1/*-
2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3 *
4 * Copyright (c) 2011 James Gritton
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/types.h>
33#include <sys/event.h>
34#include <sys/mount.h>
35#include <sys/stat.h>
36#include <sys/sysctl.h>
37#include <sys/user.h>
38#include <sys/wait.h>
39
40#include <err.h>
41#include <errno.h>
42#include <fcntl.h>
43#include <kvm.h>
44#include <login_cap.h>
45#include <paths.h>
46#include <pwd.h>
47#include <signal.h>
48#include <stdio.h>
49#include <stdlib.h>
50#include <string.h>
51#include <unistd.h>
52#include <vis.h>
53
54#include "jailp.h"
55
56#define DEFAULT_STOP_TIMEOUT	10
57#define PHASH_SIZE		256
58
59LIST_HEAD(phhead, phash);
60
61struct phash {
62	LIST_ENTRY(phash)	le;
63	struct cfjail		*j;
64	pid_t			pid;
65};
66
67int paralimit = -1;
68
69extern char **environ;
70
71static int run_command(struct cfjail *j);
72static int add_proc(struct cfjail *j, pid_t pid);
73static void clear_procs(struct cfjail *j);
74static struct cfjail *find_proc(pid_t pid);
75static int term_procs(struct cfjail *j);
76static int get_user_info(struct cfjail *j, const char *username,
77    const struct passwd **pwdp, login_cap_t **lcapp);
78static int check_path(struct cfjail *j, const char *pname, const char *path,
79    int isfile, const char *umount_type);
80
81static struct cfjails sleeping = TAILQ_HEAD_INITIALIZER(sleeping);
82static struct cfjails runnable = TAILQ_HEAD_INITIALIZER(runnable);
83static struct cfstring dummystring = { .len = 1 };
84static struct phhead phash[PHASH_SIZE];
85static int kq;
86
87/*
88 * Run the next command associated with a jail.
89 */
90int
91next_command(struct cfjail *j)
92{
93	enum intparam comparam;
94	int create_failed, stopping;
95
96	if (paralimit == 0) {
97		if (j->flags & JF_FROM_RUNQ)
98			requeue_head(j, &runnable);
99		else
100			requeue(j, &runnable);
101		return 1;
102	}
103	j->flags &= ~JF_FROM_RUNQ;
104	create_failed = (j->flags & (JF_STOP | JF_FAILED)) == JF_FAILED;
105	stopping = (j->flags & JF_STOP) != 0;
106	comparam = *j->comparam;
107	for (;;) {
108		if (j->comstring == NULL) {
109			j->comparam += create_failed ? -1 : 1;
110			switch ((comparam = *j->comparam)) {
111			case IP__NULL:
112				return 0;
113			case IP_MOUNT_DEVFS:
114				if (!bool_param(j->intparams[IP_MOUNT_DEVFS]))
115					continue;
116				j->comstring = &dummystring;
117				break;
118			case IP_MOUNT_FDESCFS:
119				if (!bool_param(j->intparams[IP_MOUNT_FDESCFS]))
120					continue;
121				j->comstring = &dummystring;
122				break;
123			case IP_MOUNT_PROCFS:
124				if (!bool_param(j->intparams[IP_MOUNT_PROCFS]))
125					continue;
126				j->comstring = &dummystring;
127				break;
128			case IP__OP:
129			case IP_STOP_TIMEOUT:
130				j->comstring = &dummystring;
131				break;
132			default:
133				if (j->intparams[comparam] == NULL)
134					continue;
135				j->comstring = create_failed || (stopping &&
136				    (j->intparams[comparam]->flags & PF_REV))
137				    ? TAILQ_LAST(&j->intparams[comparam]->val,
138					cfstrings)
139				    : TAILQ_FIRST(&j->intparams[comparam]->val);
140			}
141		} else {
142			j->comstring = j->comstring == &dummystring ? NULL :
143			    create_failed || (stopping &&
144			    (j->intparams[comparam]->flags & PF_REV))
145			    ? TAILQ_PREV(j->comstring, cfstrings, tq)
146			    : TAILQ_NEXT(j->comstring, tq);
147		}
148		if (j->comstring == NULL || j->comstring->len == 0 ||
149		    (create_failed && (comparam == IP_EXEC_PRESTART ||
150		    comparam == IP_EXEC_CREATED || comparam == IP_EXEC_START ||
151		    comparam == IP_COMMAND || comparam == IP_EXEC_POSTSTART ||
152		    comparam == IP_EXEC_PREPARE)))
153			continue;
154		switch (run_command(j)) {
155		case -1:
156			failed(j);
157			/* FALLTHROUGH */
158		case 1:
159			return 1;
160		}
161	}
162}
163
164/*
165 * Check command exit status
166 */
167int
168finish_command(struct cfjail *j)
169{
170	struct cfjail *rj;
171	int error;
172
173	if (!(j->flags & JF_SLEEPQ))
174		return 0;
175	j->flags &= ~JF_SLEEPQ;
176	if (*j->comparam == IP_STOP_TIMEOUT) {
177		j->flags &= ~JF_TIMEOUT;
178		j->pstatus = 0;
179		return 0;
180	}
181	paralimit++;
182	if (!TAILQ_EMPTY(&runnable)) {
183		rj = TAILQ_FIRST(&runnable);
184		rj->flags |= JF_FROM_RUNQ;
185		requeue(rj, &ready);
186	}
187	error = 0;
188	if (j->flags & JF_TIMEOUT) {
189		j->flags &= ~JF_TIMEOUT;
190		if (*j->comparam != IP_STOP_TIMEOUT) {
191			jail_warnx(j, "%s: timed out", j->comline);
192			failed(j);
193			error = -1;
194		} else if (verbose > 0)
195			jail_note(j, "timed out\n");
196	} else if (j->pstatus != 0) {
197		if (WIFSIGNALED(j->pstatus))
198			jail_warnx(j, "%s: exited on signal %d",
199			    j->comline, WTERMSIG(j->pstatus));
200		else
201			jail_warnx(j, "%s: failed", j->comline);
202		j->pstatus = 0;
203		failed(j);
204		error = -1;
205	}
206	free(j->comline);
207	j->comline = NULL;
208	return error;
209}
210
211/*
212 * Check for finished processes or timeouts.
213 */
214struct cfjail *
215next_proc(int nonblock)
216{
217	struct kevent ke;
218	struct timespec ts;
219	struct timespec *tsp;
220	struct cfjail *j;
221
222	if (!TAILQ_EMPTY(&sleeping)) {
223	again:
224		tsp = NULL;
225		if ((j = TAILQ_FIRST(&sleeping)) && j->timeout.tv_sec) {
226			clock_gettime(CLOCK_REALTIME, &ts);
227			ts.tv_sec = j->timeout.tv_sec - ts.tv_sec;
228			ts.tv_nsec = j->timeout.tv_nsec - ts.tv_nsec;
229			if (ts.tv_nsec < 0) {
230				ts.tv_sec--;
231				ts.tv_nsec += 1000000000;
232			}
233			if (ts.tv_sec < 0 ||
234			    (ts.tv_sec == 0 && ts.tv_nsec == 0)) {
235				j->flags |= JF_TIMEOUT;
236				clear_procs(j);
237				return j;
238			}
239			tsp = &ts;
240		}
241		if (nonblock) {
242			ts.tv_sec = 0;
243			ts.tv_nsec = 0;
244			tsp = &ts;
245		}
246		switch (kevent(kq, NULL, 0, &ke, 1, tsp)) {
247		case -1:
248			if (errno != EINTR)
249				err(1, "kevent");
250			goto again;
251		case 0:
252			if (!nonblock) {
253				j = TAILQ_FIRST(&sleeping);
254				j->flags |= JF_TIMEOUT;
255				clear_procs(j);
256				return j;
257			}
258			break;
259		case 1:
260			(void)waitpid(ke.ident, NULL, WNOHANG);
261			if ((j = find_proc(ke.ident))) {
262				j->pstatus = ke.data;
263				return j;
264			}
265			goto again;
266		}
267	}
268	return NULL;
269}
270
271/*
272 * Run a single command for a jail, possibly inside the jail.
273 */
274static int
275run_command(struct cfjail *j)
276{
277	const struct passwd *pwd;
278	const struct cfstring *comstring, *s;
279	login_cap_t *lcap;
280	const char **argv;
281	char *acs, *cs, *comcs, *devpath;
282	const char *jidstr, *conslog, *path, *ruleset, *term, *username;
283	enum intparam comparam;
284	size_t comlen;
285	pid_t pid;
286	int argc, bg, clean, consfd, down, fib, i, injail, sjuser, timeout;
287#if defined(INET) || defined(INET6)
288	char *addr, *extrap, *p, *val;
289#endif
290
291	static char *cleanenv;
292
293	/* Perform some operations that aren't actually commands */
294	comparam = *j->comparam;
295	down = j->flags & (JF_STOP | JF_FAILED);
296	switch (comparam) {
297	case IP_STOP_TIMEOUT:
298		return term_procs(j);
299
300	case IP__OP:
301		if (down) {
302			if (jail_remove(j->jid) < 0 && errno == EPERM) {
303				jail_warnx(j, "jail_remove: %s",
304					   strerror(errno));
305				return -1;
306			}
307			if (verbose > 0 || (verbose == 0 && (j->flags & JF_STOP
308			    ? note_remove : j->name != NULL)))
309			    jail_note(j, "removed\n");
310			j->jid = -1;
311			if (j->flags & JF_STOP)
312				dep_done(j, DF_LIGHT);
313			else
314				j->flags &= ~JF_PERSIST;
315		} else {
316			if (create_jail(j) < 0)
317				return -1;
318			if (iflag)
319				printf("%d\n", j->jid);
320			if (verbose >= 0 && (j->name || verbose > 0))
321				jail_note(j, "created\n");
322			dep_done(j, DF_LIGHT);
323		}
324		return 0;
325
326	default: ;
327	}
328	/*
329	 * Collect exec arguments.  Internal commands for network and
330	 * mounting build their own argument lists.
331	 */
332	comstring = j->comstring;
333	bg = 0;
334	switch (comparam) {
335#ifdef INET
336	case IP__IP4_IFADDR:
337		argc = 0;
338		val = alloca(strlen(comstring->s) + 1);
339		strcpy(val, comstring->s);
340		cs = val;
341		extrap = NULL;
342		while ((p = strchr(cs, ' ')) != NULL && strlen(p) > 1) {
343			if (extrap == NULL) {
344				*p = '\0';
345				extrap = p + 1;
346			}
347			cs = p + 1;
348			argc++;
349		}
350
351		argv = alloca((8 + argc) * sizeof(char *));
352		argv[0] = _PATH_IFCONFIG;
353		if ((cs = strchr(val, '|'))) {
354			argv[1] = acs = alloca(cs - val + 1);
355			strlcpy(acs, val, cs - val + 1);
356			addr = cs + 1;
357		} else {
358			argv[1] = string_param(j->intparams[IP_INTERFACE]);
359			addr = val;
360		}
361		argv[2] = "inet";
362		if (!(cs = strchr(addr, '/'))) {
363			argv[3] = addr;
364			argv[4] = "netmask";
365			argv[5] = "255.255.255.255";
366			argc = 6;
367		} else if (strchr(cs + 1, '.')) {
368			argv[3] = acs = alloca(cs - addr + 1);
369			strlcpy(acs, addr, cs - addr + 1);
370			argv[4] = "netmask";
371			argv[5] = cs + 1;
372			argc = 6;
373		} else {
374			argv[3] = addr;
375			argc = 4;
376		}
377
378		if (!down && extrap != NULL) {
379			for (cs = strtok(extrap, " "); cs;
380			     cs = strtok(NULL, " ")) {
381				size_t len = strlen(cs) + 1;
382				argv[argc++] = acs = alloca(len);
383				strlcpy(acs, cs, len);
384			}
385		}
386
387		argv[argc] = down ? "-alias" : "alias";
388		argv[argc + 1] = NULL;
389		break;
390#endif
391
392#ifdef INET6
393	case IP__IP6_IFADDR:
394		argc = 0;
395		val = alloca(strlen(comstring->s) + 1);
396		strcpy(val, comstring->s);
397		cs = val;
398		extrap = NULL;
399		while ((p = strchr(cs, ' ')) != NULL && strlen(p) > 1) {
400			if (extrap == NULL) {
401				*p = '\0';
402				extrap = p + 1;
403			}
404			cs = p + 1;
405			argc++;
406		}
407
408		argv = alloca((8 + argc) * sizeof(char *));
409		argv[0] = _PATH_IFCONFIG;
410		if ((cs = strchr(val, '|'))) {
411			argv[1] = acs = alloca(cs - val + 1);
412			strlcpy(acs, val, cs - val + 1);
413			addr = cs + 1;
414		} else {
415			argv[1] = string_param(j->intparams[IP_INTERFACE]);
416			addr = val;
417		}
418		argv[2] = "inet6";
419		argv[3] = addr;
420		if (!(cs = strchr(addr, '/'))) {
421			argv[4] = "prefixlen";
422			argv[5] = "128";
423			argc = 6;
424		} else
425			argc = 4;
426
427		if (!down) {
428			for (cs = strtok(extrap, " "); cs;
429			     cs = strtok(NULL, " ")) {
430				size_t len = strlen(cs) + 1;
431				argv[argc++] = acs = alloca(len);
432				strlcpy(acs, cs, len);
433			}
434		}
435
436		argv[argc] = down ? "-alias" : "alias";
437		argv[argc + 1] = NULL;
438		break;
439#endif
440
441	case IP_VNET_INTERFACE:
442		argv = alloca(5 * sizeof(char *));
443		argv[0] = _PATH_IFCONFIG;
444		argv[1] = comstring->s;
445		argv[2] = down ? "-vnet" : "vnet";
446		jidstr = string_param(j->intparams[KP_JID]);
447		argv[3] = jidstr ? jidstr : string_param(j->intparams[KP_NAME]);
448		argv[4] = NULL;
449		break;
450
451	case IP_MOUNT:
452	case IP__MOUNT_FROM_FSTAB:
453		argv = alloca(8 * sizeof(char *));
454		comcs = alloca(comstring->len + 1);
455		strcpy(comcs, comstring->s);
456		argc = 0;
457		for (cs = strtok(comcs, " \t\f\v\r\n"); cs && argc < 4;
458		     cs = strtok(NULL, " \t\f\v\r\n")) {
459			if (argc <= 1 && strunvis(cs, cs) < 0) {
460				jail_warnx(j, "%s: %s: fstab parse error",
461				    j->intparams[comparam]->name, comstring->s);
462				return -1;
463			}
464			argv[argc++] = cs;
465		}
466		if (argc == 0)
467			return 0;
468		if (argc < 3) {
469			jail_warnx(j, "%s: %s: missing information",
470			    j->intparams[comparam]->name, comstring->s);
471			return -1;
472		}
473		if (check_path(j, j->intparams[comparam]->name, argv[1], 0,
474		    down ? argv[2] : NULL) < 0)
475			return -1;
476		if (down) {
477			argv[4] = NULL;
478			argv[3] = argv[1];
479			argv[1] = "-ft";
480			argv[0] = "/sbin/umount";
481		} else {
482			if (argc == 4) {
483				argv[7] = NULL;
484				argv[6] = argv[1];
485				argv[5] = argv[0];
486				argv[4] = argv[3];
487				argv[3] = "-o";
488			} else {
489				argv[5] = NULL;
490				argv[4] = argv[1];
491				argv[3] = argv[0];
492			}
493			argv[1] = "-t";
494			argv[0] = _PATH_MOUNT;
495		}
496		break;
497
498	case IP_MOUNT_DEVFS:
499		argv = alloca(7 * sizeof(char *));
500		path = string_param(j->intparams[KP_PATH]);
501		if (path == NULL) {
502			jail_warnx(j, "mount.devfs: no jail root path defined");
503			return -1;
504		}
505		devpath = alloca(strlen(path) + 5);
506		sprintf(devpath, "%s/dev", path);
507		if (check_path(j, "mount.devfs", devpath, 0,
508		    down ? "devfs" : NULL) < 0)
509			return -1;
510		if (down) {
511			argv[0] = "/sbin/umount";
512			argv[1] = devpath;
513			argv[2] = NULL;
514		} else {
515			argv[0] = _PATH_MOUNT;
516			argv[1] = "-t";
517			argv[2] = "devfs";
518			ruleset = string_param(j->intparams[KP_DEVFS_RULESET]);
519			if (!ruleset)
520			    ruleset = "4";	/* devfsrules_jail */
521			argv[3] = acs = alloca(11 + strlen(ruleset));
522			sprintf(acs, "-oruleset=%s", ruleset);
523			argv[4] = ".";
524			argv[5] = devpath;
525			argv[6] = NULL;
526		}
527		break;
528
529	case IP_MOUNT_FDESCFS:
530		argv = alloca(7 * sizeof(char *));
531		path = string_param(j->intparams[KP_PATH]);
532		if (path == NULL) {
533			jail_warnx(j, "mount.fdescfs: no jail root path defined");
534			return -1;
535		}
536		devpath = alloca(strlen(path) + 8);
537		sprintf(devpath, "%s/dev/fd", path);
538		if (check_path(j, "mount.fdescfs", devpath, 0,
539		    down ? "fdescfs" : NULL) < 0)
540			return -1;
541		if (down) {
542			argv[0] = "/sbin/umount";
543			argv[1] = devpath;
544			argv[2] = NULL;
545		} else {
546			argv[0] = _PATH_MOUNT;
547			argv[1] = "-t";
548			argv[2] = "fdescfs";
549			argv[3] = ".";
550			argv[4] = devpath;
551			argv[5] = NULL;
552		}
553		break;
554
555	case IP_MOUNT_PROCFS:
556		argv = alloca(7 * sizeof(char *));
557		path = string_param(j->intparams[KP_PATH]);
558		if (path == NULL) {
559			jail_warnx(j, "mount.procfs: no jail root path defined");
560			return -1;
561		}
562		devpath = alloca(strlen(path) + 6);
563		sprintf(devpath, "%s/proc", path);
564		if (check_path(j, "mount.procfs", devpath, 0,
565		    down ? "procfs" : NULL) < 0)
566			return -1;
567		if (down) {
568			argv[0] = "/sbin/umount";
569			argv[1] = devpath;
570			argv[2] = NULL;
571		} else {
572			argv[0] = _PATH_MOUNT;
573			argv[1] = "-t";
574			argv[2] = "procfs";
575			argv[3] = ".";
576			argv[4] = devpath;
577			argv[5] = NULL;
578		}
579		break;
580
581	case IP_COMMAND:
582		if (j->name != NULL)
583			goto default_command;
584		argc = 0;
585		TAILQ_FOREACH(s, &j->intparams[IP_COMMAND]->val, tq)
586			argc++;
587		argv = alloca((argc + 1) * sizeof(char *));
588		argc = 0;
589		TAILQ_FOREACH(s, &j->intparams[IP_COMMAND]->val, tq)
590			argv[argc++] = s->s;
591		argv[argc] = NULL;
592		j->comstring = &dummystring;
593		break;
594
595	default:
596	default_command:
597		if ((cs = strpbrk(comstring->s, "!\"$&'()*;<>?[\\]`{|}~")) &&
598		    !(cs[0] == '&' && cs[1] == '\0')) {
599			argv = alloca(4 * sizeof(char *));
600			argv[0] = _PATH_BSHELL;
601			argv[1] = "-c";
602			argv[2] = comstring->s;
603			argv[3] = NULL;
604		} else {
605			if (cs) {
606				*cs = 0;
607				bg = 1;
608			}
609			comcs = alloca(comstring->len + 1);
610			strcpy(comcs, comstring->s);
611			argc = 0;
612			for (cs = strtok(comcs, " \t\f\v\r\n"); cs;
613			     cs = strtok(NULL, " \t\f\v\r\n"))
614				argc++;
615			argv = alloca((argc + 1) * sizeof(char *));
616			strcpy(comcs, comstring->s);
617			argc = 0;
618			for (cs = strtok(comcs, " \t\f\v\r\n"); cs;
619			     cs = strtok(NULL, " \t\f\v\r\n"))
620				argv[argc++] = cs;
621			argv[argc] = NULL;
622		}
623	}
624	if (argv[0] == NULL)
625		return 0;
626
627	if (int_param(j->intparams[IP_EXEC_TIMEOUT], &timeout) &&
628	    timeout != 0) {
629		clock_gettime(CLOCK_REALTIME, &j->timeout);
630		j->timeout.tv_sec += timeout;
631	} else
632		j->timeout.tv_sec = 0;
633
634	injail = comparam == IP_EXEC_START || comparam == IP_COMMAND ||
635	    comparam == IP_EXEC_STOP;
636	clean = bool_param(j->intparams[IP_EXEC_CLEAN]);
637	username = string_param(j->intparams[injail
638	    ? IP_EXEC_JAIL_USER : IP_EXEC_SYSTEM_USER]);
639	sjuser = bool_param(j->intparams[IP_EXEC_SYSTEM_JAIL_USER]);
640
641	consfd = 0;
642	if (injail &&
643	    (conslog = string_param(j->intparams[IP_EXEC_CONSOLELOG]))) {
644		if (check_path(j, "exec.consolelog", conslog, 1, NULL) < 0)
645			return -1;
646		consfd =
647		    open(conslog, O_WRONLY | O_CREAT | O_APPEND, DEFFILEMODE);
648		if (consfd < 0) {
649			jail_warnx(j, "open %s: %s", conslog, strerror(errno));
650			return -1;
651		}
652	}
653
654	comlen = 0;
655	for (i = 0; argv[i]; i++)
656		comlen += strlen(argv[i]) + 1;
657	j->comline = cs = emalloc(comlen);
658	for (i = 0; argv[i]; i++) {
659		strcpy(cs, argv[i]);
660		if (argv[i + 1]) {
661			cs += strlen(argv[i]) + 1;
662			cs[-1] = ' ';
663		}
664	}
665	if (verbose > 0)
666		jail_note(j, "run command%s%s%s: %s\n",
667		    injail ? " in jail" : "", username ? " as " : "",
668		    username ? username : "", j->comline);
669
670	pid = fork();
671	if (pid < 0)
672		err(1, "fork");
673	if (pid > 0) {
674		if (bg || !add_proc(j, pid)) {
675			free(j->comline);
676			j->comline = NULL;
677			return 0;
678		} else {
679			paralimit--;
680			return 1;
681		}
682	}
683	if (bg)
684		setsid();
685
686	/* Set up the environment and run the command */
687	pwd = NULL;
688	lcap = NULL;
689	if ((clean || username) && injail && sjuser &&
690	    get_user_info(j, username, &pwd, &lcap) < 0)
691		exit(1);
692	if (injail) {
693		/* jail_attach won't chdir along with its chroot. */
694		path = string_param(j->intparams[KP_PATH]);
695		if (path && chdir(path) < 0) {
696			jail_warnx(j, "chdir %s: %s", path, strerror(errno));
697			exit(1);
698		}
699		if (int_param(j->intparams[IP_EXEC_FIB], &fib) &&
700		    setfib(fib) < 0) {
701			jail_warnx(j, "setfib: %s", strerror(errno));
702			exit(1);
703		}
704		if (jail_attach(j->jid) < 0) {
705			jail_warnx(j, "jail_attach: %s", strerror(errno));
706			exit(1);
707		}
708	}
709	if (clean || username) {
710		if (!(injail && sjuser) &&
711		    get_user_info(j, username, &pwd, &lcap) < 0)
712			exit(1);
713		if (clean) {
714			term = getenv("TERM");
715			environ = &cleanenv;
716			setenv("PATH", "/bin:/usr/bin", 0);
717			if (term != NULL)
718				setenv("TERM", term, 1);
719		}
720		if (setgid(pwd->pw_gid) < 0) {
721			jail_warnx(j, "setgid %d: %s", pwd->pw_gid,
722			    strerror(errno));
723			exit(1);
724		}
725		if (setusercontext(lcap, pwd, pwd->pw_uid, username
726		    ? LOGIN_SETALL & ~LOGIN_SETGROUP & ~LOGIN_SETLOGIN
727		    : LOGIN_SETPATH | LOGIN_SETENV) < 0) {
728			jail_warnx(j, "setusercontext %s: %s", pwd->pw_name,
729			    strerror(errno));
730			exit(1);
731		}
732		login_close(lcap);
733		setenv("USER", pwd->pw_name, 1);
734		setenv("HOME", pwd->pw_dir, 1);
735		setenv("SHELL",
736		    *pwd->pw_shell ? pwd->pw_shell : _PATH_BSHELL, 1);
737		if (clean && chdir(pwd->pw_dir) < 0) {
738			jail_warnx(j, "chdir %s: %s",
739			    pwd->pw_dir, strerror(errno));
740			exit(1);
741		}
742		endpwent();
743	}
744
745	if (consfd != 0 && (dup2(consfd, 1) < 0 || dup2(consfd, 2) < 0)) {
746		jail_warnx(j, "exec.consolelog: %s", strerror(errno));
747		exit(1);
748	}
749	closefrom(3);
750	execvp(argv[0], __DECONST(char *const*, argv));
751	jail_warnx(j, "exec %s: %s", argv[0], strerror(errno));
752	exit(1);
753}
754
755/*
756 * Add a process to the hash, tied to a jail.
757 */
758static int
759add_proc(struct cfjail *j, pid_t pid)
760{
761	struct kevent ke;
762	struct cfjail *tj;
763	struct phash *ph;
764
765	if (!kq && (kq = kqueue()) < 0)
766		err(1, "kqueue");
767	EV_SET(&ke, pid, EVFILT_PROC, EV_ADD, NOTE_EXIT, 0, NULL);
768	if (kevent(kq, &ke, 1, NULL, 0, NULL) < 0) {
769		if (errno == ESRCH)
770			return 0;
771		err(1, "kevent");
772	}
773	ph = emalloc(sizeof(struct phash));
774	ph->j = j;
775	ph->pid = pid;
776	LIST_INSERT_HEAD(&phash[pid % PHASH_SIZE], ph, le);
777	j->nprocs++;
778	j->flags |= JF_SLEEPQ;
779	if (j->timeout.tv_sec == 0)
780		requeue(j, &sleeping);
781	else {
782		/* File the jail in the sleep queue according to its timeout. */
783		TAILQ_REMOVE(j->queue, j, tq);
784		TAILQ_FOREACH(tj, &sleeping, tq) {
785			if (!tj->timeout.tv_sec ||
786			    j->timeout.tv_sec < tj->timeout.tv_sec ||
787			    (j->timeout.tv_sec == tj->timeout.tv_sec &&
788			    j->timeout.tv_nsec <= tj->timeout.tv_nsec)) {
789				TAILQ_INSERT_BEFORE(tj, j, tq);
790				break;
791			}
792		}
793		if (tj == NULL)
794			TAILQ_INSERT_TAIL(&sleeping, j, tq);
795		j->queue = &sleeping;
796	}
797	return 1;
798}
799
800/*
801 * Remove any processes from the hash that correspond to a jail.
802 */
803static void
804clear_procs(struct cfjail *j)
805{
806	struct kevent ke;
807	struct phash *ph, *tph;
808	int i;
809
810	j->nprocs = 0;
811	for (i = 0; i < PHASH_SIZE; i++)
812		LIST_FOREACH_SAFE(ph, &phash[i], le, tph)
813			if (ph->j == j) {
814				EV_SET(&ke, ph->pid, EVFILT_PROC, EV_DELETE,
815				    NOTE_EXIT, 0, NULL);
816				(void)kevent(kq, &ke, 1, NULL, 0, NULL);
817				LIST_REMOVE(ph, le);
818				free(ph);
819			}
820}
821
822/*
823 * Find the jail that corresponds to an exited process.
824 */
825static struct cfjail *
826find_proc(pid_t pid)
827{
828	struct cfjail *j;
829	struct phash *ph;
830
831	LIST_FOREACH(ph, &phash[pid % PHASH_SIZE], le)
832		if (ph->pid == pid) {
833			j = ph->j;
834			LIST_REMOVE(ph, le);
835			free(ph);
836			return --j->nprocs ? NULL : j;
837		}
838	return NULL;
839}
840
841/*
842 * Send SIGTERM to all processes in a jail and wait for them to die.
843 */
844static int
845term_procs(struct cfjail *j)
846{
847	struct kinfo_proc *ki;
848	int i, noted, pcnt, timeout;
849
850	static kvm_t *kd;
851
852	if (!int_param(j->intparams[IP_STOP_TIMEOUT], &timeout))
853		timeout = DEFAULT_STOP_TIMEOUT;
854	else if (timeout == 0)
855		return 0;
856
857	if (kd == NULL) {
858		kd = kvm_open(NULL, NULL, NULL, O_RDONLY, NULL);
859		if (kd == NULL)
860			return 0;
861	}
862
863	ki = kvm_getprocs(kd, KERN_PROC_PROC, 0, &pcnt);
864	if (ki == NULL)
865		return 0;
866	noted = 0;
867	for (i = 0; i < pcnt; i++)
868		if (ki[i].ki_jid == j->jid &&
869		    kill(ki[i].ki_pid, SIGTERM) == 0) {
870			(void)add_proc(j, ki[i].ki_pid);
871			if (verbose > 0) {
872				if (!noted) {
873					noted = 1;
874					jail_note(j, "sent SIGTERM to:");
875				}
876				printf(" %d", ki[i].ki_pid);
877			}
878		}
879	if (noted)
880		printf("\n");
881	if (j->nprocs > 0) {
882		clock_gettime(CLOCK_REALTIME, &j->timeout);
883		j->timeout.tv_sec += timeout;
884		return 1;
885	}
886	return 0;
887}
888
889/*
890 * Look up a user in the passwd and login.conf files.
891 */
892static int
893get_user_info(struct cfjail *j, const char *username,
894    const struct passwd **pwdp, login_cap_t **lcapp)
895{
896	const struct passwd *pwd;
897
898	errno = 0;
899	*pwdp = pwd = username ? getpwnam(username) : getpwuid(getuid());
900	if (pwd == NULL) {
901		if (errno)
902			jail_warnx(j, "getpwnam%s%s: %s", username ? " " : "",
903			    username ? username : "", strerror(errno));
904		else if (username)
905			jail_warnx(j, "%s: no such user", username);
906		else
907			jail_warnx(j, "unknown uid %d", getuid());
908		return -1;
909	}
910	*lcapp = login_getpwclass(pwd);
911	if (*lcapp == NULL) {
912		jail_warnx(j, "getpwclass %s: %s", pwd->pw_name,
913		    strerror(errno));
914		return -1;
915	}
916	/* Set the groups while the group file is still available */
917	if (initgroups(pwd->pw_name, pwd->pw_gid) < 0) {
918		jail_warnx(j, "initgroups %s: %s", pwd->pw_name,
919		    strerror(errno));
920		return -1;
921	}
922	return 0;
923}
924
925/*
926 * Make sure a mount or consolelog path is a valid absolute pathname
927 * with no symlinks.
928 */
929static int
930check_path(struct cfjail *j, const char *pname, const char *path, int isfile,
931    const char *umount_type)
932{
933	struct stat st, mpst;
934	struct statfs stfs;
935	char *tpath, *p;
936	const char *jailpath;
937	size_t jplen;
938
939	if (path[0] != '/') {
940		jail_warnx(j, "%s: %s: not an absolute pathname",
941		    pname, path);
942		return -1;
943	}
944	/*
945	 * Only check for symlinks in components below the jail's path,
946	 * since that's where the security risk lies.
947	 */
948	jailpath = string_param(j->intparams[KP_PATH]);
949	if (jailpath == NULL)
950		jailpath = "";
951	jplen = strlen(jailpath);
952	if (!strncmp(path, jailpath, jplen) && path[jplen] == '/') {
953		tpath = alloca(strlen(path) + 1);
954		strcpy(tpath, path);
955		for (p = tpath + jplen; p != NULL; ) {
956			p = strchr(p + 1, '/');
957			if (p)
958				*p = '\0';
959			if (lstat(tpath, &st) < 0) {
960				if (errno == ENOENT && isfile && !p)
961					break;
962				jail_warnx(j, "%s: %s: %s", pname, tpath,
963				    strerror(errno));
964				return -1;
965			}
966			if (S_ISLNK(st.st_mode)) {
967				jail_warnx(j, "%s: %s is a symbolic link",
968				    pname, tpath);
969				return -1;
970			}
971			if (p)
972				*p = '/';
973		}
974	}
975	if (umount_type != NULL) {
976		if (stat(path, &st) < 0 || statfs(path, &stfs) < 0) {
977			jail_warnx(j, "%s: %s: %s", pname, path,
978			    strerror(errno));
979			return -1;
980		}
981		if (stat(stfs.f_mntonname, &mpst) < 0) {
982			jail_warnx(j, "%s: %s: %s", pname, stfs.f_mntonname,
983			    strerror(errno));
984			return -1;
985		}
986		if (st.st_ino != mpst.st_ino) {
987			jail_warnx(j, "%s: %s: not a mount point",
988			    pname, path);
989			return -1;
990		}
991		if (strcmp(stfs.f_fstypename, umount_type)) {
992			jail_warnx(j, "%s: %s: not a %s mount",
993			    pname, path, umount_type);
994			return -1;
995		}
996	}
997	return 0;
998}
999