1 /*
2 * This file and its contents are supplied under the terms of the
3 * Common Development and Distribution License ("CDDL"), version 1.0.
4 * You may only use this file in accordance with the terms of version
5 * 1.0 of the CDDL.
6 *
7 * A full copy of the text of the CDDL should have accompanied this
8 * source. A copy of the CDDL is also available via the Internet at
9 * http://www.illumos.org/license/CDDL.
10 */
11
12 /*
13 * Copyright 2014 Garrett D'Amore <garrett@damore.org>
14 */
15
16 /*
17 * This file implements the 2008 newlocale and friends handling.
18 */
19
20 #ifndef _LCONV_C99
21 #define _LCONV_C99
22 #endif
23
24 #include "lint.h"
25 #include <atomic.h>
26 #include <locale.h>
27 #include <sys/types.h>
28 #include <sys/mman.h>
29 #include <errno.h>
30 #include <string.h>
31 #include "libc.h"
32 #include "mtlib.h"
33 #include "tsd.h"
34 #include "localeimpl.h"
35 #include "lctype.h"
36
37 /*
38 * Big Theory of Locales:
39 *
40 * (It is recommended that readers familiarize themselves with the POSIX
41 * 2008 (XPG Issue 7) specifications for locales, first.)
42 *
43 * Historically, we had a bunch of global variables that stored locale
44 * data. While this worked well, it limited applications to a single locale
45 * at a time. This doesn't work well in certain server applications.
46 *
47 * Issue 7, X/Open introduced the concept of a locale_t object, along with
48 * versions of functions that can take this object as a parameter, along
49 * with functions to clone and manipulate these locale objects. The new
50 * functions are named with a _l() suffix.
51 *
52 * Additionally uselocale() is introduced which can change the locale of
53 * of a single thread. However, setlocale() can still be used to change
54 * the global locale.
55 *
56 * In our implementation, we use libc's TSD to store the locale data that
57 * was previously global. We still have global data because some applications
58 * have had those global objects compiled into them. (Such applications will
59 * be unable to benefit from uselocale(), btw.) The legacy routines are
60 * reimplemented as wrappers that use the appropriate locale object by
61 * calling uselocale(). uselocale() when passed a NULL pointer returns the
62 * thread-specific locale object if one is present, or the global locale
63 * object otherwise. Note that once the TSD data is set, the only way
64 * to revert to the global locale is to pass the global locale LC_GLOBAL_LOCALE
65 * to uselocale().
66 *
67 * We are careful to minimize performance impact of multiple calls to
68 * uselocale() or setlocale() by using a cache of locale data whenever possible.
69 * As a consequence of this, applications that iterate over all possible
70 * locales will burn through a lot of virtual memory, but we find such
71 * applications rare. (locale -a might be an exception, but it is short lived.)
72 *
73 * Category data is never released (although enclosing locale objects might be),
74 * in order to guarantee thread-safety. Calling freelocale() on an object
75 * while it is in use by another thread is a programmer error (use-after-free)
76 * and we don't bother to note it further.
77 *
78 * Locale objects (global locales) established by setlocale() are also
79 * never freed (for MT safety), but we will save previous locale objects
80 * and reuse them when we can.
81 */
82
83 typedef struct locdata *(*loadfn_t)(const char *);
84
85 static const loadfn_t loaders[LC_ALL] = {
86 __lc_ctype_load,
87 __lc_numeric_load,
88 __lc_time_load,
89 __lc_collate_load,
90 __lc_monetary_load,
91 __lc_messages_load,
92 };
93
94 extern struct lc_monetary lc_monetary_posix;
95 extern struct lc_numeric lc_numeric_posix;
96 extern struct lc_messages lc_messages_posix;
97 extern struct lc_time lc_time_posix;
98 extern struct lc_ctype lc_ctype_posix;
99 extern struct lc_collate lc_collate_posix;
100
101 static struct locale posix_locale = {
102 /* locdata */
103 .locdata = {
104 &__posix_ctype_locdata,
105 &__posix_numeric_locdata,
106 &__posix_time_locdata,
107 &__posix_collate_locdata,
108 &__posix_monetary_locdata,
109 &__posix_messages_locdata,
110 },
111 .locname = "C",
112 .ctype = &lc_ctype_posix,
113 .numeric = &lc_numeric_posix,
114 .collate = &lc_collate_posix,
115 .monetary = &lc_monetary_posix,
116 .messages = &lc_messages_posix,
117 .time = &lc_time_posix,
118 .runelocale = &_DefaultRuneLocale,
119 };
120
121 locale_t ___global_locale = &posix_locale;
122
123 locale_t
124 __global_locale(void)
125 {
126 return (___global_locale);
127 }
128
129 /*
130 * Category names for getenv() Note that this was modified
131 * for Solaris. See <iso/locale_iso.h>.
132 */
133 #define NUM_CATS 7
134 static char *categories[7] = {
135 "LC_CTYPE",
136 "LC_NUMERIC",
137 "LC_TIME",
138 "LC_COLLATE",
139 "LC_MONETARY",
140 "LC_MESSAGES",
141 "LC_ALL",
142 };
143
144 /*
145 * Prototypes.
146 */
147 static const char *get_locale_env(int);
148 static struct locdata *locdata_get(int, const const char *);
149 static struct locdata *locdata_get_cache(int, const char *);
150 static locale_t mklocname(locale_t);
151
152 /*
153 * Some utility routines.
154 */
155
156 struct locdata *
157 __locdata_alloc(const char *name, size_t memsz)
158 {
159 struct locdata *ldata;
160
161 if ((ldata = lmalloc(sizeof (*ldata))) == NULL) {
162 return (NULL);
163 }
164 if ((ldata->l_data[0] = libc_malloc(memsz)) == NULL) {
165 lfree(ldata, sizeof (*ldata));
166 errno = ENOMEM;
167 return (NULL);
168 }
169 (void) strlcpy(ldata->l_lname, name, sizeof (ldata->l_lname));
170
171 return (ldata);
172 }
173
174 /*
175 * Normally we never free locale data truly, but if we failed to load it
176 * for some reason, this routine is used to cleanup the partial mess.
177 */
178 void
179 __locdata_free(struct locdata *ldata)
180 {
181 for (int i = 0; i < NLOCDATA; i++)
182 libc_free(ldata->l_data[i]);
183 if (ldata->l_map != NULL && ldata->l_map_len)
184 (void) munmap(ldata->l_map, ldata->l_map_len);
185 lfree(ldata, sizeof (*ldata));
186 }
187
188 /*
189 * It turns out that for performance reasons we would really like to
190 * cache the most recently referenced locale data to avoid wasteful
191 * loading from files.
192 */
193
194 static struct locdata *cache_data[LC_ALL];
195 static struct locdata *cat_data[LC_ALL];
196 static mutex_t cache_lock = DEFAULTMUTEX;
197
198 /*
199 * Returns the cached data if the locale name is the same. If not,
200 * returns NULL (cache miss). The locdata is returned with a hold on
201 * it, taken on behalf of the caller. The caller should drop the hold
202 * when it is finished.
203 */
204 static struct locdata *
205 locdata_get_cache(int category, const char *locname)
206 {
207 struct locdata *loc;
208
209 if (category < 0 || category >= LC_ALL)
210 return (NULL);
211
212 /* Try cache first. */
213 lmutex_lock(&cache_lock);
214 loc = cache_data[category];
215
216 if ((loc != NULL) && (strcmp(loc->l_lname, locname) == 0)) {
217 lmutex_unlock(&cache_lock);
218 return (loc);
219 }
220
221 /*
222 * Failing that try previously loaded locales (linear search) --
223 * this could be optimized to a hash, but its unlikely that a single
224 * application will ever need to work with more than a few locales.
225 */
226 for (loc = cat_data[category]; loc != NULL; loc = loc->l_next) {
227 if (strcmp(locname, loc->l_lname) == 0) {
228 break;
229 }
230 }
231
232 /*
233 * Finally, if we still don't have one, try loading the locale
234 * data from the actual on-disk data.
235 *
236 * We drop the lock (libc wants to ensure no internal locks
237 * are held when we call other routines required to read from
238 * files, allocate memory, etc.) There is a small race here,
239 * but the consequences of the race are benign -- if multiple
240 * threads hit this at precisely the same point, we could
241 * wind up with duplicates of the locale data in the cache.
242 *
243 * This wastes the memory for an extra copy of the locale
244 * data, but there is no further harm beyond that. Its not
245 * worth the effort to recode this to something "safe"
246 * (which would require rescanning the list, etc.), given
247 * that this race will probably never actually occur.
248 */
249 if (loc == NULL) {
250 lmutex_unlock(&cache_lock);
251 loc = (*loaders[category])(locname);
252 lmutex_lock(&cache_lock);
253 if (loc != NULL)
254 (void) strlcpy(loc->l_lname, locname,
255 sizeof (loc->l_lname));
256 }
257
258 /*
259 * Assuming we got one, update the cache, and stick us on the list
260 * of loaded locale data. We insert into the head (more recent
261 * use is likely to win.)
262 */
263 if (loc != NULL) {
264 cache_data[category] = loc;
265 if (!loc->l_cached) {
266 loc->l_cached = 1;
267 loc->l_next = cat_data[category];
268 cat_data[category] = loc;
269 }
270 }
271
272 lmutex_unlock(&cache_lock);
273 return (loc);
274 }
275
276 /*
277 * Routine to get the locdata for a given category and locale.
278 * This includes retrieving it from cache, retrieving it from
279 * a file, etc.
280 */
281 static struct locdata *
282 locdata_get(int category, const char *locname)
283 {
284 char scratch[ENCODING_LEN + 1];
285 char *slash;
286 int cnt;
287 int len;
288
289 if (locname == NULL || *locname == 0) {
290 locname = get_locale_env(category);
291 }
292
293 /*
294 * Extract the locale name for the category if it is a composite
295 * locale.
296 */
297 if ((slash = strchr(locname, '/')) != NULL) {
298 for (cnt = category; cnt && slash != NULL; cnt--) {
299 locname = slash + 1;
300 slash = strchr(locname, '/');
301 }
302 if (slash) {
303 len = slash - locname + 1;
304 if (len >= sizeof (scratch)) {
305 len = sizeof (scratch);
306 }
307 } else {
308 len = sizeof (scratch);
309 }
310 (void) strlcpy(scratch, locname, len);
311 locname = scratch;
312 }
313
314 if ((strcmp(locname, "C") == 0) || (strcmp(locname, "POSIX") == 0))
315 return (posix_locale.locdata[category]);
316
317 return (locdata_get_cache(category, locname));
318 }
319
320 /* tsd destructor */
321 static void
322 freelocptr(void *arg)
323 {
324 locale_t *locptr = arg;
325 if (*locptr != NULL)
326 freelocale(*locptr);
327 }
328
329 static const char *
330 get_locale_env(int category)
331 {
332 const char *env;
333
334 /* 1. check LC_ALL. */
335 env = getenv(categories[LC_ALL]);
336
337 /* 2. check LC_* */
338 if (env == NULL || *env == '\0')
339 env = getenv(categories[category]);
340
341 /* 3. check LANG */
342 if (env == NULL || *env == '\0')
343 env = getenv("LANG");
344
345 /* 4. if none is set, fall to "C" */
346 if (env == NULL || *env == '\0')
347 env = "C";
348
349 return (env);
350 }
351
352
353 /*
354 * This routine is exposed via the MB_CUR_MAX macro. Note that legacy
355 * code will continue to use _ctype[520], but we prefer this function as
356 * it is the only way to get thread-specific information.
357 */
358 unsigned char
359 __mb_cur_max_l(locale_t loc)
360 {
361 return (loc->ctype->lc_max_mblen);
362 }
363
364 unsigned char
365 __mb_cur_max(void)
366 {
367 return (__mb_cur_max_l(uselocale(NULL)));
368 }
369
370 /*
371 * Public interfaces.
372 */
373
374 locale_t
375 duplocale(locale_t src)
376 {
377 locale_t loc;
378 int i;
379
380 loc = lmalloc(sizeof (*loc));
381 if (loc == NULL) {
382 return (NULL);
383 }
384 if (src == NULL) {
385 /* illumos extension: POSIX says LC_GLOBAL_LOCALE here */
386 src = ___global_locale;
387 }
388 for (i = 0; i < LC_ALL; i++) {
389 loc->locdata[i] = src->locdata[i];
390 loc->loaded[i] = 0;
391 }
392 loc->collate = loc->locdata[LC_COLLATE]->l_data[0];
393 loc->ctype = loc->locdata[LC_CTYPE]->l_data[0];
394 loc->runelocale = loc->locdata[LC_CTYPE]->l_data[1];
395 loc->messages = loc->locdata[LC_MESSAGES]->l_data[0];
396 loc->monetary = loc->locdata[LC_MONETARY]->l_data[0];
397 loc->numeric = loc->locdata[LC_NUMERIC]->l_data[0];
398 loc->time = loc->locdata[LC_TIME]->l_data[0];
399 return (loc);
400 }
401
402 void
403 freelocale(locale_t loc)
404 {
405 /*
406 * We take extra care never to free a saved locale created by
407 * setlocale(). This shouldn't be strictly necessary, but a little
408 * extra safety doesn't hurt here.
409 */
410 if ((loc != NULL) && (loc != &posix_locale) && (!loc->on_list))
411 lfree(loc, sizeof (*loc));
412 }
413
414 locale_t
415 newlocale(int catmask, const char *locname, locale_t base)
416 {
417 locale_t loc;
418 int i, e;
419
420 if (catmask & ~(LC_ALL_MASK)) {
421 errno = EINVAL;
422 return (NULL);
423 }
424
425 /*
426 * Technically passing LC_GLOBAL_LOCALE here is illegal,
427 * but we allow it.
428 */
429 if (base == NULL || base == ___global_locale) {
430 loc = duplocale(___global_locale);
431 } else {
432 loc = duplocale(base);
433 }
434 if (loc == NULL) {
435 return (NULL);
436 }
437
438 for (i = 0; i < LC_ALL; i++) {
439 struct locdata *ldata;
440 loc->loaded[i] = 0;
441 if (((1 << i) & catmask) == 0) {
442 /* Default to base locale if not overriding */
443 continue;
444 }
445 ldata = locdata_get(i, locname);
446 if (ldata == NULL) {
447 e = errno;
448 freelocale(loc);
449 errno = e;
450 return (NULL);
451 }
452 loc->locdata[i] = ldata;
453 }
454 loc->collate = loc->locdata[LC_COLLATE]->l_data[0];
455 loc->ctype = loc->locdata[LC_CTYPE]->l_data[0];
456 loc->runelocale = loc->locdata[LC_CTYPE]->l_data[1];
457 loc->messages = loc->locdata[LC_MESSAGES]->l_data[0];
458 loc->monetary = loc->locdata[LC_MONETARY]->l_data[0];
459 loc->numeric = loc->locdata[LC_NUMERIC]->l_data[0];
460 loc->time = loc->locdata[LC_TIME]->l_data[0];
461 freelocale(base);
462
463 return (mklocname(loc));
464 }
465
466 locale_t
467 uselocale(locale_t loc)
468 {
469 locale_t lastloc = ___global_locale;
470 locale_t *locptr;
471
472 locptr = tsdalloc(_T_SETLOCALE, sizeof (locale_t), freelocptr);
473 /* Should never occur */
474 if (locptr == NULL) {
475 errno = EINVAL;
476 return (NULL);
477 }
478
479 if (*locptr != NULL)
480 lastloc = *locptr;
481
482 /* Argument loc is NULL if we are just querying. */
483 if (loc != NULL) {
484 /*
485 * Set it to LC_GLOBAL_LOCAL to return to using
486 * the global locale (setlocale).
487 */
488 if (loc == ___global_locale) {
489 *locptr = NULL;
490 } else {
491 /* No validation of the provided locale at present */
492 *locptr = loc;
493 }
494 }
495
496 /*
497 * The caller is responsible for freeing, of course it would be
498 * gross error to call freelocale() on a locale object that is still
499 * in use.
500 */
501 return (lastloc);
502 }
503
504 static locale_t
505 mklocname(locale_t loc)
506 {
507 int composite = 0;
508
509 /* Look to see if any category is different */
510 for (int i = 1; i < LC_ALL; ++i) {
511 if (strcmp(loc->locdata[0]->l_lname,
512 loc->locdata[i]->l_lname) != 0) {
513 composite = 1;
514 break;
515 }
516 }
517
518 if (composite) {
519 /*
520 * Note ordering of these follows the numeric order,
521 * if the order is changed, then setlocale() will need
522 * to be changed as well.
523 */
524 (void) snprintf(loc->locname, sizeof (loc->locname),
525 "%s/%s/%s/%s/%s/%s",
526 loc->locdata[LC_CTYPE]->l_lname,
527 loc->locdata[LC_NUMERIC]->l_lname,
528 loc->locdata[LC_TIME]->l_lname,
529 loc->locdata[LC_COLLATE]->l_lname,
530 loc->locdata[LC_MONETARY]->l_lname,
531 loc->locdata[LC_MESSAGES]->l_lname);
532 } else {
533 (void) strlcpy(loc->locname, loc->locdata[LC_CTYPE]->l_lname,
534 sizeof (loc->locname));
535 }
536 return (loc);
537 }