d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
667285c4e3
freebsd-dev/sys/kern/kern_jail.c

586 lines
13 KiB
C
Raw Normal View History

/* -> /*- for copyright notices, minor format tweaks as necessary
2005-01-06 23:35:40 +00:00
/*-
Add beer-ware license and $Id$ Noticed by: dillon
1999-04-30 06:51:51 +00:00
* ----------------------------------------------------------------------------
* "THE BEER-WARE LICENSE" (Revision 42):
* <phk@FreeBSD.ORG> wrote this file. As long as you retain this notice you
* can do whatever you want with this stuff. If we meet some day, and you think
* this stuff is worth it, you can buy me a beer in return. Poul-Henning Kamp
* ----------------------------------------------------------------------------
*/
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Add two missing includes and remove two uneeded. This is quite serious fix, because even with MAC framework compiled in, MAC entry points in those two files were simply ignored.
2004-06-27 09:03:22 +00:00
#include "opt_mac.h"
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
#include <sys/param.h>
#include <sys/types.h>
#include <sys/kernel.h>
#include <sys/systm.h>
#include <sys/errno.h>
#include <sys/sysproto.h>
Add two missing includes and remove two uneeded. This is quite serious fix, because even with MAC framework compiled in, MAC entry points in those two files were simply ignored.
2004-06-27 09:03:22 +00:00
#include <sys/mac.h>
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
#include <sys/malloc.h>
#include <sys/proc.h>
Defer the vrele() on a jail's root vnode reference from prison_free() to a new prison_complete() task run by a task queue. This removes a requirement for grabbing Giant in crfree(). Embed the 'struct task' in 'struct prison' so that we don't have to allocate memory from prison_free() (which means we also defer the FREE()). With this change, I believe grabbing Giant from crfree() can now be removed, but need to check the uidinfo code paths. To avoid header pollution, move the definition of 'struct task' to _task.h, and recursively include from taskqueue.h and jail.h; much preferably to all files including jail.h picking up a requirement to include taskqueue.h. Bumped into by: sam Reviewed by: bde, tjr
2004-01-23 20:44:26 +00:00
#include <sys/taskqueue.h>
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
#include <sys/jail.h>
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
#include <sys/lock.h>
#include <sys/mutex.h>
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
#include <sys/namei.h>
Rename sysctl security.jail.getfsstatroot_only to security.jail.enforce_statfs and extend its functionality: value policy 0 show all mount-points without any restrictions 1 show only mount-points below jail's chroot and show only part of the mount-point's path (if jail's chroot directory is /jails/foo and mount-point is /jails/foo/usr/home only /usr/home will be shown) 2 show only mount-point where jail's chroot directory is placed. Default value is 2. Discussed with: rwatson
2005-06-09 18:49:19 +00:00
#include <sys/mount.h>
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
#include <sys/queue.h>
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
#include <sys/socket.h>
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
#include <sys/syscallsubr.h>
Yet-another-update: rename ``kern.prison'' to a new sysctl root entry, ``jail'', and move the set_hostname_allowed sysctl there, as well as fixing a bug in the sysctl that resulted in jails being over-limited (preventing them from reading as well as writing the hostname). Also, correct some formatting issues, courtesy bde :-). Reviewed by: phk Approved by: jkh
2000-02-12 13:41:56 +00:00
#include <sys/sysctl.h>
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
#include <sys/vnode.h>
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
#include <net/if.h>
#include <netinet/in.h>
MALLOC_DEFINE(M_PRISON, "prison", "Prison structures");
- Attempt to help declutter kern. sysctl by moving security out from beneath it. Reviewed by: rwatson
2002-01-16 06:55:30 +00:00
SYSCTL_DECL(_security);
SYSCTL_NODE(_security, OID_AUTO, jail, CTLFLAG_RW, 0,
Yet-another-update: rename ``kern.prison'' to a new sysctl root entry, ``jail'', and move the set_hostname_allowed sysctl there, as well as fixing a bug in the sysctl that resulted in jails being over-limited (preventing them from reading as well as writing the hostname). Also, correct some formatting issues, courtesy bde :-). Reviewed by: phk Approved by: jkh
2000-02-12 13:41:56 +00:00
"Jail rules");
int jail_set_hostname_allowed = 1;
- Attempt to help declutter kern. sysctl by moving security out from beneath it. Reviewed by: rwatson
2002-01-16 06:55:30 +00:00
SYSCTL_INT(_security_jail, OID_AUTO, set_hostname_allowed, CTLFLAG_RW,
Yet-another-update: rename ``kern.prison'' to a new sysctl root entry, ``jail'', and move the set_hostname_allowed sysctl there, as well as fixing a bug in the sysctl that resulted in jails being over-limited (preventing them from reading as well as writing the hostname). Also, correct some formatting issues, courtesy bde :-). Reviewed by: phk Approved by: jkh
2000-02-12 13:41:56 +00:00
&jail_set_hostname_allowed, 0,
"Processes in jail can set their hostnames");
o Modify jail to limit creation of sockets to UNIX domain sockets, TCP/IP (v4) sockets, and routing sockets. Previously, interaction with IPv6 was not well-defined, and might be inappropriate for some environments. Similarly, sysctl MIB entries providing interface information also give out only addresses from those protocol domains. For the time being, this functionality is enabled by default, and toggleable using the sysctl variable jail.socket_unixiproute_only. In the future, protocol domains will be able to determine whether or not they are ``jail aware''. o Further limitations on process use of getpriority() and setpriority() by jailed processes. Addresses problem described in kern/17878. Reviewed by: phk, jmg
2000-06-04 04:28:31 +00:00
int jail_socket_unixiproute_only = 1;
- Attempt to help declutter kern. sysctl by moving security out from beneath it. Reviewed by: rwatson
2002-01-16 06:55:30 +00:00
SYSCTL_INT(_security_jail, OID_AUTO, socket_unixiproute_only, CTLFLAG_RW,
o Modify jail to limit creation of sockets to UNIX domain sockets, TCP/IP (v4) sockets, and routing sockets. Previously, interaction with IPv6 was not well-defined, and might be inappropriate for some environments. Similarly, sysctl MIB entries providing interface information also give out only addresses from those protocol domains. For the time being, this functionality is enabled by default, and toggleable using the sysctl variable jail.socket_unixiproute_only. In the future, protocol domains will be able to determine whether or not they are ``jail aware''. o Further limitations on process use of getpriority() and setpriority() by jailed processes. Addresses problem described in kern/17878. Reviewed by: phk, jmg
2000-06-04 04:28:31 +00:00
&jail_socket_unixiproute_only, 0,
"Processes in jail are limited to creating UNIX/IPv4/route sockets only");
o Deny access to System V IPC from within jail by default, as in the current implementation, jail neither virtualizes the Sys V IPC namespace, nor provides inter-jail protections on IPC objects. o Support for System V IPC can be enabled by setting jail.sysvipc_allowed=1 using sysctl. o This is not the "real fix" which involves virtualizing the System V IPC namespace, but prevents processes within jail from influencing those outside of jail when not approved by the administrator. Reported by: Paulo Fragoso <paulo@nlink.com.br>
2000-10-31 01:34:00 +00:00
int jail_sysvipc_allowed = 0;
- Attempt to help declutter kern. sysctl by moving security out from beneath it. Reviewed by: rwatson
2002-01-16 06:55:30 +00:00
SYSCTL_INT(_security_jail, OID_AUTO, sysvipc_allowed, CTLFLAG_RW,
o Deny access to System V IPC from within jail by default, as in the current implementation, jail neither virtualizes the Sys V IPC namespace, nor provides inter-jail protections on IPC objects. o Support for System V IPC can be enabled by setting jail.sysvipc_allowed=1 using sysctl. o This is not the "real fix" which involves virtualizing the System V IPC namespace, but prevents processes within jail from influencing those outside of jail when not approved by the administrator. Reported by: Paulo Fragoso <paulo@nlink.com.br>
2000-10-31 01:34:00 +00:00
&jail_sysvipc_allowed, 0,
"Processes in jail can use System V IPC primitives");
Rename sysctl security.jail.getfsstatroot_only to security.jail.enforce_statfs and extend its functionality: value policy 0 show all mount-points without any restrictions 1 show only mount-points below jail's chroot and show only part of the mount-point's path (if jail's chroot directory is /jails/foo and mount-point is /jails/foo/usr/home only /usr/home will be shown) 2 show only mount-point where jail's chroot directory is placed. Default value is 2. Discussed with: rwatson
2005-06-09 18:49:19 +00:00
static int jail_enforce_statfs = 2;
SYSCTL_INT(_security_jail, OID_AUTO, enforce_statfs, CTLFLAG_RW,
&jail_enforce_statfs, 0,
"Processes in jail cannot see all mounted file systems");
By default, when a process in jail calls getfsstat(), only return the data for the file system on which the jail's root vnode is located. Previous behavior (show data for all mountpoints) can be restored by setting security.jail.getfsstatroot_only to 0. Note: this also has the effect of hiding other mounts inside a jail, such as /dev, /tmp, and /proc, but errs on the side of leaking less information.
2004-02-14 18:31:11 +00:00
Give jail(8) the feature to allow raw sockets from within a jail, which is less restrictive but allows for more flexible jail usage (for those who are willing to make the sacrifice). The default is off, but allowing raw sockets within jails can now be accomplished by tuning security.jail.allow_raw_sockets to 1. Turning this on will allow you to use things like ping(8) or traceroute(8) from within a jail. The patch being committed is not identical to the patch in the PR. The committed version is more friendly to APIs which pjd is working on, so it should integrate into his work quite nicely. This change has also been presented and addressed on the freebsd-hackers mailing list. Submitted by: Christian S.J. Peron <maneo@bsdpro.com> PR: kern/65800
2004-04-26 19:46:52 +00:00
int jail_allow_raw_sockets = 0;
SYSCTL_INT(_security_jail, OID_AUTO, allow_raw_sockets, CTLFLAG_RW,
&jail_allow_raw_sockets, 0,
"Prison root can create raw sockets");
Add a new sysctl, "security.jail.chflags_allowed", which controls the behaviour of chflags within a jail. If set to 0 (the default), then a jailed root user is treated as an unprivileged user; if set to 1, then a jailed root user is treated the same as an unjailed root user. This is necessary to allow "make installworld" to work inside a jail, since it attempts to manipulate the system immutable flag on certain files. Discussed with: csjp, rwatson MFC after: 2 weeks
2005-02-08 21:31:11 +00:00
int jail_chflags_allowed = 0;
SYSCTL_INT(_security_jail, OID_AUTO, chflags_allowed, CTLFLAG_RW,
&jail_chflags_allowed, 0,
"Processes in jail can alter system file flags");
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
/* allprison, lastprid, and prisoncount are protected by allprison_mtx. */
struct prisonlist allprison;
struct mtx allprison_mtx;
int lastprid = 0;
int prisoncount = 0;
static void init_prison(void *);
Defer the vrele() on a jail's root vnode reference from prison_free() to a new prison_complete() task run by a task queue. This removes a requirement for grabbing Giant in crfree(). Embed the 'struct task' in 'struct prison' so that we don't have to allocate memory from prison_free() (which means we also defer the FREE()). With this change, I believe grabbing Giant from crfree() can now be removed, but need to check the uidinfo code paths. To avoid header pollution, move the definition of 'struct task' to _task.h, and recursively include from taskqueue.h and jail.h; much preferably to all files including jail.h picking up a requirement to include taskqueue.h. Bumped into by: sam Reviewed by: bde, tjr
2004-01-23 20:44:26 +00:00
static void prison_complete(void *context, int pending);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
static struct prison *prison_find(int);
static int sysctl_jail_list(SYSCTL_HANDLER_ARGS);
static void
init_prison(void *data __unused)
{
mtx_init(&allprison_mtx, "allprison", NULL, MTX_DEF);
LIST_INIT(&allprison);
}
SYSINIT(prison, SI_SUB_INTRINSIC, SI_ORDER_ANY, init_prison, NULL);
Pushdown Giant for acct(), kqueue(), kevent(), execve(), fork(), vfork(), rfork(), jail().
2001-09-01 03:04:31 +00:00
/*
* MPSAFE
style(9)
2003-04-28 18:32:19 +00:00
*
* struct jail_args {
* struct jail *jail;
* };
Pushdown Giant for acct(), kqueue(), kevent(), execve(), fork(), vfork(), rfork(), jail().
2001-09-01 03:04:31 +00:00
*/
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
int
style(9)
2003-04-28 18:32:19 +00:00
jail(struct thread *td, struct jail_args *uap)
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
{
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
struct nameidata nd;
struct prison *pr, *tpr;
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
struct jail j;
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
struct jail_attach_args jaa;
int error, tryprid;
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
style(9)
2003-04-28 18:32:19 +00:00
error = copyin(uap->jail, &j, sizeof(j));
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
if (error)
- Catch up to the new ucred API. - Add proc locking to the jail() syscall. This mostly involved shuffling a few things around so that blockable things like malloc and copyin were performed before acquiring the lock and checking the existing ucred and then updating the ucred as one "atomic" change under the proc lock.
2001-10-11 23:39:43 +00:00
return (error);
if (j.version != 0)
return (EINVAL);
style(9)
2003-04-28 18:32:19 +00:00
MALLOC(pr, struct prison *, sizeof(*pr), M_PRISON, M_WAITOK | M_ZERO);
Change callers of mtx_init() to pass in an appropriate lock type name. In most cases NULL is passed, but in some cases such as network driver locks (which use the MTX_NETWORK_LOCK macro) and UMA zone locks, a name is used. Tested on: i386, alpha, sparc64
2002-04-04 21:03:38 +00:00
mtx_init(&pr->pr_mtx, "jail mutex", NULL, MTX_DEF);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
pr->pr_ref = 1;
style(9)
2003-04-28 18:32:19 +00:00
error = copyinstr(j.path, &pr->pr_path, sizeof(pr->pr_path), 0);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
if (error)
goto e_killmtx;
mtx_lock(&Giant);
NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, pr->pr_path, td);
error = namei(&nd);
if (error) {
mtx_unlock(&Giant);
goto e_killmtx;
}
pr->pr_root = nd.ni_vp;
VOP_UNLOCK(nd.ni_vp, 0, td);
NDFREE(&nd, NDF_ONLY_PNBUF);
mtx_unlock(&Giant);
style(9)
2003-04-28 18:32:19 +00:00
error = copyinstr(j.hostname, &pr->pr_host, sizeof(pr->pr_host), 0);
- Catch up to the new ucred API. - Add proc locking to the jail() syscall. This mostly involved shuffling a few things around so that blockable things like malloc and copyin were performed before acquiring the lock and checking the existing ucred and then updating the ucred as one "atomic" change under the proc lock.
2001-10-11 23:39:43 +00:00
if (error)
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
goto e_dropvnref;
pr->pr_ip = j.ip_number;
pr->pr_linux = NULL;
pr->pr_securelevel = securelevel;
/* Determine next pr_id and add prison to allprison list. */
mtx_lock(&allprison_mtx);
tryprid = lastprid + 1;
if (tryprid == JAIL_MAX)
tryprid = 1;
next:
LIST_FOREACH(tpr, &allprison, pr_list) {
if (tpr->pr_id == tryprid) {
tryprid++;
if (tryprid == JAIL_MAX) {
mtx_unlock(&allprison_mtx);
error = EAGAIN;
goto e_dropvnref;
}
goto next;
}
}
pr->pr_id = jaa.jid = lastprid = tryprid;
LIST_INSERT_HEAD(&allprison, pr, pr_list);
prisoncount++;
mtx_unlock(&allprison_mtx);
error = jail_attach(td, &jaa);
if (error)
goto e_dropprref;
mtx_lock(&pr->pr_mtx);
pr->pr_ref--;
mtx_unlock(&pr->pr_mtx);
td->td_retval[0] = jaa.jid;
return (0);
e_dropprref:
mtx_lock(&allprison_mtx);
LIST_REMOVE(pr, pr_list);
prisoncount--;
mtx_unlock(&allprison_mtx);
e_dropvnref:
The jail syscall calls chroot, which is not mpsafe, so put back a mtx_lock(&Giant) around that call. Reviewed by: arr
2002-07-01 20:46:01 +00:00
mtx_lock(&Giant);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
vrele(pr->pr_root);
The jail syscall calls chroot, which is not mpsafe, so put back a mtx_lock(&Giant) around that call. Reviewed by: arr
2002-07-01 20:46:01 +00:00
mtx_unlock(&Giant);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
e_killmtx:
mtx_destroy(&pr->pr_mtx);
FREE(pr, M_PRISON);
return (error);
}
/*
* MPSAFE
style(9)
2003-04-28 18:32:19 +00:00
*
* struct jail_attach_args {
* int jid;
* };
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
*/
int
style(9)
2003-04-28 18:32:19 +00:00
jail_attach(struct thread *td, struct jail_attach_args *uap)
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
{
struct proc *p;
struct ucred *newcred, *oldcred;
struct prison *pr;
int error;
Rework jail_attach(2) so that an already jailed process cannot hop to another jail. Submitted by: rwatson
2004-02-19 21:03:20 +00:00
/*
* XXX: Note that there is a slight race here if two threads
* in the same privileged process attempt to attach to two
* different jails at the same time. It is important for
* user processes not to do this, or they might end up with
* a process root from one prison, but attached to the jail
* of another.
*/
error = suser(td);
if (error)
return (error);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
Rework jail_attach(2) so that an already jailed process cannot hop to another jail. Submitted by: rwatson
2004-02-19 21:03:20 +00:00
p = td->td_proc;
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
mtx_lock(&allprison_mtx);
pr = prison_find(uap->jid);
if (pr == NULL) {
mtx_unlock(&allprison_mtx);
return (EINVAL);
}
pr->pr_ref++;
mtx_unlock(&pr->pr_mtx);
mtx_unlock(&allprison_mtx);
mtx_lock(&Giant);
vn_lock(pr->pr_root, LK_EXCLUSIVE | LK_RETRY, td);
if ((error = change_dir(pr->pr_root, td)) != 0)
goto e_unlock;
#ifdef MAC
if ((error = mac_check_vnode_chroot(td->td_ucred, pr->pr_root)))
goto e_unlock;
#endif
VOP_UNLOCK(pr->pr_root, 0, td);
change_root(pr->pr_root, td);
mtx_unlock(&Giant);
- Catch up to the new ucred API. - Add proc locking to the jail() syscall. This mostly involved shuffling a few things around so that blockable things like malloc and copyin were performed before acquiring the lock and checking the existing ucred and then updating the ucred as one "atomic" change under the proc lock.
2001-10-11 23:39:43 +00:00
newcred = crget();
PROC_LOCK(p);
oldcred = p->p_ucred;
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
setsugid(p);
- Catch up to the new ucred API. - Add proc locking to the jail() syscall. This mostly involved shuffling a few things around so that blockable things like malloc and copyin were performed before acquiring the lock and checking the existing ucred and then updating the ucred as one "atomic" change under the proc lock.
2001-10-11 23:39:43 +00:00
crcopy(newcred, oldcred);
- The prison mutex cannot possibly protect pointers to the prison it protects, so don't bother locking it while we assign it to a ucred's cr_prison. - Fully construct the new credential for a process before assigning it to p_ucred.
2003-04-17 22:26:53 +00:00
newcred->cr_prison = pr;
- Catch up to the new ucred API. - Add proc locking to the jail() syscall. This mostly involved shuffling a few things around so that blockable things like malloc and copyin were performed before acquiring the lock and checking the existing ucred and then updating the ucred as one "atomic" change under the proc lock.
2001-10-11 23:39:43 +00:00
p->p_ucred = newcred;
PROC_UNLOCK(p);
crfree(oldcred);
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return (0);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
e_unlock:
VOP_UNLOCK(pr->pr_root, 0, td);
mtx_unlock(&Giant);
mtx_lock(&pr->pr_mtx);
pr->pr_ref--;
mtx_unlock(&pr->pr_mtx);
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return (error);
}
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
/*
* Returns a locked prison instance, or NULL on failure.
*/
static struct prison *
prison_find(int prid)
{
struct prison *pr;
mtx_assert(&allprison_mtx, MA_OWNED);
LIST_FOREACH(pr, &allprison, pr_list) {
if (pr->pr_id == prid) {
mtx_lock(&pr->pr_mtx);
return (pr);
}
}
return (NULL);
}
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
void
prison_free(struct prison *pr)
{
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
mtx_lock(&allprison_mtx);
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
mtx_lock(&pr->pr_mtx);
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
pr->pr_ref--;
if (pr->pr_ref == 0) {
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
LIST_REMOVE(pr, pr_list);
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
mtx_unlock(&pr->pr_mtx);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
prisoncount--;
mtx_unlock(&allprison_mtx);
Defer the vrele() on a jail's root vnode reference from prison_free() to a new prison_complete() task run by a task queue. This removes a requirement for grabbing Giant in crfree(). Embed the 'struct task' in 'struct prison' so that we don't have to allocate memory from prison_free() (which means we also defer the FREE()). With this change, I believe grabbing Giant from crfree() can now be removed, but need to check the uidinfo code paths. To avoid header pollution, move the definition of 'struct task' to _task.h, and recursively include from taskqueue.h and jail.h; much preferably to all files including jail.h picking up a requirement to include taskqueue.h. Bumped into by: sam Reviewed by: bde, tjr
2004-01-23 20:44:26 +00:00
TASK_INIT(&pr->pr_task, 0, prison_complete, pr);
- Use taskqueue_thread rather than taskqueue_swi since our task is going to vrele, which may vop lock. This is not safe in a software interrupt context.
2005-04-05 08:51:45 +00:00
taskqueue_enqueue(taskqueue_thread, &pr->pr_task);
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
return;
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
}
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
mtx_unlock(&pr->pr_mtx);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
mtx_unlock(&allprison_mtx);
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
}
Defer the vrele() on a jail's root vnode reference from prison_free() to a new prison_complete() task run by a task queue. This removes a requirement for grabbing Giant in crfree(). Embed the 'struct task' in 'struct prison' so that we don't have to allocate memory from prison_free() (which means we also defer the FREE()). With this change, I believe grabbing Giant from crfree() can now be removed, but need to check the uidinfo code paths. To avoid header pollution, move the definition of 'struct task' to _task.h, and recursively include from taskqueue.h and jail.h; much preferably to all files including jail.h picking up a requirement to include taskqueue.h. Bumped into by: sam Reviewed by: bde, tjr
2004-01-23 20:44:26 +00:00
static void
prison_complete(void *context, int pending)
{
struct prison *pr;
pr = (struct prison *)context;
mtx_lock(&Giant);
vrele(pr->pr_root);
mtx_unlock(&Giant);
mtx_destroy(&pr->pr_mtx);
if (pr->pr_linux != NULL)
FREE(pr->pr_linux, M_PRISON);
FREE(pr, M_PRISON);
}
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
void
prison_hold(struct prison *pr)
{
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
mtx_lock(&pr->pr_mtx);
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
pr->pr_ref++;
o Introduce pr_mtx into struct prison, providing protection for the mutable contents of struct prison (hostname, securelevel, refcount, pr_linux, ...) o Generally introduce mtx_lock()/mtx_unlock() calls throughout kern/ so as to enforce these protections, in particular, in kern_mib.c protection sysctl access to the hostname and securelevel, as well as kern_prot.c access to the securelevel for access control purposes. o Rewrite linux emulator abstractions for accessing per-jail linux mib entries (osname, osrelease, osversion) so that they don't return a pointer to the text in the struct linux_prison, rather, a copy to an array passed into the calls. Likewise, update linprocfs to use these primitives. o Update in_pcb.c to always use prison_getip() rather than directly accessing struct prison. Reviewed by: jhb
2001-12-03 16:12:27 +00:00
mtx_unlock(&pr->pr_mtx);
}
u_int32_t
prison_getip(struct ucred *cred)
{
return (cred->cr_prison->pr_ip);
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
}
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
int
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
prison_ip(struct ucred *cred, int flag, u_int32_t *ip)
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
{
u_int32_t tmp;
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
if (!jailed(cred))
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return (0);
if (flag)
tmp = *ip;
else
tmp = ntohl(*ip);
if (tmp == INADDR_ANY) {
if (flag)
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
*ip = cred->cr_prison->pr_ip;
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
else
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
*ip = htonl(cred->cr_prison->pr_ip);
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return (0);
}
Anton kindly pointed out (and fixed) a bug in the Jail handling of the bind() call on IPv4 sockets: Currently, if one tries to bind a socket using INADDR_LOOPBACK inside a jail, it will fail because prison_ip() does not take this possibility into account. On the other hand, when one tries to connect(), for example, to localhost, prison_remote_ip() will silently convert INADDR_LOOPBACK to the jail's IP address. Therefore, it is desirable to make bind() to do this implicit conversion as well. Apart from this, the patch also replaces 0x7f000001 in prison_remote_ip() to a more correct INADDR_LOOPBACK. This is a 4.4-RELEASE "during the freeze, thanks" MFC candidate. Submitted by: Anton Berezin <tobez@FreeBSD.org> Discussed with at some point: phk MFC after: 3 days
2001-08-03 18:21:06 +00:00
if (tmp == INADDR_LOOPBACK) {
if (flag)
*ip = cred->cr_prison->pr_ip;
else
*ip = htonl(cred->cr_prison->pr_ip);
return (0);
}
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
if (cred->cr_prison->pr_ip != tmp)
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return (1);
return (0);
}
void
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
prison_remote_ip(struct ucred *cred, int flag, u_int32_t *ip)
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
{
u_int32_t tmp;
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
if (!jailed(cred))
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return;
if (flag)
tmp = *ip;
else
tmp = ntohl(*ip);
Anton kindly pointed out (and fixed) a bug in the Jail handling of the bind() call on IPv4 sockets: Currently, if one tries to bind a socket using INADDR_LOOPBACK inside a jail, it will fail because prison_ip() does not take this possibility into account. On the other hand, when one tries to connect(), for example, to localhost, prison_remote_ip() will silently convert INADDR_LOOPBACK to the jail's IP address. Therefore, it is desirable to make bind() to do this implicit conversion as well. Apart from this, the patch also replaces 0x7f000001 in prison_remote_ip() to a more correct INADDR_LOOPBACK. This is a 4.4-RELEASE "during the freeze, thanks" MFC candidate. Submitted by: Anton Berezin <tobez@FreeBSD.org> Discussed with at some point: phk MFC after: 3 days
2001-08-03 18:21:06 +00:00
if (tmp == INADDR_LOOPBACK) {
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
if (flag)
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
*ip = cred->cr_prison->pr_ip;
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
else
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
*ip = htonl(cred->cr_prison->pr_ip);
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
return;
}
return;
}
int
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
prison_if(struct ucred *cred, struct sockaddr *sa)
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
{
style(9)
2003-04-28 18:32:19 +00:00
struct sockaddr_in *sai;
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
int ok;
style(9)
2003-04-28 18:32:19 +00:00
sai = (struct sockaddr_in *)sa;
o Modify jail to limit creation of sockets to UNIX domain sockets, TCP/IP (v4) sockets, and routing sockets. Previously, interaction with IPv6 was not well-defined, and might be inappropriate for some environments. Similarly, sysctl MIB entries providing interface information also give out only addresses from those protocol domains. For the time being, this functionality is enabled by default, and toggleable using the sysctl variable jail.socket_unixiproute_only. In the future, protocol domains will be able to determine whether or not they are ``jail aware''. o Further limitations on process use of getpriority() and setpriority() by jailed processes. Addresses problem described in kern/17878. Reviewed by: phk, jmg
2000-06-04 04:28:31 +00:00
if ((sai->sin_family != AF_INET) && jail_socket_unixiproute_only)
ok = 1;
else if (sai->sin_family != AF_INET)
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
ok = 0;
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
else if (cred->cr_prison->pr_ip != ntohl(sai->sin_addr.s_addr))
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
ok = 1;
else
ok = 0;
return (ok);
}
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
/*
* Return 0 if jails permit p1 to frob p2, otherwise ESRCH.
*/
int
style(9)
2003-04-28 18:32:19 +00:00
prison_check(struct ucred *cred1, struct ucred *cred2)
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
{
if (jailed(cred1)) {
if (!jailed(cred2))
return (ESRCH);
if (cred2->cr_prison != cred1->cr_prison)
return (ESRCH);
}
return (0);
}
/*
* Return 1 if the passed credential is in a jail, otherwise 0.
*/
int
style(9)
2003-04-28 18:32:19 +00:00
jailed(struct ucred *cred)
o Move per-process jail pointer (p->pr_prison) to inside of the subject credential structure, ucred (cr->cr_prison). o Allow jail inheritence to be a function of credential inheritence. o Abstract prison structure reference counting behind pr_hold() and pr_free(), invoked by the similarly named credential reference management functions, removing this code from per-ABI fork/exit code. o Modify various jail() functions to use struct ucred arguments instead of struct proc arguments. o Introduce jailed() function to determine if a credential is jailed, rather than directly checking pointers all over the place. o Convert PRISON_CHECK() macro to prison_check() function. o Move jail() function prototypes to jail.h. o Emulate the P_JAILED flag in fill_kinfo_proc() and no longer set the flag in the process flags field itself. o Eliminate that "const" qualifier from suser/p_can/etc to reflect mutex use. Notes: o Some further cleanup of the linux/jail code is still required. o It's now possible to consider resolving some of the process vs credential based permission checking confusion in the socket code. o Mutex protection of struct prison is still not present, and is required to protect the reference count plus some fields in the structure. Reviewed by: freebsd-arch Obtained from: TrustedBSD Project
2001-02-21 06:39:57 +00:00
{
return (cred->cr_prison != NULL);
}
Add a function which returns the correct hostname for a given credential. Reviewed by: phk
2002-02-27 14:58:32 +00:00
/*
* Return the correct hostname for the passed credential.
*/
Make getcredhostname() take a buffer and the buffer's size as arguments. The correct hostname is copied into the buffer while having the prison's lock acquired in a jailed process' case. Reviewed by: jhb, rwatson
2002-02-27 16:43:20 +00:00
void
style(9)
2003-04-28 18:32:19 +00:00
getcredhostname(struct ucred *cred, char *buf, size_t size)
Add a function which returns the correct hostname for a given credential. Reviewed by: phk
2002-02-27 14:58:32 +00:00
{
Make getcredhostname() take a buffer and the buffer's size as arguments. The correct hostname is copied into the buffer while having the prison's lock acquired in a jailed process' case. Reviewed by: jhb, rwatson
2002-02-27 16:43:20 +00:00
if (jailed(cred)) {
mtx_lock(&cred->cr_prison->pr_mtx);
Use strlcpy() instead of strncpy() to copy NUL terminated strings for safety and consistency.
2002-10-17 20:03:38 +00:00
strlcpy(buf, cred->cr_prison->pr_host, size);
Make getcredhostname() take a buffer and the buffer's size as arguments. The correct hostname is copied into the buffer while having the prison's lock acquired in a jailed process' case. Reviewed by: jhb, rwatson
2002-02-27 16:43:20 +00:00
mtx_unlock(&cred->cr_prison->pr_mtx);
style(9)
2003-04-28 18:32:19 +00:00
} else
Use strlcpy() instead of strncpy() to copy NUL terminated strings for safety and consistency.
2002-10-17 20:03:38 +00:00
strlcpy(buf, hostname, size);
Add a function which returns the correct hostname for a given credential. Reviewed by: phk
2002-02-27 14:58:32 +00:00
}
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
By default, when a process in jail calls getfsstat(), only return the data for the file system on which the jail's root vnode is located. Previous behavior (show data for all mountpoints) can be restored by setting security.jail.getfsstatroot_only to 0. Note: this also has the effect of hiding other mounts inside a jail, such as /dev, /tmp, and /proc, but errs on the side of leaking less information.
2004-02-14 18:31:11 +00:00
/*
Rename sysctl security.jail.getfsstatroot_only to security.jail.enforce_statfs and extend its functionality: value policy 0 show all mount-points without any restrictions 1 show only mount-points below jail's chroot and show only part of the mount-point's path (if jail's chroot directory is /jails/foo and mount-point is /jails/foo/usr/home only /usr/home will be shown) 2 show only mount-point where jail's chroot directory is placed. Default value is 2. Discussed with: rwatson
2005-06-09 18:49:19 +00:00
* Determine whether the subject represented by cred can "see"
* status of a mount point.
* Returns: 0 for permitted, ENOENT otherwise.
* XXX: This function should be called cr_canseemount() and should be
* placed in kern_prot.c.
By default, when a process in jail calls getfsstat(), only return the data for the file system on which the jail's root vnode is located. Previous behavior (show data for all mountpoints) can be restored by setting security.jail.getfsstatroot_only to 0. Note: this also has the effect of hiding other mounts inside a jail, such as /dev, /tmp, and /proc, but errs on the side of leaking less information.
2004-02-14 18:31:11 +00:00
*/
int
Rename sysctl security.jail.getfsstatroot_only to security.jail.enforce_statfs and extend its functionality: value policy 0 show all mount-points without any restrictions 1 show only mount-points below jail's chroot and show only part of the mount-point's path (if jail's chroot directory is /jails/foo and mount-point is /jails/foo/usr/home only /usr/home will be shown) 2 show only mount-point where jail's chroot directory is placed. Default value is 2. Discussed with: rwatson
2005-06-09 18:49:19 +00:00
prison_canseemount(struct ucred *cred, struct mount *mp)
By default, when a process in jail calls getfsstat(), only return the data for the file system on which the jail's root vnode is located. Previous behavior (show data for all mountpoints) can be restored by setting security.jail.getfsstatroot_only to 0. Note: this also has the effect of hiding other mounts inside a jail, such as /dev, /tmp, and /proc, but errs on the side of leaking less information.
2004-02-14 18:31:11 +00:00
{
Rename sysctl security.jail.getfsstatroot_only to security.jail.enforce_statfs and extend its functionality: value policy 0 show all mount-points without any restrictions 1 show only mount-points below jail's chroot and show only part of the mount-point's path (if jail's chroot directory is /jails/foo and mount-point is /jails/foo/usr/home only /usr/home will be shown) 2 show only mount-point where jail's chroot directory is placed. Default value is 2. Discussed with: rwatson
2005-06-09 18:49:19 +00:00
struct prison *pr;
struct statfs *sp;
size_t len;
By default, when a process in jail calls getfsstat(), only return the data for the file system on which the jail's root vnode is located. Previous behavior (show data for all mountpoints) can be restored by setting security.jail.getfsstatroot_only to 0. Note: this also has the effect of hiding other mounts inside a jail, such as /dev, /tmp, and /proc, but errs on the side of leaking less information.
2004-02-14 18:31:11 +00:00
Rename sysctl security.jail.getfsstatroot_only to security.jail.enforce_statfs and extend its functionality: value policy 0 show all mount-points without any restrictions 1 show only mount-points below jail's chroot and show only part of the mount-point's path (if jail's chroot directory is /jails/foo and mount-point is /jails/foo/usr/home only /usr/home will be shown) 2 show only mount-point where jail's chroot directory is placed. Default value is 2. Discussed with: rwatson
2005-06-09 18:49:19 +00:00
if (!jailed(cred) || jail_enforce_statfs == 0)
return (0);
pr = cred->cr_prison;
if (pr->pr_root->v_mount == mp)
return (0);
if (jail_enforce_statfs == 2)
return (ENOENT);
/*
* If jail's chroot directory is set to "/" we should be able to see
* all mount-points from inside a jail.
* This is ugly check, but this is the only situation when jail's
* directory ends with '/'.
*/
if (strcmp(pr->pr_path, "/") == 0)
return (0);
len = strlen(pr->pr_path);
sp = &mp->mnt_stat;
if (strncmp(pr->pr_path, sp->f_mntonname, len) != 0)
return (ENOENT);
/*
* Be sure that we don't have situation where jail's root directory
* is "/some/path" and mount point is "/some/pathpath".
*/
if (sp->f_mntonname[len] != '\0' && sp->f_mntonname[len] != '/')
return (ENOENT);
return (0);
}
void
prison_enforce_statfs(struct ucred *cred, struct mount *mp, struct statfs *sp)
{
char jpath[MAXPATHLEN];
struct prison *pr;
size_t len;
if (!jailed(cred) || jail_enforce_statfs == 0)
return;
pr = cred->cr_prison;
if (prison_canseemount(cred, mp) != 0) {
bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
strlcpy(sp->f_mntonname, "[restricted]",
sizeof(sp->f_mntonname));
return;
}
if (pr->pr_root->v_mount == mp) {
/*
* Clear current buffer data, so we are sure nothing from
* the valid path left there.
*/
bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
*sp->f_mntonname = '/';
return;
}
/*
* If jail's chroot directory is set to "/" we should be able to see
* all mount-points from inside a jail.
*/
if (strcmp(pr->pr_path, "/") == 0)
return;
len = strlen(pr->pr_path);
strlcpy(jpath, sp->f_mntonname + len, sizeof(jpath));
/*
* Clear current buffer data, so we are sure nothing from
* the valid path left there.
*/
bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
if (*jpath == '\0') {
/* Should never happen. */
*sp->f_mntonname = '/';
} else {
strlcpy(sp->f_mntonname, jpath, sizeof(sp->f_mntonname));
By default, when a process in jail calls getfsstat(), only return the data for the file system on which the jail's root vnode is located. Previous behavior (show data for all mountpoints) can be restored by setting security.jail.getfsstatroot_only to 0. Note: this also has the effect of hiding other mounts inside a jail, such as /dev, /tmp, and /proc, but errs on the side of leaking less information.
2004-02-14 18:31:11 +00:00
}
}
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
static int
sysctl_jail_list(SYSCTL_HANDLER_ARGS)
{
struct xprison *xp, *sxp;
struct prison *pr;
int count, error;
mtx_assert(&Giant, MA_OWNED);
Remove sysctl security.jail.list_allowed. This functionality was a misfeature, sysctl was added and turned off by default just to check if nobody complains. Reviewed by: rwatson
2004-03-15 12:10:34 +00:00
if (jailed(req->td->td_ucred))
By default, don't allow processes in a jail to list the set of jails in the system. Previous behavior (allowed) may be restored by setting security.jail.list_allowed=1.
2004-02-14 19:19:47 +00:00
return (0);
o In struct prison, add an allprison linked list of prisons (protected by allprison_mtx), a unique prison/jail identifier field, two path fields (pr_path for reporting and pr_root vnode instance) to store the chroot() point of each jail. o Add jail_attach(2) to allow a process to bind to an existing jail. o Add change_root() to perform the chroot operation on a specified vnode. o Generalize change_dir() to accept a vnode, and move namei() calls to callers of change_dir(). o Add a new sysctl (security.jail.list) which is a group of struct xprison instances that represent a snapshot of active jails. Reviewed by: rwatson, tjr
2003-04-09 02:55:18 +00:00
retry:
mtx_lock(&allprison_mtx);
count = prisoncount;
mtx_unlock(&allprison_mtx);
if (count == 0)
return (0);
sxp = xp = malloc(sizeof(*xp) * count, M_TEMP, M_WAITOK | M_ZERO);
mtx_lock(&allprison_mtx);
if (count != prisoncount) {
mtx_unlock(&allprison_mtx);
free(sxp, M_TEMP);
goto retry;
}
LIST_FOREACH(pr, &allprison, pr_list) {
mtx_lock(&pr->pr_mtx);
xp->pr_version = XPRISON_VERSION;
xp->pr_id = pr->pr_id;
strlcpy(xp->pr_path, pr->pr_path, sizeof(xp->pr_path));
strlcpy(xp->pr_host, pr->pr_host, sizeof(xp->pr_host));
xp->pr_ip = pr->pr_ip;
mtx_unlock(&pr->pr_mtx);
xp++;
}
mtx_unlock(&allprison_mtx);
error = SYSCTL_OUT(req, sxp, sizeof(*sxp) * count);
free(sxp, M_TEMP);
if (error)
return (error);
return (0);
}
SYSCTL_OID(_security_jail, OID_AUTO, list, CTLTYPE_STRUCT | CTLFLAG_RD,
NULL, 0, sysctl_jail_list, "S", "List of active jails");
Added sysctl security.jail.jailed. It returns 1 is process is inside of jail and 0 if it is not. Information if we are in jail or not is not a secret, there is plenty of ways to discover it. Many people are using own hack to check this and this will be a legal way from now on. It will be great if our starting scripts will take advantage of this sysctl to allow clean "boot" inside jail. Approved by: rwatson, scottl (mentor)
2004-02-19 14:29:14 +00:00
static int
sysctl_jail_jailed(SYSCTL_HANDLER_ARGS)
{
int error, injail;
injail = jailed(req->td->td_ucred);
error = SYSCTL_OUT(req, &injail, sizeof(injail));
return (error);
}
SYSCTL_PROC(_security_jail, OID_AUTO, jailed, CTLTYPE_INT | CTLFLAG_RD,
NULL, 0, sysctl_jail_jailed, "I", "Process in jail?");
Reference in New Issue Copy Permalink