xref: /freebsd/sys/libkern/iconv.c (revision fdafd315ad0d0f28a11b9fb4476a9ab059c62b92)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
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/param.h>
30 #include <sys/systm.h>
31 #include <sys/kernel.h>
32 #include <sys/iconv.h>
33 #include <sys/malloc.h>
34 #include <sys/mount.h>
35 #include <sys/sx.h>
36 #include <sys/syslog.h>
37 
38 #include "iconv_converter_if.h"
39 
40 SYSCTL_DECL(_kern_iconv);
41 
42 SYSCTL_NODE(_kern, OID_AUTO, iconv, CTLFLAG_RW | CTLFLAG_MPSAFE, NULL,
43     "kernel iconv interface");
44 
45 MALLOC_DEFINE(M_ICONV, "iconv", "ICONV structures");
46 static MALLOC_DEFINE(M_ICONVDATA, "iconv_data", "ICONV data");
47 
48 MODULE_VERSION(libiconv, 2);
49 
50 static struct sx iconv_lock;
51 
52 #ifdef notnow
53 /*
54  * iconv converter instance
55  */
56 struct iconv_converter {
57 	KOBJ_FIELDS;
58 	void *			c_data;
59 };
60 #endif
61 
62 struct sysctl_oid *iconv_oid_hook = &sysctl___kern_iconv;
63 
64 /*
65  * List of loaded converters
66  */
67 static TAILQ_HEAD(iconv_converter_list, iconv_converter_class)
68     iconv_converters = TAILQ_HEAD_INITIALIZER(iconv_converters);
69 
70 /*
71  * List of supported/loaded charsets pairs
72  */
73 static TAILQ_HEAD(, iconv_cspair)
74     iconv_cslist = TAILQ_HEAD_INITIALIZER(iconv_cslist);
75 static int iconv_csid = 1;
76 
77 static char iconv_unicode_string[] = "unicode";	/* save eight bytes when possible */
78 
79 static void iconv_unregister_cspair(struct iconv_cspair *csp);
80 
81 static int
iconv_mod_unload(void)82 iconv_mod_unload(void)
83 {
84 	struct iconv_cspair *csp;
85 
86 	sx_xlock(&iconv_lock);
87 	TAILQ_FOREACH(csp, &iconv_cslist, cp_link) {
88 		if (csp->cp_refcount) {
89 			sx_xunlock(&iconv_lock);
90 			return EBUSY;
91 		}
92 	}
93 
94 	while ((csp = TAILQ_FIRST(&iconv_cslist)) != NULL)
95 		iconv_unregister_cspair(csp);
96 	sx_xunlock(&iconv_lock);
97 	sx_destroy(&iconv_lock);
98 	return 0;
99 }
100 
101 static int
iconv_mod_handler(module_t mod,int type,void * data)102 iconv_mod_handler(module_t mod, int type, void *data)
103 {
104 	int error;
105 
106 	switch (type) {
107 	    case MOD_LOAD:
108 		error = 0;
109 		sx_init(&iconv_lock, "iconv");
110 		break;
111 	    case MOD_UNLOAD:
112 		error = iconv_mod_unload();
113 		break;
114 	    default:
115 		error = EINVAL;
116 	}
117 	return error;
118 }
119 
120 static moduledata_t iconv_mod = {
121 	"iconv", iconv_mod_handler, NULL
122 };
123 
124 DECLARE_MODULE(iconv, iconv_mod, SI_SUB_DRIVERS, SI_ORDER_SECOND);
125 
126 static int
iconv_register_converter(struct iconv_converter_class * dcp)127 iconv_register_converter(struct iconv_converter_class *dcp)
128 {
129 	kobj_class_compile((struct kobj_class*)dcp);
130 	dcp->refs++;
131 	TAILQ_INSERT_TAIL(&iconv_converters, dcp, cc_link);
132 	return 0;
133 }
134 
135 static int
iconv_unregister_converter(struct iconv_converter_class * dcp)136 iconv_unregister_converter(struct iconv_converter_class *dcp)
137 {
138 	dcp->refs--;
139 	if (dcp->refs > 1) {
140 		ICDEBUG("converter has %d references left\n", dcp->refs);
141 		return EBUSY;
142 	}
143 	TAILQ_REMOVE(&iconv_converters, dcp, cc_link);
144 	kobj_class_free((struct kobj_class*)dcp);
145 	return 0;
146 }
147 
148 static int
iconv_lookupconv(const char * name,struct iconv_converter_class ** dcpp)149 iconv_lookupconv(const char *name, struct iconv_converter_class **dcpp)
150 {
151 	struct iconv_converter_class *dcp;
152 
153 	TAILQ_FOREACH(dcp, &iconv_converters, cc_link) {
154 		if (name == NULL)
155 			continue;
156 		if (strcmp(name, ICONV_CONVERTER_NAME(dcp)) == 0) {
157 			if (dcpp)
158 				*dcpp = dcp;
159 			return 0;
160 		}
161 	}
162 	return ENOENT;
163 }
164 
165 static int
iconv_lookupcs(const char * to,const char * from,struct iconv_cspair ** cspp)166 iconv_lookupcs(const char *to, const char *from, struct iconv_cspair **cspp)
167 {
168 	struct iconv_cspair *csp;
169 
170 	TAILQ_FOREACH(csp, &iconv_cslist, cp_link) {
171 		if (strcasecmp(csp->cp_to, to) == 0 &&
172 		    strcasecmp(csp->cp_from, from) == 0) {
173 			if (cspp)
174 				*cspp = csp;
175 			return 0;
176 		}
177 	}
178 	return ENOENT;
179 }
180 
181 static int
iconv_register_cspair(const char * to,const char * from,struct iconv_converter_class * dcp,void * data,struct iconv_cspair ** cspp)182 iconv_register_cspair(const char *to, const char *from,
183 	struct iconv_converter_class *dcp, void *data,
184 	struct iconv_cspair **cspp)
185 {
186 	struct iconv_cspair *csp;
187 	char *cp;
188 	int csize, ucsto, ucsfrom;
189 
190 	if (iconv_lookupcs(to, from, NULL) == 0)
191 		return EEXIST;
192 	csize = sizeof(*csp);
193 	ucsto = strcmp(to, iconv_unicode_string) == 0;
194 	if (!ucsto)
195 		csize += strlen(to) + 1;
196 	ucsfrom = strcmp(from, iconv_unicode_string) == 0;
197 	if (!ucsfrom)
198 		csize += strlen(from) + 1;
199 	csp = malloc(csize, M_ICONV, M_WAITOK);
200 	bzero(csp, csize);
201 	csp->cp_id = iconv_csid++;
202 	csp->cp_dcp = dcp;
203 	cp = (char*)(csp + 1);
204 	if (!ucsto) {
205 		strcpy(cp, to);
206 		csp->cp_to = cp;
207 		cp += strlen(cp) + 1;
208 	} else
209 		csp->cp_to = iconv_unicode_string;
210 	if (!ucsfrom) {
211 		strcpy(cp, from);
212 		csp->cp_from = cp;
213 	} else
214 		csp->cp_from = iconv_unicode_string;
215 	csp->cp_data = data;
216 
217 	TAILQ_INSERT_TAIL(&iconv_cslist, csp, cp_link);
218 	*cspp = csp;
219 	return 0;
220 }
221 
222 static void
iconv_unregister_cspair(struct iconv_cspair * csp)223 iconv_unregister_cspair(struct iconv_cspair *csp)
224 {
225 	TAILQ_REMOVE(&iconv_cslist, csp, cp_link);
226 	if (csp->cp_data)
227 		free(csp->cp_data, M_ICONVDATA);
228 	free(csp, M_ICONV);
229 }
230 
231 /*
232  * Lookup and create an instance of converter.
233  * Currently this layer didn't have associated 'instance' structure
234  * to avoid unnesessary memory allocation.
235  */
236 int
iconv_open(const char * to,const char * from,void ** handle)237 iconv_open(const char *to, const char *from, void **handle)
238 {
239 	struct iconv_cspair *csp, *cspfrom, *cspto;
240 	struct iconv_converter_class *dcp;
241 	const char *cnvname;
242 	int error;
243 
244 	/*
245 	 * First, lookup fully qualified cspairs
246 	 */
247 	error = iconv_lookupcs(to, from, &csp);
248 	if (error == 0)
249 		return ICONV_CONVERTER_OPEN(csp->cp_dcp, csp, NULL, handle);
250 
251 	/*
252 	 * Well, nothing found. Now try to construct a composite conversion
253 	 * ToDo: add a 'capability' field to converter
254 	 */
255 	TAILQ_FOREACH(dcp, &iconv_converters, cc_link) {
256 		cnvname = ICONV_CONVERTER_NAME(dcp);
257 		if (cnvname == NULL)
258 			continue;
259 		error = iconv_lookupcs(cnvname, from, &cspfrom);
260 		if (error)
261 			continue;
262 		error = iconv_lookupcs(to, cnvname, &cspto);
263 		if (error)
264 			continue;
265 		/*
266 		 * Fine, we're found a pair which can be combined together
267 		 */
268 		return ICONV_CONVERTER_OPEN(dcp, cspto, cspfrom, handle);
269 	}
270 	return ENOENT;
271 }
272 
273 int
iconv_close(void * handle)274 iconv_close(void *handle)
275 {
276 	return ICONV_CONVERTER_CLOSE(handle);
277 }
278 
279 int
iconv_conv(void * handle,const char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)280 iconv_conv(void *handle, const char **inbuf,
281 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft)
282 {
283 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 0, 0);
284 }
285 
286 int
iconv_conv_case(void * handle,const char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft,int casetype)287 iconv_conv_case(void *handle, const char **inbuf,
288 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft, int casetype)
289 {
290 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 0, casetype);
291 }
292 
293 int
iconv_convchr(void * handle,const char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)294 iconv_convchr(void *handle, const char **inbuf,
295 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft)
296 {
297 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 1, 0);
298 }
299 
300 int
iconv_convchr_case(void * handle,const char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft,int casetype)301 iconv_convchr_case(void *handle, const char **inbuf,
302 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft, int casetype)
303 {
304 	return ICONV_CONVERTER_CONV(handle, inbuf, inbytesleft, outbuf, outbytesleft, 1, casetype);
305 }
306 
307 int
towlower(int c,void * handle)308 towlower(int c, void *handle)
309 {
310 	return ICONV_CONVERTER_TOLOWER(handle, c);
311 }
312 
313 int
towupper(int c,void * handle)314 towupper(int c, void *handle)
315 {
316 	return ICONV_CONVERTER_TOUPPER(handle, c);
317 }
318 
319 /*
320  * Give a list of loaded converters. Each name terminated with 0.
321  * An empty string terminates the list.
322  */
323 static int
iconv_sysctl_drvlist(SYSCTL_HANDLER_ARGS)324 iconv_sysctl_drvlist(SYSCTL_HANDLER_ARGS)
325 {
326 	struct iconv_converter_class *dcp;
327 	const char *name;
328 	char spc;
329 	int error;
330 
331 	error = 0;
332 	sx_slock(&iconv_lock);
333 	TAILQ_FOREACH(dcp, &iconv_converters, cc_link) {
334 		name = ICONV_CONVERTER_NAME(dcp);
335 		if (name == NULL)
336 			continue;
337 		error = SYSCTL_OUT(req, name, strlen(name) + 1);
338 		if (error)
339 			break;
340 	}
341 	sx_sunlock(&iconv_lock);
342 	if (error)
343 		return error;
344 	spc = 0;
345 	error = SYSCTL_OUT(req, &spc, sizeof(spc));
346 	return error;
347 }
348 
349 SYSCTL_PROC(_kern_iconv, OID_AUTO, drvlist,
350     CTLFLAG_RD | CTLTYPE_OPAQUE | CTLFLAG_MPSAFE, NULL, 0,
351     iconv_sysctl_drvlist, "S,xlat",
352     "registered converters");
353 
354 /*
355  * List all available charset pairs.
356  */
357 static int
iconv_sysctl_cslist(SYSCTL_HANDLER_ARGS)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,
383     CTLFLAG_RD | CTLTYPE_OPAQUE | CTLFLAG_MPSAFE, NULL, 0,
384     iconv_sysctl_cslist, "S,xlat",
385     "registered charset pairs");
386 
387 int
iconv_add(const char * converter,const char * to,const char * from)388 iconv_add(const char *converter, const char *to, const char *from)
389 {
390 	struct iconv_converter_class *dcp;
391 	struct iconv_cspair *csp;
392 
393 	if (iconv_lookupconv(converter, &dcp) != 0)
394 		return EINVAL;
395 
396 	return iconv_register_cspair(to, from, dcp, NULL, &csp);
397 }
398 
399 /*
400  * Add new charset pair
401  */
402 static int
iconv_sysctl_add(SYSCTL_HANDLER_ARGS)403 iconv_sysctl_add(SYSCTL_HANDLER_ARGS)
404 {
405 	struct iconv_converter_class *dcp;
406 	struct iconv_cspair *csp;
407 	struct iconv_add_in din;
408 	struct iconv_add_out dout;
409 	int error;
410 
411 	error = SYSCTL_IN(req, &din, sizeof(din));
412 	if (error)
413 		return error;
414 	if (din.ia_version != ICONV_ADD_VER)
415 		return EINVAL;
416 	if (din.ia_datalen > ICONV_CSMAXDATALEN)
417 		return EINVAL;
418 	if (strnlen(din.ia_from, sizeof(din.ia_from)) >= ICONV_CSNMAXLEN)
419 		return EINVAL;
420 	if (strnlen(din.ia_to, sizeof(din.ia_to)) >= ICONV_CSNMAXLEN)
421 		return EINVAL;
422 	if (strnlen(din.ia_converter, sizeof(din.ia_converter)) >= ICONV_CNVNMAXLEN)
423 		return EINVAL;
424 	if (iconv_lookupconv(din.ia_converter, &dcp) != 0)
425 		return EINVAL;
426 	sx_xlock(&iconv_lock);
427 	error = iconv_register_cspair(din.ia_to, din.ia_from, dcp, NULL, &csp);
428 	if (error) {
429 		sx_xunlock(&iconv_lock);
430 		return error;
431 	}
432 	if (din.ia_datalen) {
433 		csp->cp_data = malloc(din.ia_datalen, M_ICONVDATA, M_WAITOK);
434 		error = copyin(din.ia_data, csp->cp_data, din.ia_datalen);
435 		if (error)
436 			goto bad;
437 	}
438 	dout.ia_csid = csp->cp_id;
439 	error = SYSCTL_OUT(req, &dout, sizeof(dout));
440 	if (error)
441 		goto bad;
442 	sx_xunlock(&iconv_lock);
443 	ICDEBUG("%s => %s, %d bytes\n",din.ia_from, din.ia_to, din.ia_datalen);
444 	return 0;
445 bad:
446 	iconv_unregister_cspair(csp);
447 	sx_xunlock(&iconv_lock);
448 	return error;
449 }
450 
451 SYSCTL_PROC(_kern_iconv, OID_AUTO, add,
452     CTLFLAG_RW | CTLTYPE_OPAQUE | CTLFLAG_MPSAFE, NULL, 0,
453     iconv_sysctl_add, "S,xlat",
454     "register charset pair");
455 
456 /*
457  * Default stubs for converters
458  */
459 int
iconv_converter_initstub(struct iconv_converter_class * dp)460 iconv_converter_initstub(struct iconv_converter_class *dp)
461 {
462 	return 0;
463 }
464 
465 int
iconv_converter_donestub(struct iconv_converter_class * dp)466 iconv_converter_donestub(struct iconv_converter_class *dp)
467 {
468 	return 0;
469 }
470 
471 int
iconv_converter_tolowerstub(int c,void * handle)472 iconv_converter_tolowerstub(int c, void *handle)
473 {
474 	return (c);
475 }
476 
477 int
iconv_converter_handler(module_t mod,int type,void * data)478 iconv_converter_handler(module_t mod, int type, void *data)
479 {
480 	struct iconv_converter_class *dcp = data;
481 	int error;
482 
483 	switch (type) {
484 	    case MOD_LOAD:
485 		sx_xlock(&iconv_lock);
486 		error = iconv_register_converter(dcp);
487 		if (error) {
488 			sx_xunlock(&iconv_lock);
489 			break;
490 		}
491 		error = ICONV_CONVERTER_INIT(dcp);
492 		if (error)
493 			iconv_unregister_converter(dcp);
494 		sx_xunlock(&iconv_lock);
495 		break;
496 	    case MOD_UNLOAD:
497 		sx_xlock(&iconv_lock);
498 		ICONV_CONVERTER_DONE(dcp);
499 		error = iconv_unregister_converter(dcp);
500 		sx_xunlock(&iconv_lock);
501 		break;
502 	    default:
503 		error = EINVAL;
504 	}
505 	return error;
506 }
507 
508 /*
509  * Common used functions (don't use with unicode)
510  */
511 char *
iconv_convstr(void * handle,char * dst,const char * src)512 iconv_convstr(void *handle, char *dst, const char *src)
513 {
514 	char *p = dst;
515 	size_t inlen, outlen;
516 	int error;
517 
518 	if (handle == NULL) {
519 		strcpy(dst, src);
520 		return dst;
521 	}
522 	inlen = outlen = strlen(src);
523 	error = iconv_conv(handle, NULL, NULL, &p, &outlen);
524 	if (error)
525 		return NULL;
526 	error = iconv_conv(handle, &src, &inlen, &p, &outlen);
527 	if (error)
528 		return NULL;
529 	*p = 0;
530 	return dst;
531 }
532 
533 void *
iconv_convmem(void * handle,void * dst,const void * src,int size)534 iconv_convmem(void *handle, void *dst, const void *src, int size)
535 {
536 	const char *s = src;
537 	char *d = dst;
538 	size_t inlen, outlen;
539 	int error;
540 
541 	if (size == 0)
542 		return dst;
543 	if (handle == NULL) {
544 		memcpy(dst, src, size);
545 		return dst;
546 	}
547 	inlen = outlen = size;
548 	error = iconv_conv(handle, NULL, NULL, &d, &outlen);
549 	if (error)
550 		return NULL;
551 	error = iconv_conv(handle, &s, &inlen, &d, &outlen);
552 	if (error)
553 		return NULL;
554 	return dst;
555 }
556 
557 int
iconv_lookupcp(char ** cpp,const char * s)558 iconv_lookupcp(char **cpp, const char *s)
559 {
560 	if (cpp == NULL) {
561 		ICDEBUG("warning a NULL list passed\n", "");
562 		return ENOENT;
563 	}
564 	for (; *cpp; cpp++)
565 		if (strcmp(*cpp, s) == 0)
566 			return 0;
567 	return ENOENT;
568 }
569 
570 /*
571  * Return if fsname is in use of not
572  */
573 int
iconv_vfs_refcount(const char * fsname)574 iconv_vfs_refcount(const char *fsname)
575 {
576 	struct vfsconf *vfsp;
577 
578 	vfsp = vfs_byname(fsname);
579 	if (vfsp != NULL && vfsp->vfc_refcount > 0)
580 		return (EBUSY);
581 	return (0);
582 }
583