xref: /freebsd/sys/compat/linux/linux_mib.c (revision 25dd52cdb10d223b9258836e23cc6ae4ea333b86)
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 
241 	if (vfs_flagopt(opts, "nolinux", NULL, 0))
242 		return (0);
243 	/*
244 	 * Inherit a prison's initial values from its parent
245 	 * (different from NULL which also inherits changes).
246 	 */
247 	return linux_alloc_prison(pr, NULL);
248 }
249 
250 static int
251 linux_prison_check(void *obj __unused, void *data)
252 {
253 	struct vfsoptlist *opts = data;
254 	char *osname, *osrelease;
255 	int error, len, osrel, oss_version;
256 
257 	/* Check that the parameters are correct. */
258 	(void)vfs_flagopt(opts, "linux", NULL, 0);
259 	(void)vfs_flagopt(opts, "nolinux", NULL, 0);
260 	error = vfs_getopt(opts, "linux.osname", (void **)&osname, &len);
261 	if (error != ENOENT) {
262 		if (error != 0)
263 			return (error);
264 		if (len == 0 || osname[len - 1] != '\0')
265 			return (EINVAL);
266 		if (len > LINUX_MAX_UTSNAME) {
267 			vfs_opterror(opts, "linux.osname too long");
268 			return (ENAMETOOLONG);
269 		}
270 	}
271 	error = vfs_getopt(opts, "linux.osrelease", (void **)&osrelease, &len);
272 	if (error != ENOENT) {
273 		if (error != 0)
274 			return (error);
275 		if (len == 0 || osrelease[len - 1] != '\0')
276 			return (EINVAL);
277 		if (len > LINUX_MAX_UTSNAME) {
278 			vfs_opterror(opts, "linux.osrelease too long");
279 			return (ENAMETOOLONG);
280 		}
281 		error = linux_map_osrel(osrelease, &osrel);
282 		if (error != 0) {
283 			vfs_opterror(opts, "linux.osrelease format error");
284 			return (error);
285 		}
286 	}
287 	error = vfs_copyopt(opts, "linux.oss_version", &oss_version,
288 	    sizeof(oss_version));
289 	return (error == ENOENT ? 0 : error);
290 }
291 
292 static int
293 linux_prison_set(void *obj, void *data)
294 {
295 	struct linux_prison *lpr;
296 	struct prison *pr = obj;
297 	struct vfsoptlist *opts = data;
298 	char *osname, *osrelease;
299 	int error, gotversion, len, nolinux, oss_version, yeslinux;
300 
301 	/* Set the parameters, which should be correct. */
302 	yeslinux = vfs_flagopt(opts, "linux", NULL, 0);
303 	nolinux = vfs_flagopt(opts, "nolinux", NULL, 0);
304 	error = vfs_getopt(opts, "linux.osname", (void **)&osname, &len);
305 	if (error == ENOENT)
306 		osname = NULL;
307 	else
308 		yeslinux = 1;
309 	error = vfs_getopt(opts, "linux.osrelease", (void **)&osrelease, &len);
310 	if (error == ENOENT)
311 		osrelease = NULL;
312 	else
313 		yeslinux = 1;
314 	error = vfs_copyopt(opts, "linux.oss_version", &oss_version,
315 	    sizeof(oss_version));
316 	gotversion = (error == 0);
317 	yeslinux |= gotversion;
318 	if (nolinux) {
319 		/* "nolinux": inherit the parent's Linux info. */
320 		mtx_lock(&pr->pr_mtx);
321 		osd_jail_del(pr, linux_osd_jail_slot);
322 		mtx_unlock(&pr->pr_mtx);
323 	} else if (yeslinux) {
324 		/*
325 		 * "linux" or "linux.*":
326 		 * the prison gets its own Linux info.
327 		 */
328 		error = linux_alloc_prison(pr, &lpr);
329 		if (error) {
330 			mtx_unlock(&pr->pr_mtx);
331 			return (error);
332 		}
333 		if (osrelease) {
334 			error = linux_map_osrel(osrelease, &lpr->pr_osrel);
335 			if (error) {
336 				mtx_unlock(&pr->pr_mtx);
337 				return (error);
338 			}
339 			strlcpy(lpr->pr_osrelease, osrelease,
340 			    LINUX_MAX_UTSNAME);
341 		}
342 		if (osname)
343 			strlcpy(lpr->pr_osname, osname, LINUX_MAX_UTSNAME);
344 		if (gotversion)
345 			lpr->pr_oss_version = oss_version;
346 		mtx_unlock(&pr->pr_mtx);
347 	}
348 	return (0);
349 }
350 
351 SYSCTL_JAIL_PARAM_NODE(linux, "Jail Linux parameters");
352 SYSCTL_JAIL_PARAM(, nolinux, CTLTYPE_INT | CTLFLAG_RW,
353     "BN", "Jail w/ no Linux parameters");
354 SYSCTL_JAIL_PARAM_STRING(_linux, osname, CTLFLAG_RW, LINUX_MAX_UTSNAME,
355     "Jail Linux kernel OS name");
356 SYSCTL_JAIL_PARAM_STRING(_linux, osrelease, CTLFLAG_RW, LINUX_MAX_UTSNAME,
357     "Jail Linux kernel OS release");
358 SYSCTL_JAIL_PARAM(_linux, oss_version, CTLTYPE_INT | CTLFLAG_RW,
359     "I", "Jail Linux OSS version");
360 
361 static int
362 linux_prison_get(void *obj, void *data)
363 {
364 	struct linux_prison *lpr;
365 	struct prison *ppr;
366 	struct prison *pr = obj;
367 	struct vfsoptlist *opts = data;
368 	int error, i;
369 
370 	static int version0;
371 
372 	/* See if this prison is the one with the Linux info. */
373 	lpr = linux_find_prison(pr, &ppr);
374 	i = (ppr == pr);
375 	error = vfs_setopt(opts, "linux", &i, sizeof(i));
376 	if (error != 0 && error != ENOENT)
377 		goto done;
378 	i = !i;
379 	error = vfs_setopt(opts, "nolinux", &i, sizeof(i));
380 	if (error != 0 && error != ENOENT)
381 		goto done;
382 	if (i) {
383 		/*
384 		 * If this prison is inheriting its Linux info, report
385 		 * empty/zero parameters.
386 		 */
387 		error = vfs_setopts(opts, "linux.osname", "");
388 		if (error != 0 && error != ENOENT)
389 			goto done;
390 		error = vfs_setopts(opts, "linux.osrelease", "");
391 		if (error != 0 && error != ENOENT)
392 			goto done;
393 		error = vfs_setopt(opts, "linux.oss_version", &version0,
394 		    sizeof(lpr->pr_oss_version));
395 		if (error != 0 && error != ENOENT)
396 			goto done;
397 	} else {
398 		error = vfs_setopts(opts, "linux.osname", lpr->pr_osname);
399 		if (error != 0 && error != ENOENT)
400 			goto done;
401 		error = vfs_setopts(opts, "linux.osrelease", lpr->pr_osrelease);
402 		if (error != 0 && error != ENOENT)
403 			goto done;
404 		error = vfs_setopt(opts, "linux.oss_version",
405 		    &lpr->pr_oss_version, sizeof(lpr->pr_oss_version));
406 		if (error != 0 && error != ENOENT)
407 			goto done;
408 	}
409 	error = 0;
410 
411  done:
412 	mtx_unlock(&ppr->pr_mtx);
413 	return (error);
414 }
415 
416 static void
417 linux_prison_destructor(void *data)
418 {
419 
420 	free(data, M_PRISON);
421 }
422 
423 void
424 linux_osd_jail_register(void)
425 {
426 	struct prison *pr;
427 	osd_method_t methods[PR_MAXMETHOD] = {
428 	    [PR_METHOD_CREATE] =	linux_prison_create,
429 	    [PR_METHOD_GET] =		linux_prison_get,
430 	    [PR_METHOD_SET] =		linux_prison_set,
431 	    [PR_METHOD_CHECK] =		linux_prison_check
432 	};
433 
434 	linux_osd_jail_slot =
435 	    osd_jail_register(linux_prison_destructor, methods);
436 	if (linux_osd_jail_slot > 0) {
437 		/* Copy the system linux info to any current prisons. */
438 		sx_xlock(&allprison_lock);
439 		TAILQ_FOREACH(pr, &allprison, pr_list)
440 			(void)linux_alloc_prison(pr, NULL);
441 		sx_xunlock(&allprison_lock);
442 	}
443 }
444 
445 void
446 linux_osd_jail_deregister(void)
447 {
448 
449 	if (linux_osd_jail_slot)
450 		osd_jail_deregister(linux_osd_jail_slot);
451 }
452 
453 void
454 linux_get_osname(struct thread *td, char *dst)
455 {
456 	struct prison *pr;
457 	struct linux_prison *lpr;
458 
459 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
460 	bcopy(lpr->pr_osname, dst, LINUX_MAX_UTSNAME);
461 	mtx_unlock(&pr->pr_mtx);
462 }
463 
464 int
465 linux_set_osname(struct thread *td, char *osname)
466 {
467 	struct prison *pr;
468 	struct linux_prison *lpr;
469 
470 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
471 	strlcpy(lpr->pr_osname, osname, LINUX_MAX_UTSNAME);
472 	mtx_unlock(&pr->pr_mtx);
473 	return (0);
474 }
475 
476 void
477 linux_get_osrelease(struct thread *td, char *dst)
478 {
479 	struct prison *pr;
480 	struct linux_prison *lpr;
481 
482 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
483 	bcopy(lpr->pr_osrelease, dst, LINUX_MAX_UTSNAME);
484 	mtx_unlock(&pr->pr_mtx);
485 }
486 
487 int
488 linux_kernver(struct thread *td)
489 {
490 	struct prison *pr;
491 	struct linux_prison *lpr;
492 	int osrel;
493 
494 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
495 	osrel = lpr->pr_osrel;
496 	mtx_unlock(&pr->pr_mtx);
497 	return (osrel);
498 }
499 
500 int
501 linux_set_osrelease(struct thread *td, char *osrelease)
502 {
503 	struct prison *pr;
504 	struct linux_prison *lpr;
505 	int error;
506 
507 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
508 	error = linux_map_osrel(osrelease, &lpr->pr_osrel);
509 	if (error == 0)
510 		strlcpy(lpr->pr_osrelease, osrelease, LINUX_MAX_UTSNAME);
511 	mtx_unlock(&pr->pr_mtx);
512 	return (error);
513 }
514 
515 int
516 linux_get_oss_version(struct thread *td)
517 {
518 	struct prison *pr;
519 	struct linux_prison *lpr;
520 	int version;
521 
522 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
523 	version = lpr->pr_oss_version;
524 	mtx_unlock(&pr->pr_mtx);
525 	return (version);
526 }
527 
528 int
529 linux_set_oss_version(struct thread *td, int oss_version)
530 {
531 	struct prison *pr;
532 	struct linux_prison *lpr;
533 
534 	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
535 	lpr->pr_oss_version = oss_version;
536 	mtx_unlock(&pr->pr_mtx);
537 	return (0);
538 }
539 
540 #if defined(DEBUG) || defined(KTR)
541 
542 u_char linux_debug_map[howmany(LINUX_SYS_MAXSYSCALL, sizeof(u_char))];
543 
544 static int
545 linux_debug(int syscall, int toggle, int global)
546 {
547 
548 	if (global) {
549 		char c = toggle ? 0 : 0xff;
550 
551 		memset(linux_debug_map, c, sizeof(linux_debug_map));
552 		return (0);
553 	}
554 	if (syscall < 0 || syscall >= LINUX_SYS_MAXSYSCALL)
555 		return (EINVAL);
556 	if (toggle)
557 		clrbit(linux_debug_map, syscall);
558 	else
559 		setbit(linux_debug_map, syscall);
560 	return (0);
561 }
562 
563 /*
564  * Usage: sysctl linux.debug=<syscall_nr>.<0/1>
565  *
566  *    E.g.: sysctl linux.debug=21.0
567  *
568  * As a special case, syscall "all" will apply to all syscalls globally.
569  */
570 #define LINUX_MAX_DEBUGSTR	16
571 static int
572 linux_sysctl_debug(SYSCTL_HANDLER_ARGS)
573 {
574 	char value[LINUX_MAX_DEBUGSTR], *p;
575 	int error, sysc, toggle;
576 	int global = 0;
577 
578 	value[0] = '\0';
579 	error = sysctl_handle_string(oidp, value, LINUX_MAX_DEBUGSTR, req);
580 	if (error || req->newptr == NULL)
581 		return (error);
582 	for (p = value; *p != '\0' && *p != '.'; p++);
583 	if (*p == '\0')
584 		return (EINVAL);
585 	*p++ = '\0';
586 	sysc = strtol(value, NULL, 0);
587 	toggle = strtol(p, NULL, 0);
588 	if (strcmp(value, "all") == 0)
589 		global = 1;
590 	error = linux_debug(sysc, toggle, global);
591 	return (error);
592 }
593 
594 SYSCTL_PROC(_compat_linux, OID_AUTO, debug,
595             CTLTYPE_STRING | CTLFLAG_RW,
596             0, 0, linux_sysctl_debug, "A",
597             "Linux debugging control");
598 
599 #endif /* DEBUG || KTR */
600