d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
e4b16f2fb1
freebsd-dev/sys/kern/kern_jail.c

4596 lines
115 KiB
C
Raw Normal View History

/* -> /*- for copyright notices, minor format tweaks as necessary
2005-01-06 23:35:40 +00:00
/*-
sys/kern: adoption of SPDX licensing ID tags. Mainly focus on files that use BSD 2-Clause license, however the tool I was using misidentified many licenses so this was mostly a manual - error prone - task. The Software Package Data Exchange (SPDX) group provides a specification to make it easier for automated tools to detect and summarize well known opensource licenses. We are gradually adopting the specification, noting that the tags are considered only advisory and do not, in any way, superceed or replace the license texts.
2017-11-27 15:20:12 +00:00
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
* Copyright (c) 1999 Poul-Henning Kamp.
* Copyright (c) 2008 Bjoern A. Zeeb.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
* Copyright (c) 2009 James Gritton.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
* All rights reserved.
Whitespace/spelling fixes in advance of upcoming functional changes. Approved by: bz (mentor)
2009-03-27 13:13:59 +00:00
*
With the permissions of phk@ change the license on kern_jail.c to a 2 clause BSD license.
2008-11-28 19:23:46 +00:00
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
Add beer-ware license and $Id$ Noticed by: dillon
1999-04-30 06:51:51 +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
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#include "opt_ddb.h"
#include "opt_inet.h"
#include "opt_inet6.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
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>
#include <sys/malloc.h>
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#include <sys/osd.h>
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
#include <sys/priv.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/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>
Add the support for the AT_FDCWD and fd-relative name lookups to the namei(9). Based on the submission by rdivacky, sponsored by Google Summer of Code 2007 Reviewed by: rwatson, rdivacky Tested by: pho
2008-03-31 12:01:21 +00:00
#include <sys/fcntl.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>
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
#include <sys/linker.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>
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
#include <sys/racct.h>
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
#include <sys/rctl.h>
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
#include <sys/refcount.h>
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
#include <sys/sx.h>
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
#include <sys/sysent.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>
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
#include <sys/uuid.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>
Merge the remainder of kern_vimage.c and vimage.h into vnet.c and vnet.h, we now use jails (rather than vimages) as the abstraction for virtualization management, and what remained was specific to virtual network stacks. Minor cleanups are done in the process, and comments updated to reflect these changes. Reviewed by: bz Approved by: re (vimage blanket)
2009-08-01 19:26:27 +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
#include <net/if.h>
Merge the remainder of kern_vimage.c and vimage.h into vnet.c and vnet.h, we now use jails (rather than vimages) as the abstraction for virtualization management, and what remained was specific to virtual network stacks. Minor cleanups are done in the process, and comments updated to reflect these changes. Reviewed by: bz Approved by: re (vimage blanket)
2009-08-01 19:26:27 +00:00
#include <net/vnet.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 <netinet/in.h>
Merge the remainder of kern_vimage.c and vimage.h into vnet.c and vnet.h, we now use jails (rather than vimages) as the abstraction for virtualization management, and what remained was specific to virtual network stacks. Minor cleanups are done in the process, and comments updated to reflect these changes. Reviewed by: bz Approved by: re (vimage blanket)
2009-08-01 19:26:27 +00:00
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef DDB
#include <ddb/ddb.h>
#endif /* DDB */
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
Complete break-out of sys/sys/mac.h into sys/security/mac/mac_framework.h begun with a repo-copy of mac.h to mac_framework.h. sys/mac.h now contains the userspace and user<->kernel API and definitions, with all in-kernel interfaces moved to mac_framework.h, which is now included across most of the kernel instead. This change is the first step in a larger cleanup and sweep of MAC Framework interfaces in the kernel, and will not be MFC'd. Obtained from: TrustedBSD Project Sponsored by: SPARTA
2006-10-22 11:52:19 +00:00
#include <security/mac/mac_framework.h>
Change the default value of the "ip4" and "ip6" jail parameters to "disable", which only allows access to the parent/physical system's IP addresses when specifically directed. Change the default value of "host" to "new", and don't copy the parent host values, to insulate jails from the parent hostname et al. Approved by: re (kib), bz (mentor)
2009-07-29 16:41:02 +00:00
#define DEFAULT_HOSTUUID "00000000-0000-0000-0000-000000000000"
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
#define PRISON0_HOSTUUID_MODULE "hostuuid"
Change the default value of the "ip4" and "ip6" jail parameters to "disable", which only allows access to the parent/physical system's IP addresses when specifically directed. Change the default value of "host" to "new", and don't copy the parent host values, to insulate jails from the parent hostname et al. Approved by: re (kib), bz (mentor)
2009-07-29 16:41:02 +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
MALLOC_DEFINE(M_PRISON, "prison", "Prison structures");
Mark MALLOC_DEFINEs static that have no corresponding MALLOC_DECLAREs. This means that their use is restricted to a single C file.
2011-11-07 06:44:47 +00:00
static MALLOC_DEFINE(M_PRISON_RACCT, "prison_racct", "Prison racct structures");
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
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
/* Keep struct prison prison0 and some code in kern_jail_set() readable. */
#ifdef INET
#ifdef INET6
#define _PR_IP_SADDRSEL PR_IP4_SADDRSEL|PR_IP6_SADDRSEL
#else
#define _PR_IP_SADDRSEL PR_IP4_SADDRSEL
#endif
#else /* !INET */
#ifdef INET6
#define _PR_IP_SADDRSEL PR_IP6_SADDRSEL
#else
#define _PR_IP_SADDRSEL 0
#endif
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* prison0 describes what is "real" about the system. */
struct prison prison0 = {
.pr_id = 0,
.pr_name = "0",
.pr_ref = 1,
.pr_uref = 1,
.pr_path = "/",
.pr_securelevel = -1,
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
.pr_devfs_rsnum = 0,
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
.pr_state = PRISON_STATE_ALIVE,
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
.pr_childmax = JAIL_MAX,
Change the default value of the "ip4" and "ip6" jail parameters to "disable", which only allows access to the parent/physical system's IP addresses when specifically directed. Change the default value of "host" to "new", and don't copy the parent host values, to insulate jails from the parent hostname et al. Approved by: re (kib), bz (mentor)
2009-07-29 16:41:02 +00:00
.pr_hostuuid = DEFAULT_HOSTUUID,
(S)LIST_HEAD_INITIALIZER takes a (S)LIST_HEAD as an argument. Fix some wrong usages. Note: this does not affect generated binaries as this argument is not used. PR: 137213 Submitted by: Eygene Ryabinkin (initial version) MFC after: 1 month
2009-12-28 22:56:30 +00:00
.pr_children = LIST_HEAD_INITIALIZER(prison0.pr_children),
Make it possible to change the vnet sysctl variables on jails with their own virtual network stack. Jails only inheriting a network stack cannot change anything that cannot be changed from within a prison. Reviewed by: rwatson, zec Approved by: re (kib)
2009-08-13 10:26:34 +00:00
#ifdef VIMAGE
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
.pr_flags = PR_HOST|PR_VNET|_PR_IP_SADDRSEL,
Make it possible to change the vnet sysctl variables on jails with their own virtual network stack. Jails only inheriting a network stack cannot change anything that cannot be changed from within a prison. Reviewed by: rwatson, zec Approved by: re (kib)
2009-08-13 10:26:34 +00:00
#else
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
.pr_flags = PR_HOST|_PR_IP_SADDRSEL,
Make it possible to change the vnet sysctl variables on jails with their own virtual network stack. Jails only inheriting a network stack cannot change anything that cannot be changed from within a prison. Reviewed by: rwatson, zec Approved by: re (kib)
2009-08-13 10:26:34 +00:00
#endif
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
.pr_allow = PR_ALLOW_ALL_STATIC,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
};
MTX_SYSINIT(prison0, &prison0.pr_mtx, "jail mutex", MTX_DEF);
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
struct bool_flags {
const char *name;
const char *noname;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
volatile u_int flag;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
};
struct jailsys_flags {
const char *name;
unsigned disable;
unsigned new;
};
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
/* allprison, allprison_racct and lastprid are protected by allprison_lock. */
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
struct sx allprison_lock;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
SX_SYSINIT(allprison_lock, &allprison_lock, "allprison");
struct prisonlist allprison = TAILQ_HEAD_INITIALIZER(allprison);
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
LIST_HEAD(, prison_racct) allprison_racct;
Revert rev. 178124 as requested by kris@. Having jail id not being reused too frequently is useful for script controlled environment.
2008-06-19 21:41:57 +00:00
int lastprid = 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
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
static int get_next_prid(struct prison **insprp);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
static int do_jail_attach(struct thread *td, struct prison *pr, int drflags);
Whitespace/spelling fixes in advance of upcoming functional changes. Approved by: bz (mentor)
2009-03-27 13:13:59 +00:00
static void prison_complete(void *context, int pending);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
static void prison_deref(struct prison *pr, int flags);
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
static void prison_deref_kill(struct prison *pr, struct prisonlist *freeprison);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
static int prison_lock_xlock(struct prison *pr, int flags);
static void prison_free_not_last(struct prison *pr);
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
static void prison_proc_free_not_last(struct prison *pr);
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
static void prison_set_allow_locked(struct prison *pr, unsigned flag,
int enable);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
static char *prison_path(struct prison *pr1, struct prison *pr2);
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
#ifdef RACCT
static void prison_racct_attach(struct prison *pr);
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
static void prison_racct_modify(struct prison *pr);
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
static void prison_racct_detach(struct prison *pr);
#endif
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
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Flags for prison_deref */
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
#define PD_DEREF 0x01 /* Decrement pr_ref */
#define PD_DEUREF 0x02 /* Decrement pr_uref */
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
#define PD_KILL 0x04 /* Remove jail, kill processes, etc */
#define PD_LOCKED 0x10 /* pr_mtx is held */
#define PD_LIST_SLOCKED 0x20 /* allprison_lock is held shared */
#define PD_LIST_XLOCKED 0x40 /* allprison_lock is held exclusive */
jail: Add safety around prison_deref() flags. do_jail_attach() now only uses the PD_XXX flags that refer to lock status, so make sure that something else like PD_KILL doesn't slip through. Add a KASSERT() in prison_deref() to catch any further PD_KILL misuse.
2021-02-26 04:10:42 +00:00
#define PD_OP_FLAGS 0x07 /* Operation flags */
#define PD_LOCK_FLAGS 0x70 /* Lock status 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
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
Mfp4 CH177924: Add and export constants of array sizes of jail parameters as compiled into the kernel. This is the least intrusive way to allow kvm to read the (sparse) arrays independent of the options the kernel was compiled with. Reviewed by: jhb (originally) MFC after: 1 week Sponsored by: The FreeBSD Foundation Sponsored by: CK Software GmbH
2010-12-31 22:49:13 +00:00
* Parameter names corresponding to PR_* flag values. Size values are for kvm
* as we cannot figure out the size of a sparse array, or an array without a
* terminating entry.
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
*/
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
static struct bool_flags pr_flag_bool[] = {
{"persist", "nopersist", PR_PERSIST},
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
#ifdef INET
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"ip4.saddrsel", "ip4.nosaddrsel", PR_IP4_SADDRSEL},
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
#endif
#ifdef INET6
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"ip6.saddrsel", "ip6.nosaddrsel", PR_IP6_SADDRSEL},
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
#endif
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
};
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
const size_t pr_flag_bool_size = sizeof(pr_flag_bool);
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
static struct jailsys_flags pr_flag_jailsys[] = {
{"host", 0, PR_HOST},
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
#ifdef VIMAGE
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"vnet", 0, PR_VNET},
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#ifdef INET
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"ip4", PR_IP4_USER, PR_IP4_USER},
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
#ifdef INET6
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"ip6", PR_IP6_USER, PR_IP6_USER},
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
};
Mfp4 CH177924: Add and export constants of array sizes of jail parameters as compiled into the kernel. This is the least intrusive way to allow kvm to read the (sparse) arrays independent of the options the kernel was compiled with. Reviewed by: jhb (originally) MFC after: 1 week Sponsored by: The FreeBSD Foundation Sponsored by: CK Software GmbH
2010-12-31 22:49:13 +00:00
const size_t pr_flag_jailsys_size = sizeof(pr_flag_jailsys);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
/*
* Make this array full-size so dynamic parameters can be added.
* It is protected by prison0.mtx, but lockless reading is allowed
* with an atomic check of the flag values.
*/
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
static struct bool_flags pr_flag_allow[NBBY * NBPW] = {
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"allow.set_hostname", "allow.noset_hostname", PR_ALLOW_SET_HOSTNAME},
{"allow.sysvipc", "allow.nosysvipc", PR_ALLOW_SYSVIPC},
{"allow.raw_sockets", "allow.noraw_sockets", PR_ALLOW_RAW_SOCKETS},
{"allow.chflags", "allow.nochflags", PR_ALLOW_CHFLAGS},
{"allow.mount", "allow.nomount", PR_ALLOW_MOUNT},
{"allow.quotas", "allow.noquotas", PR_ALLOW_QUOTAS},
{"allow.socket_af", "allow.nosocket_af", PR_ALLOW_SOCKET_AF},
Add allow.mlock to jail parameters It allows locking or unlocking physical pages in memory within a jail This allows running elasticsearch with "bootstrap.memory_lock" inside a jail Reviewed by: jamie@ Differential Revision: https://reviews.freebsd.org/D16342
2018-07-29 12:41:56 +00:00
{"allow.mlock", "allow.nomlock", PR_ALLOW_MLOCK},
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
{"allow.reserved_ports", "allow.noreserved_ports",
PR_ALLOW_RESERVED_PORTS},
Add a new jail permission, allow.read_msgbuf. When true, jailed processes can see the dmesg buffer (this is the current behavior). When false (the new default), dmesg will be unavailable to jailed users, whether root or not. The security.bsd.unprivileged_read_msgbuf sysctl still works as before, controlling system-wide whether non-root users can see the buffer. PR: 211580 Submitted by: bz Approved by: re@ (kib@) MFC after: 3 days
2018-10-17 16:11:43 +00:00
{"allow.read_msgbuf", "allow.noread_msgbuf", PR_ALLOW_READ_MSGBUF},
In hardened systems, where the security.bsd.unprivileged_proc_debug sysctl node is set, allow setting security.bsd.unprivileged_proc_debug per-jail. In part, this is needed to create jails in which the Address Sanitizer (ASAN) fully works as ASAN utilizes libkvm to inspect the virtual address space. Instead of having to allow unprivileged process debugging for the entire system, allow setting it on a per-jail basis. The sysctl node is still security.bsd.unprivileged_proc_debug and the jail(8) param is allow.unprivileged_proc_debug. The sysctl code is now a sysctl proc rather than a sysctl int. This allows us to determine setting the flag for the corresponding jail (or prison0). As part of the change, the dynamic allow.* API needed to be modified to take into account pr_allow flags which may now be disabled in prison0. This prevents conflicts with new pr_allow flags (like that of vmm(4)) that are added (and removed) dynamically. Also teach the jail creation KPI to allow differences for certain pr_allow flags between the parent and child jail. This can happen when unprivileged process debugging is disabled in the parent prison, but enabled in the child. Submitted by: Shawn Webb <lattera at gmail.com> Obtained from: HardenedBSD (45b3625edba0f73b3e3890b1ec3d0d1e95fd47e1, deba0b5078cef0faae43cbdafed3035b16587afc, ab21eeb3b4c72f2500987c96ff603ccf3b6e7de8) Relnotes: yes Sponsored by: HardenedBSD and G2, Inc Differential Revision: https://reviews.freebsd.org/D18319
2018-11-27 17:51:50 +00:00
{"allow.unprivileged_proc_debug", "allow.nounprivileged_proc_debug",
PR_ALLOW_UNPRIV_DEBUG},
jail: introduce per jail suser_enabled setting The suser_enable sysctl allows to remove a privileged rights from uid 0. This change introduce per jail setting which allow to make root a normal user. Reviewed by: jamie Previous version reviewed by: kevans, emaste, markj, me_igalic.co Discussed with: pjd Differential Revision: https://reviews.freebsd.org/D27128
2020-11-18 21:07:08 +00:00
{"allow.suser", "allow.nosuser", PR_ALLOW_SUSER},
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
};
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
static unsigned pr_allow_all = PR_ALLOW_ALL_STATIC;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
const size_t pr_flag_allow_size = sizeof(pr_flag_allow);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
In hardened systems, where the security.bsd.unprivileged_proc_debug sysctl node is set, allow setting security.bsd.unprivileged_proc_debug per-jail. In part, this is needed to create jails in which the Address Sanitizer (ASAN) fully works as ASAN utilizes libkvm to inspect the virtual address space. Instead of having to allow unprivileged process debugging for the entire system, allow setting it on a per-jail basis. The sysctl node is still security.bsd.unprivileged_proc_debug and the jail(8) param is allow.unprivileged_proc_debug. The sysctl code is now a sysctl proc rather than a sysctl int. This allows us to determine setting the flag for the corresponding jail (or prison0). As part of the change, the dynamic allow.* API needed to be modified to take into account pr_allow flags which may now be disabled in prison0. This prevents conflicts with new pr_allow flags (like that of vmm(4)) that are added (and removed) dynamically. Also teach the jail creation KPI to allow differences for certain pr_allow flags between the parent and child jail. This can happen when unprivileged process debugging is disabled in the parent prison, but enabled in the child. Submitted by: Shawn Webb <lattera at gmail.com> Obtained from: HardenedBSD (45b3625edba0f73b3e3890b1ec3d0d1e95fd47e1, deba0b5078cef0faae43cbdafed3035b16587afc, ab21eeb3b4c72f2500987c96ff603ccf3b6e7de8) Relnotes: yes Sponsored by: HardenedBSD and G2, Inc Differential Revision: https://reviews.freebsd.org/D18319
2018-11-27 17:51:50 +00:00
#define JAIL_DEFAULT_ALLOW (PR_ALLOW_SET_HOSTNAME | \
PR_ALLOW_RESERVED_PORTS | \
jail: introduce per jail suser_enabled setting The suser_enable sysctl allows to remove a privileged rights from uid 0. This change introduce per jail setting which allow to make root a normal user. Reviewed by: jamie Previous version reviewed by: kevans, emaste, markj, me_igalic.co Discussed with: pjd Differential Revision: https://reviews.freebsd.org/D27128
2020-11-18 21:07:08 +00:00
PR_ALLOW_UNPRIV_DEBUG | \
PR_ALLOW_SUSER)
Make the "enforce_statfs" default 2 (most restrictive) in jail_set(2), instead of whatever the parent/system has (which is generally 0). This mirrors the old-style default used for jail(2) in conjunction with the security.jail.enforce_statfs sysctl. Approved by: re (kib), bz (mentor)
2009-07-31 16:00:41 +00:00
#define JAIL_DEFAULT_ENFORCE_STATFS 2
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
#define JAIL_DEFAULT_DEVFS_RSNUM 0
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
static unsigned jail_default_allow = JAIL_DEFAULT_ALLOW;
Make the "enforce_statfs" default 2 (most restrictive) in jail_set(2), instead of whatever the parent/system has (which is generally 0). This mirrors the old-style default used for jail(2) in conjunction with the security.jail.enforce_statfs sysctl. Approved by: re (kib), bz (mentor)
2009-07-31 16:00:41 +00:00
static int jail_default_enforce_statfs = JAIL_DEFAULT_ENFORCE_STATFS;
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
static int jail_default_devfs_rsnum = JAIL_DEFAULT_DEVFS_RSNUM;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#if defined(INET) || defined(INET6)
Fix some overflow errors: a signed allocation and an insufficiant array size. Reported by: pho Tested by: pho Approved by: bz (mentor)
2009-06-09 22:09:29 +00:00
static unsigned jail_max_af_ips = 255;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
/*
* Initialize the parts of prison0 that can't be static-initialized with
* constants. This is called from proc0_init() after creating thread0 cpuset.
*/
void
prison0_init(void)
{
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
uint8_t *file, *data;
size_t size;
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
prison0.pr_cpuset = cpuset_ref(thread0.td_cpuset);
prison0.pr_osreldate = osreldate;
strlcpy(prison0.pr_osrelease, osrelease, sizeof(prison0.pr_osrelease));
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
/* If we have a preloaded hostuuid, use it. */
file = preload_search_by_type(PRISON0_HOSTUUID_MODULE);
if (file != NULL) {
data = preload_fetch_addr(file);
size = preload_fetch_size(file);
if (data != NULL) {
/*
* The preloaded data may include trailing whitespace, almost
* certainly a newline; skip over any whitespace or
* non-printable characters to be safe.
*/
while (size > 0 && data[size - 1] <= 0x20) {
Fix buffer overflow in preloaded hostuuid cleaning When a module of type "hostuuid" is provided by the loader, prison0_init strips any trailing whitespace and ASCII control characters by (a) adjusting the buffer length, and (b) zeroing out the characters in question, before storing it as the system's hostuuid. The buffer length adjustment was correct, but the zeroing overwrote one byte higher in memory than intended -- in the typical case, zeroing one byte past the end of the hostuuid buffer. Due to the layout of buffers passed by the boot loader to the kernel, this will be the first byte of a subsequent buffer. This was *probably* harmless; prison0_init runs after preloaded kernel modules have been linked and after the preloaded /boot/entropy cache has been processed, so in both cases having the first byte overwritten will not cause problems. We cannot however rule out the possibility that other objects which are preloaded by the loader could suffer from having the first byte overwritten. Since the zeroing does not in fact serve any purpose, remove it and trim trailing whitespace and ASCII control characters by adjusting the buffer length alone. Fixes: c3188289 Preload hostuuid for early-boot use Reviewed by: kevans, markj MFC after: 3 days
2021-05-14 18:07:37 +00:00
size--;
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
}
if (validate_uuid(data, size, NULL, 0) == 0) {
(void)strlcpy(prison0.pr_hostuuid, data,
size + 1);
} else if (bootverbose) {
Fix 'hostuuid: preload data malformed' warning If the preloaded hostuuid value is invalid and verbose booting is enabled, a warning is printed. This printf had two bugs: 1. It was missing a trailing \n character. 2. The malformed UUID is printed with %s even though it is not known to be NUL-terminated. This commit adds the missing \n and uses %.*s with the (already known) length of the preloaded UUID to ensure that we don't read past the end of the buffer. Reported by: kevans Fixes: c3188289 Preload hostuuid for early-boot use MFC after: 3 days
2021-05-15 05:57:38 +00:00
printf("hostuuid: preload data malformed: '%.*s'\n",
(int)size, data);
Preload hostuuid for early-boot use prison0's hostuuid will get set by the hostid rc script, either after generating it and saving it to /etc/hostid or by simply reading /etc/hostid. Some things (e.g. arbitrary MAC address generation) may use the hostuuid as a factor in early boot, so providing a way to read /etc/hostid (if it's available) and using it before userland starts up is desirable. The code is written such that the preload doesn't *have* to be /etc/hostid, thus not assuming that there will be newline at the end of the buffer or even the exact shape of the newline. White trailing whitespace/non-printables trimmed, the result will be validated as a valid uuid before it's used for early boot purposes. The preload can be turned off with hostuuid_load="NO" in /boot/loader.conf, just as other preloads; it's worth noting that this is a 37-byte file, the overhead is believed to be generally minimal. It doesn't seem necessary at this time to be concerned with kern.hostid. One does wonder if we should consider validating hostuuids coming in via jail_set(2); some bits seem to care about uuid form and we bother validating format of smbios-provided uuid and in-fact whatever uuid comes from /etc/hostid. Reviewed by: karels, delphij, jamie MFC after: 1 week (don't preload by default, probably) Differential Revision: https://reviews.freebsd.org/D24288
2020-04-16 00:54:06 +00:00
}
}
}
if (bootverbose)
printf("hostuuid: using %s\n", prison0.pr_hostuuid);
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* struct jail_args {
* struct jail *jail;
* };
*/
int
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
sys_jail(struct thread *td, struct jail_args *uap)
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
{
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
uint32_t version;
int error;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
struct jail j;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = copyin(uap->jail, &version, sizeof(uint32_t));
if (error)
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
switch (version) {
case 0:
{
struct jail_v0 j0;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* FreeBSD single IPv4 jails. */
bzero(&j, sizeof(struct jail));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = copyin(uap->jail, &j0, sizeof(struct jail_v0));
if (error)
return (error);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
j.version = j0.version;
j.path = j0.path;
j.hostname = j0.hostname;
jail_v0.ip_number was always in host byte order. This was handled in one of the many layers of indirection and shims through stable/7 in jail_handle_ips(). When it was cleaned up and unified through kern_jail() for 8.x, the byte order swap was lost. This only matters for ancient binaries that call jail(2) themselves internally.
2013-11-28 19:40:33 +00:00
j.ip4s = htonl(j0.ip_number); /* jail_v0 is host order */
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
break;
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
case 1:
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
/*
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
* Version 1 was used by multi-IPv4 jail implementations
* that never made it into the official kernel.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
*/
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
return (EINVAL);
case 2: /* JAIL_API_VERSION */
/* FreeBSD multi-IPv4/IPv6,noIP jails. */
error = copyin(uap->jail, &j, sizeof(struct jail));
if (error)
return (error);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
break;
default:
/* Sci-Fi jails are not supported, sorry. */
return (EINVAL);
}
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
return (kern_jail(td, &j));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
int
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
kern_jail(struct thread *td, struct jail *j)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
{
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
struct iovec optiov[2 * (4 + nitems(pr_flag_allow)
Fix some overflow errors: a signed allocation and an insufficiant array size. Reported by: pho Tested by: pho Approved by: bz (mentor)
2009-06-09 22:09:29 +00:00
#ifdef INET
+ 1
#endif
#ifdef INET6
+ 1
#endif
)];
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
struct uio opt;
char *u_path, *u_hostname, *u_name;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
struct bool_flags *bf;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Fix some overflow errors: a signed allocation and an insufficiant array size. Reported by: pho Tested by: pho Approved by: bz (mentor)
2009-06-09 22:09:29 +00:00
uint32_t ip4s;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
struct in_addr *u_ip4;
#endif
#ifdef INET6
struct in6_addr *u_ip6;
#endif
size_t tmplen;
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
int error, enforce_statfs;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
bzero(&optiov, sizeof(optiov));
opt.uio_iov = optiov;
opt.uio_iovcnt = 0;
opt.uio_offset = -1;
opt.uio_resid = -1;
opt.uio_segflg = UIO_SYSSPACE;
opt.uio_rw = UIO_READ;
opt.uio_td = td;
/* Set permissions for top-level jails from sysctls. */
if (!jailed(td->td_ucred)) {
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_allow;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
bf < pr_flag_allow + nitems(pr_flag_allow) &&
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
atomic_load_int(&bf->flag) != 0;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
bf++) {
optiov[opt.uio_iovcnt].iov_base = __DECONST(char *,
(jail_default_allow & bf->flag)
? bf->name : bf->noname);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
optiov[opt.uio_iovcnt].iov_len =
strlen(optiov[opt.uio_iovcnt].iov_base) + 1;
opt.uio_iovcnt += 2;
}
optiov[opt.uio_iovcnt].iov_base = "enforce_statfs";
optiov[opt.uio_iovcnt].iov_len = sizeof("enforce_statfs");
opt.uio_iovcnt++;
enforce_statfs = jail_default_enforce_statfs;
optiov[opt.uio_iovcnt].iov_base = &enforce_statfs;
optiov[opt.uio_iovcnt].iov_len = sizeof(enforce_statfs);
opt.uio_iovcnt++;
}
tmplen = MAXPATHLEN + MAXHOSTNAMELEN + MAXHOSTNAMELEN;
#ifdef INET
ip4s = (j->version == 0) ? 1 : j->ip4s;
if (ip4s > jail_max_af_ips)
return (EINVAL);
tmplen += ip4s * sizeof(struct in_addr);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#else
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (j->ip4s > 0)
return (EINVAL);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
#ifdef INET6
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (j->ip6s > jail_max_af_ips)
return (EINVAL);
tmplen += j->ip6s * sizeof(struct in6_addr);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#else
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (j->ip6s > 0)
return (EINVAL);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
u_path = malloc(tmplen, M_TEMP, M_WAITOK);
u_hostname = u_path + MAXPATHLEN;
u_name = u_hostname + MAXHOSTNAMELEN;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
u_ip4 = (struct in_addr *)(u_name + MAXHOSTNAMELEN);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
#ifdef INET6
#ifdef INET
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
u_ip6 = (struct in6_addr *)(u_ip4 + ip4s);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#else
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
u_ip6 = (struct in6_addr *)(u_name + MAXHOSTNAMELEN);
#endif
#endif
optiov[opt.uio_iovcnt].iov_base = "path";
optiov[opt.uio_iovcnt].iov_len = sizeof("path");
opt.uio_iovcnt++;
optiov[opt.uio_iovcnt].iov_base = u_path;
error = copyinstr(j->path, u_path, MAXPATHLEN,
&optiov[opt.uio_iovcnt].iov_len);
if (error) {
free(u_path, M_TEMP);
return (error);
}
opt.uio_iovcnt++;
optiov[opt.uio_iovcnt].iov_base = "host.hostname";
optiov[opt.uio_iovcnt].iov_len = sizeof("host.hostname");
opt.uio_iovcnt++;
optiov[opt.uio_iovcnt].iov_base = u_hostname;
error = copyinstr(j->hostname, u_hostname, MAXHOSTNAMELEN,
&optiov[opt.uio_iovcnt].iov_len);
if (error) {
free(u_path, M_TEMP);
return (error);
}
opt.uio_iovcnt++;
if (j->jailname != NULL) {
optiov[opt.uio_iovcnt].iov_base = "name";
optiov[opt.uio_iovcnt].iov_len = sizeof("name");
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
opt.uio_iovcnt++;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
optiov[opt.uio_iovcnt].iov_base = u_name;
error = copyinstr(j->jailname, u_name, MAXHOSTNAMELEN,
&optiov[opt.uio_iovcnt].iov_len);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error) {
free(u_path, M_TEMP);
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
opt.uio_iovcnt++;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
#ifdef INET
optiov[opt.uio_iovcnt].iov_base = "ip4.addr";
optiov[opt.uio_iovcnt].iov_len = sizeof("ip4.addr");
opt.uio_iovcnt++;
optiov[opt.uio_iovcnt].iov_base = u_ip4;
optiov[opt.uio_iovcnt].iov_len = ip4s * sizeof(struct in_addr);
if (j->version == 0)
u_ip4->s_addr = j->ip4s;
else {
error = copyin(j->ip4, u_ip4, optiov[opt.uio_iovcnt].iov_len);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error) {
free(u_path, M_TEMP);
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
opt.uio_iovcnt++;
#endif
#ifdef INET6
optiov[opt.uio_iovcnt].iov_base = "ip6.addr";
optiov[opt.uio_iovcnt].iov_len = sizeof("ip6.addr");
opt.uio_iovcnt++;
optiov[opt.uio_iovcnt].iov_base = u_ip6;
optiov[opt.uio_iovcnt].iov_len = j->ip6s * sizeof(struct in6_addr);
error = copyin(j->ip6, u_ip6, optiov[opt.uio_iovcnt].iov_len);
if (error) {
free(u_path, M_TEMP);
return (error);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
opt.uio_iovcnt++;
#endif
kernel: use our nitems() macro when it is available through param.h. No functional change, only trivial cases are done in this sweep, Discussed in: freebsd-current
2016-04-19 23:48:27 +00:00
KASSERT(opt.uio_iovcnt <= nitems(optiov),
("kern_jail: too many iovecs (%d)", opt.uio_iovcnt));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = kern_jail_set(td, &opt, JAIL_CREATE | JAIL_ATTACH);
free(u_path, M_TEMP);
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* struct jail_set_args {
* struct iovec *iovp;
* unsigned int iovcnt;
* int flags;
* };
*/
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_jail_set(struct thread *td, struct jail_set_args *uap)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
{
struct uio *auio;
int error;
/* Check that we have an even number of iovecs. */
if (uap->iovcnt & 1)
return (EINVAL);
error = copyinuio(uap->iovp, uap->iovcnt, &auio);
if (error)
return (error);
error = kern_jail_set(td, auio, uap->flags);
free(auio, M_IOV);
return (error);
}
int
kern_jail_set(struct thread *td, struct uio *optuio, int flags)
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
{
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
struct nameidata nd;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Whitespace/spelling fixes in advance of upcoming functional changes. Approved by: bz (mentor)
2009-03-27 13:13:59 +00:00
struct in_addr *ip4;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
#ifdef INET6
struct in6_addr *ip6;
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
struct vfsopt *opt;
struct vfsoptlist *opts;
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
struct prison *pr, *deadpr, *inspr, *mypr, *ppr, *tpr;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
struct vnode *root;
Allow a jail's name to be the same as its jid (which is the default if no name is specified), but still disallow other numeric names. Reviewed by: zec Approved by: bz (mentor) MFC after: 3 days
2009-09-04 19:00:48 +00:00
char *domain, *errmsg, *host, *name, *namelc, *p, *path, *uuid;
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
char *g_path, *osrelstr;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
struct bool_flags *bf;
struct jailsys_flags *jsf;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#if defined(INET) || defined(INET6)
Make the kernel compile without IP networking by moving a variable under a proper #ifdef. Approved by: re (rwatson)
2009-08-12 12:12:23 +00:00
struct prison *tppr;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
void *op;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
unsigned long hid;
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
size_t namelen, onamelen, pnamelen;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
int born, created, cuflags, descend, drflags, enforce;
Add a new jail OSD method, PR_METHOD_REMOVE. It's called when a jail is removed from the user perspective, i.e. when the last pr_uref goes away, even though the jail mail still exist in the dying state. It will also be called if either PR_METHOD_CREATE or PR_METHOD_SET fail. PR: 48471 MFC after: 5 days
2016-04-25 04:24:00 +00:00
int error, errmsg_len, errmsg_pos;
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
int gotchildmax, gotenforce, gothid, gotrsnum, gotslevel;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
int jid, jsys, len, level;
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
int childmax, osreldt, rsnum, slevel;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#if defined(INET) || defined(INET6)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
int ii, ij;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
#ifdef INET
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
int ip4s, redo_ip4;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
#ifdef INET6
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
int ip6s, redo_ip6;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
This update changes the mnt_flag field in the mount structure from 32 bits to 64 bits and eliminates the unused mnt_xflag field. The existing mnt_flag field is completely out of bits, so this update gives us room to expand. Note that the f_flags field in the statfs structure is already 64 bits, so the expanded mnt_flag field can be exported without having to make any changes in the statfs structure. Approved by: re (bz)
2011-07-24 17:43:09 +00:00
uint64_t pr_allow, ch_allow, pr_flags, ch_flags;
In hardened systems, where the security.bsd.unprivileged_proc_debug sysctl node is set, allow setting security.bsd.unprivileged_proc_debug per-jail. In part, this is needed to create jails in which the Address Sanitizer (ASAN) fully works as ASAN utilizes libkvm to inspect the virtual address space. Instead of having to allow unprivileged process debugging for the entire system, allow setting it on a per-jail basis. The sysctl node is still security.bsd.unprivileged_proc_debug and the jail(8) param is allow.unprivileged_proc_debug. The sysctl code is now a sysctl proc rather than a sysctl int. This allows us to determine setting the flag for the corresponding jail (or prison0). As part of the change, the dynamic allow.* API needed to be modified to take into account pr_allow flags which may now be disabled in prison0. This prevents conflicts with new pr_allow flags (like that of vmm(4)) that are added (and removed) dynamically. Also teach the jail creation KPI to allow differences for certain pr_allow flags between the parent and child jail. This can happen when unprivileged process debugging is disabled in the parent prison, but enabled in the child. Submitted by: Shawn Webb <lattera at gmail.com> Obtained from: HardenedBSD (45b3625edba0f73b3e3890b1ec3d0d1e95fd47e1, deba0b5078cef0faae43cbdafed3035b16587afc, ab21eeb3b4c72f2500987c96ff603ccf3b6e7de8) Relnotes: yes Sponsored by: HardenedBSD and G2, Inc Differential Revision: https://reviews.freebsd.org/D18319
2018-11-27 17:51:50 +00:00
uint64_t pr_allow_diff;
This update changes the mnt_flag field in the mount structure from 32 bits to 64 bits and eliminates the unused mnt_xflag field. The existing mnt_flag field is completely out of bits, so this update gives us room to expand. Note that the f_flags field in the statfs structure is already 64 bits, so the expanded mnt_flag field can be exported without having to make any changes in the statfs structure. Approved by: re (bz)
2011-07-24 17:43:09 +00:00
unsigned tallow;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
char numbuf[12];
error = priv_check(td, PRIV_JAIL_SET);
if (!error && (flags & JAIL_ATTACH))
error = priv_check(td, PRIV_JAIL_ATTACH);
if (error)
return (error);
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
mypr = td->td_ucred->cr_prison;
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
if ((flags & JAIL_CREATE) && mypr->pr_childmax == 0)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return (EPERM);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (flags & ~JAIL_SET_MASK)
return (EINVAL);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
/*
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
* Check all the parameters before committing to anything. Not all
* errors can be caught early, but we may as well try. Also, this
* takes care of some expensive stuff (path lookup) before getting
* the allprison lock.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
*
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
* XXX Jails are not filesystems, and jail parameters are not mount
* options. But it makes more sense to re-use the vfsopt code
* than duplicate it under a different name.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
*/
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_buildopts(optuio, &opts);
if (error)
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
ip4 = NULL;
#endif
#ifdef INET6
ip6 = NULL;
#endif
Use separate buffer for global path to avoid overflow of path buffer. Reviewed by: jamie@ MFC after: 3 weeks
2012-01-21 00:06:21 +00:00
g_path = NULL;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
cuflags = flags & (JAIL_CREATE | JAIL_UPDATE);
if (!cuflags) {
error = EINVAL;
vfs_opterror(opts, "no valid operation (create or update)");
goto done_errmsg;
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_copyopt(opts, "jid", &jid, sizeof(jid));
if (error == ENOENT)
jid = 0;
else if (error != 0)
goto done_free;
error = vfs_copyopt(opts, "securelevel", &slevel, sizeof(slevel));
if (error == ENOENT)
gotslevel = 0;
else if (error != 0)
goto done_free;
else
gotslevel = 1;
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
error =
vfs_copyopt(opts, "children.max", &childmax, sizeof(childmax));
if (error == ENOENT)
gotchildmax = 0;
else if (error != 0)
goto done_free;
else
gotchildmax = 1;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = vfs_copyopt(opts, "enforce_statfs", &enforce, sizeof(enforce));
Don't exit kern_jail_set without freeing options when enforce_statfs has an illegal value. MFC after: 3 days
2010-09-10 21:45:42 +00:00
if (error == ENOENT)
gotenforce = 0;
else if (error != 0)
goto done_free;
else if (enforce < 0 || enforce > 2) {
error = EINVAL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
goto done_free;
Don't exit kern_jail_set without freeing options when enforce_statfs has an illegal value. MFC after: 3 days
2010-09-10 21:45:42 +00:00
} else
gotenforce = 1;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
error = vfs_copyopt(opts, "devfs_ruleset", &rsnum, sizeof(rsnum));
if (error == ENOENT)
gotrsnum = 0;
else if (error != 0)
goto done_free;
else
gotrsnum = 1;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
pr_flags = ch_flags = 0;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_bool;
bf < pr_flag_bool + nitems(pr_flag_bool);
bf++) {
vfs_flagopt(opts, bf->name, &pr_flags, bf->flag);
vfs_flagopt(opts, bf->noname, &ch_flags, bf->flag);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
ch_flags |= pr_flags;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (jsf = pr_flag_jailsys;
jsf < pr_flag_jailsys + nitems(pr_flag_jailsys);
jsf++) {
error = vfs_copyopt(opts, jsf->name, &jsys, sizeof(jsys));
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
if (error == ENOENT)
continue;
if (error != 0)
goto done_free;
switch (jsys) {
case JAIL_SYS_DISABLE:
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
if (!jsf->disable) {
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
error = EINVAL;
goto done_free;
}
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
pr_flags |= jsf->disable;
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
break;
case JAIL_SYS_NEW:
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
pr_flags |= jsf->new;
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
break;
case JAIL_SYS_INHERIT:
break;
default:
error = EINVAL;
goto done_free;
}
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
ch_flags |= jsf->new | jsf->disable;
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
}
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
if ((flags & (JAIL_CREATE | JAIL_ATTACH)) == JAIL_CREATE
Back out r210974. Any convenience of not typing "persist" is outweighed by the possibility of unintended partially-formed jails.
2010-08-08 23:22:55 +00:00
&& !(pr_flags & PR_PERSIST)) {
error = EINVAL;
vfs_opterror(opts, "new jail must persist or attach");
goto done_errmsg;
}
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#ifdef VIMAGE
if ((flags & JAIL_UPDATE) && (ch_flags & PR_VNET)) {
error = EINVAL;
vfs_opterror(opts, "vnet cannot be changed after creation");
goto done_errmsg;
}
#endif
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
#ifdef INET
if ((flags & JAIL_UPDATE) && (ch_flags & PR_IP4_USER)) {
error = EINVAL;
vfs_opterror(opts, "ip4 cannot be changed after creation");
goto done_errmsg;
}
#endif
#ifdef INET6
if ((flags & JAIL_UPDATE) && (ch_flags & PR_IP6_USER)) {
error = EINVAL;
vfs_opterror(opts, "ip6 cannot be changed after creation");
goto done_errmsg;
}
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr_allow = ch_allow = 0;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_allow;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
bf < pr_flag_allow + nitems(pr_flag_allow) &&
atomic_load_int(&bf->flag) != 0;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
bf++) {
vfs_flagopt(opts, bf->name, &pr_allow, bf->flag);
vfs_flagopt(opts, bf->noname, &ch_allow, bf->flag);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
ch_allow |= pr_allow;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_getopt(opts, "name", (void **)&name, &len);
if (error == ENOENT)
name = NULL;
else if (error != 0)
goto done_free;
else {
if (len == 0 || name[len - 1] != '\0') {
error = EINVAL;
goto done_free;
}
if (len > MAXHOSTNAMELEN) {
error = ENAMETOOLONG;
goto done_free;
}
}
error = vfs_getopt(opts, "host.hostname", (void **)&host, &len);
if (error == ENOENT)
host = NULL;
else if (error != 0)
goto done_free;
else {
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
ch_flags |= PR_HOST;
pr_flags |= PR_HOST;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (len == 0 || host[len - 1] != '\0') {
error = EINVAL;
goto done_free;
}
if (len > MAXHOSTNAMELEN) {
error = ENAMETOOLONG;
goto done_free;
}
}
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
error = vfs_getopt(opts, "host.domainname", (void **)&domain, &len);
if (error == ENOENT)
domain = NULL;
else if (error != 0)
goto done_free;
else {
ch_flags |= PR_HOST;
pr_flags |= PR_HOST;
if (len == 0 || domain[len - 1] != '\0') {
error = EINVAL;
goto done_free;
}
if (len > MAXHOSTNAMELEN) {
error = ENAMETOOLONG;
goto done_free;
}
}
error = vfs_getopt(opts, "host.hostuuid", (void **)&uuid, &len);
if (error == ENOENT)
uuid = NULL;
else if (error != 0)
goto done_free;
else {
ch_flags |= PR_HOST;
pr_flags |= PR_HOST;
if (len == 0 || uuid[len - 1] != '\0') {
error = EINVAL;
goto done_free;
}
if (len > HOSTUUIDLEN) {
error = ENAMETOOLONG;
goto done_free;
}
}
Provide groundwork for 32-bit binary compatibility on non-x86 platforms, for upcoming 64-bit PowerPC and MIPS support. This renames the COMPAT_IA32 option to COMPAT_FREEBSD32, removes some IA32-specific code from MI parts of the kernel and enhances the freebsd32 compatibility code to support big-endian platforms. Reviewed by: kib, jhb
2010-03-11 14:49:06 +00:00
#ifdef COMPAT_FREEBSD32
Add macro to test the sv_flags of any process. Change some places to test the flags instead of explicit comparing with address of known sysentvec structures. MFC after: 1 month
2011-01-26 20:03:58 +00:00
if (SV_PROC_FLAG(td->td_proc, SV_ILP32)) {
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
uint32_t hid32;
error = vfs_copyopt(opts, "host.hostid", &hid32, sizeof(hid32));
hid = hid32;
} else
#endif
error = vfs_copyopt(opts, "host.hostid", &hid, sizeof(hid));
if (error == ENOENT)
gothid = 0;
else if (error != 0)
goto done_free;
else {
gothid = 1;
ch_flags |= PR_HOST;
pr_flags |= PR_HOST;
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET
error = vfs_getopt(opts, "ip4.addr", &op, &ip4s);
if (error == ENOENT)
Don't set prison's pr_ip4s or pr_ip6s to -1. PR: 196474 MFC after: 3 days
2015-01-14 03:52:41 +00:00
ip4s = 0;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
else if (error != 0)
goto done_free;
else if (ip4s & (sizeof(*ip4) - 1)) {
error = EINVAL;
goto done_free;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
} else {
Remove the prison flags PR_IP4_DISABLE and PR_IP6_DISABLE, which have been write-only for as long as they've existed.
2015-01-14 04:50:28 +00:00
ch_flags |= PR_IP4_USER;
pr_flags |= PR_IP4_USER;
if (ip4s > 0) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
ip4s /= sizeof(*ip4);
if (ip4s > jail_max_af_ips) {
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
error = EINVAL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
vfs_opterror(opts, "too many IPv4 addresses");
goto done_errmsg;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
ip4 = malloc(ip4s * sizeof(*ip4), M_PRISON, M_WAITOK);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
bcopy(op, ip4, ip4s * sizeof(*ip4));
/*
* IP addresses are all sorted but ip[0] to preserve
* the primary IP address as given from userland.
* This special IP is used for unbound outgoing
Adjust a comment to reflect reality, as we have proper source address selection, even for IPv4, since r183571. Pointed out by: Jase Thew (bazerka beardz.net) MFC after: 3 days
2010-01-11 21:21:30 +00:00
* connections as well for "loopback" traffic in case
* source address selection cannot find any more fitting
* address to connect from.
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
*/
if (ip4s > 1)
Move IPv4-specific jail functions to new file netinet/in_jail.c _prison_check_ip4 renamed to prison_check_ip4_locked Move IPv6-specific jail functions to new file netinet6/in6_jail.c _prison_check_ip6 renamed to prison_check_ip6_locked Add appropriate prototypes to sys/sys/jail.h Adjust kern_jail.c to call prison_check_ip4_locked and prison_check_ip6_locked accordingly. Add netinet/in_jail.c and netinet6/in6_jail.c to the list of files that need to be built when INET and INET6, respectively, are configured in the kernel configuration file. Reviewed by: jtl Approved by: sjg (mentor) Sponsored by: Juniper Networks, Inc. Differential Revision: https://reviews.freebsd.org/D6799
2016-08-09 02:16:21 +00:00
qsort(ip4 + 1, ip4s - 1, sizeof(*ip4),
prison_qcmp_v4);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
* Check for duplicate addresses and do some simple
* zero and broadcast checks. If users give other bogus
* addresses it is their problem.
*
* We do not have to care about byte order for these
* checks so we will do them in NBO.
*/
for (ii = 0; ii < ip4s; ii++) {
if (ip4[ii].s_addr == INADDR_ANY ||
ip4[ii].s_addr == INADDR_BROADCAST) {
error = EINVAL;
goto done_free;
}
if ((ii+1) < ip4s &&
(ip4[0].s_addr == ip4[ii+1].s_addr ||
ip4[ii].s_addr == ip4[ii+1].s_addr)) {
error = EINVAL;
goto done_free;
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET6
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_getopt(opts, "ip6.addr", &op, &ip6s);
if (error == ENOENT)
Don't set prison's pr_ip4s or pr_ip6s to -1. PR: 196474 MFC after: 3 days
2015-01-14 03:52:41 +00:00
ip6s = 0;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
else if (error != 0)
goto done_free;
else if (ip6s & (sizeof(*ip6) - 1)) {
error = EINVAL;
goto done_free;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
} else {
Remove the prison flags PR_IP4_DISABLE and PR_IP6_DISABLE, which have been write-only for as long as they've existed.
2015-01-14 04:50:28 +00:00
ch_flags |= PR_IP6_USER;
pr_flags |= PR_IP6_USER;
if (ip6s > 0) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
ip6s /= sizeof(*ip6);
if (ip6s > jail_max_af_ips) {
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
error = EINVAL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
vfs_opterror(opts, "too many IPv6 addresses");
goto done_errmsg;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
ip6 = malloc(ip6s * sizeof(*ip6), M_PRISON, M_WAITOK);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
bcopy(op, ip6, ip6s * sizeof(*ip6));
if (ip6s > 1)
Move IPv4-specific jail functions to new file netinet/in_jail.c _prison_check_ip4 renamed to prison_check_ip4_locked Move IPv6-specific jail functions to new file netinet6/in6_jail.c _prison_check_ip6 renamed to prison_check_ip6_locked Add appropriate prototypes to sys/sys/jail.h Adjust kern_jail.c to call prison_check_ip4_locked and prison_check_ip6_locked accordingly. Add netinet/in_jail.c and netinet6/in6_jail.c to the list of files that need to be built when INET and INET6, respectively, are configured in the kernel configuration file. Reviewed by: jtl Approved by: sjg (mentor) Sponsored by: Juniper Networks, Inc. Differential Revision: https://reviews.freebsd.org/D6799
2016-08-09 02:16:21 +00:00
qsort(ip6 + 1, ip6s - 1, sizeof(*ip6),
prison_qcmp_v6);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
for (ii = 0; ii < ip6s; ii++) {
if (IN6_IS_ADDR_UNSPECIFIED(&ip6[ii])) {
error = EINVAL;
goto done_free;
}
if ((ii+1) < ip6s &&
(IN6_ARE_ADDR_EQUAL(&ip6[0], &ip6[ii+1]) ||
IN6_ARE_ADDR_EQUAL(&ip6[ii], &ip6[ii+1])))
{
error = EINVAL;
goto done_free;
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#if defined(VIMAGE) && (defined(INET) || defined(INET6))
if ((ch_flags & PR_VNET) && (ch_flags & (PR_IP4_USER | PR_IP6_USER))) {
error = EINVAL;
vfs_opterror(opts,
"vnet jails cannot have IP address restrictions");
goto done_errmsg;
}
#endif
Re-order some jail parameter reading to prevent a vnode leak.
2016-06-09 20:43:14 +00:00
error = vfs_getopt(opts, "osrelease", (void **)&osrelstr, &len);
if (error == ENOENT)
osrelstr = NULL;
else if (error != 0)
goto done_free;
else {
if (flags & JAIL_UPDATE) {
error = EINVAL;
vfs_opterror(opts,
"osrelease cannot be changed after creation");
goto done_errmsg;
}
kern_jail: missing \0 termination check on osrelease parameter If a user spplies a non-\0 terminated osrelease parameter reading it back may disclose kernel memory. This is a problem in case of nested jails (children.max > 0, which is not the default). Otherwise root outside the jail has access to kernel memory by other means and root inside a jail cannot create a child jail. Add the proper \0 check at the end of a supplied osrelease parameter and make sure any copies of the field will be \0-terminated. Submitted by: Hans Christian Woithe (chwoithe yahoo.com) MFC after: 3 days
2020-03-14 14:04:55 +00:00
if (len == 0 || osrelstr[len - 1] != '\0') {
Re-order some jail parameter reading to prevent a vnode leak.
2016-06-09 20:43:14 +00:00
error = EINVAL;
kern_jail: missing \0 termination check on osrelease parameter If a user spplies a non-\0 terminated osrelease parameter reading it back may disclose kernel memory. This is a problem in case of nested jails (children.max > 0, which is not the default). Otherwise root outside the jail has access to kernel memory by other means and root inside a jail cannot create a child jail. Add the proper \0 check at the end of a supplied osrelease parameter and make sure any copies of the field will be \0-terminated. Submitted by: Hans Christian Woithe (chwoithe yahoo.com) MFC after: 3 days
2020-03-14 14:04:55 +00:00
goto done_free;
}
if (len >= OSRELEASELEN) {
error = ENAMETOOLONG;
Re-order some jail parameter reading to prevent a vnode leak.
2016-06-09 20:43:14 +00:00
vfs_opterror(opts,
"osrelease string must be 1-%d bytes long",
OSRELEASELEN - 1);
goto done_errmsg;
}
}
error = vfs_copyopt(opts, "osreldate", &osreldt, sizeof(osreldt));
if (error == ENOENT)
osreldt = 0;
else if (error != 0)
goto done_free;
else {
if (flags & JAIL_UPDATE) {
error = EINVAL;
vfs_opterror(opts,
"osreldate cannot be changed after creation");
goto done_errmsg;
}
if (osreldt == 0) {
error = EINVAL;
vfs_opterror(opts, "osreldate cannot be 0");
goto done_errmsg;
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
root = NULL;
error = vfs_getopt(opts, "path", (void **)&path, &len);
if (error == ENOENT)
path = NULL;
else if (error != 0)
goto done_free;
else {
if (flags & JAIL_UPDATE) {
error = EINVAL;
vfs_opterror(opts,
"path cannot be changed after creation");
goto done_errmsg;
}
if (len == 0 || path[len - 1] != '\0') {
error = EINVAL;
goto done_free;
}
Remove the support for using non-mpsafe filesystem modules. In particular, do not lock Giant conditionally when calling into the filesystem module, remove the VFS_LOCK_GIANT() and related macros. Stop handling buffers belonging to non-mpsafe filesystems. The VFS_VERSION is bumped to indicate the interface change which does not result in the interface signatures changes. Conducted and reviewed by: attilio Tested by: pho
2012-10-22 17:50:54 +00:00
NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE,
Introduce vn_path_to_global_path() This function updates path string to vnode's full global path and checks the size of the new path string against the pathlen argument. In vfs_domount(), sys_unmount() and kern_jail_set() this new function is used to update the supplied path argument to the respective global path. Unbreaks jailed zfs(8) with enforce_statfs set to 1. Reviewed by: kib MFC after: 1 month
2012-01-15 12:08:20 +00:00
path, td);
error = namei(&nd);
if (error)
goto done_free;
root = nd.ni_vp;
NDFREE(&nd, NDF_ONLY_PNBUF);
Use separate buffer for global path to avoid overflow of path buffer. Reviewed by: jamie@ MFC after: 3 weeks
2012-01-21 00:06:21 +00:00
g_path = malloc(MAXPATHLEN, M_TEMP, M_WAITOK);
strlcpy(g_path, path, MAXPATHLEN);
error = vn_path_to_global_path(td, root, g_path, MAXPATHLEN);
vfs: remove now useless ENODEV handling from vn_fullpath consumers Noted by: ngie
2020-02-08 15:51:08 +00:00
if (error == 0) {
Use separate buffer for global path to avoid overflow of path buffer. Reviewed by: jamie@ MFC after: 3 weeks
2012-01-21 00:06:21 +00:00
path = g_path;
} else {
Introduce vn_path_to_global_path() This function updates path string to vnode's full global path and checks the size of the new path string against the pathlen argument. In vfs_domount(), sys_unmount() and kern_jail_set() this new function is used to update the supplied path argument to the respective global path. Unbreaks jailed zfs(8) with enforce_statfs set to 1. Reviewed by: kib MFC after: 1 month
2012-01-15 12:08:20 +00:00
/* exit on other errors */
goto done_free;
}
if (root->v_type != VDIR) {
error = ENOTDIR;
vput(root);
goto done_free;
}
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(root);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
* Find the specified jail, or at least its parent.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
* This abuses the file error codes ENOENT and EEXIST.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
*/
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
pr = NULL;
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
inspr = NULL;
Allow a jail's name to be the same as its jid (which is the default if no name is specified), but still disallow other numeric names. Reviewed by: zec Approved by: bz (mentor) MFC after: 3 days
2009-09-04 19:00:48 +00:00
if (cuflags == JAIL_CREATE && jid == 0 && name != NULL) {
namelc = strrchr(name, '.');
jid = strtoul(namelc != NULL ? namelc + 1 : name, &p, 10);
if (*p != '\0')
jid = 0;
}
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
sx_xlock(&allprison_lock);
jail: Fix locking on an early jail_set error. I had locked allprison_lock without immediately setting PD_LIST_LOCKED.
2021-02-26 03:52:58 +00:00
drflags = PD_LIST_XLOCKED;
jail: Don't allow jails under dying parents If a jail is created with jail_set(...JAIL_DYING), and it has a parent currently in a dying state, that will bring the parent jail back to life. Restrict that to require that the parent itself be explicitly brought back first, and not implicitly created along with the new child jail. Differential Revision: https://reviews.freebsd.org/D28515
2021-02-23 01:04:06 +00:00
ppr = mypr;
if (!prison_isalive(ppr)) {
/* This jail is dying. This process will surely follow. */
error = EAGAIN;
goto done_deref;
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (jid != 0) {
if (jid < 0) {
error = EINVAL;
vfs_opterror(opts, "negative jid");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
/*
* See if a requested jid already exists. Keep track of
* where it can be inserted later.
*/
TAILQ_FOREACH(inspr, &allprison, pr_list) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (inspr->pr_id < jid)
continue;
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
if (inspr->pr_id > jid)
break;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
pr = inspr;
mtx_lock(&pr->pr_mtx);
drflags |= PD_LOCKED;
inspr = NULL;
break;
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (pr != NULL) {
/* Create: jid must not exist. */
if (cuflags == JAIL_CREATE) {
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
/*
* Even creators that cannot see the jail will
* get EEXIST.
*/
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = EEXIST;
vfs_opterror(opts, "jail %d already exists",
jid);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (!prison_ischild(mypr, pr)) {
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
/*
* Updaters get ENOENT if they cannot see the
* jail. This is true even for CREATE | UPDATE,
* which normally cannot give this error.
*/
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
error = ENOENT;
vfs_opterror(opts, "jail %d not found", jid);
goto done_deref;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
}
jail: Don't allow jails under dying parents If a jail is created with jail_set(...JAIL_DYING), and it has a parent currently in a dying state, that will bring the parent jail back to life. Restrict that to require that the parent itself be explicitly brought back first, and not implicitly created along with the new child jail. Differential Revision: https://reviews.freebsd.org/D28515
2021-02-23 01:04:06 +00:00
ppr = pr->pr_parent;
if (!prison_isalive(ppr)) {
error = ENOENT;
vfs_opterror(opts, "jail %d is dying",
ppr->pr_id);
goto done_deref;
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (!prison_isalive(pr)) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (!(flags & JAIL_DYING)) {
error = ENOENT;
vfs_opterror(opts, "jail %d is dying",
jid);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
}
if ((flags & JAIL_ATTACH) ||
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
(pr_flags & PR_PERSIST)) {
/*
* A dying jail might be resurrected
* (via attach or persist), but first
* it must determine if another jail
* has claimed its name. Accomplish
* this by implicitly re-setting the
* name.
*/
if (name == NULL)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
name = prison_name(mypr, pr);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
} else {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Update: jid must exist. */
if (cuflags == JAIL_UPDATE) {
error = ENOENT;
vfs_opterror(opts, "jail %d not found", jid);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
}
/*
* If the caller provided a name, look for a jail by that name.
* This has different semantics for creates and updates keyed by jid
* (where the name must not already exist in a different jail),
* and updates keyed by the name itself (where the name must exist
* because that is the jail being updated).
*/
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
namelc = NULL;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (name != NULL) {
Allow a jail's name to be the same as its jid (which is the default if no name is specified), but still disallow other numeric names. Reviewed by: zec Approved by: bz (mentor) MFC after: 3 days
2009-09-04 19:00:48 +00:00
namelc = strrchr(name, '.');
if (namelc == NULL)
namelc = name;
else {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
* This is a hierarchical name. Split it into the
* parent and child names, and make sure the parent
* exists or matches an already found jail.
*/
if (pr != NULL) {
Allow a jail's name to be the same as its jid (which is the default if no name is specified), but still disallow other numeric names. Reviewed by: zec Approved by: bz (mentor) MFC after: 3 days
2009-09-04 19:00:48 +00:00
if (strncmp(name, ppr->pr_name, namelc - name)
|| ppr->pr_name[namelc - name] != '\0') {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = EINVAL;
vfs_opterror(opts,
"cannot change jail's parent");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
} else {
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
*namelc = '\0';
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
ppr = prison_find_name(mypr, name);
if (ppr == NULL) {
error = ENOENT;
vfs_opterror(opts,
"jail \"%s\" not found", name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
jail: Don't allow jails under dying parents If a jail is created with jail_set(...JAIL_DYING), and it has a parent currently in a dying state, that will bring the parent jail back to life. Restrict that to require that the parent itself be explicitly brought back first, and not implicitly created along with the new child jail. Differential Revision: https://reviews.freebsd.org/D28515
2021-02-23 01:04:06 +00:00
mtx_unlock(&ppr->pr_mtx);
if (!prison_isalive(ppr)) {
jail: Handle a parent jail when a child is added to it It's possible when adding a jail that its dying parent comes back to life. Only allow that to happen when JAIL_DYING is specified. And if it does happen, call PR_METHOD_CREATE on it.
2021-01-29 05:51:09 +00:00
error = ENOENT;
vfs_opterror(opts,
"jail \"%s\" is dying", name);
goto done_deref;
}
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
*namelc = '.';
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
namelc++;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
if (namelc[0] != '\0') {
pnamelen =
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
(ppr == &prison0) ? 0 : strlen(ppr->pr_name) + 1;
deadpr = NULL;
FOREACH_PRISON_CHILD(ppr, tpr) {
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
if (tpr != pr &&
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
!strcmp(tpr->pr_name + pnamelen, namelc)) {
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
if (prison_isalive(tpr)) {
if (pr == NULL &&
cuflags != JAIL_CREATE) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* Use this jail
* for updates.
*/
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
pr = tpr;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
mtx_lock(&pr->pr_mtx);
jail: Fix a stray mutex from 76ad42abf9d4.
2021-01-18 23:47:09 +00:00
drflags |= PD_LOCKED;
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
break;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
/*
* Create, or update(jid):
* name must not exist in an
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
* active sibling jail.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
*/
error = EEXIST;
vfs_opterror(opts,
"jail \"%s\" already exists",
name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
if (pr == NULL &&
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
cuflags != JAIL_CREATE) {
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
deadpr = tpr;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
/* If no active jail is found, use a dying one. */
if (deadpr != NULL && pr == NULL) {
if (flags & JAIL_DYING) {
pr = deadpr;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
mtx_lock(&pr->pr_mtx);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_LOCKED;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
} else if (cuflags == JAIL_UPDATE) {
error = ENOENT;
vfs_opterror(opts,
"jail \"%s\" is dying", name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
/* Update: name must exist if no jid. */
else if (cuflags == JAIL_UPDATE && pr == NULL) {
error = ENOENT;
vfs_opterror(opts, "jail \"%s\" not found",
name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
}
/* Update: must provide a jid or name. */
else if (cuflags == JAIL_UPDATE && pr == NULL) {
error = ENOENT;
vfs_opterror(opts, "update specified no jail");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* If there's no prison to update, create a new one and link it in. */
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
created = pr == NULL;
if (created) {
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
for (tpr = mypr; tpr != NULL; tpr = tpr->pr_parent)
if (tpr->pr_childcount >= tpr->pr_childmax) {
error = EPERM;
vfs_opterror(opts, "prison limit exceeded");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
}
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
if (jid == 0 && (jid = get_next_prid(&inspr)) == 0) {
error = EAGAIN;
vfs_opterror(opts, "no available jail IDs");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
pr = malloc(sizeof(*pr), M_PRISON, M_WAITOK | M_ZERO);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
pr->pr_state = PRISON_STATE_INVALID;
refcount_init(&pr->pr_ref, 1);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
refcount_init(&pr->pr_uref, 0);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
drflags |= PD_DEREF;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
LIST_INIT(&pr->pr_children);
mtx_init(&pr->pr_mtx, "jail mutex", NULL, MTX_DEF | MTX_DUPOK);
TASK_INIT(&pr->pr_task, 0, prison_complete, pr);
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
pr->pr_id = jid;
if (inspr != NULL)
TAILQ_INSERT_BEFORE(inspr, pr, pr_list);
else
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
TAILQ_INSERT_TAIL(&allprison, pr, pr_list);
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
pr->pr_parent = ppr;
jail: Don't allow jails under dying parents If a jail is created with jail_set(...JAIL_DYING), and it has a parent currently in a dying state, that will bring the parent jail back to life. Restrict that to require that the parent itself be explicitly brought back first, and not implicitly created along with the new child jail. Differential Revision: https://reviews.freebsd.org/D28515
2021-02-23 01:04:06 +00:00
prison_hold(ppr);
prison_proc_hold(ppr);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
LIST_INSERT_HEAD(&ppr->pr_children, pr, pr_sibling);
for (tpr = ppr; tpr != NULL; tpr = tpr->pr_parent)
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
tpr->pr_childcount++;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* Set some default values, and inherit some from the parent. */
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
if (namelc == NULL)
namelc = "";
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (path == NULL) {
path = "/";
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
root = mypr->pr_root;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
vref(root);
}
Change the default value of the "ip4" and "ip6" jail parameters to "disable", which only allows access to the parent/physical system's IP addresses when specifically directed. Change the default value of "host" to "new", and don't copy the parent host values, to insulate jails from the parent hostname et al. Approved by: re (kib), bz (mentor)
2009-07-29 16:41:02 +00:00
strlcpy(pr->pr_hostuuid, DEFAULT_HOSTUUID, HOSTUUIDLEN);
pr->pr_flags |= PR_HOST;
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#if defined(INET) || defined(INET6)
#ifdef VIMAGE
if (!(pr_flags & PR_VNET))
#endif
{
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#ifdef INET
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (!(ch_flags & PR_IP4_USER))
Remove the prison flags PR_IP4_DISABLE and PR_IP6_DISABLE, which have been write-only for as long as they've existed.
2015-01-14 04:50:28 +00:00
pr->pr_flags |= PR_IP4 | PR_IP4_USER;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
else if (!(pr_flags & PR_IP4_USER)) {
pr->pr_flags |= ppr->pr_flags & PR_IP4;
if (ppr->pr_ip4 != NULL) {
pr->pr_ip4s = ppr->pr_ip4s;
pr->pr_ip4 = malloc(pr->pr_ip4s *
sizeof(struct in_addr), M_PRISON,
M_WAITOK);
bcopy(ppr->pr_ip4, pr->pr_ip4,
pr->pr_ip4s * sizeof(*pr->pr_ip4));
}
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
#ifdef INET6
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (!(ch_flags & PR_IP6_USER))
Remove the prison flags PR_IP4_DISABLE and PR_IP6_DISABLE, which have been write-only for as long as they've existed.
2015-01-14 04:50:28 +00:00
pr->pr_flags |= PR_IP6 | PR_IP6_USER;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
else if (!(pr_flags & PR_IP6_USER)) {
pr->pr_flags |= ppr->pr_flags & PR_IP6;
if (ppr->pr_ip6 != NULL) {
pr->pr_ip6s = ppr->pr_ip6s;
pr->pr_ip6 = malloc(pr->pr_ip6s *
sizeof(struct in6_addr), M_PRISON,
M_WAITOK);
bcopy(ppr->pr_ip6, pr->pr_ip6,
pr->pr_ip6s * sizeof(*pr->pr_ip6));
}
}
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#endif
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
/* Source address selection is always on by default. */
pr->pr_flags |= _PR_IP_SADDRSEL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr->pr_securelevel = ppr->pr_securelevel;
pr->pr_allow = JAIL_DEFAULT_ALLOW & ppr->pr_allow;
Initialise pr_enforce_statfs from the "default" sysctl value and not from the compile time constant. The sysctl value is seeded from the compile time constant. MFC after: 2 weeks
2015-06-17 13:15:54 +00:00
pr->pr_enforce_statfs = jail_default_enforce_statfs;
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
pr->pr_devfs_rsnum = ppr->pr_devfs_rsnum;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
pr->pr_osreldate = osreldt ? osreldt : ppr->pr_osreldate;
if (osrelstr == NULL)
kern_jail: missing \0 termination check on osrelease parameter If a user spplies a non-\0 terminated osrelease parameter reading it back may disclose kernel memory. This is a problem in case of nested jails (children.max > 0, which is not the default). Otherwise root outside the jail has access to kernel memory by other means and root inside a jail cannot create a child jail. Add the proper \0 check at the end of a supplied osrelease parameter and make sure any copies of the field will be \0-terminated. Submitted by: Hans Christian Woithe (chwoithe yahoo.com) MFC after: 3 days
2020-03-14 14:04:55 +00:00
strlcpy(pr->pr_osrelease, ppr->pr_osrelease,
sizeof(pr->pr_osrelease));
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
else
kern_jail: missing \0 termination check on osrelease parameter If a user spplies a non-\0 terminated osrelease parameter reading it back may disclose kernel memory. This is a problem in case of nested jails (children.max > 0, which is not the default). Otherwise root outside the jail has access to kernel memory by other means and root inside a jail cannot create a child jail. Add the proper \0 check at the end of a supplied osrelease parameter and make sure any copies of the field will be \0-terminated. Submitted by: Hans Christian Woithe (chwoithe yahoo.com) MFC after: 3 days
2020-03-14 14:04:55 +00:00
strlcpy(pr->pr_osrelease, osrelstr,
sizeof(pr->pr_osrelease));
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#ifdef VIMAGE
/* Allocate a new vnet if specified. */
pr->pr_vnet = (pr_flags & PR_VNET)
? vnet_alloc() : ppr->pr_vnet;
#endif
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
/*
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
* Allocate a dedicated cpuset for each jail.
* Unlike other initial settings, this may return an erorr.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
*/
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = cpuset_create_root(ppr, &pr->pr_cpuset);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
if (error)
goto done_deref;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
mtx_lock(&pr->pr_mtx);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_LOCKED;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
} else {
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
/*
* Grab a reference for existing prisons, to ensure they
* continue to exist for the duration of the call.
*/
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
prison_hold(pr);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_DEREF;
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#if defined(VIMAGE) && (defined(INET) || defined(INET6))
if ((pr->pr_flags & PR_VNET) &&
(ch_flags & (PR_IP4_USER | PR_IP6_USER))) {
error = EINVAL;
vfs_opterror(opts,
"vnet jails cannot have IP address restrictions");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
}
#endif
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
#ifdef INET
if (PR_IP4_USER & ch_flags & (pr_flags ^ pr->pr_flags)) {
error = EINVAL;
vfs_opterror(opts,
"ip4 cannot be changed after creation");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
}
#endif
#ifdef INET6
if (PR_IP6_USER & ch_flags & (pr_flags ^ pr->pr_flags)) {
error = EINVAL;
vfs_opterror(opts,
"ip6 cannot be changed after creation");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
}
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Do final error checking before setting anything. */
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (gotslevel) {
if (slevel < ppr->pr_securelevel) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
if (gotchildmax) {
if (childmax >= ppr->pr_childmax) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
}
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (gotenforce) {
if (enforce < ppr->pr_enforce_statfs) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
if (gotrsnum) {
/*
* devfs_rsnum is a uint16_t
*/
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
if (rsnum < 0 || rsnum > 65535) {
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
error = EINVAL;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
}
/*
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
* Nested jails always inherit parent's devfs ruleset
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
*/
if (jailed(td->td_ucred)) {
if (rsnum > 0 && rsnum != ppr->pr_devfs_rsnum) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
} else
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
rsnum = ppr->pr_devfs_rsnum;
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (ip4s > 0) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (ppr->pr_flags & PR_IP4) {
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
/*
* Make sure the new set of IP addresses is a
* subset of the parent's list. Don't worry
* about the parent being unlocked, as any
* setting is done with allprison_lock held.
*/
for (ij = 0; ij < ppr->pr_ip4s; ij++)
if (ip4[0].s_addr == ppr->pr_ip4[ij].s_addr)
break;
if (ij == ppr->pr_ip4s) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
}
if (ip4s > 1) {
for (ii = ij = 1; ii < ip4s; ii++) {
if (ip4[ii].s_addr ==
ppr->pr_ip4[0].s_addr)
continue;
for (; ij < ppr->pr_ip4s; ij++)
if (ip4[ii].s_addr ==
ppr->pr_ip4[ij].s_addr)
break;
if (ij == ppr->pr_ip4s)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
break;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (ij == ppr->pr_ip4s) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
}
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
/*
* Check for conflicting IP addresses. We permit them
* if there is no more than one IP on each jail. If
* there is a duplicate on a jail with more than one
* IP stop checking and return error.
*/
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
Fix the test prohibiting jails from sharing IP addresses. It's not supposed to be legal for two jails to contain the same IP address, unless both jails contain only that one address. This is the behavior documented in jail(8), and is there to prevent confusion when multiple jails are listening on IADDR_ANY. VIMAGE jails (now the default for GENERIC kernels) test this correctly, but non-VIMAGE jails have been performing an incomplete test when nested jails are used. Approved by: re@ (kib@) MFC after: 5 days
2018-10-06 02:10:32 +00:00
for (tppr = ppr; tppr != &prison0; tppr = tppr->pr_parent)
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (tppr->pr_flags & PR_VNET)
break;
Fix the test prohibiting jails from sharing IP addresses. It's not supposed to be legal for two jails to contain the same IP address, unless both jails contain only that one address. This is the behavior documented in jail(8), and is there to prevent confusion when multiple jails are listening on IADDR_ANY. VIMAGE jails (now the default for GENERIC kernels) test this correctly, but non-VIMAGE jails have been performing an incomplete test when nested jails are used. Approved by: re@ (kib@) MFC after: 5 days
2018-10-06 02:10:32 +00:00
#else
tppr = &prison0;
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#endif
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
FOREACH_PRISON_DESCENDANT(tppr, tpr, descend) {
if (tpr == pr ||
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
(tpr != tppr && (tpr->pr_flags & PR_VNET)) ||
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#endif
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
!prison_isalive(tpr)) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
descend = 0;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
continue;
}
if (!(tpr->pr_flags & PR_IP4_USER))
continue;
descend = 0;
if (tpr->pr_ip4 == NULL ||
(ip4s == 1 && tpr->pr_ip4s == 1))
continue;
for (ii = 0; ii < ip4s; ii++) {
Move IPv4-specific jail functions to new file netinet/in_jail.c _prison_check_ip4 renamed to prison_check_ip4_locked Move IPv6-specific jail functions to new file netinet6/in6_jail.c _prison_check_ip6 renamed to prison_check_ip6_locked Add appropriate prototypes to sys/sys/jail.h Adjust kern_jail.c to call prison_check_ip4_locked and prison_check_ip6_locked accordingly. Add netinet/in_jail.c and netinet6/in6_jail.c to the list of files that need to be built when INET and INET6, respectively, are configured in the kernel configuration file. Reviewed by: jtl Approved by: sjg (mentor) Sponsored by: Juniper Networks, Inc. Differential Revision: https://reviews.freebsd.org/D6799
2016-08-09 02:16:21 +00:00
if (prison_check_ip4_locked(tpr, &ip4[ii]) ==
0) {
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
error = EADDRINUSE;
vfs_opterror(opts,
"IPv4 addresses clash");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
}
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET6
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (ip6s > 0) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (ppr->pr_flags & PR_IP6) {
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
/*
* Make sure the new set of IP addresses is a
* subset of the parent's list.
*/
for (ij = 0; ij < ppr->pr_ip6s; ij++)
if (IN6_ARE_ADDR_EQUAL(&ip6[0],
&ppr->pr_ip6[ij]))
break;
if (ij == ppr->pr_ip6s) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
}
if (ip6s > 1) {
for (ii = ij = 1; ii < ip6s; ii++) {
if (IN6_ARE_ADDR_EQUAL(&ip6[ii],
&ppr->pr_ip6[0]))
continue;
for (; ij < ppr->pr_ip6s; ij++)
if (IN6_ARE_ADDR_EQUAL(
&ip6[ii], &ppr->pr_ip6[ij]))
break;
if (ij == ppr->pr_ip6s)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
break;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (ij == ppr->pr_ip6s) {
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
}
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
/* Check for conflicting IP addresses. */
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
Fix the test prohibiting jails from sharing IP addresses. It's not supposed to be legal for two jails to contain the same IP address, unless both jails contain only that one address. This is the behavior documented in jail(8), and is there to prevent confusion when multiple jails are listening on IADDR_ANY. VIMAGE jails (now the default for GENERIC kernels) test this correctly, but non-VIMAGE jails have been performing an incomplete test when nested jails are used. Approved by: re@ (kib@) MFC after: 5 days
2018-10-06 02:10:32 +00:00
for (tppr = ppr; tppr != &prison0; tppr = tppr->pr_parent)
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (tppr->pr_flags & PR_VNET)
break;
Fix the test prohibiting jails from sharing IP addresses. It's not supposed to be legal for two jails to contain the same IP address, unless both jails contain only that one address. This is the behavior documented in jail(8), and is there to prevent confusion when multiple jails are listening on IADDR_ANY. VIMAGE jails (now the default for GENERIC kernels) test this correctly, but non-VIMAGE jails have been performing an incomplete test when nested jails are used. Approved by: re@ (kib@) MFC after: 5 days
2018-10-06 02:10:32 +00:00
#else
tppr = &prison0;
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#endif
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
FOREACH_PRISON_DESCENDANT(tppr, tpr, descend) {
if (tpr == pr ||
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
(tpr != tppr && (tpr->pr_flags & PR_VNET)) ||
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#endif
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
!prison_isalive(tpr)) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
descend = 0;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
continue;
}
if (!(tpr->pr_flags & PR_IP6_USER))
continue;
descend = 0;
if (tpr->pr_ip6 == NULL ||
(ip6s == 1 && tpr->pr_ip6s == 1))
continue;
for (ii = 0; ii < ip6s; ii++) {
Move IPv4-specific jail functions to new file netinet/in_jail.c _prison_check_ip4 renamed to prison_check_ip4_locked Move IPv6-specific jail functions to new file netinet6/in6_jail.c _prison_check_ip6 renamed to prison_check_ip6_locked Add appropriate prototypes to sys/sys/jail.h Adjust kern_jail.c to call prison_check_ip4_locked and prison_check_ip6_locked accordingly. Add netinet/in_jail.c and netinet6/in6_jail.c to the list of files that need to be built when INET and INET6, respectively, are configured in the kernel configuration file. Reviewed by: jtl Approved by: sjg (mentor) Sponsored by: Juniper Networks, Inc. Differential Revision: https://reviews.freebsd.org/D6799
2016-08-09 02:16:21 +00:00
if (prison_check_ip6_locked(tpr, &ip6[ii]) ==
0) {
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
error = EADDRINUSE;
vfs_opterror(opts,
"IPv6 addresses clash");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
onamelen = namelen = 0;
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
if (namelc != NULL) {
Fix jail name checking that disallowed anything that starts with '0'. The intention was to just limit leading zeroes on numeric names. That check is now improved to also catch the leading spaces and '+' that strtoul can pass through. PR: 204897 MFC after: 3 days
2015-12-15 17:25:00 +00:00
/* Give a default name of the jid. Also allow the name to be
* explicitly the jid - but not any other number, and only in
* normal form (no leading zero/etc).
*/
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
if (namelc[0] == '\0')
snprintf(namelc = numbuf, sizeof(numbuf), "%d", jid);
Fix jail name checking that disallowed anything that starts with '0'. The intention was to just limit leading zeroes on numeric names. That check is now improved to also catch the leading spaces and '+' that strtoul can pass through. PR: 204897 MFC after: 3 days
2015-12-15 17:25:00 +00:00
else if ((strtoul(namelc, &p, 10) != jid ||
namelc[0] < '1' || namelc[0] > '9') && *p == '\0') {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = EINVAL;
Allow a jail's name to be the same as its jid (which is the default if no name is specified), but still disallow other numeric names. Reviewed by: zec Approved by: bz (mentor) MFC after: 3 days
2009-09-04 19:00:48 +00:00
vfs_opterror(opts,
"name cannot be numeric (unless it is the jid)");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
* Make sure the name isn't too long for the prison or its
* children.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
*/
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
pnamelen = (ppr == &prison0) ? 0 : strlen(ppr->pr_name) + 1;
onamelen = strlen(pr->pr_name + pnamelen);
namelen = strlen(namelc);
if (pnamelen + namelen + 1 > sizeof(pr->pr_name)) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = ENAMETOOLONG;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
FOREACH_PRISON_DESCENDANT(pr, tpr, descend) {
if (strlen(tpr->pr_name) + (namelen - onamelen) >=
sizeof(pr->pr_name)) {
error = ENAMETOOLONG;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
}
In hardened systems, where the security.bsd.unprivileged_proc_debug sysctl node is set, allow setting security.bsd.unprivileged_proc_debug per-jail. In part, this is needed to create jails in which the Address Sanitizer (ASAN) fully works as ASAN utilizes libkvm to inspect the virtual address space. Instead of having to allow unprivileged process debugging for the entire system, allow setting it on a per-jail basis. The sysctl node is still security.bsd.unprivileged_proc_debug and the jail(8) param is allow.unprivileged_proc_debug. The sysctl code is now a sysctl proc rather than a sysctl int. This allows us to determine setting the flag for the corresponding jail (or prison0). As part of the change, the dynamic allow.* API needed to be modified to take into account pr_allow flags which may now be disabled in prison0. This prevents conflicts with new pr_allow flags (like that of vmm(4)) that are added (and removed) dynamically. Also teach the jail creation KPI to allow differences for certain pr_allow flags between the parent and child jail. This can happen when unprivileged process debugging is disabled in the parent prison, but enabled in the child. Submitted by: Shawn Webb <lattera at gmail.com> Obtained from: HardenedBSD (45b3625edba0f73b3e3890b1ec3d0d1e95fd47e1, deba0b5078cef0faae43cbdafed3035b16587afc, ab21eeb3b4c72f2500987c96ff603ccf3b6e7de8) Relnotes: yes Sponsored by: HardenedBSD and G2, Inc Differential Revision: https://reviews.freebsd.org/D18319
2018-11-27 17:51:50 +00:00
pr_allow_diff = pr_allow & ~ppr->pr_allow;
if (pr_allow_diff & ~PR_ALLOW_DIFFERENCES) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = EPERM;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
/*
* Let modules check their parameters. This requires unlocking and
* then re-locking the prison, but this is still a valid state as long
* as allprison_lock remains xlocked.
*/
mtx_unlock(&pr->pr_mtx);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags &= ~PD_LOCKED;
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
error = osd_jail_call(pr, PR_METHOD_CHECK, opts);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
if (error != 0)
goto done_deref;
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
mtx_lock(&pr->pr_mtx);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_LOCKED;
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
/* At this point, all valid parameters should have been noted. */
TAILQ_FOREACH(opt, opts, link) {
if (!opt->seen && strcmp(opt->name, "errmsg")) {
error = EINVAL;
vfs_opterror(opts, "unknown parameter: %s", opt->name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Set the parameters of the prison. */
#ifdef INET
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
redo_ip4 = 0;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (pr_flags & PR_IP4_USER) {
pr->pr_flags |= PR_IP4;
free(pr->pr_ip4, M_PRISON);
pr->pr_ip4s = ip4s;
pr->pr_ip4 = ip4;
ip4 = NULL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend) {
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
if (tpr->pr_flags & PR_VNET) {
descend = 0;
continue;
}
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (prison_restrict_ip4(tpr, NULL)) {
redo_ip4 = 1;
descend = 0;
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
#endif
#ifdef INET6
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
redo_ip6 = 0;
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
if (pr_flags & PR_IP6_USER) {
pr->pr_flags |= PR_IP6;
free(pr->pr_ip6, M_PRISON);
pr->pr_ip6s = ip6s;
pr->pr_ip6 = ip6;
ip6 = NULL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend) {
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
if (tpr->pr_flags & PR_VNET) {
descend = 0;
continue;
}
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (prison_restrict_ip6(tpr, NULL)) {
redo_ip6 = 1;
descend = 0;
}
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (gotslevel) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
pr->pr_securelevel = slevel;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* Set all child jails to be at least this level. */
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend)
if (tpr->pr_securelevel < slevel)
tpr->pr_securelevel = slevel;
}
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
if (gotchildmax) {
pr->pr_childmax = childmax;
/* Set all child jails to under this limit. */
FOREACH_PRISON_DESCENDANT_LOCKED_LEVEL(pr, tpr, descend, level)
if (tpr->pr_childmax > childmax - level)
tpr->pr_childmax = childmax > level
? childmax - level : 0;
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (gotenforce) {
pr->pr_enforce_statfs = enforce;
/* Pass this restriction on to the children. */
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend)
if (tpr->pr_enforce_statfs < enforce)
tpr->pr_enforce_statfs = enforce;
}
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
if (gotrsnum) {
pr->pr_devfs_rsnum = rsnum;
/* Pass this restriction on to the children. */
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend)
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
tpr->pr_devfs_rsnum = rsnum;
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
}
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
if (namelc != NULL) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (ppr == &prison0)
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
strlcpy(pr->pr_name, namelc, sizeof(pr->pr_name));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
else
snprintf(pr->pr_name, sizeof(pr->pr_name), "%s.%s",
Pass the current/new jail to PR_METHOD_CHECK, which pushes the call until after the jail is found or created. This requires unlocking the jail for the call and re-locking it afterward, but that works because nothing in the jail has been changed yet, and other processes won't change the important fields as long as allprison_lock remains held. Keep better track of name vs namelc in kern_jail_set. Name should always be the hierarchical name (relative to the caller), and namelc the last component. PR: 48471 MFC after: 5 days
2016-04-25 04:27:58 +00:00
ppr->pr_name, namelc);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* Change this component of child names. */
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend) {
bcopy(tpr->pr_name + onamelen, tpr->pr_name + namelen,
strlen(tpr->pr_name + onamelen) + 1);
bcopy(pr->pr_name, tpr->pr_name, namelen);
}
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (path != NULL) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* Try to keep a real-rooted full pathname. */
vfs: remove now useless ENODEV handling from vn_fullpath consumers Noted by: ngie
2020-02-08 15:51:08 +00:00
strlcpy(pr->pr_path, path, sizeof(pr->pr_path));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
pr->pr_root = root;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
root = NULL;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (PR_HOST & ch_flags & ~pr_flags) {
if (pr->pr_flags & PR_HOST) {
/*
* Copy the parent's host info. As with pr_ip4 above,
* the lack of a lock on the parent is not a problem;
* it is always set with allprison_lock at least
* shared, and is held exclusively here.
*/
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(pr->pr_hostname, pr->pr_parent->pr_hostname,
sizeof(pr->pr_hostname));
strlcpy(pr->pr_domainname, pr->pr_parent->pr_domainname,
sizeof(pr->pr_domainname));
strlcpy(pr->pr_hostuuid, pr->pr_parent->pr_hostuuid,
sizeof(pr->pr_hostuuid));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
pr->pr_hostid = pr->pr_parent->pr_hostid;
}
} else if (host != NULL || domain != NULL || uuid != NULL || gothid) {
/* Set this prison, and any descendants without PR_HOST. */
if (host != NULL)
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(pr->pr_hostname, host, sizeof(pr->pr_hostname));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (domain != NULL)
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(pr->pr_domainname, domain,
sizeof(pr->pr_domainname));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (uuid != NULL)
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(pr->pr_hostuuid, uuid, sizeof(pr->pr_hostuuid));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (gothid)
pr->pr_hostid = hid;
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend) {
if (tpr->pr_flags & PR_HOST)
descend = 0;
else {
if (host != NULL)
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(tpr->pr_hostname,
pr->pr_hostname,
sizeof(tpr->pr_hostname));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (domain != NULL)
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(tpr->pr_domainname,
pr->pr_domainname,
sizeof(tpr->pr_domainname));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (uuid != NULL)
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(tpr->pr_hostuuid,
pr->pr_hostuuid,
sizeof(tpr->pr_hostuuid));
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (gothid)
tpr->pr_hostid = hid;
}
}
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr->pr_allow = (pr->pr_allow & ~ch_allow) | pr_allow;
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
if ((tallow = ch_allow & ~pr_allow))
prison_set_allow_locked(pr, tallow, 0);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* Persistent prisons get an extra reference, and prisons losing their
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
* persist flag lose that reference.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
*/
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
born = !prison_isalive(pr);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
if (ch_flags & PR_PERSIST & (pr_flags ^ pr->pr_flags)) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (pr_flags & PR_PERSIST) {
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
prison_hold(pr);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
/*
* This may make a dead prison alive again, but wait
* to label it as such until after OSD calls have had
* a chance to run (and perhaps to fail).
*/
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
refcount_acquire(&pr->pr_uref);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
} else {
jail: A jail could be removed without calling OSD methods Fix a long-standing bug where setting nopersist on a process-less jail would remove it without calling the the OSD PR_METHOD_REMOVE methods.
2021-01-22 18:50:10 +00:00
drflags |= PD_DEUREF;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
prison_free_not_last(pr);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
pr->pr_flags = (pr->pr_flags & ~ch_flags) | pr_flags;
mtx_unlock(&pr->pr_mtx);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags &= ~PD_LOCKED;
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
/*
* Any errors past this point will need to de-persist newly created
* prisons, as well as call remove methods.
*/
if (born)
drflags |= PD_KILL;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
#ifdef RACCT
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
if (racct_enable && created)
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
prison_racct_attach(pr);
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* Locks may have prevented a complete restriction of child IP
* addresses. If so, allocate some more memory and try again.
*/
#ifdef INET
while (redo_ip4) {
ip4s = pr->pr_ip4s;
ip4 = malloc(ip4s * sizeof(*ip4), M_PRISON, M_WAITOK);
mtx_lock(&pr->pr_mtx);
redo_ip4 = 0;
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend) {
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
if (tpr->pr_flags & PR_VNET) {
descend = 0;
continue;
}
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (prison_restrict_ip4(tpr, ip4)) {
if (ip4 != NULL)
ip4 = NULL;
else
redo_ip4 = 1;
}
}
mtx_unlock(&pr->pr_mtx);
}
#endif
#ifdef INET6
while (redo_ip6) {
ip6s = pr->pr_ip6s;
ip6 = malloc(ip6s * sizeof(*ip6), M_PRISON, M_WAITOK);
mtx_lock(&pr->pr_mtx);
redo_ip6 = 0;
FOREACH_PRISON_DESCENDANT_LOCKED(pr, tpr, descend) {
Don't allow mixing the "vnet" and "ip4/6" jail parameters, since vnet jails have their own IP stack and don't have access to the parent IP addresses anyway. Note that a virtual network stack forms a break between prisons with regard to the list of allowed IP addresses. Approved by: re (kib), bz (mentor)
2009-07-29 16:46:59 +00:00
#ifdef VIMAGE
if (tpr->pr_flags & PR_VNET) {
descend = 0;
continue;
}
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (prison_restrict_ip6(tpr, ip6)) {
if (ip6 != NULL)
ip6 = NULL;
else
redo_ip6 = 1;
}
}
mtx_unlock(&pr->pr_mtx);
}
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Let the modules do their work. */
Add a new jail OSD method, PR_METHOD_REMOVE. It's called when a jail is removed from the user perspective, i.e. when the last pr_uref goes away, even though the jail mail still exist in the dying state. It will also be called if either PR_METHOD_CREATE or PR_METHOD_SET fail. PR: 48471 MFC after: 5 days
2016-04-25 04:24:00 +00:00
if (born) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = osd_jail_call(pr, PR_METHOD_CREATE, opts);
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
if (error)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
error = osd_jail_call(pr, PR_METHOD_SET, opts);
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
if (error)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
/*
* A new prison is now ready to be seen; either it has gained a user
* reference via persistence, or is about to gain one via attachment.
*/
if (born) {
drflags = prison_lock_xlock(pr, drflags);
pr->pr_state = PRISON_STATE_ALIVE;
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Attach this process to the prison if requested. */
if (flags & JAIL_ATTACH) {
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
error = do_jail_attach(td, pr,
jail: Add safety around prison_deref() flags. do_jail_attach() now only uses the PD_XXX flags that refer to lock status, so make sure that something else like PD_KILL doesn't slip through. Add a KASSERT() in prison_deref() to catch any further PD_KILL misuse.
2021-02-26 04:10:42 +00:00
prison_lock_xlock(pr, drflags & PD_LOCK_FLAGS));
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags &= ~(PD_LOCKED | PD_LIST_XLOCKED);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error) {
vfs_opterror(opts, "attach failed");
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done_deref;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
Fix use-after-free in kern_jail_set() triggered e.g. by attempts to clear "persist" flag from empty persistent jail, like this: jail -c persist=1 jail -n 1 -m persist=0 Submitted by: Mateusz Guzik <mjguzik at gmail dot com> MFC after: 2 weeks
2012-05-22 19:43:20 +00:00
#ifdef RACCT
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
if (racct_enable && !created) {
jail: Fix a LOR introduced in 1158508a8086
2021-02-22 23:51:10 +00:00
if (drflags & PD_LOCKED) {
mtx_unlock(&pr->pr_mtx);
drflags &= ~PD_LOCKED;
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (drflags & PD_LIST_XLOCKED) {
sx_xunlock(&allprison_lock);
drflags &= ~PD_LIST_XLOCKED;
jail: Clean up allprison_lock handing in kern_jail_set Keep explicit track of the allprison_lock state during the final part of kern_jail_set, instead of deducing it from the JAIL_ATTACH flag. Also properly clean up when the attachment fails, fixing a long- standing (though minor) memory leak.
2020-12-31 23:18:43 +00:00
}
Fix use-after-free in kern_jail_set() triggered e.g. by attempts to clear "persist" flag from empty persistent jail, like this: jail -c persist=1 jail -n 1 -m persist=0 Submitted by: Mateusz Guzik <mjguzik at gmail dot com> MFC after: 2 weeks
2012-05-22 19:43:20 +00:00
prison_racct_modify(pr);
}
#endif
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
drflags &= ~PD_KILL;
Fix use-after-free in kern_jail_set() triggered e.g. by attempts to clear "persist" flag from empty persistent jail, like this: jail -c persist=1 jail -n 1 -m persist=0 Submitted by: Mateusz Guzik <mjguzik at gmail dot com> MFC after: 2 weeks
2012-05-22 19:43:20 +00:00
td->td_retval[0] = pr->pr_id;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
done_deref:
/* Release any temporary prison holds and/or locks. */
if (pr != NULL)
prison_deref(pr, drflags);
else if (drflags & PD_LIST_SLOCKED)
sx_sunlock(&allprison_lock);
else if (drflags & PD_LIST_XLOCKED)
sx_xunlock(&allprison_lock);
Remove the support for using non-mpsafe filesystem modules. In particular, do not lock Giant conditionally when calling into the filesystem module, remove the VFS_LOCK_GIANT() and related macros. Stop handling buffers belonging to non-mpsafe filesystems. The VFS_VERSION is bumped to indicate the interface change which does not result in the interface signatures changes. Conducted and reviewed by: attilio Tested by: pho
2012-10-22 17:50:54 +00:00
if (root != NULL)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
vrele(root);
done_errmsg:
if (error) {
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
/* Write the error message back to userspace. */
Clean up some logic in jail error messages, replacing a missing test and a redundant test with a single correct test.
2016-06-09 20:39:57 +00:00
if (vfs_getopt(opts, "errmsg", (void **)&errmsg,
&errmsg_len) == 0 && errmsg_len > 0) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
errmsg_pos = 2 * vfs_getopt_pos(opts, "errmsg") + 1;
Clean up some logic in jail error messages, replacing a missing test and a redundant test with a single correct test.
2016-06-09 20:39:57 +00:00
if (optuio->uio_segflg == UIO_SYSSPACE)
bcopy(errmsg,
optuio->uio_iov[errmsg_pos].iov_base,
errmsg_len);
else
copyout(errmsg,
optuio->uio_iov[errmsg_pos].iov_base,
errmsg_len);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
done_free:
#ifdef INET
free(ip4, M_PRISON);
#endif
#ifdef INET6
free(ip6, M_PRISON);
#endif
Use separate buffer for global path to avoid overflow of path buffer. Reviewed by: jamie@ MFC after: 3 weeks
2012-01-21 00:06:21 +00:00
if (g_path != NULL)
free(g_path, M_TEMP);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
vfs_freeopts(opts);
return (error);
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
/*
* Find the next available prison ID. Return the ID on success, or zero
* on failure. Also set a pointer to the allprison list entry the prison
* should be inserted before.
*/
static int
get_next_prid(struct prison **insprp)
{
struct prison *inspr;
int jid, maxid;
jid = lastprid % JAIL_MAX + 1;
if (TAILQ_EMPTY(&allprison) ||
TAILQ_LAST(&allprison, prisonlist)->pr_id < jid) {
/*
* A common case is for all jails to be implicitly numbered,
* which means they'll go on the end of the list, at least
* for the first JAIL_MAX times.
*/
inspr = NULL;
} else {
/*
* Take two passes through the allprison list: first starting
* with the proposed jid, then ending with it.
*/
for (maxid = JAIL_MAX; maxid != 0; ) {
TAILQ_FOREACH(inspr, &allprison, pr_list) {
if (inspr->pr_id < jid)
continue;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (inspr->pr_id > jid) {
/* Found an opening. */
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
maxid = 0;
break;
}
if (++jid > maxid) {
if (lastprid == maxid || lastprid == 0)
{
/*
* The entire legal range
* has been traversed
*/
return 0;
}
/* Try again from the start. */
jid = 1;
maxid = lastprid;
break;
}
}
if (inspr == NULL) {
/* Found room at the end of the list. */
break;
}
}
}
*insprp = inspr;
lastprid = jid;
return (jid);
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* struct jail_get_args {
* struct iovec *iovp;
* unsigned int iovcnt;
* int flags;
* };
*/
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_jail_get(struct thread *td, struct jail_get_args *uap)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
{
struct uio *auio;
int error;
/* Check that we have an even number of iovecs. */
if (uap->iovcnt & 1)
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
return (EINVAL);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = copyinuio(uap->iovp, uap->iovcnt, &auio);
if (error)
return (error);
error = kern_jail_get(td, auio, uap->flags);
if (error == 0)
error = copyout(auio->uio_iov, uap->iovp,
uap->iovcnt * sizeof (struct iovec));
free(auio, M_IOV);
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
int
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
kern_jail_get(struct thread *td, struct uio *optuio, int flags)
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
{
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
struct bool_flags *bf;
struct jailsys_flags *jsf;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
struct prison *pr, *mypr;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
struct vfsopt *opt;
struct vfsoptlist *opts;
char *errmsg, *name;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
int drflags, error, errmsg_len, errmsg_pos, i, jid, len, pos;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
unsigned f;
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
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (flags & ~JAIL_GET_MASK)
return (EINVAL);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Get the parameter list. */
error = vfs_buildopts(optuio, &opts);
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);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
errmsg_pos = vfs_getopt_pos(opts, "errmsg");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
mypr = td->td_ucred->cr_prison;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
pr = NULL;
Delay an error message until the variable it uses gets initialized. Found with: Coverity Prevent(tm) CID: 4316 Reported by: trasz Approved by: bz (mentor)
2009-05-23 16:13:26 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* Find the prison specified by one of: lastjid, jid, name.
*/
sx_slock(&allprison_lock);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags = PD_LIST_SLOCKED;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_copyopt(opts, "lastjid", &jid, sizeof(jid));
if (error == 0) {
TAILQ_FOREACH(pr, &allprison, pr_list) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (pr->pr_id > jid &&
((flags & JAIL_DYING) || prison_isalive(pr)) &&
prison_ischild(mypr, pr)) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
mtx_lock(&pr->pr_mtx);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags |= PD_LOCKED;
goto found_prison;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
error = ENOENT;
vfs_opterror(opts, "no jail after %d", jid);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
} else if (error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_copyopt(opts, "jid", &jid, sizeof(jid));
if (error == 0) {
if (jid != 0) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr = prison_find_child(mypr, jid);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (pr != NULL) {
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_LOCKED;
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
if (!(prison_isalive(pr) ||
(flags & JAIL_DYING))) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = ENOENT;
vfs_opterror(opts, "jail %d is dying",
jid);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
goto found_prison;
}
error = ENOENT;
vfs_opterror(opts, "jail %d not found", jid);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
} else if (error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_getopt(opts, "name", (void **)&name, &len);
if (error == 0) {
if (len == 0 || name[len - 1] != '\0') {
error = EINVAL;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr = prison_find_name(mypr, name);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (pr != NULL) {
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_LOCKED;
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
if (!(prison_isalive(pr) || (flags & JAIL_DYING))) {
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = ENOENT;
vfs_opterror(opts, "jail \"%s\" is dying",
name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
goto found_prison;
}
error = ENOENT;
vfs_opterror(opts, "jail \"%s\" not found", name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
} else if (error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
vfs_opterror(opts, "no jail specified");
error = ENOENT;
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
found_prison:
/* Get the parameters of the prison. */
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
prison_hold(pr);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags |= PD_DEREF;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
td->td_retval[0] = pr->pr_id;
error = vfs_setopt(opts, "jid", &pr->pr_id, sizeof(pr->pr_id));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
i = (pr->pr_parent == mypr) ? 0 : pr->pr_parent->pr_id;
error = vfs_setopt(opts, "parent", &i, sizeof(i));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = vfs_setopts(opts, "name", prison_name(mypr, pr));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = vfs_setopt(opts, "cpuset.id", &pr->pr_cpuset->cs_id,
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
sizeof(pr->pr_cpuset->cs_id));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = vfs_setopts(opts, "path", prison_path(mypr, pr));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET
error = vfs_setopt_part(opts, "ip4.addr", pr->pr_ip4,
pr->pr_ip4s * sizeof(*pr->pr_ip4));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
#ifdef INET6
error = vfs_setopt_part(opts, "ip6.addr", pr->pr_ip6,
pr->pr_ip6s * sizeof(*pr->pr_ip6));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
error = vfs_setopt(opts, "securelevel", &pr->pr_securelevel,
sizeof(pr->pr_securelevel));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
error = vfs_setopt(opts, "children.cur", &pr->pr_childcount,
sizeof(pr->pr_childcount));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
error = vfs_setopt(opts, "children.max", &pr->pr_childmax,
sizeof(pr->pr_childmax));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
error = vfs_setopts(opts, "host.hostname", pr->pr_hostname);
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
error = vfs_setopts(opts, "host.domainname", pr->pr_domainname);
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
error = vfs_setopts(opts, "host.hostuuid", pr->pr_hostuuid);
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Provide groundwork for 32-bit binary compatibility on non-x86 platforms, for upcoming 64-bit PowerPC and MIPS support. This renames the COMPAT_IA32 option to COMPAT_FREEBSD32, removes some IA32-specific code from MI parts of the kernel and enhances the freebsd32 compatibility code to support big-endian platforms. Reviewed by: kib, jhb
2010-03-11 14:49:06 +00:00
#ifdef COMPAT_FREEBSD32
Add macro to test the sv_flags of any process. Change some places to test the flags instead of explicit comparing with address of known sysentvec structures. MFC after: 1 month
2011-01-26 20:03:58 +00:00
if (SV_PROC_FLAG(td->td_proc, SV_ILP32)) {
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
uint32_t hid32 = pr->pr_hostid;
error = vfs_setopt(opts, "host.hostid", &hid32, sizeof(hid32));
} else
#endif
error = vfs_setopt(opts, "host.hostid", &pr->pr_hostid,
sizeof(pr->pr_hostid));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
error = vfs_setopt(opts, "enforce_statfs", &pr->pr_enforce_statfs,
sizeof(pr->pr_enforce_statfs));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
error = vfs_setopt(opts, "devfs_ruleset", &pr->pr_devfs_rsnum,
sizeof(pr->pr_devfs_rsnum));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_bool;
bf < pr_flag_bool + nitems(pr_flag_bool);
bf++) {
i = (pr->pr_flags & bf->flag) ? 1 : 0;
error = vfs_setopt(opts, bf->name, &i, sizeof(i));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
i = !i;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
error = vfs_setopt(opts, bf->noname, &i, sizeof(i));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (jsf = pr_flag_jailsys;
jsf < pr_flag_jailsys + nitems(pr_flag_jailsys);
jsf++) {
f = pr->pr_flags & (jsf->disable | jsf->new);
i = (f != 0 && f == jsf->disable) ? JAIL_SYS_DISABLE
: (f == jsf->new) ? JAIL_SYS_NEW
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
: JAIL_SYS_INHERIT;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
error = vfs_setopt(opts, jsf->name, &i, sizeof(i));
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
}
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_allow;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
bf < pr_flag_allow + nitems(pr_flag_allow) &&
atomic_load_int(&bf->flag) != 0;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
bf++) {
i = (pr->pr_allow & bf->flag) ? 1 : 0;
error = vfs_setopt(opts, bf->name, &i, sizeof(i));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
i = !i;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
error = vfs_setopt(opts, bf->noname, &i, sizeof(i));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
i = !prison_isalive(pr);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = vfs_setopt(opts, "dying", &i, sizeof(i));
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
i = !i;
error = vfs_setopt(opts, "nodying", &i, sizeof(i));
Export the new osreldate and osrelease jail parms in jail_get(2).
2015-02-28 17:32:31 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Format the line properly (wrap before column 80).
2015-02-28 17:44:31 +00:00
error = vfs_setopt(opts, "osreldate", &pr->pr_osreldate,
sizeof(pr->pr_osreldate));
Export the new osreldate and osrelease jail parms in jail_get(2).
2015-02-28 17:32:31 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Export the new osreldate and osrelease jail parms in jail_get(2).
2015-02-28 17:32:31 +00:00
error = vfs_setopts(opts, "osrelease", pr->pr_osrelease);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error != 0 && error != ENOENT)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Get the module parameters. */
mtx_unlock(&pr->pr_mtx);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
drflags &= ~PD_LOCKED;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = osd_jail_call(pr, PR_METHOD_GET, opts);
- 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)
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
prison_deref(pr, drflags);
pr = NULL;
drflags = 0;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* By now, all parameters should have been noted. */
TAILQ_FOREACH(opt, opts, link) {
if (!opt->seen && strcmp(opt->name, "errmsg")) {
error = EINVAL;
vfs_opterror(opts, "unknown parameter: %s", opt->name);
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
goto done;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Write the fetched parameters back to userspace. */
error = 0;
TAILQ_FOREACH(opt, opts, link) {
if (opt->pos >= 0 && opt->pos != errmsg_pos) {
pos = 2 * opt->pos + 1;
optuio->uio_iov[pos].iov_len = opt->len;
if (opt->value != NULL) {
if (optuio->uio_segflg == UIO_SYSSPACE) {
bcopy(opt->value,
optuio->uio_iov[pos].iov_base,
opt->len);
} else {
error = copyout(opt->value,
optuio->uio_iov[pos].iov_base,
opt->len);
if (error)
break;
}
}
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
done:
/* Release any temporary prison holds and/or locks. */
if (pr != NULL)
prison_deref(pr, drflags);
else if (drflags & PD_LIST_SLOCKED)
sx_sunlock(&allprison_lock);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (error && errmsg_pos >= 0) {
jail: Simplify handling of prison_deref() Track the the current lock/reference state in a single variable, rather than deducing the proper prison_deref() flags from a combination of equations and hard-coded values.
2021-01-10 05:05:06 +00:00
/* Write the error message back to userspace. */
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
vfs_getopt(opts, "errmsg", (void **)&errmsg, &errmsg_len);
errmsg_pos = 2 * errmsg_pos + 1;
if (errmsg_len > 0) {
if (optuio->uio_segflg == UIO_SYSSPACE)
bcopy(errmsg,
optuio->uio_iov[errmsg_pos].iov_base,
errmsg_len);
else
copyout(errmsg,
optuio->uio_iov[errmsg_pos].iov_base,
errmsg_len);
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
vfs_freeopts(opts);
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
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
* struct jail_remove_args {
* int jid;
* };
*/
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_jail_remove(struct thread *td, struct jail_remove_args *uap)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
{
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
struct prison *pr;
int error;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = priv_check(td, PRIV_JAIL_REMOVE);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
if (error)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
return (error);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
sx_xlock(&allprison_lock);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr = prison_find_child(td->td_ucred->cr_prison, uap->jid);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (pr == NULL) {
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
sx_xunlock(&allprison_lock);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
return (EINVAL);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
if (!prison_isalive(pr)) {
/* Silently ignore already-dying prisons. */
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
mtx_unlock(&pr->pr_mtx);
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
sx_xunlock(&allprison_lock);
return (0);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
prison_deref(pr, PD_KILL | PD_LOCKED | PD_LIST_XLOCKED);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +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
/*
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
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_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 prison *pr;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
int error;
White space nits.
2007-03-07 21:24:51 +00:00
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
error = priv_check(td, PRIV_JAIL_ATTACH);
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
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
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
sx_slock(&allprison_lock);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr = prison_find_child(td->td_ucred->cr_prison, uap->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
if (pr == NULL) {
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
sx_sunlock(&allprison_lock);
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
return (EINVAL);
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
/* Do not allow a process to attach to a prison that is not alive. */
if (!prison_isalive(pr)) {
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
mtx_unlock(&pr->pr_mtx);
sx_sunlock(&allprison_lock);
return (EINVAL);
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
return (do_jail_attach(td, pr, PD_LOCKED | PD_LIST_SLOCKED));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
static int
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
do_jail_attach(struct thread *td, struct prison *pr, int drflags)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
{
struct proc *p;
struct ucred *newcred, *oldcred;
Remove the support for using non-mpsafe filesystem modules. In particular, do not lock Giant conditionally when calling into the filesystem module, remove the VFS_LOCK_GIANT() and related macros. Stop handling buffers belonging to non-mpsafe filesystems. The VFS_VERSION is bumped to indicate the interface change which does not result in the interface signatures changes. Conducted and reviewed by: attilio Tested by: pho
2012-10-22 17:50:54 +00:00
int error;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
mtx_assert(&pr->pr_mtx, MA_OWNED);
sx_assert(&allprison_lock, SX_LOCKED);
jail: Add safety around prison_deref() flags. do_jail_attach() now only uses the PD_XXX flags that refer to lock status, so make sure that something else like PD_KILL doesn't slip through. Add a KASSERT() in prison_deref() to catch any further PD_KILL misuse.
2021-02-26 04:10:42 +00:00
drflags &= PD_LOCK_FLAGS;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +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.
*/
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
prison_hold(pr);
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
refcount_acquire(&pr->pr_uref);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags |= PD_DEREF | PD_DEUREF;
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(&pr->pr_mtx);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags &= ~PD_LOCKED;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/* Let modules do whatever they need to prepare for attaching. */
error = osd_jail_call(pr, PR_METHOD_ATTACH, td);
if (error) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
prison_deref(pr, drflags);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
return (error);
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
sx_unlock(&allprison_lock);
drflags &= ~(PD_LIST_SLOCKED | PD_LIST_XLOCKED);
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
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
/*
* Reparent the newly attached process to this jail.
*/
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
p = td->td_proc;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
error = cpuset_setproc_update_set(p, pr->pr_cpuset);
if (error)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
goto e_revert_osd;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
vn_lock() is currently only used with the 'curthread' passed as argument. Remove this argument and pass curthread directly to underlying VOP_LOCK1() VFS method. This modify makes the code cleaner and in particular remove an annoying dependence helping next lockmgr() cleanup. KPI results, obviously, changed. Manpage and FreeBSD_version will be updated through further commits. As a side note, would be valuable to say that next commits will address a similar cleanup about VFS methods, in particular vop_lock1 and vop_unlock. Tested by: Diego Sardina <siarodx at gmail dot com>, Andrea Di Pasquale <whyx dot it at gmail dot com>
2008-01-10 01:10:58 +00:00
vn_lock(pr->pr_root, LK_EXCLUSIVE | LK_RETRY);
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 = change_dir(pr->pr_root, td)) != 0)
goto e_unlock;
#ifdef MAC
Merge first in a series of TrustedBSD MAC Framework KPI changes from Mac OS X Leopard--rationalize naming for entry points to the following general forms: mac_<object>_<method/action> mac_<object>_check_<method/action> The previous naming scheme was inconsistent and mostly reversed from the new scheme. Also, make object types more consistent and remove spaces from object types that contain multiple parts ("posix_sem" -> "posixsem") to make mechanical parsing easier. Introduce a new "netinet" object type for certain IPv4/IPv6-related methods. Also simplify, slightly, some entry point names. All MAC policy modules will need to be recompiled, and modules not updates as part of this commit will need to be modified to conform to the new KPI. Sponsored by: SPARTA (original patches against Mac OS X) Obtained from: TrustedBSD Project, Apple Computer
2007-10-24 19:04:04 +00:00
if ((error = mac_vnode_check_chroot(td->td_ucred, pr->pr_root)))
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_unlock;
#endif
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(pr->pr_root);
jail: Change both root and working directories in jail_attach(2) jail_attach(2) performs an internal chroot operation, leaving it up to the calling process to assure the working directory is inside the jail. Add a matching internal chdir operation to the jail's root. Also ignore kern.chroot_allow_open_directories, and always disallow the operation if there are any directory descriptors open. Reported by: mjg Approved by: markj, kib MFC after: 3 days
2021-02-19 22:13:35 +00:00
if ((error = pwd_chroot_chdir(td, pr->pr_root)))
Remove the support for using non-mpsafe filesystem modules. In particular, do not lock Giant conditionally when calling into the filesystem module, remove the VFS_LOCK_GIANT() and related macros. Stop handling buffers belonging to non-mpsafe filesystems. The VFS_VERSION is bumped to indicate the interface change which does not result in the interface signatures changes. Conducted and reviewed by: attilio Tested by: pho
2012-10-22 17:50:54 +00:00
goto e_revert_osd;
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
- 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);
Use crcopysafe in jail_attach.
2016-04-26 21:19:12 +00:00
oldcred = crcopysafe(p, newcred);
- 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;
cred: add proc_set_cred helper The goal here is to provide one place altering process credentials. This eases debugging and opens up posibilities to do additional work when such an action is performed.
2015-03-16 00:10:03 +00:00
proc_set_cred(p, newcred);
Use crcopysafe in jail_attach.
2016-04-26 21:19:12 +00:00
setsugid(p);
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
#ifdef RACCT
racct_proc_ucred_changed(p, oldcred, newcred);
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
crhold(newcred);
#endif
PROC_UNLOCK(p);
#ifdef RCTL
rctl_proc_ucred_changed(p, newcred);
crfree(newcred);
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
#endif
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
prison_deref(oldcred->cr_prison, drflags);
- 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
crfree(oldcred);
jail: Handle a possible race between jail_remove(2) and fork(2) jail_remove(2) includes a loop that sends SIGKILL to all processes in a jail, but skips processes in PRS_NEW state. Thus it is possible the a process in mid-fork(2) during jail removal can survive the jail being removed. Add a prison flag PR_REMOVE, which is checked before the new process returns. If the jail is being removed, the process will then exit. Also check this flag in jail_attach(2) which has a similar issue. Reported by: trasz Approved by: kib MFC after: 3 days
2021-02-16 19:19:13 +00:00
/*
* If the prison was killed while changing credentials, die along
* with it.
*/
if (!prison_isalive(pr)) {
PROC_LOCK(p);
kern_psignal(p, SIGKILL);
PROC_UNLOCK(p);
}
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);
Use crcopysafe in jail_attach.
2016-04-26 21:19:12 +00:00
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
e_unlock:
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(pr->pr_root);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
e_revert_osd:
/* Tell modules this thread is still in its old jail after all. */
jail: add a missing lock around an osd_jail_call(). allprison_lock should be at least held shared when jail OSD methods are called. Add a shared lock around one such call where that wasn't the case. In another such call, change an exclusive lock grab to be shared in what is likely the more common case.
2020-12-27 04:49:30 +00:00
sx_slock(&allprison_lock);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags |= PD_LIST_SLOCKED;
Use crcopysafe in jail_attach.
2016-04-26 21:19:12 +00:00
(void)osd_jail_call(td->td_ucred->cr_prison, PR_METHOD_ATTACH, td);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
prison_deref(pr, drflags);
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.
*/
Make prison_find() globally accessible.
2007-04-05 21:34:54 +00:00
struct prison *
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
prison_find(int prid)
{
struct prison *pr;
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
sx_assert(&allprison_lock, SX_LOCKED);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
TAILQ_FOREACH(pr, &allprison, pr_list) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (pr->pr_id < prid)
continue;
jail: Fix an O(n^2) loop when adding jails When a jail is added using the default (system-chosen) JID, and non-default-JID jails already exist, a loop through the allprison list could restart and result in unnecessary O(n^2) behaviour. There should never be more than two list passes required. Also clean up inefficient (though still O(n)) allprison list traversal when finding jails by ID, or when adding jails in the common case of all default JIDs.
2020-12-26 18:39:34 +00:00
if (pr->pr_id > prid)
break;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
KASSERT(prison_isvalid(pr), ("Found invalid prison %p", pr));
mtx_lock(&pr->pr_mtx);
return (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
}
return (NULL);
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
* Find a prison that is a descendant of mypr. Returns a locked prison or NULL.
*/
struct prison *
prison_find_child(struct prison *mypr, int prid)
{
struct prison *pr;
int descend;
sx_assert(&allprison_lock, SX_LOCKED);
FOREACH_PRISON_DESCENDANT(mypr, pr, descend) {
if (pr->pr_id == prid) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
KASSERT(prison_isvalid(pr),
("Found invalid prison %p", pr));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
mtx_lock(&pr->pr_mtx);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
return (pr);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
}
return (NULL);
}
/*
* Look for the name relative to mypr. Returns a locked prison or NULL.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
*/
struct prison *
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
prison_find_name(struct prison *mypr, const char *name)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
{
struct prison *pr, *deadpr;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
size_t mylen;
int descend;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
sx_assert(&allprison_lock, SX_LOCKED);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
mylen = (mypr == &prison0) ? 0 : strlen(mypr->pr_name) + 1;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
deadpr = NULL;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
FOREACH_PRISON_DESCENDANT(mypr, pr, descend) {
if (!strcmp(pr->pr_name + mylen, name)) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
KASSERT(prison_isvalid(pr),
("Found invalid prison %p", pr));
if (prison_isalive(pr)) {
mtx_lock(&pr->pr_mtx);
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
return (pr);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
}
deadpr = pr;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* There was no valid prison - perhaps there was a dying one. */
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (deadpr != NULL)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
mtx_lock(&deadpr->pr_mtx);
return (deadpr);
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
* See if a prison has the specific flag set. The prison should be locked,
* unless checking for flags that are only set at jail creation (such as
* PR_IP4 and PR_IP6), or only the single bit is examined, without regard
* to any other prison data.
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
*/
int
prison_flag(struct ucred *cred, unsigned flag)
{
return (cred->cr_prison->pr_flags & flag);
}
int
prison_allow(struct ucred *cred, unsigned flag)
{
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
return ((cred->cr_prison->pr_allow & flag) != 0);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
* Hold a prison reference, by incrementing pr_ref. It is generally
* an error to hold a prison that does not already have a reference.
* A prison record will remain valid as long as it has at least one
* reference, and will not be removed as long as either the prison
* mutex or the allprison lock is held (allprison_lock may be shared).
*/
void
prison_hold_locked(struct prison *pr)
{
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
/* Locking is no longer required. */
prison_hold(pr);
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
}
void
prison_hold(struct prison *pr)
{
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
#ifdef INVARIANTS
int was_valid = refcount_acquire_if_not_zero(&pr->pr_ref);
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
KASSERT(was_valid,
("Trying to hold dead prison %p (jid=%d).", pr, pr->pr_id));
#else
refcount_acquire(&pr->pr_ref);
#endif
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
}
/*
* Remove a prison reference. If that was the last reference, the
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
* prison will be removed (at a later time).
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
*/
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
Update ZFS from version 6 to 13 and bring some FreeBSD-specific changes. This bring huge amount of changes, I'll enumerate only user-visible changes: - Delegated Administration Allows regular users to perform ZFS operations, like file system creation, snapshot creation, etc. - L2ARC Level 2 cache for ZFS - allows to use additional disks for cache. Huge performance improvements mostly for random read of mostly static content. - slog Allow to use additional disks for ZFS Intent Log to speed up operations like fsync(2). - vfs.zfs.super_owner Allows regular users to perform privileged operations on files stored on ZFS file systems owned by him. Very careful with this one. - chflags(2) Not all the flags are supported. This still needs work. - ZFSBoot Support to boot off of ZFS pool. Not finished, AFAIK. Submitted by: dfr - Snapshot properties - New failure modes Before if write requested failed, system paniced. Now one can select from one of three failure modes: - panic - panic on write error - wait - wait for disk to reappear - continue - serve read requests if possible, block write requests - Refquota, refreservation properties Just quota and reservation properties, but don't count space consumed by children file systems, clones and snapshots. - Sparse volumes ZVOLs that don't reserve space in the pool. - External attributes Compatible with extattr(2). - NFSv4-ACLs Not sure about the status, might not be complete yet. Submitted by: trasz - Creation-time properties - Regression tests for zpool(8) command. Obtained from: OpenSolaris
2008-11-17 20:49:29 +00:00
prison_free_locked(struct prison *pr)
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
{
Update ZFS from version 6 to 13 and bring some FreeBSD-specific changes. This bring huge amount of changes, I'll enumerate only user-visible changes: - Delegated Administration Allows regular users to perform ZFS operations, like file system creation, snapshot creation, etc. - L2ARC Level 2 cache for ZFS - allows to use additional disks for cache. Huge performance improvements mostly for random read of mostly static content. - slog Allow to use additional disks for ZFS Intent Log to speed up operations like fsync(2). - vfs.zfs.super_owner Allows regular users to perform privileged operations on files stored on ZFS file systems owned by him. Very careful with this one. - chflags(2) Not all the flags are supported. This still needs work. - ZFSBoot Support to boot off of ZFS pool. Not finished, AFAIK. Submitted by: dfr - Snapshot properties - New failure modes Before if write requested failed, system paniced. Now one can select from one of three failure modes: - panic - panic on write error - wait - wait for disk to reappear - continue - serve read requests if possible, block write requests - Refquota, refreservation properties Just quota and reservation properties, but don't count space consumed by children file systems, clones and snapshots. - Sparse volumes ZVOLs that don't reserve space in the pool. - External attributes Compatible with extattr(2). - NFSv4-ACLs Not sure about the status, might not be complete yet. Submitted by: trasz - Creation-time properties - Regression tests for zpool(8) command. Obtained from: OpenSolaris
2008-11-17 20:49:29 +00:00
mtx_assert(&pr->pr_mtx, MA_OWNED);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
/*
* Locking is no longer required, but unlock because the caller
* expects it.
*/
mtx_unlock(&pr->pr_mtx);
prison_free(pr);
}
void
prison_free(struct prison *pr)
{
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
KASSERT(refcount_load(&pr->pr_ref) > 0,
("Trying to free dead prison %p (jid=%d).",
pr, pr->pr_id));
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (!refcount_release_if_not_last(&pr->pr_ref)) {
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
/*
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
* Don't remove the last reference in this context,
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
* in case there are locks held.
*/
Delay revmoing the last jail reference in prison_proc_free, and instead put it off into the pr_task. This is similar to prison_free, and in fact uses the same task even though they do something slightly different. This resolves a LOR between the process lock and allprison_lock, which came about in r298565. PR: 48471
2016-04-27 02:25:21 +00:00
taskqueue_enqueue(taskqueue_thread, &pr->pr_task);
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
}
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
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
static void
prison_free_not_last(struct prison *pr)
Update ZFS from version 6 to 13 and bring some FreeBSD-specific changes. This bring huge amount of changes, I'll enumerate only user-visible changes: - Delegated Administration Allows regular users to perform ZFS operations, like file system creation, snapshot creation, etc. - L2ARC Level 2 cache for ZFS - allows to use additional disks for cache. Huge performance improvements mostly for random read of mostly static content. - slog Allow to use additional disks for ZFS Intent Log to speed up operations like fsync(2). - vfs.zfs.super_owner Allows regular users to perform privileged operations on files stored on ZFS file systems owned by him. Very careful with this one. - chflags(2) Not all the flags are supported. This still needs work. - ZFSBoot Support to boot off of ZFS pool. Not finished, AFAIK. Submitted by: dfr - Snapshot properties - New failure modes Before if write requested failed, system paniced. Now one can select from one of three failure modes: - panic - panic on write error - wait - wait for disk to reappear - continue - serve read requests if possible, block write requests - Refquota, refreservation properties Just quota and reservation properties, but don't count space consumed by children file systems, clones and snapshots. - Sparse volumes ZVOLs that don't reserve space in the pool. - External attributes Compatible with extattr(2). - NFSv4-ACLs Not sure about the status, might not be complete yet. Submitted by: trasz - Creation-time properties - Regression tests for zpool(8) command. Obtained from: OpenSolaris
2008-11-17 20:49:29 +00:00
{
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
#ifdef INVARIANTS
int lastref;
Update ZFS from version 6 to 13 and bring some FreeBSD-specific changes. This bring huge amount of changes, I'll enumerate only user-visible changes: - Delegated Administration Allows regular users to perform ZFS operations, like file system creation, snapshot creation, etc. - L2ARC Level 2 cache for ZFS - allows to use additional disks for cache. Huge performance improvements mostly for random read of mostly static content. - slog Allow to use additional disks for ZFS Intent Log to speed up operations like fsync(2). - vfs.zfs.super_owner Allows regular users to perform privileged operations on files stored on ZFS file systems owned by him. Very careful with this one. - chflags(2) Not all the flags are supported. This still needs work. - ZFSBoot Support to boot off of ZFS pool. Not finished, AFAIK. Submitted by: dfr - Snapshot properties - New failure modes Before if write requested failed, system paniced. Now one can select from one of three failure modes: - panic - panic on write error - wait - wait for disk to reappear - continue - serve read requests if possible, block write requests - Refquota, refreservation properties Just quota and reservation properties, but don't count space consumed by children file systems, clones and snapshots. - Sparse volumes ZVOLs that don't reserve space in the pool. - External attributes Compatible with extattr(2). - NFSv4-ACLs Not sure about the status, might not be complete yet. Submitted by: trasz - Creation-time properties - Regression tests for zpool(8) command. Obtained from: OpenSolaris
2008-11-17 20:49:29 +00:00
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
KASSERT(refcount_load(&pr->pr_ref) > 0,
("Trying to free dead prison %p (jid=%d).",
pr, pr->pr_id));
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
lastref = refcount_release(&pr->pr_ref);
KASSERT(!lastref,
("prison_free_not_last freed last ref on prison %p (jid=%d).",
pr, pr->pr_id));
#else
jail: fix build after the previous commit Noted by: Michael Butler <imb protected-networks.net>
2021-02-21 19:57:21 +00:00
refcount_release(&pr->pr_ref);
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
#endif
Update ZFS from version 6 to 13 and bring some FreeBSD-specific changes. This bring huge amount of changes, I'll enumerate only user-visible changes: - Delegated Administration Allows regular users to perform ZFS operations, like file system creation, snapshot creation, etc. - L2ARC Level 2 cache for ZFS - allows to use additional disks for cache. Huge performance improvements mostly for random read of mostly static content. - slog Allow to use additional disks for ZFS Intent Log to speed up operations like fsync(2). - vfs.zfs.super_owner Allows regular users to perform privileged operations on files stored on ZFS file systems owned by him. Very careful with this one. - chflags(2) Not all the flags are supported. This still needs work. - ZFSBoot Support to boot off of ZFS pool. Not finished, AFAIK. Submitted by: dfr - Snapshot properties - New failure modes Before if write requested failed, system paniced. Now one can select from one of three failure modes: - panic - panic on write error - wait - wait for disk to reappear - continue - serve read requests if possible, block write requests - Refquota, refreservation properties Just quota and reservation properties, but don't count space consumed by children file systems, clones and snapshots. - Sparse volumes ZVOLs that don't reserve space in the pool. - External attributes Compatible with extattr(2). - NFSv4-ACLs Not sure about the status, might not be complete yet. Submitted by: trasz - Creation-time properties - Regression tests for zpool(8) command. Obtained from: OpenSolaris
2008-11-17 20:49:29 +00:00
}
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
/*
* Hold a a prison for user visibility, by incrementing pr_uref.
* It is generally an error to hold a prison that isn't already
* user-visible, except through the the jail system calls. It is also
* an error to hold an invalid prison. A prison record will remain
* alive as long as it has at least one user reference, and will not
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
* be set to the dying state until the prison mutex and allprison_lock
* are both freed.
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
*/
void
prison_proc_hold(struct prison *pr)
{
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
#ifdef INVARIANTS
int was_alive = refcount_acquire_if_not_zero(&pr->pr_uref);
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
KASSERT(was_alive,
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
("Cannot add a process to a non-alive prison (jid=%d)", pr->pr_id));
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
#else
refcount_acquire(&pr->pr_uref);
#endif
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
}
/*
* Remove a prison user reference. If it was the last reference, the
* prison will be considered "dying", and may be removed once all of
* its references are dropped.
*/
void
prison_proc_free(struct prison *pr)
{
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
/*
* Locking is only required when releasing the last reference.
* This allows assurance that a locked prison will remain alive
* until it is unlocked.
*/
KASSERT(refcount_load(&pr->pr_uref) > 0,
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
("Trying to kill a process in a dead prison (jid=%d)", pr->pr_id));
jail: fix dangling reference bug from 6754ae2572eb The change to use refcounts for pr_uref was mishandled in prison_proc_free, so killing a jail's last process could add an extra reference, leaving it an unkillable zombie.
2021-01-22 18:56:24 +00:00
if (!refcount_release_if_not_last(&pr->pr_uref)) {
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
/*
* Don't remove the last user reference in this context,
* which is expected to be a process that is not only locked,
* but also half dead. Add a reference so any calls to
* prison_free() won't re-submit the task.
*/
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
prison_hold(pr);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
mtx_lock(&pr->pr_mtx);
KASSERT(!(pr->pr_flags & PR_COMPLETE_PROC),
("Redundant last reference in prison_proc_free (jid=%d)",
pr->pr_id));
pr->pr_flags |= PR_COMPLETE_PROC;
mtx_unlock(&pr->pr_mtx);
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
taskqueue_enqueue(taskqueue_thread, &pr->pr_task);
}
}
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
static void
prison_proc_free_not_last(struct prison *pr)
{
#ifdef INVARIANTS
int lastref;
KASSERT(refcount_load(&pr->pr_uref) > 0,
("Trying to free dead prison %p (jid=%d).",
pr, pr->pr_id));
lastref = refcount_release(&pr->pr_uref);
KASSERT(!lastref,
("prison_proc_free_not_last freed last uref on prison %p (jid=%d).",
pr, pr->pr_id));
#else
refcount_release(&pr->pr_uref);
#endif
}
Delay revmoing the last jail reference in prison_proc_free, and instead put it off into the pr_task. This is similar to prison_free, and in fact uses the same task even though they do something slightly different. This resolves a LOR between the process lock and allprison_lock, which came about in r298565. PR: 48471
2016-04-27 02:25:21 +00:00
/*
* Complete a call to either prison_free or prison_proc_free.
*/
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)
{
Delay revmoing the last jail reference in prison_proc_free, and instead put it off into the pr_task. This is similar to prison_free, and in fact uses the same task even though they do something slightly different. This resolves a LOR between the process lock and allprison_lock, which came about in r298565. PR: 48471
2016-04-27 02:25:21 +00:00
struct prison *pr = context;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
int drflags;
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
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
/*
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
* This could be called to release the last reference, or the last
* user reference (plus the reference held in prison_proc_free).
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
*/
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags = prison_lock_xlock(pr, PD_DEREF);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
if (pr->pr_flags & PR_COMPLETE_PROC) {
pr->pr_flags &= ~PR_COMPLETE_PROC;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
drflags |= PD_DEUREF;
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
prison_deref(pr, drflags);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +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
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
/*
jail: Clean up some function placement and improve comments. Move prison_hold, prison_hold_locked ,prison_proc_hold, and prison_proc_free to a more intuitive part of the file (together with with prison_free and prison_free_locked), and add or improve comments to these and others, to better describe what's going in the prison reference cycle. No functional changes.
2021-01-19 01:23:51 +00:00
* Remove a prison reference and/or user reference (usually).
* This assumes context that allows sleeping (for allprison_lock),
* with no non-sleeping locks held, except perhaps the prison itself.
* If there are no more references, release and delist the prison.
* On completion, the prison lock and the allprison lock are both
* unlocked.
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
*/
static void
prison_deref(struct prison *pr, int flags)
{
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
struct prisonlist freeprison;
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
struct prison *killpr, *rpr, *ppr, *tpr;
struct proc *p;
Update ZFS from version 6 to 13 and bring some FreeBSD-specific changes. This bring huge amount of changes, I'll enumerate only user-visible changes: - Delegated Administration Allows regular users to perform ZFS operations, like file system creation, snapshot creation, etc. - L2ARC Level 2 cache for ZFS - allows to use additional disks for cache. Huge performance improvements mostly for random read of mostly static content. - slog Allow to use additional disks for ZFS Intent Log to speed up operations like fsync(2). - vfs.zfs.super_owner Allows regular users to perform privileged operations on files stored on ZFS file systems owned by him. Very careful with this one. - chflags(2) Not all the flags are supported. This still needs work. - ZFSBoot Support to boot off of ZFS pool. Not finished, AFAIK. Submitted by: dfr - Snapshot properties - New failure modes Before if write requested failed, system paniced. Now one can select from one of three failure modes: - panic - panic on write error - wait - wait for disk to reappear - continue - serve read requests if possible, block write requests - Refquota, refreservation properties Just quota and reservation properties, but don't count space consumed by children file systems, clones and snapshots. - Sparse volumes ZVOLs that don't reserve space in the pool. - External attributes Compatible with extattr(2). - NFSv4-ACLs Not sure about the status, might not be complete yet. Submitted by: trasz - Creation-time properties - Regression tests for zpool(8) command. Obtained from: OpenSolaris
2008-11-17 20:49:29 +00:00
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
killpr = NULL;
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
TAILQ_INIT(&freeprison);
/*
* Release this prison as requested, which may cause its parent
* to be released, and then maybe its grandparent, etc.
*/
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
for (;;) {
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
if (flags & PD_KILL) {
/* Kill the prison and its descendents. */
jail: Add safety around prison_deref() flags. do_jail_attach() now only uses the PD_XXX flags that refer to lock status, so make sure that something else like PD_KILL doesn't slip through. Add a KASSERT() in prison_deref() to catch any further PD_KILL misuse.
2021-02-26 04:10:42 +00:00
KASSERT(pr != &prison0,
("prison_deref trying to kill prison0"));
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
if (!(flags & PD_DEREF)) {
prison_hold(pr);
flags |= PD_DEREF;
}
flags = prison_lock_xlock(pr, flags);
prison_deref_kill(pr, &freeprison);
}
Delay the recursive decrement of pr_uref when jails are made invisible but not removed; decrement it instead when the child jail actually goes away. This avoids letting the counter go below zero in the case where dying (pr_uref==0) jails are "resurrected", and an associated KASSERT panic. Submitted by: Steven Hartland Approved by: re (bz) MFC after: 1 week
2011-08-26 16:03:34 +00:00
if (flags & PD_DEUREF) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
/* Drop a user reference. */
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
KASSERT(refcount_load(&pr->pr_uref) > 0,
Delay revmoing the last jail reference in prison_proc_free, and instead put it off into the pr_task. This is similar to prison_free, and in fact uses the same task even though they do something slightly different. This resolves a LOR between the process lock and allprison_lock, which came about in r298565. PR: 48471
2016-04-27 02:25:21 +00:00
("prison_deref PD_DEUREF on a dead prison (jid=%d)",
pr->pr_id));
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (!refcount_release_if_not_last(&pr->pr_uref)) {
if (!(flags & PD_DEREF)) {
prison_hold(pr);
flags |= PD_DEREF;
}
flags = prison_lock_xlock(pr, flags);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
if (refcount_release(&pr->pr_uref) &&
pr->pr_state == PRISON_STATE_ALIVE) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
/*
* When the last user references goes,
* this becomes a dying prison.
*/
KASSERT(
refcount_load(&prison0.pr_uref) > 0,
("prison0 pr_uref=0"));
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
pr->pr_state = PRISON_STATE_DYING;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
mtx_unlock(&pr->pr_mtx);
flags &= ~PD_LOCKED;
(void)osd_jail_call(pr,
PR_METHOD_REMOVE, NULL);
}
}
}
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
if (flags & PD_KILL) {
/*
* Any remaining user references are probably processes
* that need to be killed, either in this prison or its
* descendants.
*/
if (refcount_load(&pr->pr_uref) > 0)
killpr = pr;
jail: Add safety around prison_deref() flags. do_jail_attach() now only uses the PD_XXX flags that refer to lock status, so make sure that something else like PD_KILL doesn't slip through. Add a KASSERT() in prison_deref() to catch any further PD_KILL misuse.
2021-02-26 04:10:42 +00:00
/* Make sure the parent prison doesn't get killed. */
flags &= ~PD_KILL;
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
}
Delay revmoing the last jail reference in prison_proc_free, and instead put it off into the pr_task. This is similar to prison_free, and in fact uses the same task even though they do something slightly different. This resolves a LOR between the process lock and allprison_lock, which came about in r298565. PR: 48471
2016-04-27 02:25:21 +00:00
if (flags & PD_DEREF) {
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
/* Drop a reference. */
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
KASSERT(refcount_load(&pr->pr_ref) > 0,
Delay revmoing the last jail reference in prison_proc_free, and instead put it off into the pr_task. This is similar to prison_free, and in fact uses the same task even though they do something slightly different. This resolves a LOR between the process lock and allprison_lock, which came about in r298565. PR: 48471
2016-04-27 02:25:21 +00:00
("prison_deref PD_DEREF on a dead prison (jid=%d)",
pr->pr_id));
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (!refcount_release_if_not_last(&pr->pr_ref)) {
flags = prison_lock_xlock(pr, flags);
if (refcount_release(&pr->pr_ref)) {
/*
* When the last reference goes,
* unlink the prison and set it aside.
*/
KASSERT(
refcount_load(&pr->pr_uref) == 0,
("prison_deref: last ref, "
"but still has %d urefs (jid=%d)",
pr->pr_uref, pr->pr_id));
KASSERT(
refcount_load(&prison0.pr_ref) != 0,
("prison0 pr_ref=0"));
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
pr->pr_state = PRISON_STATE_INVALID;
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
TAILQ_REMOVE(&allprison, pr, pr_list);
LIST_REMOVE(pr, pr_sibling);
TAILQ_INSERT_TAIL(&freeprison, pr,
pr_list);
for (ppr = pr->pr_parent;
ppr != NULL;
ppr = ppr->pr_parent)
ppr->pr_childcount--;
/*
* Removing a prison frees references
* from its parent.
*/
mtx_unlock(&pr->pr_mtx);
flags &= ~PD_LOCKED;
pr = pr->pr_parent;
flags |= PD_DEREF | PD_DEUREF;
continue;
jail: add a missing lock around an osd_jail_call(). allprison_lock should be at least held shared when jail OSD methods are called. Add a shared lock around one such call where that wasn't the case. In another such call, change an exclusive lock grab to be shared in what is likely the more common case.
2020-12-27 04:49:30 +00:00
}
Add a new jail OSD method, PR_METHOD_REMOVE. It's called when a jail is removed from the user perspective, i.e. when the last pr_uref goes away, even though the jail mail still exist in the dying state. It will also be called if either PR_METHOD_CREATE or PR_METHOD_SET fail. PR: 48471 MFC after: 5 days
2016-04-25 04:24:00 +00:00
}
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
break;
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
}
/* Release all the prison locks. */
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
if (flags & PD_LOCKED)
mtx_unlock(&pr->pr_mtx);
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
if (flags & PD_LIST_SLOCKED)
sx_sunlock(&allprison_lock);
else if (flags & PD_LIST_XLOCKED)
Fix a LOR between allprison_lock and vnode locks by releasing allprison_lock before releasing a prison's root vnode. PR: kern/138004 Reviewed by: kib Approved by: bz (mentor) MFC after: 3 days
2009-08-27 16:15:51 +00:00
sx_xunlock(&allprison_lock);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
/* Kill any processes attached to a killed prison. */
if (killpr != NULL) {
sx_slock(&allproc_lock);
FOREACH_PROC_IN_SYSTEM(p) {
PROC_LOCK(p);
if (p->p_state != PRS_NEW && p->p_ucred != NULL) {
for (ppr = p->p_ucred->cr_prison;
ppr != &prison0;
ppr = ppr->pr_parent)
if (ppr == killpr) {
kern_psignal(p, SIGKILL);
break;
}
}
PROC_UNLOCK(p);
}
sx_sunlock(&allproc_lock);
}
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
/*
* Finish removing any unreferenced prisons, which couldn't happen
* while allprison_lock was held (to avoid a LOR on vrele).
*/
TAILQ_FOREACH_SAFE(rpr, &freeprison, pr_list, tpr) {
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#ifdef VIMAGE
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
if (rpr->pr_vnet != rpr->pr_parent->pr_vnet)
vnet_destroy(rpr->pr_vnet);
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#endif
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
if (rpr->pr_root != NULL)
vrele(rpr->pr_root);
mtx_destroy(&rpr->pr_mtx);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
free(rpr->pr_ip4, M_PRISON);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET6
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
free(rpr->pr_ip6, M_PRISON);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#endif
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
if (rpr->pr_cpuset != NULL)
cpuset_rel(rpr->pr_cpuset);
osd_jail_exit(rpr);
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
#ifdef RACCT
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
if (racct_enable)
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
prison_racct_detach(rpr);
Add rctl. It's used by racct to take user-configurable actions based on the set of rules it maintains and the current resource usage. It also privides userland API to manage that ruleset. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-30 17:48:15 +00:00
#endif
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
TAILQ_REMOVE(&freeprison, rpr, pr_list);
jail: Improve locking when removing prisons Change the flow of prison_deref() so it doesn't let go of allprison_lock until it's completely done using it (except for a possible drop as part of an upgrade on its first try). Differential Revision: https://reviews.freebsd.org/D28458 MFC after: 3 days
2021-02-20 22:38:58 +00:00
free(rpr, M_PRISON);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +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
}
jail: re-commit 811e27fa3c44 with fixes Make sure PD_KILL isn't passed to do_jail_attach, where it might end up trying to kill the caller's prison (even prison0). Fix the child jail loop in prison_deref_kill, which was doing the post-order part during the pre-order part. That's not a system- killer, but make jails not always die correctly.
2021-02-25 05:54:49 +00:00
/*
* Kill the prison and its descendants. Mark them as dying, clear the
* persist flag, and call module remove methods.
*/
static void
prison_deref_kill(struct prison *pr, struct prisonlist *freeprison)
{
struct prison *cpr, *ppr, *rpr;
bool descend;
/*
* Unlike the descendants, the target prison can be killed
* even if it is currently dying. This is useful for failed
* creation in jail_set(2).
*/
KASSERT(refcount_load(&pr->pr_ref) > 0,
("Trying to kill dead prison %p (jid=%d).",
pr, pr->pr_id));
refcount_acquire(&pr->pr_uref);
pr->pr_state = PRISON_STATE_DYING;
mtx_unlock(&pr->pr_mtx);
rpr = NULL;
FOREACH_PRISON_DESCENDANT_PRE_POST(pr, cpr, descend) {
if (descend) {
if (!prison_isalive(cpr)) {
descend = false;
continue;
}
prison_hold(cpr);
prison_proc_hold(cpr);
mtx_lock(&cpr->pr_mtx);
cpr->pr_state = PRISON_STATE_DYING;
cpr->pr_flags |= PR_REMOVE;
mtx_unlock(&cpr->pr_mtx);
continue;
}
if (!(cpr->pr_flags & PR_REMOVE))
continue;
(void)osd_jail_call(cpr, PR_METHOD_REMOVE, NULL);
mtx_lock(&cpr->pr_mtx);
cpr->pr_flags &= ~PR_REMOVE;
if (cpr->pr_flags & PR_PERSIST) {
cpr->pr_flags &= ~PR_PERSIST;
prison_proc_free_not_last(cpr);
prison_free_not_last(cpr);
}
(void)refcount_release(&cpr->pr_uref);
if (refcount_release(&cpr->pr_ref)) {
/*
* When the last reference goes, unlink the prison
* and set it aside for prison_deref() to handle.
* Delay unlinking the sibling list to keep the loop
* safe.
*/
if (rpr != NULL)
LIST_REMOVE(rpr, pr_sibling);
rpr = cpr;
rpr->pr_state = PRISON_STATE_INVALID;
TAILQ_REMOVE(&allprison, rpr, pr_list);
TAILQ_INSERT_TAIL(freeprison, rpr, pr_list);
/*
* Removing a prison frees references from its parent.
*/
ppr = rpr->pr_parent;
prison_proc_free_not_last(ppr);
prison_free_not_last(ppr);
for (; ppr != NULL; ppr = ppr->pr_parent)
ppr->pr_childcount--;
}
mtx_unlock(&cpr->pr_mtx);
}
if (rpr != NULL)
LIST_REMOVE(rpr, pr_sibling);
(void)osd_jail_call(pr, PR_METHOD_REMOVE, NULL);
mtx_lock(&pr->pr_mtx);
if (pr->pr_flags & PR_PERSIST) {
pr->pr_flags &= ~PR_PERSIST;
prison_proc_free_not_last(pr);
prison_free_not_last(pr);
}
(void)refcount_release(&pr->pr_uref);
}
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
/*
* Given the current locking state in the flags, make sure allprison_lock
* is held exclusive, and the prison is locked. Return flags indicating
* the new state.
*/
static int
prison_lock_xlock(struct prison *pr, int flags)
{
if (!(flags & PD_LIST_XLOCKED)) {
/*
* Get allprison_lock, which may be an upgrade,
* and may require unlocking the prison.
*/
if (flags & PD_LOCKED) {
mtx_unlock(&pr->pr_mtx);
flags &= ~PD_LOCKED;
}
if (flags & PD_LIST_SLOCKED) {
if (!sx_try_upgrade(&allprison_lock)) {
sx_sunlock(&allprison_lock);
sx_xlock(&allprison_lock);
}
flags &= ~PD_LIST_SLOCKED;
} else
sx_xlock(&allprison_lock);
flags |= PD_LIST_XLOCKED;
}
if (!(flags & PD_LOCKED)) {
/* Lock the prison mutex. */
mtx_lock(&pr->pr_mtx);
flags |= PD_LOCKED;
}
return flags;
}
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
/*
* Set or clear a permission bit in the pr_allow field, passing restrictions
* (cleared permission) down to child jails.
*/
void
prison_set_allow(struct ucred *cred, unsigned flag, int enable)
{
struct prison *pr;
pr = cred->cr_prison;
sx_slock(&allprison_lock);
mtx_lock(&pr->pr_mtx);
prison_set_allow_locked(pr, flag, enable);
mtx_unlock(&pr->pr_mtx);
sx_sunlock(&allprison_lock);
}
static void
prison_set_allow_locked(struct prison *pr, unsigned flag, int enable)
{
struct prison *cpr;
int descend;
if (enable != 0)
pr->pr_allow |= flag;
else {
pr->pr_allow &= ~flag;
FOREACH_PRISON_DESCENDANT_LOCKED(pr, cpr, descend)
cpr->pr_allow &= ~flag;
}
}
Don't allow creating a socket with a protocol family that the current jail doesn't support. This involves a new function prison_check_af, like prison_check_ip[46] but that checks only the family. With this change, most of the errors generated by jailed sockets shouldn't ever occur, at least until jails are changeable. Approved by: bz (mentor)
2009-02-05 14:15:18 +00:00
/*
* Check if a jail supports the given address family.
*
* Returns 0 if not jailed or the address family is supported, EAFNOSUPPORT
* if not.
*/
int
prison_check_af(struct ucred *cred, int af)
{
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
struct prison *pr;
Don't allow creating a socket with a protocol family that the current jail doesn't support. This involves a new function prison_check_af, like prison_check_ip[46] but that checks only the family. With this change, most of the errors generated by jailed sockets shouldn't ever occur, at least until jails are changeable. Approved by: bz (mentor)
2009-02-05 14:15:18 +00:00
int error;
KASSERT(cred != NULL, ("%s: cred is NULL", __func__));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr = cred->cr_prison;
Wrap a PR_VNET inside "#ifdef VIMAGE" since that the only place it applies. bz wants the blame for this. Noticed by: rwatson Approved by: bz (mentor)
2009-06-24 22:06:56 +00:00
#ifdef VIMAGE
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
/* Prisons with their own network stack are not limited. */
Throughout the network stack we have a few places of if (jailed(cred)) left. If you are running with a vnet (virtual network stack) those will return true and defer you to classic IP-jails handling and thus things will be "denied" or returned with an error. Work around this problem by introducing another "jailed()" function, jailed_without_vnet(), that also takes vnets into account, and permits the calls, should the jail from the given cred have its own virtual network stack. We cannot change the classic jailed() call to do that, as it is used outside the network stack as well. Discussed with: julian, zec, jamie, rwatson (back in Sept) MFC after: 5 days
2009-12-13 13:57:32 +00:00
if (prison_owns_vnet(cred))
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
return (0);
Wrap a PR_VNET inside "#ifdef VIMAGE" since that the only place it applies. bz wants the blame for this. Noticed by: rwatson Approved by: bz (mentor)
2009-06-24 22:06:56 +00:00
#endif
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
Don't allow creating a socket with a protocol family that the current jail doesn't support. This involves a new function prison_check_af, like prison_check_ip[46] but that checks only the family. With this change, most of the errors generated by jailed sockets shouldn't ever occur, at least until jails are changeable. Approved by: bz (mentor)
2009-02-05 14:15:18 +00:00
error = 0;
switch (af)
{
#ifdef INET
case AF_INET:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr->pr_flags & PR_IP4)
{
mtx_lock(&pr->pr_mtx);
if ((pr->pr_flags & PR_IP4) && pr->pr_ip4 == NULL)
error = EAFNOSUPPORT;
mtx_unlock(&pr->pr_mtx);
}
Don't allow creating a socket with a protocol family that the current jail doesn't support. This involves a new function prison_check_af, like prison_check_ip[46] but that checks only the family. With this change, most of the errors generated by jailed sockets shouldn't ever occur, at least until jails are changeable. Approved by: bz (mentor)
2009-02-05 14:15:18 +00:00
break;
#endif
#ifdef INET6
case AF_INET6:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr->pr_flags & PR_IP6)
{
mtx_lock(&pr->pr_mtx);
if ((pr->pr_flags & PR_IP6) && pr->pr_ip6 == NULL)
error = EAFNOSUPPORT;
mtx_unlock(&pr->pr_mtx);
}
Don't allow creating a socket with a protocol family that the current jail doesn't support. This involves a new function prison_check_af, like prison_check_ip[46] but that checks only the family. With this change, most of the errors generated by jailed sockets shouldn't ever occur, at least until jails are changeable. Approved by: bz (mentor)
2009-02-05 14:15:18 +00:00
break;
#endif
case AF_LOCAL:
case AF_ROUTE:
break;
default:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (!(pr->pr_allow & PR_ALLOW_SOCKET_AF))
Don't allow creating a socket with a protocol family that the current jail doesn't support. This involves a new function prison_check_af, like prison_check_ip[46] but that checks only the family. With this change, most of the errors generated by jailed sockets shouldn't ever occur, at least until jails are changeable. Approved by: bz (mentor)
2009-02-05 14:15:18 +00:00
error = EAFNOSUPPORT;
}
return (error);
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
/*
* Check if given address belongs to the jail referenced by cred (wrapper to
* prison_check_ip[46]).
*
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
* Returns 0 if jail doesn't restrict the address family or if address belongs
* to jail, EADDRNOTAVAIL if the address doesn't belong, or EAFNOSUPPORT if
* the jail doesn't allow the address family. IPv4 Address passed in in NBO.
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +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
Split gigantic rtsock route_output() into smaller functions. Amount of changes to the original code has been intentionally minimised to ease diffing. The changes are mostly mechanical, with the following exceptions: * lltable handler is now called directly based of RTF_LLINFO flag presense. * "report" logic for updating rtm in RTM_GET/RTM_DELETE has been simplified, fixing several potential use-after-free cases in rt_addrinfo. * llable asserts has been replaced with error-returning, preventing kernel crashes when lltable gw af family is invalid (root required). MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D22864
2019-12-31 17:26:53 +00:00
prison_if(struct ucred *cred, const 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
{
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Split gigantic rtsock route_output() into smaller functions. Amount of changes to the original code has been intentionally minimised to ease diffing. The changes are mostly mechanical, with the following exceptions: * lltable handler is now called directly based of RTF_LLINFO flag presense. * "report" logic for updating rtm in RTM_GET/RTM_DELETE has been simplified, fixing several potential use-after-free cases in rt_addrinfo. * llable asserts has been replaced with error-returning, preventing kernel crashes when lltable gw af family is invalid (root required). MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D22864
2019-12-31 17:26:53 +00:00
const struct sockaddr_in *sai;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
#ifdef INET6
Split gigantic rtsock route_output() into smaller functions. Amount of changes to the original code has been intentionally minimised to ease diffing. The changes are mostly mechanical, with the following exceptions: * lltable handler is now called directly based of RTF_LLINFO flag presense. * "report" logic for updating rtm in RTM_GET/RTM_DELETE has been simplified, fixing several potential use-after-free cases in rt_addrinfo. * llable asserts has been replaced with error-returning, preventing kernel crashes when lltable gw af family is invalid (root required). MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D22864
2019-12-31 17:26:53 +00:00
const struct sockaddr_in6 *sai6;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Standardize the various prison_foo_ip[46] functions and prison_if to return zero on success and an error code otherwise. The possible errors are EADDRNOTAVAIL if an address being checked for doesn't match the prison, and EAFNOSUPPORT if the prison doesn't have any addresses in that address family. For most callers of these functions, use the returned error code instead of e.g. a hard-coded EADDRNOTAVAIL or EINVAL. Always include a jailed() check in these functions, where a non-jailed cred always returns success (and makes no changes). Remove the explicit jailed() checks that preceded many of the function calls. Approved by: bz (mentor)
2009-02-05 14:06:09 +00:00
int error;
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
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
KASSERT(cred != NULL, ("%s: cred is NULL", __func__));
KASSERT(sa != NULL, ("%s: sa is NULL", __func__));
Throughout the network stack we have a few places of if (jailed(cred)) left. If you are running with a vnet (virtual network stack) those will return true and defer you to classic IP-jails handling and thus things will be "denied" or returned with an error. Work around this problem by introducing another "jailed()" function, jailed_without_vnet(), that also takes vnets into account, and permits the calls, should the jail from the given cred have its own virtual network stack. We cannot change the classic jailed() call to do that, as it is used outside the network stack as well. Discussed with: julian, zec, jamie, rwatson (back in Sept) MFC after: 5 days
2009-12-13 13:57:32 +00:00
#ifdef VIMAGE
if (prison_owns_vnet(cred))
return (0);
#endif
Standardize the various prison_foo_ip[46] functions and prison_if to return zero on success and an error code otherwise. The possible errors are EADDRNOTAVAIL if an address being checked for doesn't match the prison, and EAFNOSUPPORT if the prison doesn't have any addresses in that address family. For most callers of these functions, use the returned error code instead of e.g. a hard-coded EADDRNOTAVAIL or EINVAL. Always include a jailed() check in these functions, where a non-jailed cred always returns success (and makes no changes). Remove the explicit jailed() checks that preceded many of the function calls. Approved by: bz (mentor)
2009-02-05 14:06:09 +00:00
error = 0;
switch (sa->sa_family)
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
{
#ifdef INET
case AF_INET:
Split gigantic rtsock route_output() into smaller functions. Amount of changes to the original code has been intentionally minimised to ease diffing. The changes are mostly mechanical, with the following exceptions: * lltable handler is now called directly based of RTF_LLINFO flag presense. * "report" logic for updating rtm in RTM_GET/RTM_DELETE has been simplified, fixing several potential use-after-free cases in rt_addrinfo. * llable asserts has been replaced with error-returning, preventing kernel crashes when lltable gw af family is invalid (root required). MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D22864
2019-12-31 17:26:53 +00:00
sai = (const struct sockaddr_in *)sa;
Standardize the various prison_foo_ip[46] functions and prison_if to return zero on success and an error code otherwise. The possible errors are EADDRNOTAVAIL if an address being checked for doesn't match the prison, and EAFNOSUPPORT if the prison doesn't have any addresses in that address family. For most callers of these functions, use the returned error code instead of e.g. a hard-coded EADDRNOTAVAIL or EINVAL. Always include a jailed() check in these functions, where a non-jailed cred always returns success (and makes no changes). Remove the explicit jailed() checks that preceded many of the function calls. Approved by: bz (mentor)
2009-02-05 14:06:09 +00:00
error = prison_check_ip4(cred, &sai->sin_addr);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
break;
#endif
#ifdef INET6
case AF_INET6:
Split gigantic rtsock route_output() into smaller functions. Amount of changes to the original code has been intentionally minimised to ease diffing. The changes are mostly mechanical, with the following exceptions: * lltable handler is now called directly based of RTF_LLINFO flag presense. * "report" logic for updating rtm in RTM_GET/RTM_DELETE has been simplified, fixing several potential use-after-free cases in rt_addrinfo. * llable asserts has been replaced with error-returning, preventing kernel crashes when lltable gw af family is invalid (root required). MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D22864
2019-12-31 17:26:53 +00:00
sai6 = (const struct sockaddr_in6 *)sa;
Standardize the various prison_foo_ip[46] functions and prison_if to return zero on success and an error code otherwise. The possible errors are EADDRNOTAVAIL if an address being checked for doesn't match the prison, and EAFNOSUPPORT if the prison doesn't have any addresses in that address family. For most callers of these functions, use the returned error code instead of e.g. a hard-coded EADDRNOTAVAIL or EINVAL. Always include a jailed() check in these functions, where a non-jailed cred always returns success (and makes no changes). Remove the explicit jailed() checks that preceded many of the function calls. Approved by: bz (mentor)
2009-02-05 14:06:09 +00:00
error = prison_check_ip6(cred, &sai6->sin6_addr);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
break;
#endif
default:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (!(cred->cr_prison->pr_allow & PR_ALLOW_SOCKET_AF))
Standardize the various prison_foo_ip[46] functions and prison_if to return zero on success and an error code otherwise. The possible errors are EADDRNOTAVAIL if an address being checked for doesn't match the prison, and EAFNOSUPPORT if the prison doesn't have any addresses in that address family. For most callers of these functions, use the returned error code instead of e.g. a hard-coded EADDRNOTAVAIL or EINVAL. Always include a jailed() check in these functions, where a non-jailed cred always returns success (and makes no changes). Remove the explicit jailed() checks that preceded many of the function calls. Approved by: bz (mentor)
2009-02-05 14:06:09 +00:00
error = EAFNOSUPPORT;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Standardize the various prison_foo_ip[46] functions and prison_if to return zero on success and an error code otherwise. The possible errors are EADDRNOTAVAIL if an address being checked for doesn't match the prison, and EAFNOSUPPORT if the prison doesn't have any addresses in that address family. For most callers of these functions, use the returned error code instead of e.g. a hard-coded EADDRNOTAVAIL or EINVAL. Always include a jailed() check in these functions, where a non-jailed cred always returns success (and makes no changes). Remove the explicit jailed() checks that preceded many of the function calls. Approved by: bz (mentor)
2009-02-05 14:06:09 +00:00
return (error);
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 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
{
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return ((cred1->cr_prison == cred2->cr_prison ||
prison_ischild(cred1->cr_prison, cred2->cr_prison)) ? 0 : ESRCH);
}
/*
* Return 1 if p2 is a child of p1, otherwise 0.
*/
int
prison_ischild(struct prison *pr1, struct prison *pr2)
{
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
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
for (pr2 = pr2->pr_parent; pr2 != NULL; pr2 = pr2->pr_parent)
if (pr1 == pr2)
return (1);
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);
}
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
/*
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
* Return true if the prison is currently alive. A prison is alive if it
* holds user references and it isn't being removed.
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
*/
bool
prison_isalive(struct prison *pr)
{
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
if (__predict_false(pr->pr_state != PRISON_STATE_ALIVE))
jail: Handle a possible race between jail_remove(2) and fork(2) jail_remove(2) includes a loop that sends SIGKILL to all processes in a jail, but skips processes in PRS_NEW state. Thus it is possible the a process in mid-fork(2) during jail removal can survive the jail being removed. Add a prison flag PR_REMOVE, which is checked before the new process returns. If the jail is being removed, the process will then exit. Also check this flag in jail_attach(2) which has a similar issue. Reported by: trasz Approved by: kib MFC after: 3 days
2021-02-16 19:19:13 +00:00
return (false);
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
return (true);
}
/*
* Return true if the prison is currently valid. A prison is valid if it has
* been fully created, and is not being destroyed. Note that dying prisons
jail: Change the locking around pr_ref and pr_uref Require both the prison mutex and allprison_lock when pr_ref or pr_uref go to/from zero. Adding a non-first or removing a non-last reference remain lock-free. This means that a shared hold on allprison_lock is sufficient for prison_isalive() to be useful, which removes a number of cases of lock/check/unlock on the prison mutex. Expand the locking in kern_jail_set() to keep allprison_lock held exclusive until the new prison is valid, thus making invalid prisons invisible to any thread holding allprison_lock (except of course the one creating or destroying the prison). This renders prison_isvalid() nearly redundant, now used only in asserts. Differential Revision: https://reviews.freebsd.org/D28419 Differential Revision: https://reviews.freebsd.org/D28458
2021-02-21 18:55:44 +00:00
* are still considered valid. Invalid prisons won't be found under normal
* circumstances, as they're only put in that state by functions that have
* an exclusive hold on allprison_lock.
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
*/
bool
prison_isvalid(struct prison *pr)
{
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
if (__predict_false(pr->pr_state == PRISON_STATE_INVALID))
return (false);
jail: Use refcount(9) for prison references. Use refcount(9) for both pr_ref and pr_uref in struct prison. This allows prisons to held and freed without requiring the prison mutex. An exception to this is that dropping the last reference will still lock the prison, to keep the guarantee that a locked prison remains valid and alive (provided it was at the time it was locked). Among other things, this honors the promise made in a comment in crcopy(9), that it will not block, which hasn't been true for two decades.
2021-01-20 23:08:27 +00:00
if (__predict_false(refcount_load(&pr->pr_ref) == 0))
jail: Add prison_isvalid() and prison_isalive() prison_isvalid() checks if a prison record can be used at all, i.e. pr_ref > 0. This filters out prisons that aren't fully created, and those that are either in the process of being dismantled, or will be at the next opportunity. While the check for pr_ref > 0 is simple enough to make without a convenience function, this prepares the way for other measures of prison validity. prison_isalive() checks not only validity as far as the useablity of the prison structure, but also whether the prison is visible to user space. It replaces a test for pr_uref > 0, which is currently only used within kern_jail.c, and not often there. Both of these functions also assert that either the prison mutex or allprison_lock is held, since it's generally the case that unlocked prisons aren't guaranteed to remain useable for any length of time. This isn't entirely true, for example a thread can assume its own prison is good, but most exceptions will exist inside of kern_jail.c.
2021-01-18 18:56:20 +00:00
return (false);
return (true);
}
Throughout the network stack we have a few places of if (jailed(cred)) left. If you are running with a vnet (virtual network stack) those will return true and defer you to classic IP-jails handling and thus things will be "denied" or returned with an error. Work around this problem by introducing another "jailed()" function, jailed_without_vnet(), that also takes vnets into account, and permits the calls, should the jail from the given cred have its own virtual network stack. We cannot change the classic jailed() call to do that, as it is used outside the network stack as well. Discussed with: julian, zec, jamie, rwatson (back in Sept) MFC after: 5 days
2009-12-13 13:57:32 +00:00
/*
* Return 1 if the passed credential is in a jail and that jail does not
* have its own virtual network stack, otherwise 0.
*/
int
jailed_without_vnet(struct ucred *cred)
{
if (!jailed(cred))
return (0);
#ifdef VIMAGE
if (prison_owns_vnet(cred))
return (0);
#endif
return (1);
}
Add a function which returns the correct hostname for a given credential. Reviewed by: phk
2002-02-27 14:58:32 +00:00
/*
Add counterparts to getcredhostname: getcreddomainname, getcredhostuuid, getcredhostid Suggested by: rmacklem Approved by: bz
2009-06-13 00:12:02 +00:00
* Return the correct hostname (domainname, et al) for the passed credential.
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
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
{
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
struct prison *pr;
Add a function which returns the correct hostname for a given credential. Reviewed by: phk
2002-02-27 14:58:32 +00:00
Add counterparts to getcredhostname: getcreddomainname, getcredhostuuid, getcredhostid Suggested by: rmacklem Approved by: bz
2009-06-13 00:12:02 +00:00
/*
* A NULL credential can be used to shortcut to the physical
* system's hostname.
*/
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
pr = (cred != NULL) ? cred->cr_prison : &prison0;
mtx_lock(&pr->pr_mtx);
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(buf, pr->pr_hostname, size);
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
mtx_unlock(&pr->pr_mtx);
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
Add counterparts to getcredhostname: getcreddomainname, getcredhostuuid, getcredhostid Suggested by: rmacklem Approved by: bz
2009-06-13 00:12:02 +00:00
void
getcreddomainname(struct ucred *cred, char *buf, size_t size)
{
mtx_lock(&cred->cr_prison->pr_mtx);
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(buf, cred->cr_prison->pr_domainname, size);
Add counterparts to getcredhostname: getcreddomainname, getcredhostuuid, getcredhostid Suggested by: rmacklem Approved by: bz
2009-06-13 00:12:02 +00:00
mtx_unlock(&cred->cr_prison->pr_mtx);
}
void
getcredhostuuid(struct ucred *cred, char *buf, size_t size)
{
mtx_lock(&cred->cr_prison->pr_mtx);
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(buf, cred->cr_prison->pr_hostuuid, size);
Add counterparts to getcredhostname: getcreddomainname, getcredhostuuid, getcredhostid Suggested by: rmacklem Approved by: bz
2009-06-13 00:12:02 +00:00
mtx_unlock(&cred->cr_prison->pr_mtx);
}
void
getcredhostid(struct ucred *cred, unsigned long *hostid)
{
mtx_lock(&cred->cr_prison->pr_mtx);
*hostid = cred->cr_prison->pr_hostid;
mtx_unlock(&cred->cr_prison->pr_mtx);
}
ethersubr: Make the mac address generation more robust If we create two (vnet) jails and create a bridge interface in each we end up with the same mac address on both bridge interfaces. These very often conflicts, resulting in same mac address in both jails. Mitigate this problem by including the jail name in the mac address. Reviewed by: kevans, melifaro MFC after: 1 week Differential Revision: https://reviews.freebsd.org/D24383
2020-04-18 07:50:30 +00:00
void
getjailname(struct ucred *cred, char *name, size_t len)
{
mtx_lock(&cred->cr_prison->pr_mtx);
strlcpy(name, cred->cr_prison->pr_name, len);
mtx_unlock(&cred->cr_prison->pr_mtx);
}
Make it possible to change the vnet sysctl variables on jails with their own virtual network stack. Jails only inheriting a network stack cannot change anything that cannot be changed from within a prison. Reviewed by: rwatson, zec Approved by: re (kib)
2009-08-13 10:26:34 +00:00
#ifdef VIMAGE
/*
* Determine whether the prison represented by cred owns
* its vnet rather than having it inherited.
*
* Returns 1 in case the prison owns the vnet, 0 otherwise.
*/
int
prison_owns_vnet(struct ucred *cred)
{
/*
* vnets cannot be added/removed after jail creation,
* so no need to lock here.
*/
return (cred->cr_prison->pr_flags & PR_VNET ? 1 : 0);
}
#endif
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
pr = cred->cr_prison;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr->pr_enforce_statfs == 0)
return (0);
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 (pr->pr_root->v_mount == mp)
return (0);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr->pr_enforce_statfs == 2)
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
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;
pr = cred->cr_prison;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr->pr_enforce_statfs == 0)
return;
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 (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
}
}
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
* Check with permission for a specific privilege is granted within jail. We
* have a specific list of accepted privileges; the rest are denied.
*/
int
prison_priv_check(struct ucred *cred, int priv)
{
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
struct prison *pr;
int error;
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
Partially decompose priv_check by adding priv_check_cred_vfs_generation During buildkernel there are very frequent calls to priv_check and they all are for PRIV_VFS_GENERATION (coming from stat/fstat). This results in branching on several potential privileges checking if perhaps that's the one which has to be evaluated. Instead of the kitchen-sink approach provide a way to have commonly used privs directly evaluated.
2020-02-13 22:22:15 +00:00
/*
* Some policies have custom handlers. This routine should not be
* called for them. See priv_check_cred().
*/
switch (priv) {
vfs: respect PRIV_VFS_LOOKUP in vaccess_smr Reported by: novel
2020-08-25 14:18:50 +00:00
case PRIV_VFS_LOOKUP:
Partially decompose priv_check by adding priv_check_cred_vfs_generation During buildkernel there are very frequent calls to priv_check and they all are for PRIV_VFS_GENERATION (coming from stat/fstat). This results in branching on several potential privileges checking if perhaps that's the one which has to be evaluated. Instead of the kitchen-sink approach provide a way to have commonly used privs directly evaluated.
2020-02-13 22:22:15 +00:00
case PRIV_VFS_GENERATION:
KASSERT(0, ("prison_priv_check instead of a custom handler "
"called for %d\n", priv));
}
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
if (!jailed(cred))
return (0);
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
#ifdef VIMAGE
/*
* Privileges specific to prisons with a virtual network stack.
* There might be a duplicate entry here in case the privilege
* is only granted conditionally in the legacy jail case.
*/
switch (priv) {
#ifdef notyet
/*
* NFS-specific privileges.
*/
case PRIV_NFS_DAEMON:
case PRIV_NFS_LOCKD:
#endif
/*
* Network stack privileges.
*/
case PRIV_NET_BRIDGE:
case PRIV_NET_GRE:
case PRIV_NET_BPF:
case PRIV_NET_RAW: /* Dup, cond. in legacy jail case. */
case PRIV_NET_ROUTE:
case PRIV_NET_TAP:
case PRIV_NET_SETIFMTU:
case PRIV_NET_SETIFFLAGS:
case PRIV_NET_SETIFCAP:
Revised revision 199201 (add interface description capability as inspired by OpenBSD), based on comments from many, including rwatson, jhb, brooks and others. Sponsored by: iXsystems, Inc. MFC after: 1 month
2010-01-27 00:30:07 +00:00
case PRIV_NET_SETIFDESCR:
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
case PRIV_NET_SETIFNAME :
case PRIV_NET_SETIFMETRIC:
case PRIV_NET_SETIFPHYS:
case PRIV_NET_SETIFMAC:
Allow set ether/vlan PCP operation from the VNET jails. The vlan interfaces can be created from vnet jails, it seems, so it sounds logical to allow pcp configuration as well. Reviewed by: bz, hselasky (previous version) Sponsored by: Mellanox Technologies MFC after: 1 week Differential revision: https://reviews.freebsd.org/D17777
2018-11-12 15:59:32 +00:00
case PRIV_NET_SETLANPCP:
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
case PRIV_NET_ADDMULTI:
case PRIV_NET_DELMULTI:
case PRIV_NET_HWIOCTL:
case PRIV_NET_SETLLADDR:
case PRIV_NET_ADDIFGROUP:
case PRIV_NET_DELIFGROUP:
case PRIV_NET_IFCREATE:
case PRIV_NET_IFDESTROY:
case PRIV_NET_ADDIFADDR:
case PRIV_NET_DELIFADDR:
case PRIV_NET_LAGG:
case PRIV_NET_GIF:
case PRIV_NET_SETIFVNET:
Add infrastructure to allow all frames/packets received on an interface to be assigned to a non-default FIB instance. You may need to recompile world or ports due to the change of struct ifnet. Submitted by: cjsp Submitted by: Alexander V. Chernikov (melifaro ipfw.ru) (original versions) Reviewed by: julian Reviewed by: Alexander V. Chernikov (melifaro ipfw.ru) MFC after: 2 weeks X-MFC: use spare in struct ifnet
2011-07-03 12:22:02 +00:00
case PRIV_NET_SETIFFIB:
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
/*
* 802.11-related privileges.
*/
[net80211] Add new privileges; restrict what can be done in a jail. Split the MANAGE privilege into MANAGE, SETMAC and CREATE_VAP. + VAP_MANAGE is everything but setting the MAC and creating a VAP. + VAP_SETMAC is setting the MAC address of the VAP. Typically you wouldn't want the jail to be able to modify this. + CREATE_VAP is to create a new VAP. Again, you don't want to be doing this in a jail, but this DOES stop being able to run some corner cases like Dynamic WDS (DWDS) AP in a jail/vnet. We can figure this bit out later. This allows me to run wpa_supplicant in a jail after transferring a STA VAP into it. I unfortunately can't currently set the wlan debugging inside the jail; that would be super useful! Reviewed by: bz Differential Revision: https://reviews.freebsd.org/D25630
2020-07-19 15:16:27 +00:00
case PRIV_NET80211_VAP_GETKEY:
case PRIV_NET80211_VAP_MANAGE:
In case of prisons with their own network stack, permit additional privileges as well as not restricting the type of sockets a user can open. Note: the VIMAGE/vnet fetaure of of jails is still considered experimental and cannot guarantee that privileged users can be kept imprisoned if enabled. Reviewed by: rwatson Approved by: bz (mentor)
2009-06-24 21:39:50 +00:00
#ifdef notyet
/*
* ATM privileges.
*/
case PRIV_NETATM_CFG:
case PRIV_NETATM_ADD:
case PRIV_NETATM_DEL:
case PRIV_NETATM_SET:
/*
* Bluetooth privileges.
*/
case PRIV_NETBLUETOOTH_RAW:
#endif
/*
* Netgraph and netgraph module privileges.
*/
case PRIV_NETGRAPH_CONTROL:
#ifdef notyet
case PRIV_NETGRAPH_TTY:
#endif
/*
* IPv4 and IPv6 privileges.
*/
case PRIV_NETINET_IPFW:
case PRIV_NETINET_DIVERT:
case PRIV_NETINET_PF:
case PRIV_NETINET_DUMMYNET:
case PRIV_NETINET_CARP:
case PRIV_NETINET_MROUTE:
case PRIV_NETINET_RAW:
case PRIV_NETINET_ADDRCTRL6:
case PRIV_NETINET_ND6:
case PRIV_NETINET_SCOPE6:
case PRIV_NETINET_ALIFETIME6:
case PRIV_NETINET_IPSEC:
case PRIV_NETINET_BINDANY:
#ifdef notyet
/*
* NCP privileges.
*/
case PRIV_NETNCP:
/*
* SMB privileges.
*/
case PRIV_NETSMB:
#endif
/*
* No default: or deny here.
* In case of no permit fall through to next switch().
*/
if (cred->cr_prison->pr_flags & PR_VNET)
return (0);
}
#endif /* VIMAGE */
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
switch (priv) {
/*
* Allow ktrace privileges for root in jail.
*/
case PRIV_KTRACE:
For now, reflect practical reality that Audit system calls aren't allowed in Jail: return a privilege error.
2007-02-19 13:10:29 +00:00
#if 0
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
* Allow jailed processes to configure audit identity and
* submit audit records (login, etc). In the future we may
* want to further refine the relationship between audit and
* jail.
*/
case PRIV_AUDIT_GETAUDIT:
case PRIV_AUDIT_SETAUDIT:
case PRIV_AUDIT_SUBMIT:
For now, reflect practical reality that Audit system calls aren't allowed in Jail: return a privilege error.
2007-02-19 13:10:29 +00:00
#endif
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
* Allow jailed processes to manipulate process UNIX
* credentials in any way they see fit.
*/
case PRIV_CRED_SETUID:
case PRIV_CRED_SETEUID:
case PRIV_CRED_SETGID:
case PRIV_CRED_SETEGID:
case PRIV_CRED_SETGROUPS:
case PRIV_CRED_SETREUID:
case PRIV_CRED_SETREGID:
case PRIV_CRED_SETRESUID:
case PRIV_CRED_SETRESGID:
/*
* Jail implements visibility constraints already, so allow
* jailed root to override uid/gid-based constraints.
*/
case PRIV_SEEOTHERGIDS:
case PRIV_SEEOTHERUIDS:
/*
* Jail implements inter-process debugging limits already, so
* allow jailed root various debugging privileges.
*/
case PRIV_DEBUG_DIFFCRED:
case PRIV_DEBUG_SUGID:
case PRIV_DEBUG_UNPRIV:
/*
* Allow jail to set various resource limits and login
* properties, and for now, exceed process resource limits.
*/
case PRIV_PROC_LIMIT:
case PRIV_PROC_SETLOGIN:
case PRIV_PROC_SETRLIMIT:
/*
* System V and POSIX IPC privileges are granted in jail.
*/
case PRIV_IPC_READ:
case PRIV_IPC_WRITE:
case PRIV_IPC_ADMIN:
case PRIV_IPC_MSGSIZE:
case PRIV_MQ_ADMIN:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
* Jail operations within a jail work on child jails.
*/
case PRIV_JAIL_ATTACH:
case PRIV_JAIL_SET:
case PRIV_JAIL_REMOVE:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
* Jail implements its own inter-process limits, so allow
* root processes in jail to change scheduling on other
* processes in the same jail. Likewise for signalling.
*/
case PRIV_SCHED_DIFFCRED:
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
case PRIV_SCHED_CPUSET:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
case PRIV_SIGNAL_DIFFCRED:
case PRIV_SIGNAL_SUGID:
/*
* Allow jailed processes to write to sysctls marked as jail
* writable.
*/
case PRIV_SYSCTL_WRITEJAIL:
/*
* Allow root in jail to manage a variety of quota
Limit quota privileges in jail to PRIV_UFS_GETQUOTA and PRIV_UFS_SETQUOTA.
2007-02-19 13:26:39 +00:00
* properties. These should likely be conditional on a
* configuration option.
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
*/
Rename three quota privileges from the UFS privilege namespace to the VFS privilege namespace: exceedquota, getquota, and setquota. Leave UFS-specific quota configuration privileges in the UFS name space. This renumbers VFS and UFS privileges, so requires rebuilding modules if you are using security policies aware of privilege identifiers. This is likely no one at this point since none of the committed MAC policies use the privilege checks.
2007-02-19 13:33:10 +00:00
case PRIV_VFS_GETQUOTA:
case PRIV_VFS_SETQUOTA:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
* Since Jail relies on chroot() to implement file system
* protections, grant many VFS privileges to root in jail.
* Be careful to exclude mount-related and NFS-related
* privileges.
*/
case PRIV_VFS_READ:
case PRIV_VFS_WRITE:
case PRIV_VFS_ADMIN:
case PRIV_VFS_EXEC:
case PRIV_VFS_BLOCKRESERVE: /* XXXRW: Slightly surprising. */
case PRIV_VFS_CHFLAGS_DEV:
case PRIV_VFS_CHOWN:
case PRIV_VFS_CHROOT:
Rename PRIV_VFS_CLEARSUGID to PRIV_VFS_RETAINSUGID, which seems to better describe the privilege. OK'ed by: rwatson
2007-03-01 20:47:42 +00:00
case PRIV_VFS_RETAINSUGID:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
case PRIV_VFS_FCHROOT:
case PRIV_VFS_LINK:
case PRIV_VFS_SETGID:
Add PRIV_VFS_STAT privilege, which will allow overriding policy limits on the right to stat() a file, such as in mac_bsdextended. Obtained from: TrustedBSD Project MFC after: 3 months
2007-10-21 22:50:11 +00:00
case PRIV_VFS_STAT:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
case PRIV_VFS_STICKYFILE:
Keep PRIV_KMEM_READ permitted inside jails as it is on the outside.
2013-09-06 17:32:29 +00:00
/*
* As in the non-jail case, non-root users are expected to be
* able to read kernel/phyiscal memory (provided /dev/[k]mem
* exists in the jail and they have permission to access it).
*/
case PRIV_KMEM_READ:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
return (0);
/*
* Depending on the global setting, allow privilege of
* setting system flags.
*/
case PRIV_VFS_SYSFLAGS:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (cred->cr_prison->pr_allow & PR_ALLOW_CHFLAGS)
Add security.jail.mount_allowed sysctl, which allows to mount and unmount jail-friendly file systems from within a jail. Precisely it grants PRIV_VFS_MOUNT, PRIV_VFS_UNMOUNT and PRIV_VFS_MOUNT_NONUSER privileges for a jailed super-user. It is turned off by default. A jail-friendly file system is a file system which driver registers itself with VFCF_JAIL flag via VFS_SET(9) API. The lsvfs(1) command can be used to see which file systems are jail-friendly ones. There currently no jail-friendly file systems, ZFS will be the first one. In the future we may consider marking file systems like nullfs as jail-friendly. Reviewed by: rwatson
2007-04-05 21:03:05 +00:00
return (0);
else
return (EPERM);
/*
* Depending on the global setting, allow privilege of
* mounting/unmounting file systems.
*/
case PRIV_VFS_MOUNT:
case PRIV_VFS_UNMOUNT:
case PRIV_VFS_MOUNT_NONUSER:
Fix jails and jail-friendly file systems handling: - We need to allow for PRIV_VFS_MOUNT_OWNER inside a jail. - Move security checks to vfs_suser() and deny unmounting and updating for jailed root from different jails, etc. OK'ed by: rwatson
2007-04-13 23:54:22 +00:00
case PRIV_VFS_MOUNT_OWNER:
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
pr = cred->cr_prison;
prison_lock(pr);
if (pr->pr_allow & PR_ALLOW_MOUNT && pr->pr_enforce_statfs < 2)
error = 0;
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
else
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
error = EPERM;
prison_unlock(pr);
return (error);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
vfs: add restrictions to read(2) of a directory [2/2] This commit adds the priv(9) that waters down the sysctl to make it only allow read(2) of a dirfd by the system root. Jailed root is not allowed, but jail policy and superuser policy will abstain from allowing/denying it so that a MAC module can fully control the policy. Such a MAC module has been written, and can be found at: https://people.freebsd.org/~kevans/mac_read_dir-0.1.0.tar.gz It is expected that the MAC module won't be needed by many, as most only need to do such diagnostics that require this behavior as system root anyways. Interested parties are welcome to grab the MAC module above and create a port or locally integrate it, and with enough support it could see introduction to base. As noted in mac_read_dir.c, it is released under the BSD 2 clause license and allows the restrictions to be lifted for only jailed root or for all unprivileged users. PR: 246412 Reviewed by: mckusick, kib, emaste, jilles, cy, phk, imp (all previous) Reviewed by: rgrimes (latest version) Differential Revision: https://reviews.freebsd.org/D24596
2020-06-04 18:17:25 +00:00
/*
* Jails should hold no disposition on the PRIV_VFS_READ_DIR
* policy. priv_check_cred will not specifically allow it, and
* we may want a MAC policy to allow it.
*/
case PRIV_VFS_READ_DIR:
return (0);
Add allow.mlock to jail parameters It allows locking or unlocking physical pages in memory within a jail This allows running elasticsearch with "bootstrap.memory_lock" inside a jail Reviewed by: jamie@ Differential Revision: https://reviews.freebsd.org/D16342
2018-07-29 12:41:56 +00:00
/*
* Conditionnaly allow locking (unlocking) physical pages
* in memory.
*/
case PRIV_VM_MLOCK:
case PRIV_VM_MUNLOCK:
if (cred->cr_prison->pr_allow & PR_ALLOW_MLOCK)
return (0);
else
return (EPERM);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
Jails: Optionally prevent jailed root from binding to privileged ports You may now optionally specify allow.noreserved_ports to prevent root inside a jail from using privileged ports (less than 1024) PR: 217728 Submitted by: Matt Miller <mattm916@pulsar.neomailbox.ch> Reviewed by: jamie, cem, smh Relnotes: yes Differential Revision: https://reviews.freebsd.org/D10202
2017-06-06 02:15:00 +00:00
* Conditionally allow jailed root to bind reserved ports.
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
*/
case PRIV_NETINET_RESERVEDPORT:
Jails: Optionally prevent jailed root from binding to privileged ports You may now optionally specify allow.noreserved_ports to prevent root inside a jail from using privileged ports (less than 1024) PR: 217728 Submitted by: Matt Miller <mattm916@pulsar.neomailbox.ch> Reviewed by: jamie, cem, smh Relnotes: yes Differential Revision: https://reviews.freebsd.org/D10202
2017-06-06 02:15:00 +00:00
if (cred->cr_prison->pr_allow & PR_ALLOW_RESERVED_PORTS)
return (0);
else
return (EPERM);
/*
* Allow jailed root to reuse in-use ports.
*/
Allow PRIV_NETINET_REUSEPORT in jail.
2007-04-10 15:59:49 +00:00
case PRIV_NETINET_REUSEPORT:
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
return (0);
Replace the last susers calls in netinet6/ with privilege checks. Introduce a new privilege allowing to set certain IP header options (hop-by-hop, routing headers). Leave a few comments to be addressed later. Reviewed by: rwatson (older version, before addressing his comments)
2008-01-24 08:25:59 +00:00
/*
sys/kern: spelling fixes in comments. No functional change.
2016-04-29 22:15:33 +00:00
* Allow jailed root to set certain IPv4/6 (option) headers.
Replace the last susers calls in netinet6/ with privilege checks. Introduce a new privilege allowing to set certain IP header options (hop-by-hop, routing headers). Leave a few comments to be addressed later. Reviewed by: rwatson (older version, before addressing his comments)
2008-01-24 08:25:59 +00:00
*/
case PRIV_NETINET_SETHDROPTS:
return (0);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/*
* Conditionally allow creating raw sockets in jail.
*/
case PRIV_NETINET_RAW:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (cred->cr_prison->pr_allow & PR_ALLOW_RAW_SOCKETS)
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
return (0);
else
return (EPERM);
/*
* Since jail implements its own visibility limits on netstat
* sysctls, allow getcred. This allows identd to work in
* jail.
*/
case PRIV_NETINET_GETCRED:
return (0);
Add two new system calls, setloginclass(2) and getloginclass(2). This makes it possible for the kernel to track login class the process is assigned to, which is required for RCTL. This change also make setusercontext(3) call setloginclass(2) and makes it possible to retrieve current login class using id(1). Reviewed by: kib (as part of a larger patch)
2011-03-05 12:40:35 +00:00
/*
* Allow jailed root to set loginclass.
*/
case PRIV_PROC_SETLOGINCLASS:
return (0);
Add a new jail permission, allow.read_msgbuf. When true, jailed processes can see the dmesg buffer (this is the current behavior). When false (the new default), dmesg will be unavailable to jailed users, whether root or not. The security.bsd.unprivileged_read_msgbuf sysctl still works as before, controlling system-wide whether non-root users can see the buffer. PR: 211580 Submitted by: bz Approved by: re@ (kib@) MFC after: 3 days
2018-10-17 16:11:43 +00:00
/*
Fix typos from r339409. Reported by: maxim Approved by: re (gjb)
2018-10-18 15:02:57 +00:00
* Do not allow a process inside a jail to read the kernel
Add a new jail permission, allow.read_msgbuf. When true, jailed processes can see the dmesg buffer (this is the current behavior). When false (the new default), dmesg will be unavailable to jailed users, whether root or not. The security.bsd.unprivileged_read_msgbuf sysctl still works as before, controlling system-wide whether non-root users can see the buffer. PR: 211580 Submitted by: bz Approved by: re@ (kib@) MFC after: 3 days
2018-10-17 16:11:43 +00:00
* message buffer unless explicitly permitted.
*/
case PRIV_MSGBUF:
if (cred->cr_prison->pr_allow & PR_ALLOW_READ_MSGBUF)
return (0);
return (EPERM);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
default:
/*
* In all remaining cases, deny the privilege request. This
* includes almost all network privileges, many system
* configuration privileges.
*/
return (EPERM);
}
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
* Return the part of pr2's name that is relative to pr1, or the whole name
* if it does not directly follow.
*/
char *
prison_name(struct prison *pr1, struct prison *pr2)
{
char *name;
/* Jails see themselves as "0" (if they see themselves at all). */
if (pr1 == pr2)
return "0";
name = pr2->pr_name;
if (prison_ischild(pr1, pr2)) {
/*
* pr1 isn't locked (and allprison_lock may not be either)
* so its length can't be counted on. But the number of dots
* can be counted on - and counted.
*/
for (; pr1 != &prison0; pr1 = pr1->pr_parent)
name = strchr(name, '.') + 1;
}
return (name);
}
/*
* Return the part of pr2's path that is relative to pr1, or the whole path
* if it does not directly follow.
*/
static char *
prison_path(struct prison *pr1, struct prison *pr2)
{
char *path1, *path2;
int len1;
path1 = pr1->pr_path;
path2 = pr2->pr_path;
if (!strcmp(path1, "/"))
return (path2);
len1 = strlen(path1);
if (strncmp(path1, path2, len1))
return (path2);
if (path2[len1] == '\0')
return "/";
if (path2[len1] == '/')
return (path2 + len1);
return (path2);
}
/*
* Jail-related sysctls.
*/
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
static SYSCTL_NODE(_security, OID_AUTO, jail, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
"Jails");
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)
{
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
struct xprison *xp;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
struct prison *pr, *cpr;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#ifdef INET
struct in_addr *ip4 = NULL;
int ip4s = 0;
#endif
#ifdef INET6
Declare ip6 as (struct in6_addr *) instead of (struct in_addr *). This is a harmless bug since we never actually use ip6 as anything other than an opaque pointer. Found with: Coverty Prevent(tm) CID: 4319 MFC after: 1 month
2010-06-04 14:38:24 +00:00
struct in6_addr *ip6 = NULL;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
int ip6s = 0;
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
int descend, 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
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
xp = malloc(sizeof(*xp), M_TEMP, M_WAITOK);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
pr = req->td->td_ucred->cr_prison;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = 0;
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
sx_slock(&allprison_lock);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
FOREACH_PRISON_DESCENDANT(pr, cpr, descend) {
#if defined(INET) || defined(INET6)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
again:
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
mtx_lock(&cpr->pr_mtx);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (cpr->pr_ip4s > 0) {
if (ip4s < cpr->pr_ip4s) {
ip4s = cpr->pr_ip4s;
mtx_unlock(&cpr->pr_mtx);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
ip4 = realloc(ip4, ip4s *
sizeof(struct in_addr), M_TEMP, M_WAITOK);
goto again;
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
bcopy(cpr->pr_ip4, ip4,
cpr->pr_ip4s * sizeof(struct in_addr));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
#ifdef INET6
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (cpr->pr_ip6s > 0) {
if (ip6s < cpr->pr_ip6s) {
ip6s = cpr->pr_ip6s;
mtx_unlock(&cpr->pr_mtx);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
ip6 = realloc(ip6, ip6s *
sizeof(struct in6_addr), M_TEMP, M_WAITOK);
goto again;
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
bcopy(cpr->pr_ip6, ip6,
cpr->pr_ip6s * sizeof(struct in6_addr));
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
bzero(xp, sizeof(*xp));
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
xp->pr_version = XPRISON_VERSION;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
xp->pr_id = cpr->pr_id;
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
xp->pr_state = cpr->pr_state;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
strlcpy(xp->pr_path, prison_path(pr, cpr), sizeof(xp->pr_path));
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
strlcpy(xp->pr_host, cpr->pr_hostname, sizeof(xp->pr_host));
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
strlcpy(xp->pr_name, prison_name(pr, cpr), sizeof(xp->pr_name));
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
xp->pr_ip4s = cpr->pr_ip4s;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
#ifdef INET6
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
xp->pr_ip6s = cpr->pr_ip6s;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
mtx_unlock(&cpr->pr_mtx);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
error = SYSCTL_OUT(req, xp, sizeof(*xp));
if (error)
break;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (xp->pr_ip4s > 0) {
error = SYSCTL_OUT(req, ip4,
xp->pr_ip4s * sizeof(struct in_addr));
if (error)
break;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
#endif
#ifdef INET6
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
if (xp->pr_ip6s > 0) {
error = SYSCTL_OUT(req, ip6,
xp->pr_ip6s * sizeof(struct in6_addr));
if (error)
break;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
#endif
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
}
Implement functionality I called 'jail services'. It may be used for external modules to attach some data to jail's in-kernel structure. - Change allprison_mtx mutex to allprison_sx sx(9) lock. We will need to call external functions while holding this lock, which may want to allocate memory. Make use of the fact that this is shared-exclusive lock and use shared version when possible. - Implement the following functions: prison_service_register() - registers a service that wants to be noticed when a jail is created and destroyed prison_service_deregister() - deregisters service prison_service_data_add() - adds service-specific data to the jail structure prison_service_data_get() - takes service-specific data from the jail structure prison_service_data_del() - removes service-specific data from the jail structure Reviewed by: rwatson
2007-04-05 23:19:13 +00:00
sx_sunlock(&allprison_lock);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
free(xp, M_TEMP);
#ifdef INET
free(ip4, M_TEMP);
#endif
#ifdef INET6
free(ip6, M_TEMP);
#endif
Minor simplification.
2007-03-09 05:22:10 +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
}
Mark most often used sysctl's as MPSAFE. After running a `make buildkernel', I noticed most of the Giant locks in sysctl are only caused by a very small amount of sysctl's: - sysctl.name2oid. This one is locked by SYSCTL_LOCK, just like sysctl.oidfmt. - kern.ident, kern.osrelease, kern.version, etc. These are just constant strings. - kern.arandom, used by the stack protector. It is already protected by arc4_mtx. I also saw the following sysctl's show up. Not as often as the ones above, but still quite often: - security.jail.jailed. Also mark security.jail.list as MPSAFE. They don't need locking or already use allprison_lock. - kern.devname, used by devname(3), ttyname(3), etc. This seems to reduce Giant locking inside sysctl by ~75% in my primitive test setup.
2009-01-28 19:58:05 +00:00
SYSCTL_OID(_security_jail, OID_AUTO, list,
CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_MPSAFE, 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);
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
Mark most often used sysctl's as MPSAFE. After running a `make buildkernel', I noticed most of the Giant locks in sysctl are only caused by a very small amount of sysctl's: - sysctl.name2oid. This one is locked by SYSCTL_LOCK, just like sysctl.oidfmt. - kern.ident, kern.osrelease, kern.version, etc. These are just constant strings. - kern.arandom, used by the stack protector. It is already protected by arc4_mtx. I also saw the following sysctl's show up. Not as often as the ones above, but still quite often: - security.jail.jailed. Also mark security.jail.list as MPSAFE. They don't need locking or already use allprison_lock. - kern.devname, used by devname(3), ttyname(3), etc. This seems to reduce Giant locking inside sysctl by ~75% in my primitive test setup.
2009-01-28 19:58:05 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, jailed,
CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0,
sysctl_jail_jailed, "I", "Process in jail?");
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
Refine the "nojail" rc keyword, adding "nojailvnet" for files that don't apply to most jails but do apply to vnet jails. This includes adding a new sysctl "security.jail.vnet" to identify vnet jails. PR: conf/149050 Submitted by: mdodd MFC after: 3 days
2013-05-19 04:10:34 +00:00
static int
sysctl_jail_vnet(SYSCTL_HANDLER_ARGS)
{
int error, havevnet;
#ifdef VIMAGE
struct ucred *cred = req->td->td_ucred;
havevnet = jailed(cred) && prison_owns_vnet(cred);
#else
havevnet = 0;
#endif
error = SYSCTL_OUT(req, &havevnet, sizeof(havevnet));
return (error);
}
SYSCTL_PROC(_security_jail, OID_AUTO, vnet,
CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0,
Try to be consistent and spell "vnet" lower case like all the other options (and as we do on command line). Sponsored by: iXsystems, Inc.
2018-05-24 15:31:05 +00:00
sysctl_jail_vnet, "I", "Jail owns vnet?");
Refine the "nojail" rc keyword, adding "nojailvnet" for files that don't apply to most jails but do apply to vnet jails. This includes adding a new sysctl "security.jail.vnet" to identify vnet jails. PR: conf/149050 Submitted by: mdodd MFC after: 3 days
2013-05-19 04:10:34 +00:00
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#if defined(INET) || defined(INET6)
Fix some overflow errors: a signed allocation and an insufficiant array size. Reported by: pho Tested by: pho Approved by: bz (mentor)
2009-06-09 22:09:29 +00:00
SYSCTL_UINT(_security_jail, OID_AUTO, jail_max_af_ips, CTLFLAG_RW,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
&jail_max_af_ips, 0,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Number of IP addresses a jail may have at most per address family (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
/*
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
* Default parameters for jail(2) compatibility. For historical reasons,
* the sysctl names have varying similarity to the parameter names. Prisons
* just see their own parameters, and can't change them.
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
*/
static int
sysctl_jail_default_allow(SYSCTL_HANDLER_ARGS)
{
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
int error, i;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/* Get the current flag value, and convert it to a boolean. */
jail: Consistently handle the pr_allow bitmask Return a boolean (i.e. 0 or 1) from prison_allow, instead of the flag value itself, which is what sysctl expects. Add prison_set_allow(), which can set or clear a permission bit, and propagates cleared bits down to child jails. Use prison_allow() and prison_set_allow() in the various jail.allow.* sysctls, and others that depend on thoe permissions. Add locking around checking both pr_allow and pr_enforce_statfs in prison_priv_check().
2020-12-27 04:25:02 +00:00
if (req->td->td_ucred->cr_prison == &prison0) {
mtx_lock(&prison0.pr_mtx);
i = (jail_default_allow & arg2) != 0;
mtx_unlock(&prison0.pr_mtx);
} else
i = prison_allow(req->td->td_ucred, arg2);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (arg1 != NULL)
i = !i;
error = sysctl_handle_int(oidp, &i, 0, req);
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
if (error || !req->newptr)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return (error);
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
i = i ? arg2 : 0;
if (arg1 != NULL)
i ^= arg2;
/*
* The sysctls don't have CTLFLAGS_PRISON, so assume prison0
* for writing.
*/
mtx_lock(&prison0.pr_mtx);
jail_default_allow = (jail_default_allow & ~arg2) | i;
mtx_unlock(&prison0.pr_mtx);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return (0);
}
SYSCTL_PROC(_security_jail, OID_AUTO, set_hostname_allowed,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
NULL, PR_ALLOW_SET_HOSTNAME, sysctl_jail_default_allow, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Processes in jail can set their hostnames (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, socket_unixiproute_only,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
(void *)1, PR_ALLOW_SOCKET_AF, sysctl_jail_default_allow, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Processes in jail are limited to creating UNIX/IP/route sockets only (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, sysvipc_allowed,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
NULL, PR_ALLOW_SYSVIPC, sysctl_jail_default_allow, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Processes in jail can use System V IPC primitives (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, allow_raw_sockets,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
NULL, PR_ALLOW_RAW_SOCKETS, sysctl_jail_default_allow, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Prison root can create raw sockets (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, chflags_allowed,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
NULL, PR_ALLOW_CHFLAGS, sysctl_jail_default_allow, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Processes in jail can alter system file flags (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, mount_allowed,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
NULL, PR_ALLOW_MOUNT, sysctl_jail_default_allow, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Processes in jail can mount/unmount jail-friendly file systems (deprecated)");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
static int
sysctl_jail_default_level(SYSCTL_HANDLER_ARGS)
{
struct prison *pr;
int level, error;
pr = req->td->td_ucred->cr_prison;
level = (pr == &prison0) ? *(int *)arg1 : *(int *)((char *)pr + arg2);
error = sysctl_handle_int(oidp, &level, 0, req);
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
if (error || !req->newptr)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return (error);
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
*(int *)arg1 = level;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return (0);
}
SYSCTL_PROC(_security_jail, OID_AUTO, enforce_statfs,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
&jail_default_enforce_statfs, offsetof(struct prison, pr_enforce_statfs),
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
sysctl_jail_default_level, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Processes in jail cannot see all mounted file systems (deprecated)");
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
SYSCTL_PROC(_security_jail, OID_AUTO, devfs_ruleset,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_MPSAFE,
&jail_default_devfs_rsnum, offsetof(struct prison, pr_devfs_rsnum),
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
sysctl_jail_default_level, "I",
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
"Ruleset for the devfs filesystem in jail (deprecated)");
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
* Nodes to describe jail parameters. Maximum length of string parameters
* is returned in the string itself, and the other parameters exist merely
* to make themselves and their types known.
*/
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
SYSCTL_NODE(_security_jail, OID_AUTO, param, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
"Jail parameters");
int
sysctl_jail_param(SYSCTL_HANDLER_ARGS)
{
int i;
long l;
size_t s;
char numbuf[12];
switch (oidp->oid_kind & CTLTYPE)
{
case CTLTYPE_LONG:
case CTLTYPE_ULONG:
l = 0;
#ifdef SCTL_MASK32
if (!(req->flags & SCTL_MASK32))
#endif
return (SYSCTL_OUT(req, &l, sizeof(l)));
case CTLTYPE_INT:
case CTLTYPE_UINT:
i = 0;
return (SYSCTL_OUT(req, &i, sizeof(i)));
case CTLTYPE_STRING:
- Merge changes to the base system to support OFED. These include a wider arg2 for sysctl, updates to vlan code, IFT_INFINIBAND, and other miscellaneous small features.
2011-03-21 09:40:01 +00:00
snprintf(numbuf, sizeof(numbuf), "%jd", (intmax_t)arg2);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
return
(sysctl_handle_string(oidp, numbuf, sizeof(numbuf), req));
case CTLTYPE_STRUCT:
s = (size_t)arg2;
return (SYSCTL_OUT(req, &s, sizeof(s)));
}
return (0);
}
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
/*
* CTLFLAG_RDTUN in the following indicates jail parameters that can be set at
* jail creation time but cannot be changed in an existing jail.
*/
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM(, jid, CTLTYPE_INT | CTLFLAG_RDTUN, "I", "Jail ID");
SYSCTL_JAIL_PARAM(, parent, CTLTYPE_INT | CTLFLAG_RD, "I", "Jail parent ID");
SYSCTL_JAIL_PARAM_STRING(, name, CTLFLAG_RW, MAXHOSTNAMELEN, "Jail name");
SYSCTL_JAIL_PARAM_STRING(, path, CTLFLAG_RDTUN, MAXPATHLEN, "Jail root path");
SYSCTL_JAIL_PARAM(, securelevel, CTLTYPE_INT | CTLFLAG_RW,
"I", "Jail secure level");
Fix style nits.
2020-11-18 20:59:58 +00:00
SYSCTL_JAIL_PARAM(, osreldate, CTLTYPE_INT | CTLFLAG_RDTUN, "I",
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
"Jail value for kern.osreldate and uname -K");
Fix style nits.
2020-11-18 20:59:58 +00:00
SYSCTL_JAIL_PARAM_STRING(, osrelease, CTLFLAG_RDTUN, OSRELEASELEN,
Allow the kern.osrelease and kern.osreldate sysctl values to be set in a jail's creation parameters. This allows the kernel version to be reliably spoofed within the jail whether examined directly with sysctl or indirectly with the uname -r and -K options. The values can only be set at jail creation time, to eliminate the need for any locking when accessing the values via sysctl. The overridden values are inherited by nested jails (unless the config for the nested jails also overrides the values). There is no sanity or range checking, other than disallowing an empty release string or a zero release date, by design. The system administrator is trusted to set sane values. Setting values that are newer than the actual running kernel will likely cause compatibility problems. Differential Revision: https://reviews.freebsd.org/D1948 Relnotes: yes
2015-02-27 16:28:55 +00:00
"Jail value for kern.osrelease and uname -r");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM(, enforce_statfs, CTLTYPE_INT | CTLFLAG_RW,
"I", "Jail cannot see all mounted file systems");
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
SYSCTL_JAIL_PARAM(, devfs_ruleset, CTLTYPE_INT | CTLFLAG_RW,
"I", "Ruleset for in-jail devfs mounts");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM(, persist, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail persistence");
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#ifdef VIMAGE
SYSCTL_JAIL_PARAM(, vnet, CTLTYPE_INT | CTLFLAG_RDTUN,
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
"E,jailsys", "Virtual network stack");
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#endif
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM(, dying, CTLTYPE_INT | CTLFLAG_RD,
"B", "Jail is in the process of shutting down");
Add a limit for child jails via the "children.cur" and "children.max" parameters. This replaces the simple "allow.jails" permission. Approved by: bz (mentor)
2009-06-23 20:35:51 +00:00
SYSCTL_JAIL_PARAM_NODE(children, "Number of child jails");
SYSCTL_JAIL_PARAM(_children, cur, CTLTYPE_INT | CTLFLAG_RD,
"I", "Current number of child jails");
SYSCTL_JAIL_PARAM(_children, max, CTLTYPE_INT | CTLFLAG_RW,
"I", "Maximum number of child jails");
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
SYSCTL_JAIL_PARAM_SYS_NODE(host, CTLFLAG_RW, "Jail host info");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM_STRING(_host, hostname, CTLFLAG_RW, MAXHOSTNAMELEN,
"Jail hostname");
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
SYSCTL_JAIL_PARAM_STRING(_host, domainname, CTLFLAG_RW, MAXHOSTNAMELEN,
"Jail NIS domainname");
SYSCTL_JAIL_PARAM_STRING(_host, hostuuid, CTLFLAG_RW, HOSTUUIDLEN,
"Jail host UUID");
SYSCTL_JAIL_PARAM(_host, hostid, CTLTYPE_ULONG | CTLFLAG_RW,
"LU", "Jail host ID");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM_NODE(cpuset, "Jail cpuset");
SYSCTL_JAIL_PARAM(_cpuset, id, CTLTYPE_INT | CTLFLAG_RD, "I", "Jail cpuset ID");
#ifdef INET
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
SYSCTL_JAIL_PARAM_SYS_NODE(ip4, CTLFLAG_RDTUN,
"Jail IPv4 address virtualization");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM_STRUCT(_ip4, addr, CTLFLAG_RW, sizeof(struct in_addr),
"S,in_addr,a", "Jail IPv4 addresses");
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
SYSCTL_JAIL_PARAM(_ip4, saddrsel, CTLTYPE_INT | CTLFLAG_RW,
"B", "Do (not) use IPv4 source address selection rather than the "
"primary jail IPv4 address.");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
#ifdef INET6
Remove a LOR, where the the sleepable allprison_lock was being obtained in prison_equal_ip4/6 while an inp mutex was held. Locking allprison_lock can be avoided by making a restriction on the IP addresses associated with jails: Don't allow the "ip4" and "ip6" parameters to be changed after a jail is created. Setting the "ip4.addr" and "ip6.addr" parameters is allowed, but only if the jail was already created with either ip4/6=new or ip4/6=disable. With this restriction, the prison flags in question (PR_IP4_USER and PR_IP6_USER) become read-only and can be checked without locking. This also allows the simplification of a messy code path that was needed to handle an existing prison gaining an IP address list. PR: kern/136899 Reported by: Dirk Meyer Approved by: re (kib), bz (mentor)
2009-07-30 14:28:56 +00:00
SYSCTL_JAIL_PARAM_SYS_NODE(ip6, CTLFLAG_RDTUN,
"Jail IPv6 address virtualization");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
SYSCTL_JAIL_PARAM_STRUCT(_ip6, addr, CTLFLAG_RW, sizeof(struct in6_addr),
"S,in6_addr,a", "Jail IPv6 addresses");
Add ip4.saddrsel/ip4.nosaddrsel (and equivalent for ip6) to control whether to use source address selection (default) or the primary jail address for unbound outgoing connections. This is intended to be used by people upgrading from single-IP jails to multi-IP jails but not having to change firewall rules, application ACLs, ... but to force their connections (unless otherwise changed) to the primry jail IP they had been used for years, as well as for people prefering to implement similar policies. Note that for IPv6, if configured incorrectly, this might lead to scope violations, which single-IPv6 jails could as well, as by the design of jails. [1] Reviewed by: jamie, hrs (ipv6 part) Pointed out by: hrs [1] MFC After: 2 weeks Asked for by: Jase Thew (bazerka beardz.net)
2010-01-17 12:57:11 +00:00
SYSCTL_JAIL_PARAM(_ip6, saddrsel, CTLTYPE_INT | CTLFLAG_RW,
"B", "Do (not) use IPv6 source address selection rather than the "
"primary jail IPv6 address.");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
#endif
SYSCTL_JAIL_PARAM_NODE(allow, "Jail permission flags");
SYSCTL_JAIL_PARAM(_allow, set_hostname, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may set hostname");
SYSCTL_JAIL_PARAM(_allow, sysvipc, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may use SYSV IPC");
SYSCTL_JAIL_PARAM(_allow, raw_sockets, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may create raw sockets");
SYSCTL_JAIL_PARAM(_allow, chflags, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may alter system file flags");
SYSCTL_JAIL_PARAM(_allow, quotas, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may set file quotas");
SYSCTL_JAIL_PARAM(_allow, socket_af, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may create sockets other than just UNIX/IPv4/IPv6/route");
Add allow.mlock to jail parameters It allows locking or unlocking physical pages in memory within a jail This allows running elasticsearch with "bootstrap.memory_lock" inside a jail Reviewed by: jamie@ Differential Revision: https://reviews.freebsd.org/D16342
2018-07-29 12:41:56 +00:00
SYSCTL_JAIL_PARAM(_allow, mlock, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may lock (unlock) physical pages in memory");
Jails: Optionally prevent jailed root from binding to privileged ports You may now optionally specify allow.noreserved_ports to prevent root inside a jail from using privileged ports (less than 1024) PR: 217728 Submitted by: Matt Miller <mattm916@pulsar.neomailbox.ch> Reviewed by: jamie, cem, smh Relnotes: yes Differential Revision: https://reviews.freebsd.org/D10202
2017-06-06 02:15:00 +00:00
SYSCTL_JAIL_PARAM(_allow, reserved_ports, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may bind sockets to reserved ports");
Add a new jail permission, allow.read_msgbuf. When true, jailed processes can see the dmesg buffer (this is the current behavior). When false (the new default), dmesg will be unavailable to jailed users, whether root or not. The security.bsd.unprivileged_read_msgbuf sysctl still works as before, controlling system-wide whether non-root users can see the buffer. PR: 211580 Submitted by: bz Approved by: re@ (kib@) MFC after: 3 days
2018-10-17 16:11:43 +00:00
SYSCTL_JAIL_PARAM(_allow, read_msgbuf, CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may read the kernel message buffer");
In hardened systems, where the security.bsd.unprivileged_proc_debug sysctl node is set, allow setting security.bsd.unprivileged_proc_debug per-jail. In part, this is needed to create jails in which the Address Sanitizer (ASAN) fully works as ASAN utilizes libkvm to inspect the virtual address space. Instead of having to allow unprivileged process debugging for the entire system, allow setting it on a per-jail basis. The sysctl node is still security.bsd.unprivileged_proc_debug and the jail(8) param is allow.unprivileged_proc_debug. The sysctl code is now a sysctl proc rather than a sysctl int. This allows us to determine setting the flag for the corresponding jail (or prison0). As part of the change, the dynamic allow.* API needed to be modified to take into account pr_allow flags which may now be disabled in prison0. This prevents conflicts with new pr_allow flags (like that of vmm(4)) that are added (and removed) dynamically. Also teach the jail creation KPI to allow differences for certain pr_allow flags between the parent and child jail. This can happen when unprivileged process debugging is disabled in the parent prison, but enabled in the child. Submitted by: Shawn Webb <lattera at gmail.com> Obtained from: HardenedBSD (45b3625edba0f73b3e3890b1ec3d0d1e95fd47e1, deba0b5078cef0faae43cbdafed3035b16587afc, ab21eeb3b4c72f2500987c96ff603ccf3b6e7de8) Relnotes: yes Sponsored by: HardenedBSD and G2, Inc Differential Revision: https://reviews.freebsd.org/D18319
2018-11-27 17:51:50 +00:00
SYSCTL_JAIL_PARAM(_allow, unprivileged_proc_debug, CTLTYPE_INT | CTLFLAG_RW,
"B", "Unprivileged processes may use process debugging facilities");
jail: introduce per jail suser_enabled setting The suser_enable sysctl allows to remove a privileged rights from uid 0. This change introduce per jail setting which allow to make root a normal user. Reviewed by: jamie Previous version reviewed by: kevans, emaste, markj, me_igalic.co Discussed with: pjd Differential Revision: https://reviews.freebsd.org/D27128
2020-11-18 21:07:08 +00:00
SYSCTL_JAIL_PARAM(_allow, suser, CTLTYPE_INT | CTLFLAG_RW,
"B", "Processes in jail with uid 0 have privilege");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
SYSCTL_JAIL_PARAM_SUBNODE(allow, mount, "Jail mount/unmount permission flags");
SYSCTL_JAIL_PARAM(_allow_mount, , CTLTYPE_INT | CTLFLAG_RW,
"B", "Jail may mount/unmount jail-friendly file systems in general");
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
/*
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
* Add a dynamic parameter allow.<name>, or allow.<prefix>.<name>. Return
* its associated bit in the pr_allow bitmask, or zero if the parameter was
* not created.
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
*/
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
unsigned
prison_add_allow(const char *prefix, const char *name, const char *prefix_descr,
const char *descr)
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
{
struct bool_flags *bf;
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
struct sysctl_oid *parent;
char *allow_name, *allow_noname, *allowed;
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
#ifndef NO_SYSCTL_DESCR
char *descr_deprecated;
#endif
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
u_int allow_flag;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
if (prefix
? asprintf(&allow_name, M_PRISON, "allow.%s.%s", prefix, name)
< 0 ||
asprintf(&allow_noname, M_PRISON, "allow.%s.no%s", prefix, name)
< 0
: asprintf(&allow_name, M_PRISON, "allow.%s", name) < 0 ||
asprintf(&allow_noname, M_PRISON, "allow.no%s", name) < 0) {
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
free(allow_name, M_PRISON);
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
return 0;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
}
/*
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
* See if this parameter has already beed added, i.e. a module was
* previously loaded/unloaded.
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
*/
mtx_lock(&prison0.pr_mtx);
for (bf = pr_flag_allow;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
bf < pr_flag_allow + nitems(pr_flag_allow) &&
atomic_load_int(&bf->flag) != 0;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
bf++) {
if (strcmp(bf->name, allow_name) == 0) {
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
allow_flag = bf->flag;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
goto no_add;
}
}
/*
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
* Find a free bit in pr_allow_all, failing if there are none
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
* (which shouldn't happen as long as we keep track of how many
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
* potential dynamic flags exist).
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
*/
for (allow_flag = 1;; allow_flag <<= 1) {
if (allow_flag == 0)
goto no_add;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
if ((pr_allow_all & allow_flag) == 0)
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
break;
}
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
/* Note the parameter in the next open slot in pr_flag_allow. */
for (bf = pr_flag_allow; ; bf++) {
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
if (bf == pr_flag_allow + nitems(pr_flag_allow)) {
/* This should never happen, but is not fatal. */
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
allow_flag = 0;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
goto no_add;
}
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
if (atomic_load_int(&bf->flag) == 0)
break;
}
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
bf->name = allow_name;
bf->noname = allow_noname;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
pr_allow_all |= allow_flag;
/*
* prison0 always has permission for the new parameter.
* Other jails must have it granted to them.
*/
prison0.pr_allow |= allow_flag;
/* The flag indicates a valid entry, so make sure it is set last. */
atomic_store_rel_int(&bf->flag, allow_flag);
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
mtx_unlock(&prison0.pr_mtx);
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
/*
* Create sysctls for the paramter, and the back-compat global
* permission.
*/
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
parent = prefix
? SYSCTL_ADD_NODE(NULL,
SYSCTL_CHILDREN(&sysctl___security_jail_param_allow),
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
OID_AUTO, prefix, CTLFLAG_MPSAFE, 0, prefix_descr)
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
: &sysctl___security_jail_param_allow;
(void)SYSCTL_ADD_PROC(NULL, SYSCTL_CHILDREN(parent), OID_AUTO,
name, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
NULL, 0, sysctl_jail_param, "B", descr);
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
if ((prefix
? asprintf(&allowed, M_TEMP, "%s_%s_allowed", prefix, name)
: asprintf(&allowed, M_TEMP, "%s_allowed", name)) >= 0) {
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
#ifndef NO_SYSCTL_DESCR
(void)asprintf(&descr_deprecated, M_TEMP, "%s (deprecated)",
descr);
#endif
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
(void)SYSCTL_ADD_PROC(NULL,
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
SYSCTL_CHILDREN(&sysctl___security_jail), OID_AUTO, allowed,
Revert r337922, except for some documention-only bits. This needs to wait until user is changed to stop using jail(2). Differential Revision: D14791
2018-08-16 19:09:43 +00:00
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, NULL, allow_flag,
sysctl_jail_default_allow, "I", descr_deprecated);
#ifndef NO_SYSCTL_DESCR
free(descr_deprecated, M_TEMP);
#endif
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
free(allowed, M_TEMP);
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
}
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
return allow_flag;
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
no_add:
mtx_unlock(&prison0.pr_mtx);
free(allow_name, M_PRISON);
free(allow_noname, M_PRISON);
Change prison_add_vfs() to the more generic prison_add_allow(), which can add any dynamic allow.* or allow.*.* parameter. Also keep prison_add_vfs() as a wrapper. Differential Revision: D16146
2018-07-06 18:50:22 +00:00
return allow_flag;
}
/*
* The VFS system will register jail-aware filesystems here. They each get
* a parameter allow.mount.xxxfs and a flag to check when a jailed user
* attempts to mount.
*/
void
prison_add_vfs(struct vfsconf *vfsp)
{
#ifdef NO_SYSCTL_DESCR
vfsp->vfc_prison_flag = prison_add_allow("mount", vfsp->vfc_name,
NULL, NULL);
#else
char *descr;
(void)asprintf(&descr, M_TEMP, "Jail may mount the %s file system",
vfsp->vfc_name);
vfsp->vfc_prison_flag = prison_add_allow("mount", vfsp->vfc_name,
NULL, descr);
free(descr, M_TEMP);
#endif
Make it easier for filesystems to count themselves as jail-enabled, by doing most of the work in a new function prison_add_vfs in kern_jail.c Now a jail-enabled filesystem need only mark itself with VFCF_JAIL, and the rest is taken care of. This includes adding a jail parameter like allow.mount.foofs, and a sysctl like security.jail.mount_foofs_allowed. Both of these used to be a static list of known filesystems, with predefined permission bits. Reviewed by: kib Differential Revision: D14681
2018-05-04 20:54:27 +00:00
}
To improve control over the use of mount(8) inside a jail(8), introduce a new jail parameter node with the following parameters: allow.mount.devfs: allow mounting the devfs filesystem inside a jail allow.mount.nullfs: allow mounting the nullfs filesystem inside a jail Both parameters are disabled by default (equals the behavior before devfs and nullfs in jails). Administrators have to explicitly allow mounting devfs and nullfs for each jail. The value "-1" of the devfs_ruleset parameter is removed in favor of the new allow setting. Reviewed by: jamie Suggested by: pjd MFC after: 2 weeks
2012-02-23 18:51:24 +00:00
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
#ifdef RACCT
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
void
prison_racct_foreach(void (*callback)(struct racct *racct,
Speed up rctl operation with large rulesets, by holding the lock during iteration instead of relocking it for each traversed rule. Reviewed by: mjg@ MFC after: 1 month Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D4110
2015-11-15 12:10:51 +00:00
void *arg2, void *arg3), void (*pre)(void), void (*post)(void),
void *arg2, void *arg3)
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
{
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
struct prison_racct *prr;
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
sx_slock(&allprison_lock);
Speed up rctl operation with large rulesets, by holding the lock during iteration instead of relocking it for each traversed rule. Reviewed by: mjg@ MFC after: 1 month Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D4110
2015-11-15 12:10:51 +00:00
if (pre != NULL)
(pre)();
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
LIST_FOREACH(prr, &allprison_racct, prr_next)
(callback)(prr->prr_racct, arg2, arg3);
Speed up rctl operation with large rulesets, by holding the lock during iteration instead of relocking it for each traversed rule. Reviewed by: mjg@ MFC after: 1 month Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D4110
2015-11-15 12:10:51 +00:00
if (post != NULL)
(post)();
Add racct. It's an API to keep per-process, per-jail, per-loginclass and per-loginclass resource accounting information, to be used by the new resource limits code. It's connected to the build, but the code that actually calls the new functions will come later. Sponsored by: The FreeBSD Foundation Reviewed by: kib (earlier version)
2011-03-29 17:47:25 +00:00
sx_sunlock(&allprison_lock);
}
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
static struct prison_racct *
prison_racct_find_locked(const char *name)
{
struct prison_racct *prr;
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
sx_assert(&allprison_lock, SA_XLOCKED);
if (name[0] == '\0' || strlen(name) >= MAXHOSTNAMELEN)
return (NULL);
LIST_FOREACH(prr, &allprison_racct, prr_next) {
if (strcmp(name, prr->prr_name) != 0)
continue;
/* Found prison_racct with a matching name? */
prison_racct_hold(prr);
return (prr);
}
/* Add new prison_racct. */
prr = malloc(sizeof(*prr), M_PRISON_RACCT, M_ZERO | M_WAITOK);
racct_create(&prr->prr_racct);
strcpy(prr->prr_name, name);
refcount_init(&prr->prr_refcount, 1);
LIST_INSERT_HEAD(&allprison_racct, prr, prr_next);
return (prr);
}
struct prison_racct *
prison_racct_find(const char *name)
{
struct prison_racct *prr;
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
sx_xlock(&allprison_lock);
prr = prison_racct_find_locked(name);
sx_xunlock(&allprison_lock);
return (prr);
}
void
prison_racct_hold(struct prison_racct *prr)
{
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
refcount_acquire(&prr->prr_refcount);
}
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
static void
prison_racct_free_locked(struct prison_racct *prr)
{
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
sx_assert(&allprison_lock, SA_XLOCKED);
if (refcount_release(&prr->prr_refcount)) {
racct_destroy(&prr->prr_racct);
LIST_REMOVE(prr, prr_next);
free(prr, M_PRISON_RACCT);
}
}
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
void
prison_racct_free(struct prison_racct *prr)
{
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
sx_assert(&allprison_lock, SA_UNLOCKED);
Replace hand-rolled unrefs if > 1 with refcount_release_if_not_last Sponsored by: The FreeBSD Foundation
2018-12-07 16:11:45 +00:00
if (refcount_release_if_not_last(&prr->prr_refcount))
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
return;
sx_xlock(&allprison_lock);
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
prison_racct_free_locked(prr);
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
sx_xunlock(&allprison_lock);
}
static void
prison_racct_attach(struct prison *pr)
{
struct prison_racct *prr;
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
sx_assert(&allprison_lock, SA_XLOCKED);
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
prr = prison_racct_find_locked(pr->pr_name);
KASSERT(prr != NULL, ("cannot find prison_racct"));
pr->pr_prison_racct = prr;
}
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
/*
* Handle jail renaming. From the racct point of view, renaming means
* moving from one prison_racct to another.
*/
static void
prison_racct_modify(struct prison *pr)
{
Silence warnings about unused variables when RACCT is defined but RCTL is not. Reported by: Dries Michiels <driesm.michiels@gmail.com> Sponsored by: The FreeBSD Foundation MFC after: 3 days
2018-07-05 13:37:31 +00:00
#ifdef RCTL
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
struct proc *p;
struct ucred *cred;
Silence warnings about unused variables when RACCT is defined but RCTL is not. Reported by: Dries Michiels <driesm.michiels@gmail.com> Sponsored by: The FreeBSD Foundation MFC after: 3 days
2018-07-05 13:37:31 +00:00
#endif
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
struct prison_racct *oldprr;
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
sx_slock(&allproc_lock);
sx_xlock(&allprison_lock);
Don't leak locks in prison_racct_modify(). Submitted by: Mateusz Guzik <mjguzik at gmail dot com> MFC after: 2 weeks
2012-05-22 17:30:02 +00:00
if (strcmp(pr->pr_name, pr->pr_prison_racct->prr_name) == 0) {
sx_xunlock(&allprison_lock);
sx_sunlock(&allproc_lock);
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
return;
Don't leak locks in prison_racct_modify(). Submitted by: Mateusz Guzik <mjguzik at gmail dot com> MFC after: 2 weeks
2012-05-22 17:30:02 +00:00
}
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
oldprr = pr->pr_prison_racct;
pr->pr_prison_racct = NULL;
prison_racct_attach(pr);
/*
* Move resource utilisation records.
*/
racct_move(pr->pr_prison_racct->prr_racct, oldprr->prr_racct);
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
#ifdef RCTL
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
/*
* Force rctl to reattach rules to processes.
*/
FOREACH_PROC_IN_SYSTEM(p) {
PROC_LOCK(p);
cred = crhold(p->p_ucred);
PROC_UNLOCK(p);
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
rctl_proc_ucred_changed(p, cred);
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
crfree(cred);
}
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
#endif
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
sx_sunlock(&allproc_lock);
prison_racct_free_locked(oldprr);
sx_xunlock(&allprison_lock);
}
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
static void
prison_racct_detach(struct prison *pr)
{
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
Add kern.racct.enable tunable and RACCT_DISABLED config option. The point of this is to be able to add RACCT (with RACCT_DISABLED) to GENERIC, to avoid having to rebuild the kernel to use rctl(8). Differential Revision: https://reviews.freebsd.org/D2369 Reviewed by: kib@ MFC after: 1 month Relnotes: yes Sponsored by: The FreeBSD Foundation
2015-04-29 10:23:02 +00:00
ASSERT_RACCT_ENABLED();
Make racct and rctl correctly handle jail renaming. Previously they would continue using old name, the one jail was created with. PR: bin/165207
2012-03-06 11:05:50 +00:00
sx_assert(&allprison_lock, SA_UNLOCKED);
prison_racct_detach can be called for not fully initialized jail, so make it check that the jail has racct before doing anything PR: kern/174436 Reviewed by: trasz MFC after: 3 days
2012-12-18 18:34:36 +00:00
if (pr->pr_prison_racct == NULL)
return;
Change the way rctl interfaces with jails by introducing prison_racct structure, which acts as a proxy between them. This makes jail rules persistent, i.e. they can be added before jail gets created, and they don't disappear when the jail gets destroyed.
2011-05-03 07:32:58 +00:00
prison_racct_free(pr->pr_prison_racct);
pr->pr_prison_racct = NULL;
}
#endif /* RACCT */
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef DDB
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
static void
db_show_prison(struct prison *pr)
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
{
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
struct bool_flags *bf;
struct jailsys_flags *jsf;
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
#if defined(INET) || defined(INET6)
int ii;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
unsigned f;
Use inet_ntoa_r() instead of inet_ntoa() throughout the kernel inet_ntoa() cannot be used safely in a multithreaded environment because it uses a static local buffer. Instead, use inet_ntoa_r() with a buffer on the caller's stack. Suggested by: glebius, emaste Reviewed by: gnn MFC after: 2 weeks Sponsored by: Dell EMC Differential Revision: https://reviews.freebsd.org/D9625
2017-02-16 20:47:41 +00:00
#ifdef INET
char ip4buf[INET_ADDRSTRLEN];
#endif
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET6
char ip6buf[INET6_ADDRSTRLEN];
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf("prison %p:\n", pr);
db_printf(" jid = %d\n", pr->pr_id);
db_printf(" name = %s\n", pr->pr_name);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
db_printf(" parent = %p\n", pr->pr_parent);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf(" ref = %d\n", pr->pr_ref);
db_printf(" uref = %d\n", pr->pr_uref);
jail: Add pr_state to struct prison Rather that using references (pr_ref and pr_uref) to deduce the state of a prison, keep track of its state explicitly. A prison is either "invalid" (pr_ref == 0), "alive" (pr_uref > 0) or "dying" (pr_uref == 0). State transitions are generally tied to the reference counts, but with some flexibility: a new prison is "invalid" even though it now starts with a reference, and jail_remove(2) sets the state to "dying" before the user reference count drops to zero (which was prviously accomplished via the PR_REMOVE flag). pr_state is protected by both the prison mutex and allprison_lock, so it has the same availablity guarantees as the reference counts do. Differential Revision: https://reviews.freebsd.org/D27876
2021-02-21 21:24:47 +00:00
db_printf(" state = %s\n",
pr->pr_state == PRISON_STATE_ALIVE ? "alive" :
pr->pr_state == PRISON_STATE_DYING ? "dying" :
"invalid");
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf(" path = %s\n", pr->pr_path);
db_printf(" cpuset = %d\n", pr->pr_cpuset
? pr->pr_cpuset->cs_id : -1);
Manage vnets via the jail system. If a jail is given the boolean parameter "vnet" when it is created, a new vnet instance will be created along with the jail. Networks interfaces can be moved between prisons with an ioctl similar to the one that moves them between vimages. For now vnets will co-exist under both jails and vimages, but soon struct vimage will be going away. Reviewed by: zec, julian Approved by: bz (mentor)
2009-06-15 18:59:29 +00:00
#ifdef VIMAGE
db_printf(" vnet = %p\n", pr->pr_vnet);
#endif
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf(" root = %p\n", pr->pr_root);
db_printf(" securelevel = %d\n", pr->pr_securelevel);
Add support for mounting devfs inside jails. A new jail(8) option "devfs_ruleset" defines the ruleset enforcement for mounting devfs inside jails. A value of -1 disables mounting devfs in jails, a value of zero means no restrictions. Nested jails can only have mounting devfs disabled or inherit parent's enforcement as jails are not allowed to view or manipulate devfs(8) rules. Utilizes new functions introduced in r231265. Reviewed by: jamie MFC after: 1 month
2012-02-09 10:22:08 +00:00
db_printf(" devfs_rsnum = %d\n", pr->pr_devfs_rsnum);
Change DDB show prison: - name some columns more closely to the user space variables, as we do for host.* or allow.* (in the listing) already. - print pr_childmax (children.max). - prefix hex values with 0x. MFC after: 3 weeks
2010-01-11 22:34:25 +00:00
db_printf(" children.max = %d\n", pr->pr_childmax);
db_printf(" children.cur = %d\n", pr->pr_childcount);
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
db_printf(" child = %p\n", LIST_FIRST(&pr->pr_children));
db_printf(" sibling = %p\n", LIST_NEXT(pr, pr_sibling));
Change DDB show prison: - name some columns more closely to the user space variables, as we do for host.* or allow.* (in the listing) already. - print pr_childmax (children.max). - prefix hex values with 0x. MFC after: 3 weeks
2010-01-11 22:34:25 +00:00
db_printf(" flags = 0x%x", pr->pr_flags);
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_bool; bf < pr_flag_bool + nitems(pr_flag_bool); bf++)
if (pr->pr_flags & bf->flag)
db_printf(" %s", bf->name);
for (jsf = pr_flag_jailsys;
jsf < pr_flag_jailsys + nitems(pr_flag_jailsys);
jsf++) {
f = pr->pr_flags & (jsf->disable | jsf->new);
db_printf(" %-16s= %s\n", jsf->name,
(f != 0 && f == jsf->disable) ? "disable"
: (f == jsf->new) ? "new"
Some jail parameters (in particular, "ip4" and "ip6" for IP address restrictions) were found to be inadequately described by a boolean. Define a new parameter type with three values (disable, new, inherit) to handle these and future cases. Approved by: re (kib), bz (mentor) Discussed with: rwatson
2009-07-25 14:48:57 +00:00
: "inherit");
}
Change DDB show prison: - name some columns more closely to the user space variables, as we do for host.* or allow.* (in the listing) already. - print pr_childmax (children.max). - prefix hex values with 0x. MFC after: 3 weeks
2010-01-11 22:34:25 +00:00
db_printf(" allow = 0x%x", pr->pr_allow);
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
for (bf = pr_flag_allow;
jail: Fix lock-free access to dynamic pr.allow flags Use atomic access and a memory barrier to ensure that the flag parameter in pr_flag_allow is indeed set after the rest of the structure is valid. Simplify adding flag bits with pr_allow_all, a dynamic version of PR_ALLOW_ALL_STATIC.
2020-12-26 20:53:28 +00:00
bf < pr_flag_allow + nitems(pr_flag_allow) &&
atomic_load_int(&bf->flag) != 0;
Represent boolean jail options as an array of structures containing the flag and both the regular and "no" names, instead of two different string arrays whose indices need to match the flag's bit position. This makes them similar to the say "jailsys" options are represented. Loop through either kind of option array with a structure pointer rather then an integer index.
2018-03-20 23:08:42 +00:00
bf++)
if (pr->pr_allow & bf->flag)
db_printf(" %s", bf->name);
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf("\n");
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
db_printf(" enforce_statfs = %d\n", pr->pr_enforce_statfs);
Rename the host-related prison fields to be the same as the host.* parameters they represent, and the variables they replaced, instead of abbreviated versions of them. Approved by: bz (mentor)
2009-06-13 15:39:12 +00:00
db_printf(" host.hostname = %s\n", pr->pr_hostname);
db_printf(" host.domainname = %s\n", pr->pr_domainname);
db_printf(" host.hostuuid = %s\n", pr->pr_hostuuid);
Place hostnames and similar information fully under the prison system. The system hostname is now stored in prison0, and the global variable "hostname" has been removed, as has the hostname_mtx mutex. Jails may have their own host information, or they may inherit it from the parent/system. The proper way to read the hostname is via getcredhostname(), which will copy either the hostname associated with the passed cred, or the system hostname if you pass NULL. The system hostname can still be accessed directly (and without locking) at prison0.pr_host, but that should be avoided where possible. The "similar information" referred to is domainname, hostid, and hostuuid, which have also become prison parameters and had their associated global variables removed. Approved by: bz (mentor)
2009-05-29 21:27:12 +00:00
db_printf(" host.hostid = %lu\n", pr->pr_hostid);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#ifdef INET
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf(" ip4s = %d\n", pr->pr_ip4s);
for (ii = 0; ii < pr->pr_ip4s; ii++)
db_printf(" %s %s\n",
Change DDB show prison: - name some columns more closely to the user space variables, as we do for host.* or allow.* (in the listing) already. - print pr_childmax (children.max). - prefix hex values with 0x. MFC after: 3 weeks
2010-01-11 22:34:25 +00:00
ii == 0 ? "ip4.addr =" : " ",
Use inet_ntoa_r() instead of inet_ntoa() throughout the kernel inet_ntoa() cannot be used safely in a multithreaded environment because it uses a static local buffer. Instead, use inet_ntoa_r() with a buffer on the caller's stack. Suggested by: glebius, emaste Reviewed by: gnn MFC after: 2 weeks Sponsored by: Dell EMC Differential Revision: https://reviews.freebsd.org/D9625
2017-02-16 20:47:41 +00:00
inet_ntoa_r(pr->pr_ip4[ii], ip4buf));
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif
#ifdef INET6
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_printf(" ip6s = %d\n", pr->pr_ip6s);
for (ii = 0; ii < pr->pr_ip6s; ii++)
db_printf(" %s %s\n",
Change DDB show prison: - name some columns more closely to the user space variables, as we do for host.* or allow.* (in the listing) already. - print pr_childmax (children.max). - prefix hex values with 0x. MFC after: 3 weeks
2010-01-11 22:34:25 +00:00
ii == 0 ? "ip6.addr =" : " ",
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
ip6_sprintf(ip6buf, &pr->pr_ip6[ii]));
#endif
}
DB_SHOW_COMMAND(prison, db_show_prison_command)
{
struct prison *pr;
if (!have_addr) {
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
/*
* Show all prisons in the list, and prison0 which is not
* listed.
*/
db_show_prison(&prison0);
if (!db_pager_quit) {
TAILQ_FOREACH(pr, &allprison, pr_list) {
db_show_prison(pr);
if (db_pager_quit)
break;
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
}
return;
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (addr == 0)
pr = &prison0;
else {
/* Look for a prison with the ID and with references. */
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
TAILQ_FOREACH(pr, &allprison, pr_list)
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr->pr_id == addr && pr->pr_ref > 0)
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
break;
Add hierarchical jails. A jail may further virtualize its environment by creating a child jail, which is visible to that jail and to any parent jails. Child jails may be restricted more than their parents, but never less. Jail names reflect this hierarchy, being MIB-style dot-separated strings. Every thread now points to a jail, the default being prison0, which contains information about the physical system. Prison0's root directory is the same as rootvnode; its hostname is the same as the global hostname, and its securelevel replaces the global securelevel. Note that the variable "securelevel" has actually gone away, which should not cause any problems for code that properly uses securelevel_gt() and securelevel_ge(). Some jail-related permissions that were kept in global variables and set via sysctls are now per-jail settings. The sysctls still exist for backward compatibility, used only by the now-deprecated jail(2) system call. Approved by: bz (mentor)
2009-05-27 14:11:23 +00:00
if (pr == NULL)
/* Look again, without requiring a reference. */
TAILQ_FOREACH(pr, &allprison, pr_list)
if (pr->pr_id == addr)
break;
if (pr == NULL)
/* Assume address points to a valid prison. */
pr = (struct prison *)addr;
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
db_show_prison(pr);
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
}
Introduce the extensible jail framework, using the same "name=value" interface as nmount(2). Three new system calls are added: * jail_set, to create jails and change the parameters of existing jails. This replaces jail(2). * jail_get, to read the parameters of existing jails. This replaces the security.jail.list sysctl. * jail_remove to kill off a jail's processes and remove the jail. Most jail parameters may now be changed after creation, and jails may be set to exist without any attached processes. The current jail(2) system call still exists, though it is now a stub to jail_set(2). Approved by: bz (mentor)
2009-04-29 21:14:15 +00:00
MFp4: Bring in updated jail support from bz_jail branch. This enhances the current jail implementation to permit multiple addresses per jail. In addtion to IPv4, IPv6 is supported as well. Due to updated checks it is even possible to have jails without an IP address at all, which basically gives one a chroot with restricted process view, no networking,.. SCTP support was updated and supports IPv6 in jails as well. Cpuset support permits jails to be bound to specific processor sets after creation. Jails can have an unrestricted (no duplicate protection, etc.) name in addition to the hostname. The jail name cannot be changed from within a jail and is considered to be used for management purposes or as audit-token in the future. DDB 'show jails' command was added to aid debugging. Proper compat support permits 32bit jail binaries to be used on 64bit systems to manage jails. Also backward compatibility was preserved where possible: for jail v1 syscalls, as well as with user space management utilities. Both jail as well as prison version were updated for the new features. A gap was intentionally left as the intermediate versions had been used by various patches floating around the last years. Bump __FreeBSD_version for the afore mentioned and in kernel changes. Special thanks to: - Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches and Olivier Houchard (cognet) for initial single-IPv6 patches. - Jeff Roberson (jeff) and Randall Stewart (rrs) for their help, ideas and review on cpuset and SCTP support. - Robert Watson (rwatson) for lots and lots of help, discussions, suggestions and review of most of the patch at various stages. - John Baldwin (jhb) for his help. - Simon L. Nielsen (simon) as early adopter testing changes on cluster machines as well as all the testers and people who provided feedback the last months on freebsd-jail and other channels. - My employer, CK Software GmbH, for the support so I could work on this. Reviewed by: (see above) MFC after: 3 months (this is just so that I get the mail) X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
#endif /* DDB */
Reference in New Issue Copy Permalink