dhangul_1[3], dhangul_2[3]; |
| 501 | |
| 502 | casefold = test_bit(HFSPLUS_SB_CASEFOLD, &HFSPLUS_SB(sb)->flags); |
| 503 | decompose = !test_bit(HFSPLUS_SB_NODECOMPOSE, &HFSPLUS_SB(sb)->flags); |
| 504 | astr1 = str; |
| 505 | len1 = len; |
| 506 | astr2 = name->name; |
| 507 | len2 = name->len; |
| 508 | dsize1 = dsize2 = 0; |
| 509 | dstr1 = dstr2 = NULL; |
| 510 | |
| 511 | while (len1 > 0 && len2 > 0) { |
| 512 | if (!dsize1) { |
| 513 | size = asc2unichar(sb, astr: astr1, len: len1, uc: &c); |
| 514 | astr1 += size; |
| 515 | len1 -= size; |
| 516 | |
| 517 | if (decompose) |
| 518 | dstr1 = decompose_unichar(uc: c, size: &dsize1, |
| 519 | hangul_buffer: dhangul_1); |
| 520 | if (!decompose || !dstr1) { |
| 521 | c1 = c; |
| 522 | dstr1 = &c1; |
| 523 | dsize1 = 1; |
| 524 | } |
| 525 | } |
| 526 | |
| 527 | if (!dsize2) { |
| 528 | size = asc2unichar(sb, astr: astr2, len: len2, uc: &c); |
| 529 | astr2 += size; |
| 530 | len2 -= size; |
| 531 | |
| 532 | if (decompose) |
| 533 | dstr2 = decompose_unichar(uc: c, size: &dsize2, |
| 534 | hangul_buffer: dhangul_2); |
| 535 | if (!decompose || !dstr2) { |
| 536 | c2 = c; |
| 537 | dstr2 = &c2; |
| 538 | dsize2 = 1; |
| 539 | } |
| 540 | } |
| 541 | |
| 542 | c1 = *dstr1; |
| 543 | c2 = *dstr2; |
| 544 | if (casefold) { |
| 545 | c1 = case_fold(c: c1); |
| 546 | if (!c1) { |
| 547 | dstr1++; |
| 548 | dsize1--; |
| 549 | continue; |
| 550 | } |
| 551 | c2 = case_fold(c: c2); |
| 552 | if (!c2) { |
| 553 | dstr2++; |
| 554 | dsize2--; |
| 555 | continue; |
| 556 | } |
| 557 | } |
| 558 | if (c1 < c2) |
| 559 | return -1; |
| 560 | else if (c1 > c2) |
| 561 | return 1; |
| 562 | |
| 563 | dstr1++; |
| 564 | dsize1--; |
| 565 | dstr2++; |
| 566 | dsize2--; |
| 567 | } |
| 568 | |
| 569 | if (len1 < len2) |
| 570 | return -1; |
| 571 | if (len1 > len2) |
| 572 | return 1; |
| 573 | return 0; |
| 574 | } |
| 575 | EXPORT_SYMBOL_IF_KUNIT(hfsplus_compare_dentry); |
| 576 | |