xref: /freebsd/sys/libkern/iconv.c (revision cd0d51baaa4509a1db83251a601d34404d20c990)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2000-2001 Boris Popov
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26  * SUCH DAMAGE.
27  */
28 
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
31 
32 #include <sys/param.h>
33 #include <sys/systm.h>
34 #include <sys/kernel.h>
35 #include <sys/iconv.h>
36 #include <sys/malloc.h>
37 #include <sys/mount.h>
38 #include <sys/sx.h>
39 #include <sys/syslog.h>
40 
41 #include "iconv_converter_if.h"
42 
43 SYSCTL_DECL(_kern_iconv);
44 
45 SYSCTL_NODE(_kern, OID_AUTO, iconv, CTLFLAG_RW, NULL, "kernel iconv interface");
46 
47 MALLOC_DEFINE(M_ICONV, "iconv", "ICONV structures");
48 static MALLOC_DEFINE(M_ICONVDATA, "iconv_data", "ICONV data");
49 
50 MODULE_VERSION(libiconv, 2);
51 
52 static struct sx iconv_lock;
53 
54 #ifdef notnow
55 /*
56  * iconv converter instance
57  */
58 struct iconv_converter {
59 	KOBJ_FIELDS;
60 	void *			c_data;
61 };
62 #endif
63 
64 struct sysctl_oid *iconv_oid_hook = &sysctl___kern_iconv;
65 
66 /*
67  * List of loaded converters
68  */
69 static TAILQ_HEAD(iconv_converter_list, iconv_converter_class)
70     iconv_converters = TAILQ_HEAD_INITIALIZER(iconv_converters);
71 
72 /*
73  * List of supported/loaded charsets pairs
74  */
75 static TAILQ_HEAD(, iconv_cspair)
76     iconv_cslist = TAILQ_HEAD_INITIALIZER(iconv_cslist);
77 static int iconv_csid = 1;
78 
79 static char iconv_unicode_string[] = "unicode";	/* save eight bytes when possible */
80 
81 static void iconv_unregister_cspair(struct iconv_cspair *csp);
82 
83 static int
84 iconv_mod_unload(void)
85 {
86 	struct iconv_cspair *csp;
87 
88 	sx_xlock(&iconv_lock);
89 	TAILQ_FOREACH(csp, &iconv_cslist, cp_link) {
90 		if (csp->cp_refcount) {
91 			sx_xunlock(&iconv_lock);
92 			return EBUSY;
93 		}
94 	}
95 
96 	while ((csp = TAILQ_FIRST(&iconv_cslist)) != NULL)
97 		iconv_unregister_cspair(csp);
98 	sx_xunlock(&iconv_lock);
99 	sx_destroy(&iconv_lock);
100 	return 0;
101 }
102 
103 static int
104 iconv_mod_handler(module_t mod, int type, void *data)
105 {
106 	int error;
107 
108 	switch (type) {
109 	    case MOD_LOAD:
110 		error = 0;
111 		sx_init(&iconv_lock, "iconv");
112 		break;
113 	    case MOD_UNLOAD:
114 		error = iconv_mod_unload();
115 		break;
116 	    default:
117 		error = EINVAL;
118 	}
119 	return error;
120 }
121 
122 static moduledata_t iconv_mod = {
123 	"iconv", iconv_mod_handler, NULL
124 };
125 
126 DECLARE_MODULE(iconv, iconv_mod, SI_SUB_DRIVERS, SI_ORDER_SECOND);
127 
128 static int
129 iconv_register_converter(struct iconv_converter_class *dcp)
130 {
131 	kobj_class_compile((struct kobj_class*)dcp);
132 	dcp->refs++;
133 	TAILQ_INSERT_TAIL(&iconv_converters, dcp, cc_link);
134 	return 0;
135 }
136 
137 static int
138 iconv_unregister_converter(struct iconv_converter_class *dcp)
139 {
140 	dcp->refs--;
141 	if (dcp->refs > 1) {
142 		ICDEBUG("converter has %d references left\n", dcp->refs);
143 		return EBUSY;
144 	}
145 	TAILQ_REMOVE(&iconv_converters, dcp, cc_link);
146 	kobj_class_free((struct kobj_class*)dcp);
147 	return 0;
148 }
149 
150 static int
151 iconv_lookupconv(const char *name, struct iconv_converter_class **dcpp)
152 {
153 	struct iconv_converter_class *dcp;
154 
155 	TAILQ_FOREACH(dcp, &iconv_converters, cc_link) {
156 		if (name == NULL)
157 			continue;
158 		if (strcmp(name, ICONV_CONVERTER_NAME(dcp)) == 0) {
159 			if (dcpp)
160 				*dcpp = dcp;
161 			return 0;
162 		}
163 	}
164 	return ENOENT;
165 }
166 
167 static int
168 iconv_lookupcs(const char *to, const char *from, struct iconv_cspair **cspp)
169 {
170 	struct iconv_cspair *csp;
171 
172 	TAILQ_FOREACH(csp, &iconv_cslist, cp_link) {
173 		if (strcasecmp(csp->cp_to, to) == 0 &&
174 		    strcasecmp(csp->cp_from, from) == 0) {
175 			if (cspp)
176 				*cspp = csp;
177 			return 0;
178 		}
179 	}
180 	return ENOENT;
181 }
182 
183 static int
184 iconv_register_cspair(const char *to, const char *from,
185 	struct iconv_converter_class *dcp, void *data,
186 	struct iconv_cspair **cspp)
187 {
188 	struct iconv_cspair *csp;
189 	char *cp;
190 	int csize, ucsto, ucsfrom;
191 
192 	if (iconv_lookupcs(to, from, NULL) == 0)
193 		return EEXIST;
194 	csize = sizeof(*csp);
195 	ucsto = strcmp(to, iconv_unicode_string) == 0;
196 	if (!ucsto)
197 		csize += strlen(to) + 1;
198 	ucsfrom = strcmp(from, iconv_unicode_string) == 0;
199 	if (!ucsfrom)
200 		csize += strlen(from) + 1;
201 	csp = malloc(csize, M_ICONV, M_WAITOK);
202 	bzero(csp, csize);
203 	csp->cp_id = iconv_csid++;
204 	csp->cp_dcp = dcp;
205 	cp = (char*)(csp + 1);
206 	if (!ucsto) {
207 		strcpy(cp, to);
208 		csp->cp_to = cp;
209 		cp += strlen(cp) + 1;
210 	} else
211 		csp->cp_to = iconv_unicode_string;
212 	if (!ucsfrom) {
213 		strcpy(cp, from);
214 		csp->cp_from = cp;
215 	} else
216 		csp->cp_from = iconv_unicode_string;
217 	csp->cp_data = data;
218 
219 	TAILQ_INSERT_TAIL(&iconv_cslist, csp, cp_link);
220 	*cspp = csp;
221 	return 0;
222 }
223 
224 static void
225 iconv_unregister_cspair(struct iconv_cspair *csp)
226 {
227 	TAILQ_REMOVE(&iconv_cslist, csp, cp_link);
228 	if (csp->cp_data)
229 		free(csp->cp_data, M_ICONVDATA);
230 	free(csp, M_ICONV);
231 }
232 
233 /*
234  * Lookup and create an instance of converter.
235  * Currently this layer didn't have associated 'instance' structure
236  * to avoid unnesessary memory allocation.
237  */
238 int
239 iconv_open(const char *to, const char *from, void **handle)
240 {
241 	struct iconv_cspair *csp, *cspfrom, *cspto;
242 	struct iconv_converter_class *dcp;
243 	const char *cnvname;
244 	int error;
245 
246 	/*
247 	 * First, lookup fully qualified cspairs
248 	 */
249 	error = iconv_lookupcs(to, from, &csp);
250 	if (error == 0)
251 		return ICONV_CONVERTER_OPEN(csp->cp_dcp, csp, NULL, handle);
252 
253 	/*
254 	 * Well, nothing found. Now try to construct a composite conversion
255 	 * ToDo: add a 'capability' field to converter
256 	 */
257 	TAILQ_FOREACH(dcp, &iconv_converters, cc_link) {
258 		cnvname = ICONV_CONVERTER_NAME(dcp);
259 		if (cnvname == NULL)
260 			continue;
261 		error = iconv_lookupcs(cnvname, from, &cspfrom);
262 		if (error)
263 			continue;
264 		error = iconv_lookupcs(to, cnvname, &cspto);
265 		if (error)
266 			continue;
267 		/*
268 		 * Fine, we're found a pair which can be combined together
269 		 */
270 		return ICONV_CONVERTER_OPEN(dcp, cspto, cspfrom, handle);
271 	}
272 	return ENOENT;
273 }
274 
275 int
276 iconv_close(void *handle)
277 {
278 	return ICONV_CONVERTER_CLOSE(handle);
279 }
280 
281 int
282 iconv_conv(void *handle, const char **inbuf,
283 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft)
284 {
285 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 0, 0);
286 }
287 
288 int
289 iconv_conv_case(void *handle, const char **inbuf,
290 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft, int casetype)
291 {
292 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 0, casetype);
293 }
294 
295 int
296 iconv_convchr(void *handle, const char **inbuf,
297 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft)
298 {
299 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 1, 0);
300 }
301 
302 int
303 iconv_convchr_case(void *handle, const char **inbuf,
304 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft, int casetype)
305 {
306 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 1, casetype);
307 }
308 
309 int
310 towlower(int c, void *handle)
311 {
312 	return ICONV_CONVERTER_TOLOWER(handle, c);
313 }
314 
315 int
316 towupper(int c, void *handle)
317 {
318 	return ICONV_CONVERTER_TOUPPER(handle, c);
319 }
320 
321 /*
322  * Give a list of loaded converters. Each name terminated with 0.
323  * An empty string terminates the list.
324  */
325 static int
326 iconv_sysctl_drvlist(SYSCTL_HANDLER_ARGS)
327 {
328 	struct iconv_converter_class *dcp;
329 	const char *name;
330 	char spc;
331 	int error;
332 
333 	error = 0;
334 	sx_slock(&iconv_lock);
335 	TAILQ_FOREACH(dcp, &iconv_converters, cc_link) {
336 		name = ICONV_CONVERTER_NAME(dcp);
337 		if (name == NULL)
338 			continue;
339 		error = SYSCTL_OUT(req, name, strlen(name) + 1);
340 		if (error)
341 			break;
342 	}
343 	sx_sunlock(&iconv_lock);
344 	if (error)
345 		return error;
346 	spc = 0;
347 	error = SYSCTL_OUT(req, &spc, sizeof(spc));
348 	return error;
349 }
350 
351 SYSCTL_PROC(_kern_iconv, OID_AUTO, drvlist, CTLFLAG_RD | CTLTYPE_OPAQUE,
352 	    NULL, 0, iconv_sysctl_drvlist, "S,xlat", "registered converters");
353 
354 /*
355  * List all available charset pairs.
356  */
357 static int
358 iconv_sysctl_cslist(SYSCTL_HANDLER_ARGS)
359 {
360 	struct iconv_cspair *csp;
361 	struct iconv_cspair_info csi;
362 	int error;
363 
364 	error = 0;
365 	bzero(&csi, sizeof(csi));
366 	csi.cs_version = ICONV_CSPAIR_INFO_VER;
367 	sx_slock(&iconv_lock);
368 	TAILQ_FOREACH(csp, &iconv_cslist, cp_link) {
369 		csi.cs_id = csp->cp_id;
370 		csi.cs_refcount = csp->cp_refcount;
371 		csi.cs_base = csp->cp_base ? csp->cp_base->cp_id : 0;
372 		strcpy(csi.cs_to, csp->cp_to);
373 		strcpy(csi.cs_from, csp->cp_from);
374 		error = SYSCTL_OUT(req, &csi, sizeof(csi));
375 		if (error)
376 			break;
377 	}
378 	sx_sunlock(&iconv_lock);
379 	return error;
380 }
381 
382 SYSCTL_PROC(_kern_iconv, OID_AUTO, cslist, CTLFLAG_RD | CTLTYPE_OPAQUE,
383 	    NULL, 0, iconv_sysctl_cslist, "S,xlat", "registered charset pairs");
384 
385 int
386 iconv_add(const char *converter, const char *to, const char *from)
387 {
388 	struct iconv_converter_class *dcp;
389 	struct iconv_cspair *csp;
390 
391 	if (iconv_lookupconv(converter, &dcp) != 0)
392 		return EINVAL;
393 
394 	return iconv_register_cspair(to, from, dcp, NULL, &csp);
395 }
396 
397 /*
398  * Add new charset pair
399  */
400 static int
401 iconv_sysctl_add(SYSCTL_HANDLER_ARGS)
402 {
403 	struct iconv_converter_class *dcp;
404 	struct iconv_cspair *csp;
405 	struct iconv_add_in din;
406 	struct iconv_add_out dout;
407 	int error;
408 
409 	error = SYSCTL_IN(req, &din, sizeof(din));
410 	if (error)
411 		return error;
412 	if (din.ia_version != ICONV_ADD_VER)
413 		return EINVAL;
414 	if (din.ia_datalen > ICONV_CSMAXDATALEN)
415 		return EINVAL;
416 	if (strnlen(din.ia_from, sizeof(din.ia_from)) >= ICONV_CSNMAXLEN)
417 		return EINVAL;
418 	if (strnlen(din.ia_to, sizeof(din.ia_to)) >= ICONV_CSNMAXLEN)
419 		return EINVAL;
420 	if (strnlen(din.ia_converter, sizeof(din.ia_converter)) >= ICONV_CNVNMAXLEN)
421 		return EINVAL;
422 	if (iconv_lookupconv(din.ia_converter, &dcp) != 0)
423 		return EINVAL;
424 	sx_xlock(&iconv_lock);
425 	error = iconv_register_cspair(din.ia_to, din.ia_from, dcp, NULL, &csp);
426 	if (error) {
427 		sx_xunlock(&iconv_lock);
428 		return error;
429 	}
430 	if (din.ia_datalen) {
431 		csp->cp_data = malloc(din.ia_datalen, M_ICONVDATA, M_WAITOK);
432 		error = copyin(din.ia_data, csp->cp_data, din.ia_datalen);
433 		if (error)
434 			goto bad;
435 	}
436 	dout.ia_csid = csp->cp_id;
437 	error = SYSCTL_OUT(req, &dout, sizeof(dout));
438 	if (error)
439 		goto bad;
440 	sx_xunlock(&iconv_lock);
441 	ICDEBUG("%s => %s, %d bytes\n",din.ia_from, din.ia_to, din.ia_datalen);
442 	return 0;
443 bad:
444 	iconv_unregister_cspair(csp);
445 	sx_xunlock(&iconv_lock);
446 	return error;
447 }
448 
449 SYSCTL_PROC(_kern_iconv, OID_AUTO, add, CTLFLAG_RW | CTLTYPE_OPAQUE,
450 	    NULL, 0, iconv_sysctl_add, "S,xlat", "register charset pair");
451 
452 /*
453  * Default stubs for converters
454  */
455 int
456 iconv_converter_initstub(struct iconv_converter_class *dp)
457 {
458 	return 0;
459 }
460 
461 int
462 iconv_converter_donestub(struct iconv_converter_class *dp)
463 {
464 	return 0;
465 }
466 
467 int
468 iconv_converter_tolowerstub(int c, void *handle)
469 {
470 	return (c);
471 }
472 
473 int
474 iconv_converter_handler(module_t mod, int type, void *data)
475 {
476 	struct iconv_converter_class *dcp = data;
477 	int error;
478 
479 	switch (type) {
480 	    case MOD_LOAD:
481 		sx_xlock(&iconv_lock);
482 		error = iconv_register_converter(dcp);
483 		if (error) {
484 			sx_xunlock(&iconv_lock);
485 			break;
486 		}
487 		error = ICONV_CONVERTER_INIT(dcp);
488 		if (error)
489 			iconv_unregister_converter(dcp);
490 		sx_xunlock(&iconv_lock);
491 		break;
492 	    case MOD_UNLOAD:
493 		sx_xlock(&iconv_lock);
494 		ICONV_CONVERTER_DONE(dcp);
495 		error = iconv_unregister_converter(dcp);
496 		sx_xunlock(&iconv_lock);
497 		break;
498 	    default:
499 		error = EINVAL;
500 	}
501 	return error;
502 }
503 
504 /*
505  * Common used functions (don't use with unicode)
506  */
507 char *
508 iconv_convstr(void *handle, char *dst, const char *src)
509 {
510 	char *p = dst;
511 	size_t inlen, outlen;
512 	int error;
513 
514 	if (handle == NULL) {
515 		strcpy(dst, src);
516 		return dst;
517 	}
518 	inlen = outlen = strlen(src);
519 	error = iconv_conv(handle, NULL, NULL, &p, &outlen);
520 	if (error)
521 		return NULL;
522 	error = iconv_conv(handle, &src, &inlen, &p, &outlen);
523 	if (error)
524 		return NULL;
525 	*p = 0;
526 	return dst;
527 }
528 
529 void *
530 iconv_convmem(void *handle, void *dst, const void *src, int size)
531 {
532 	const char *s = src;
533 	char *d = dst;
534 	size_t inlen, outlen;
535 	int error;
536 
537 	if (size == 0)
538 		return dst;
539 	if (handle == NULL) {
540 		memcpy(dst, src, size);
541 		return dst;
542 	}
543 	inlen = outlen = size;
544 	error = iconv_conv(handle, NULL, NULL, &d, &outlen);
545 	if (error)
546 		return NULL;
547 	error = iconv_conv(handle, &s, &inlen, &d, &outlen);
548 	if (error)
549 		return NULL;
550 	return dst;
551 }
552 
553 int
554 iconv_lookupcp(char **cpp, const char *s)
555 {
556 	if (cpp == NULL) {
557 		ICDEBUG("warning a NULL list passed\n", "");
558 		return ENOENT;
559 	}
560 	for (; *cpp; cpp++)
561 		if (strcmp(*cpp, s) == 0)
562 			return 0;
563 	return ENOENT;
564 }
565 
566 /*
567  * Return if fsname is in use of not
568  */
569 int
570 iconv_vfs_refcount(const char *fsname)
571 {
572 	struct vfsconf *vfsp;
573 
574 	vfsp = vfs_byname(fsname);
575 	if (vfsp != NULL && vfsp->vfc_refcount > 0)
576 		return (EBUSY);
577 	return (0);
578 }
579