xref: /freebsd/sys/compat/linux/linux_mib.c (revision 7aa383846770374466b1dcb2cefd71bde9acf463)
1 /*-
2  * Copyright (c) 1999 Marcel Moolenaar
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer
10  *    in this position and unchanged.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  * 3. The name of the author may not be used to endorse or promote products
15  *    derived from this software without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27  */
28 
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
31 
32 #include <sys/param.h>
33 #include <sys/kernel.h>
34 #include <sys/systm.h>
35 #include <sys/sysctl.h>
36 #include <sys/proc.h>
37 #include <sys/malloc.h>
38 #include <sys/mount.h>
39 #include <sys/jail.h>
40 #include <sys/lock.h>
41 #include <sys/mutex.h>
42 #include <sys/sx.h>
43 
44 #include "opt_compat.h"
45 
46 #ifdef COMPAT_LINUX32
47 #include <machine/../linux32/linux.h>
48 #else
49 #include <machine/../linux/linux.h>
50 #endif
51 #include <compat/linux/linux_mib.h>
52 
53 struct linux_prison {
54 	char	pr_osname[LINUX_MAX_UTSNAME];
55 	char	pr_osrelease[LINUX_MAX_UTSNAME];
56 	int	pr_oss_version;
57 	int	pr_osrel;
58 };
59 
60 static struct linux_prison lprison0 = {
61 	.pr_osname =		"Linux",
62 	.pr_osrelease =		"2.6.16",
63 	.pr_oss_version =	0x030600,
64 	.pr_osrel =		2006016
65 };
66 
67 static unsigned linux_osd_jail_slot;
68 
69 SYSCTL_NODE(_compat, OID_AUTO, linux, CTLFLAG_RW, 0,
70 	    "Linux mode");
71 
72 static int
73 linux_sysctl_osname(SYSCTL_HANDLER_ARGS)
74 {
75 	char osname[LINUX_MAX_UTSNAME];
76 	int error;
77 
78 	linux_get_osname(req->td, osname);
79 	error = sysctl_handle_string(oidp, osname, LINUX_MAX_UTSNAME, req);
80 	if (error || req->newptr == NULL)
81 		return (error);
82 	error = linux_set_osname(req->td, osname);
83 	return (error);
84 }
85 
86 SYSCTL_PROC(_compat_linux, OID_AUTO, osname,
87 	    CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_MPSAFE,
88 	    0, 0, linux_sysctl_osname, "A",
89 	    "Linux kernel OS name");
90 
91 static int
92 linux_sysctl_osrelease(SYSCTL_HANDLER_ARGS)
93 {
94 	char osrelease[LINUX_MAX_UTSNAME];
95 	int error;
96 
97 	linux_get_osrelease(req->td, osrelease);
98 	error = sysctl_handle_string(oidp, osrelease, LINUX_MAX_UTSNAME, req);
99 	if (error || req->newptr == NULL)
100 		return (error);
101 	error = linux_set_osrelease(req->td, osrelease);
102 	return (error);
103 }
104 
105 SYSCTL_PROC(_compat_linux, OID_AUTO, osrelease,
106 	    CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_MPSAFE,
107 	    0, 0, linux_sysctl_osrelease, "A",
108 	    "Linux kernel OS release");
109 
110 static int
111 linux_sysctl_oss_version(SYSCTL_HANDLER_ARGS)
112 {
113 	int oss_version;
114 	int error;
115 
116 	oss_version = linux_get_oss_version(req->td);
117 	error = sysctl_handle_int(oidp, &oss_version, 0, req);
118 	if (error || req->newptr == NULL)
119 		return (error);
120 	error = linux_set_oss_version(req->td, oss_version);
121 	return (error);
122 }
123 
124 SYSCTL_PROC(_compat_linux, OID_AUTO, oss_version,
125 	    CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_MPSAFE,
126 	    0, 0, linux_sysctl_oss_version, "I",
127 	    "Linux OSS version");
128 
129 /*
130  * Map the osrelease into integer
131  */
132 static int
133 linux_map_osrel(char *osrelease, int *osrel)
134 {
135 	char *sep, *eosrelease;
136 	int len, v0, v1, v2, v;
137 
138 	len = strlen(osrelease);
139 	eosrelease = osrelease + len;
140 	v0 = strtol(osrelease, &sep, 10);
141 	if (osrelease == sep || sep + 1 >= eosrelease || *sep != '.')
142 		return (EINVAL);
143 	osrelease = sep + 1;
144 	v1 = strtol(osrelease, &sep, 10);
145 	if (osrelease == sep || sep + 1 >= eosrelease || *sep != '.')
146 		return (EINVAL);
147 	osrelease = sep + 1;
148 	v2 = strtol(osrelease, &sep, 10);
149 	if (osrelease == sep || sep != eosrelease)
150 		return (EINVAL);
151 
152 	v = v0 * 1000000 + v1 * 1000 + v2;
153 	if (v < 1000000)
154 		return (EINVAL);
155 
156 	*osrel = v;
157 	return (0);
158 }
159 
160 /*
161  * Find a prison with Linux info.
162  * Return the Linux info and the (locked) prison.
163  */
164 static struct linux_prison *
165 linux_find_prison(struct prison *spr, struct prison **prp)
166 {
167 	struct prison *pr;
168 	struct linux_prison *lpr;
169 
170 	if (!linux_osd_jail_slot)
171 		/* In case osd_register failed. */
172 		spr = &prison0;
173 	for (pr = spr;; pr = pr->pr_parent) {
174 		mtx_lock(&pr->pr_mtx);
175 		lpr = (pr == &prison0)
176 		    ? &lprison0
177 		    : osd_jail_get(pr, linux_osd_jail_slot);
178 		if (lpr != NULL)
179 			break;
180 		mtx_unlock(&pr->pr_mtx);
181 	}
182 	*prp = pr;
183 	return (lpr);
184 }
185 
186 /*
187  * Ensure a prison has its own Linux info.  If lprp is non-null, point it to
188  * the Linux info and lock the prison.
189  */
190 static int
191 linux_alloc_prison(struct prison *pr, struct linux_prison **lprp)
192 {
193 	struct prison *ppr;
194 	struct linux_prison *lpr, *nlpr;
195 	int error;
196 
197 	/* If this prison already has Linux info, return that. */
198 	error = 0;
199 	lpr = linux_find_prison(pr, &ppr);
200 	if (ppr == pr)
201 		goto done;
202 	/*
203 	 * Allocate a new info record.  Then check again, in case something
204 	 * changed during the allocation.
205 	 */
206 	mtx_unlock(&ppr->pr_mtx);
207 	nlpr = malloc(sizeof(struct linux_prison), M_PRISON, M_WAITOK);
208 	lpr = linux_find_prison(pr, &ppr);
209 	if (ppr == pr) {
210 		free(nlpr, M_PRISON);
211 		goto done;
212 	}
213 	/* Inherit the initial values from the ancestor. */
214 	mtx_lock(&pr->pr_mtx);
215 	error = osd_jail_set(pr, linux_osd_jail_slot, nlpr);
216 	if (error == 0) {
217 		bcopy(lpr, nlpr, sizeof(*lpr));
218 		lpr = nlpr;
219 	} else {
220 		free(nlpr, M_PRISON);
221 		lpr = NULL;
222 	}
223 	mtx_unlock(&ppr->pr_mtx);
224  done:
225 	if (lprp != NULL)
226 		*lprp = lpr;
227 	else
228 		mtx_unlock(&pr->pr_mtx);
229 	return (error);
230 }
231 
232 /*
233  * Jail OSD methods for Linux prison data.
234  */
235 static int
236 linux_prison_create(void *obj, void *data)
237 {
238 	struct prison *pr = obj;
239 	struct vfsoptlist *opts = data;
240 	int jsys;
241 
242 	if (vfs_copyopt(opts, "linux", &jsys, sizeof(jsys)) == 0 &&
243 	    jsys == JAIL_SYS_INHERIT)
244 		return (0);
245 	/*
246 	 * Inherit a prison's initial values from its parent
247 	 * (different from JAIL_SYS_INHERIT which also inherits changes).
248 	 */
249 	return linux_alloc_prison(pr, NULL);
250 }
251 
252 static int
253 linux_prison_check(void *obj __unused, void *data)
254 {
255 	struct vfsoptlist *opts = data;
256 	char *osname, *osrelease;
257 	int error, jsys, len, osrel, oss_version;
258 
259 	/* Check that the parameters are correct. */
260 	error = vfs_copyopt(opts, "linux", &jsys, sizeof(jsys));
261 	if (error != ENOENT) {
262 		if (error != 0)
263 			return (error);
264 		if (jsys != JAIL_SYS_NEW && jsys != JAIL_SYS_INHERIT)
265 			return (EINVAL);
266 	}
267 	error = vfs_getopt(opts, "linux.osname", (void **)&osname, &len);
268 	if (error != ENOENT) {
269 		if (error != 0)
270 			return (error);
271 		if (len == 0 || osname[len - 1] != '\0')
272 			return (EINVAL);
273 		if (len > LINUX_MAX_UTSNAME) {
274 			vfs_opterror(opts, "linux.osname too long");
275 			return (ENAMETOOLONG);
276 		}
277 	}
278 	error = vfs_getopt(opts, "linux.osrelease", (void **)&osrelease, &len);
279 	if (error != ENOENT) {
280 		if (error != 0)
281 			return (error);
282 		if (len == 0 || osrelease[len - 1] != '\0')
283 			return (EINVAL);
284 		if (len > LINUX_MAX_UTSNAME) {
285 			vfs_opterror(opts, "linux.osrelease too long");
286 			return (ENAMETOOLONG);
287 		}
288 		error = linux_map_osrel(osrelease, &osrel);
289 		if (error != 0) {
290 			vfs_opterror(opts, "linux.osrelease format error");
291 			return (error);
292 		}
293 	}
294 	error = vfs_copyopt(opts, "linux.oss_version", &oss_version,
295 	    sizeof(oss_version));
296 	return (error == ENOENT ? 0 : error);
297 }
298 
299 static int
300 linux_prison_set(void *obj, void *data)
301 {
302 	struct linux_prison *lpr;
303 	struct prison *pr = obj;
304 	struct vfsoptlist *opts = data;
305 	char *osname, *osrelease;
306 	int error, gotversion, jsys, len, oss_version;
307 
308 	/* Set the parameters, which should be correct. */
309 	error = vfs_copyopt(opts, "linux", &jsys, sizeof(jsys));
310 	if (error == ENOENT)
311 		jsys = -1;
312 	error = vfs_getopt(opts, "linux.osname", (void **)&osname, &len);
313 	if (error == ENOENT)
314 		osname = NULL;
315 	else
316 		jsys = JAIL_SYS_NEW;
317 	error = vfs_getopt(opts, "linux.osrelease", (void **)&osrelease, &len);
318 	if (error == ENOENT)
319 		osrelease = NULL;
320 	else
321 		jsys = JAIL_SYS_NEW;
322 	error = vfs_copyopt(opts, "linux.oss_version", &oss_version,
323 	    sizeof(oss_version));
324 	if (error == ENOENT)
325 		gotversion = 0;
326 	else {
327 		gotversion = 1;
328 		jsys = JAIL_SYS_NEW;
329 	}
330 	switch (jsys) {
331 	case JAIL_SYS_INHERIT:
332 		/* "linux=inherit": inherit the parent's Linux info. */
333 		mtx_lock(&pr->pr_mtx);
334 		osd_jail_del(pr, linux_osd_jail_slot);
335 		mtx_unlock(&pr->pr_mtx);
336 		break;
337 	case JAIL_SYS_NEW:
338 		/*
339 		 * "linux=new" or "linux.*":
340 		 * the prison gets its own Linux info.
341 		 */
342 		error = linux_alloc_prison(pr, &lpr);
343 		if (error) {
344 			mtx_unlock(&pr->pr_mtx);
345 			return (error);
346 		}
347 		if (osrelease) {
348 			error = linux_map_osrel(osrelease, &lpr->pr_osrel);
349 			if (error) {
350 				mtx_unlock(&pr->pr_mtx);
351 				return (error);
352 			}
353 			strlcpy(lpr->pr_osrelease, osrelease,
354 			    LINUX_MAX_UTSNAME);
355 		}
356 		if (osname)
357 			strlcpy(lpr->pr_osname, osname, LINUX_MAX_UTSNAME);
358 		if (gotversion)
359 			lpr->pr_oss_version = oss_version;
360 		mtx_unlock(&pr->pr_mtx);
361 	}
362 	return (0);
363 }
364 
365 SYSCTL_JAIL_PARAM_SYS_NODE(linux, CTLFLAG_RW, "Jail Linux parameters");
366 SYSCTL_JAIL_PARAM_STRING(_linux, osname, CTLFLAG_RW, LINUX_MAX_UTSNAME,
367     "Jail Linux kernel OS name");
368 SYSCTL_JAIL_PARAM_STRING(_linux, osrelease, CTLFLAG_RW, LINUX_MAX_UTSNAME,
369     "Jail Linux kernel OS release");
370 SYSCTL_JAIL_PARAM(_linux, oss_version, CTLTYPE_INT | CTLFLAG_RW,
371     "I", "Jail Linux OSS version");
372 
373 static int
374 linux_prison_get(void *obj, void *data)
375 {
376 	struct linux_prison *lpr;
377 	struct prison *ppr;
378 	struct prison *pr = obj;
379 	struct vfsoptlist *opts = data;
380 	int error, i;
381 
382 	static int version0;
383 
384 	/* See if this prison is the one with the Linux info. */
385 	lpr = linux_find_prison(pr, &ppr);
386 	i = (ppr == pr) ? JAIL_SYS_NEW : JAIL_SYS_INHERIT;
387 	error = vfs_setopt(opts, "linux", &i, sizeof(i));
388 	if (error != 0 && error != ENOENT)
389 		goto done;
390 	if (i) {
391 		error = vfs_setopts(opts, "linux.osname", lpr->pr_osname);
392 		if (error != 0 && error != ENOENT)
393 			goto done;
394 		error = vfs_setopts(opts, "linux.osrelease", lpr->pr_osrelease);
395 		if (error != 0 && error != ENOENT)
396 			goto done;
397 		error = vfs_setopt(opts, "linux.oss_version",
398 		    &lpr->pr_oss_version, sizeof(lpr->pr_oss_version));
399 		if (error != 0 && error != ENOENT)
400 			goto done;
401 	} else {
402 		/*
403 		 * If this prison is inheriting its Linux info, report
404 		 * empty/zero parameters.
405 		 */
406 		error = vfs_setopts(opts, "linux.osname", "");
407 		if (error != 0 && error != ENOENT)
408 			goto done;
409 		error = vfs_setopts(opts, "linux.osrelease", "");
410 		if (error != 0 && error != ENOENT)
411 			goto done;
412 		error = vfs_setopt(opts, "linux.oss_version", &version0,
413 		    sizeof(lpr->pr_oss_version));
414 		if (error != 0 && error != ENOENT)
415 			goto done;
416 	}
417 	error = 0;
418 
419  done:
420 	mtx_unlock(&ppr->pr_mtx);
421 	return (error);
422 }
423 
424 static void
425 linux_prison_destructor(void *data)
426 {
427 
428 	free(data, M_PRISON);
429 }
430 
431 void
432 linux_osd_jail_register(void)
433 {
434 	struct prison *pr;
435 	osd_method_t methods[PR_MAXMETHOD] = {
436 	    [PR_METHOD_CREATE] =	linux_prison_create,
437 	    [PR_METHOD_GET] =		linux_prison_get,
438 	    [PR_METHOD_SET] =		linux_prison_set,
439 	    [PR_METHOD_CHECK] =		linux_prison_check
440 	};
441 
442 	linux_osd_jail_slot =
443 	    osd_jail_register(linux_prison_destructor, methods);
444 	if (linux_osd_jail_slot > 0) {
445 		/* Copy the system linux info to any current prisons. */
446 		sx_xlock(&allprison_lock);
447 		TAILQ_FOREACH(pr, &allprison, pr_list)
448 			(void)linux_alloc_prison(pr, NULL);
449 		sx_xunlock(&allprison_lock);
450 	}
451 }
452 
453 void
454 linux_osd_jail_deregister(void)
455 {
456 
457 	if (linux_osd_jail_slot)
458 		osd_jail_deregister(linux_osd_jail_slot);
459 }
460 
461 void
462 linux_get_osname(struct thread *td, char *dst)
463 {
464 	struct prison *pr;
465 	struct linux_prison *lpr;
466 
467 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
468 	bcopy(lpr->pr_osname, dst, LINUX_MAX_UTSNAME);
469 	mtx_unlock(&pr->pr_mtx);
470 }
471 
472 int
473 linux_set_osname(struct thread *td, char *osname)
474 {
475 	struct prison *pr;
476 	struct linux_prison *lpr;
477 
478 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
479 	strlcpy(lpr->pr_osname, osname, LINUX_MAX_UTSNAME);
480 	mtx_unlock(&pr->pr_mtx);
481 	return (0);
482 }
483 
484 void
485 linux_get_osrelease(struct thread *td, char *dst)
486 {
487 	struct prison *pr;
488 	struct linux_prison *lpr;
489 
490 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
491 	bcopy(lpr->pr_osrelease, dst, LINUX_MAX_UTSNAME);
492 	mtx_unlock(&pr->pr_mtx);
493 }
494 
495 int
496 linux_kernver(struct thread *td)
497 {
498 	struct prison *pr;
499 	struct linux_prison *lpr;
500 	int osrel;
501 
502 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
503 	osrel = lpr->pr_osrel;
504 	mtx_unlock(&pr->pr_mtx);
505 	return (osrel);
506 }
507 
508 int
509 linux_set_osrelease(struct thread *td, char *osrelease)
510 {
511 	struct prison *pr;
512 	struct linux_prison *lpr;
513 	int error;
514 
515 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
516 	error = linux_map_osrel(osrelease, &lpr->pr_osrel);
517 	if (error == 0)
518 		strlcpy(lpr->pr_osrelease, osrelease, LINUX_MAX_UTSNAME);
519 	mtx_unlock(&pr->pr_mtx);
520 	return (error);
521 }
522 
523 int
524 linux_get_oss_version(struct thread *td)
525 {
526 	struct prison *pr;
527 	struct linux_prison *lpr;
528 	int version;
529 
530 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
531 	version = lpr->pr_oss_version;
532 	mtx_unlock(&pr->pr_mtx);
533 	return (version);
534 }
535 
536 int
537 linux_set_oss_version(struct thread *td, int oss_version)
538 {
539 	struct prison *pr;
540 	struct linux_prison *lpr;
541 
542 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
543 	lpr->pr_oss_version = oss_version;
544 	mtx_unlock(&pr->pr_mtx);
545 	return (0);
546 }
547 
548 #if defined(DEBUG) || defined(KTR)
549 
550 u_char linux_debug_map[howmany(LINUX_SYS_MAXSYSCALL, sizeof(u_char))];
551 
552 static int
553 linux_debug(int syscall, int toggle, int global)
554 {
555 
556 	if (global) {
557 		char c = toggle ? 0 : 0xff;
558 
559 		memset(linux_debug_map, c, sizeof(linux_debug_map));
560 		return (0);
561 	}
562 	if (syscall < 0 || syscall >= LINUX_SYS_MAXSYSCALL)
563 		return (EINVAL);
564 	if (toggle)
565 		clrbit(linux_debug_map, syscall);
566 	else
567 		setbit(linux_debug_map, syscall);
568 	return (0);
569 }
570 
571 /*
572  * Usage: sysctl linux.debug=<syscall_nr>.<0/1>
573  *
574  *    E.g.: sysctl linux.debug=21.0
575  *
576  * As a special case, syscall "all" will apply to all syscalls globally.
577  */
578 #define LINUX_MAX_DEBUGSTR	16
579 static int
580 linux_sysctl_debug(SYSCTL_HANDLER_ARGS)
581 {
582 	char value[LINUX_MAX_DEBUGSTR], *p;
583 	int error, sysc, toggle;
584 	int global = 0;
585 
586 	value[0] = '\0';
587 	error = sysctl_handle_string(oidp, value, LINUX_MAX_DEBUGSTR, req);
588 	if (error || req->newptr == NULL)
589 		return (error);
590 	for (p = value; *p != '\0' && *p != '.'; p++);
591 	if (*p == '\0')
592 		return (EINVAL);
593 	*p++ = '\0';
594 	sysc = strtol(value, NULL, 0);
595 	toggle = strtol(p, NULL, 0);
596 	if (strcmp(value, "all") == 0)
597 		global = 1;
598 	error = linux_debug(sysc, toggle, global);
599 	return (error);
600 }
601 
602 SYSCTL_PROC(_compat_linux, OID_AUTO, debug,
603             CTLTYPE_STRING | CTLFLAG_RW,
604             0, 0, linux_sysctl_debug, "A",
605             "Linux debugging control");
606 
607 #endif /* DEBUG || KTR */
608