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/sdt.h>
35#include <sys/systm.h>
36#include <sys/sysctl.h>
37#include <sys/proc.h>
38#include <sys/malloc.h>
39#include <sys/mount.h>
40#include <sys/jail.h>
41#include <sys/lock.h>
42#include <sys/sx.h>
43
44#include <compat/linux/linux_mib.h>
45#include <compat/linux/linux_misc.h>
46
47struct linux_prison {
48	char	pr_osname[LINUX_MAX_UTSNAME];
49	char	pr_osrelease[LINUX_MAX_UTSNAME];
50	int	pr_oss_version;
51	int	pr_osrel;
52};
53
54static struct linux_prison lprison0 = {
55	.pr_osname =		"Linux",
56	.pr_osrelease =		LINUX_VERSION_STR,
57	.pr_oss_version =	0x030600,
58	.pr_osrel =		LINUX_VERSION_CODE
59};
60
61static unsigned linux_osd_jail_slot;
62
63SYSCTL_NODE(_compat, OID_AUTO, linux, CTLFLAG_RW, 0, "Linux mode");
64
65static int	linux_set_osname(struct thread *td, char *osname);
66static int	linux_set_osrelease(struct thread *td, char *osrelease);
67static int	linux_set_oss_version(struct thread *td, int oss_version);
68
69static int
70linux_sysctl_osname(SYSCTL_HANDLER_ARGS)
71{
72	char osname[LINUX_MAX_UTSNAME];
73	int error;
74
75	linux_get_osname(req->td, osname);
76	error = sysctl_handle_string(oidp, osname, LINUX_MAX_UTSNAME, req);
77	if (error != 0 || req->newptr == NULL)
78		return (error);
79	error = linux_set_osname(req->td, osname);
80
81	return (error);
82}
83
84SYSCTL_PROC(_compat_linux, OID_AUTO, osname,
85	    CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_MPSAFE,
86	    0, 0, linux_sysctl_osname, "A",
87	    "Linux kernel OS name");
88
89static int
90linux_sysctl_osrelease(SYSCTL_HANDLER_ARGS)
91{
92	char osrelease[LINUX_MAX_UTSNAME];
93	int error;
94
95	linux_get_osrelease(req->td, osrelease);
96	error = sysctl_handle_string(oidp, osrelease, LINUX_MAX_UTSNAME, req);
97	if (error != 0 || req->newptr == NULL)
98		return (error);
99	error = linux_set_osrelease(req->td, osrelease);
100
101	return (error);
102}
103
104SYSCTL_PROC(_compat_linux, OID_AUTO, osrelease,
105	    CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_MPSAFE,
106	    0, 0, linux_sysctl_osrelease, "A",
107	    "Linux kernel OS release");
108
109static int
110linux_sysctl_oss_version(SYSCTL_HANDLER_ARGS)
111{
112	int oss_version;
113	int error;
114
115	oss_version = linux_get_oss_version(req->td);
116	error = sysctl_handle_int(oidp, &oss_version, 0, req);
117	if (error != 0 || req->newptr == NULL)
118		return (error);
119	error = linux_set_oss_version(req->td, oss_version);
120
121	return (error);
122}
123
124SYSCTL_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 */
132static int
133linux_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
158	return (0);
159}
160
161/*
162 * Find a prison with Linux info.
163 * Return the Linux info and the (locked) prison.
164 */
165static struct linux_prison *
166linux_find_prison(struct prison *spr, struct prison **prp)
167{
168	struct prison *pr;
169	struct linux_prison *lpr;
170
171	if (!linux_osd_jail_slot)
172		/* In case osd_register failed. */
173		spr = &prison0;
174	for (pr = spr;; pr = pr->pr_parent) {
175		mtx_lock(&pr->pr_mtx);
176		lpr = (pr == &prison0)
177		    ? &lprison0
178		    : osd_jail_get(pr, linux_osd_jail_slot);
179		if (lpr != NULL)
180			break;
181		mtx_unlock(&pr->pr_mtx);
182	}
183	*prp = pr;
184
185	return (lpr);
186}
187
188/*
189 * Ensure a prison has its own Linux info.  If lprp is non-null, point it to
190 * the Linux info and lock the prison.
191 */
192static int
193linux_alloc_prison(struct prison *pr, struct linux_prison **lprp)
194{
195	struct prison *ppr;
196	struct linux_prison *lpr, *nlpr;
197	int error;
198
199	/* If this prison already has Linux info, return that. */
200	error = 0;
201	lpr = linux_find_prison(pr, &ppr);
202	if (ppr == pr)
203		goto done;
204	/*
205	 * Allocate a new info record.  Then check again, in case something
206	 * changed during the allocation.
207	 */
208	mtx_unlock(&ppr->pr_mtx);
209	nlpr = malloc(sizeof(struct linux_prison), M_PRISON, M_WAITOK);
210	lpr = linux_find_prison(pr, &ppr);
211	if (ppr == pr) {
212		free(nlpr, M_PRISON);
213		goto done;
214	}
215	/* Inherit the initial values from the ancestor. */
216	mtx_lock(&pr->pr_mtx);
217	error = osd_jail_set(pr, linux_osd_jail_slot, nlpr);
218	if (error == 0) {
219		bcopy(lpr, nlpr, sizeof(*lpr));
220		lpr = nlpr;
221	} else {
222		free(nlpr, M_PRISON);
223		lpr = NULL;
224	}
225	mtx_unlock(&ppr->pr_mtx);
226 done:
227	if (lprp != NULL)
228		*lprp = lpr;
229	else
230		mtx_unlock(&pr->pr_mtx);
231
232	return (error);
233}
234
235/*
236 * Jail OSD methods for Linux prison data.
237 */
238static int
239linux_prison_create(void *obj, void *data)
240{
241	struct prison *pr = obj;
242	struct vfsoptlist *opts = data;
243	int jsys;
244
245	if (vfs_copyopt(opts, "linux", &jsys, sizeof(jsys)) == 0 &&
246	    jsys == JAIL_SYS_INHERIT)
247		return (0);
248	/*
249	 * Inherit a prison's initial values from its parent
250	 * (different from JAIL_SYS_INHERIT which also inherits changes).
251	 */
252	return (linux_alloc_prison(pr, NULL));
253}
254
255static int
256linux_prison_check(void *obj __unused, void *data)
257{
258	struct vfsoptlist *opts = data;
259	char *osname, *osrelease;
260	int error, jsys, len, osrel, oss_version;
261
262	/* Check that the parameters are correct. */
263	error = vfs_copyopt(opts, "linux", &jsys, sizeof(jsys));
264	if (error != ENOENT) {
265		if (error != 0)
266			return (error);
267		if (jsys != JAIL_SYS_NEW && jsys != JAIL_SYS_INHERIT)
268			return (EINVAL);
269	}
270	error = vfs_getopt(opts, "linux.osname", (void **)&osname, &len);
271	if (error != ENOENT) {
272		if (error != 0)
273			return (error);
274		if (len == 0 || osname[len - 1] != '\0')
275			return (EINVAL);
276		if (len > LINUX_MAX_UTSNAME) {
277			vfs_opterror(opts, "linux.osname too long");
278			return (ENAMETOOLONG);
279		}
280	}
281	error = vfs_getopt(opts, "linux.osrelease", (void **)&osrelease, &len);
282	if (error != ENOENT) {
283		if (error != 0)
284			return (error);
285		if (len == 0 || osrelease[len - 1] != '\0')
286			return (EINVAL);
287		if (len > LINUX_MAX_UTSNAME) {
288			vfs_opterror(opts, "linux.osrelease too long");
289			return (ENAMETOOLONG);
290		}
291		error = linux_map_osrel(osrelease, &osrel);
292		if (error != 0) {
293			vfs_opterror(opts, "linux.osrelease format error");
294			return (error);
295		}
296	}
297	error = vfs_copyopt(opts, "linux.oss_version", &oss_version,
298	    sizeof(oss_version));
299
300	if (error == ENOENT)
301		error = 0;
302	return (error);
303}
304
305static int
306linux_prison_set(void *obj, void *data)
307{
308	struct linux_prison *lpr;
309	struct prison *pr = obj;
310	struct vfsoptlist *opts = data;
311	char *osname, *osrelease;
312	int error, gotversion, jsys, len, oss_version;
313
314	/* Set the parameters, which should be correct. */
315	error = vfs_copyopt(opts, "linux", &jsys, sizeof(jsys));
316	if (error == ENOENT)
317		jsys = -1;
318	error = vfs_getopt(opts, "linux.osname", (void **)&osname, &len);
319	if (error == ENOENT)
320		osname = NULL;
321	else
322		jsys = JAIL_SYS_NEW;
323	error = vfs_getopt(opts, "linux.osrelease", (void **)&osrelease, &len);
324	if (error == ENOENT)
325		osrelease = NULL;
326	else
327		jsys = JAIL_SYS_NEW;
328	error = vfs_copyopt(opts, "linux.oss_version", &oss_version,
329	    sizeof(oss_version));
330	if (error == ENOENT)
331		gotversion = 0;
332	else {
333		gotversion = 1;
334		jsys = JAIL_SYS_NEW;
335	}
336	switch (jsys) {
337	case JAIL_SYS_INHERIT:
338		/* "linux=inherit": inherit the parent's Linux info. */
339		mtx_lock(&pr->pr_mtx);
340		osd_jail_del(pr, linux_osd_jail_slot);
341		mtx_unlock(&pr->pr_mtx);
342		break;
343	case JAIL_SYS_NEW:
344		/*
345		 * "linux=new" or "linux.*":
346		 * the prison gets its own Linux info.
347		 */
348		error = linux_alloc_prison(pr, &lpr);
349		if (error) {
350			mtx_unlock(&pr->pr_mtx);
351			return (error);
352		}
353		if (osrelease) {
354			error = linux_map_osrel(osrelease, &lpr->pr_osrel);
355			if (error) {
356				mtx_unlock(&pr->pr_mtx);
357				return (error);
358			}
359			strlcpy(lpr->pr_osrelease, osrelease,
360			    LINUX_MAX_UTSNAME);
361		}
362		if (osname)
363			strlcpy(lpr->pr_osname, osname, LINUX_MAX_UTSNAME);
364		if (gotversion)
365			lpr->pr_oss_version = oss_version;
366		mtx_unlock(&pr->pr_mtx);
367	}
368
369	return (0);
370}
371
372SYSCTL_JAIL_PARAM_SYS_NODE(linux, CTLFLAG_RW, "Jail Linux parameters");
373SYSCTL_JAIL_PARAM_STRING(_linux, osname, CTLFLAG_RW, LINUX_MAX_UTSNAME,
374    "Jail Linux kernel OS name");
375SYSCTL_JAIL_PARAM_STRING(_linux, osrelease, CTLFLAG_RW, LINUX_MAX_UTSNAME,
376    "Jail Linux kernel OS release");
377SYSCTL_JAIL_PARAM(_linux, oss_version, CTLTYPE_INT | CTLFLAG_RW,
378    "I", "Jail Linux OSS version");
379
380static int
381linux_prison_get(void *obj, void *data)
382{
383	struct linux_prison *lpr;
384	struct prison *ppr;
385	struct prison *pr = obj;
386	struct vfsoptlist *opts = data;
387	int error, i;
388
389	static int version0;
390
391	/* See if this prison is the one with the Linux info. */
392	lpr = linux_find_prison(pr, &ppr);
393	i = (ppr == pr) ? JAIL_SYS_NEW : JAIL_SYS_INHERIT;
394	error = vfs_setopt(opts, "linux", &i, sizeof(i));
395	if (error != 0 && error != ENOENT)
396		goto done;
397	if (i) {
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	} else {
409		/*
410		 * If this prison is inheriting its Linux info, report
411		 * empty/zero parameters.
412		 */
413		error = vfs_setopts(opts, "linux.osname", "");
414		if (error != 0 && error != ENOENT)
415			goto done;
416		error = vfs_setopts(opts, "linux.osrelease", "");
417		if (error != 0 && error != ENOENT)
418			goto done;
419		error = vfs_setopt(opts, "linux.oss_version", &version0,
420		    sizeof(lpr->pr_oss_version));
421		if (error != 0 && error != ENOENT)
422			goto done;
423	}
424	error = 0;
425
426 done:
427	mtx_unlock(&ppr->pr_mtx);
428
429	return (error);
430}
431
432static void
433linux_prison_destructor(void *data)
434{
435
436	free(data, M_PRISON);
437}
438
439void
440linux_osd_jail_register(void)
441{
442	struct prison *pr;
443	osd_method_t methods[PR_MAXMETHOD] = {
444	    [PR_METHOD_CREATE] =	linux_prison_create,
445	    [PR_METHOD_GET] =		linux_prison_get,
446	    [PR_METHOD_SET] =		linux_prison_set,
447	    [PR_METHOD_CHECK] =		linux_prison_check
448	};
449
450	linux_osd_jail_slot =
451	    osd_jail_register(linux_prison_destructor, methods);
452	if (linux_osd_jail_slot > 0) {
453		/* Copy the system linux info to any current prisons. */
454		sx_xlock(&allprison_lock);
455		TAILQ_FOREACH(pr, &allprison, pr_list)
456			(void)linux_alloc_prison(pr, NULL);
457		sx_xunlock(&allprison_lock);
458	}
459}
460
461void
462linux_osd_jail_deregister(void)
463{
464
465	if (linux_osd_jail_slot)
466		osd_jail_deregister(linux_osd_jail_slot);
467}
468
469void
470linux_get_osname(struct thread *td, char *dst)
471{
472	struct prison *pr;
473	struct linux_prison *lpr;
474
475	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
476	bcopy(lpr->pr_osname, dst, LINUX_MAX_UTSNAME);
477	mtx_unlock(&pr->pr_mtx);
478}
479
480static int
481linux_set_osname(struct thread *td, char *osname)
482{
483	struct prison *pr;
484	struct linux_prison *lpr;
485
486	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
487	strlcpy(lpr->pr_osname, osname, LINUX_MAX_UTSNAME);
488	mtx_unlock(&pr->pr_mtx);
489
490	return (0);
491}
492
493void
494linux_get_osrelease(struct thread *td, char *dst)
495{
496	struct prison *pr;
497	struct linux_prison *lpr;
498
499	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
500	bcopy(lpr->pr_osrelease, dst, LINUX_MAX_UTSNAME);
501	mtx_unlock(&pr->pr_mtx);
502}
503
504int
505linux_kernver(struct thread *td)
506{
507	struct prison *pr;
508	struct linux_prison *lpr;
509	int osrel;
510
511	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
512	osrel = lpr->pr_osrel;
513	mtx_unlock(&pr->pr_mtx);
514
515	return (osrel);
516}
517
518static int
519linux_set_osrelease(struct thread *td, char *osrelease)
520{
521	struct prison *pr;
522	struct linux_prison *lpr;
523	int error;
524
525	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
526	error = linux_map_osrel(osrelease, &lpr->pr_osrel);
527	if (error == 0)
528		strlcpy(lpr->pr_osrelease, osrelease, LINUX_MAX_UTSNAME);
529	mtx_unlock(&pr->pr_mtx);
530
531	return (error);
532}
533
534int
535linux_get_oss_version(struct thread *td)
536{
537	struct prison *pr;
538	struct linux_prison *lpr;
539	int version;
540
541	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
542	version = lpr->pr_oss_version;
543	mtx_unlock(&pr->pr_mtx);
544
545	return (version);
546}
547
548static int
549linux_set_oss_version(struct thread *td, int oss_version)
550{
551	struct prison *pr;
552	struct linux_prison *lpr;
553
554	lpr = linux_find_prison(td->td_ucred->cr_prison, &pr);
555	lpr->pr_oss_version = oss_version;
556	mtx_unlock(&pr->pr_mtx);
557
558	return (0);
559}
560