d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
d21258022e
freebsd-dev/sys/security/mac/mac_net.c

3988 lines
89 KiB
C
Raw Normal View History

Stubs for the TrustedBSD MAC system calls to permit TrustedBSD MAC userland code to operate on kernel's from the main tree. Not much in this file yet. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 02:04:05 +00:00
/*-
* Copyright (c) 1999, 2000, 2001, 2002 Robert N. M. Watson
* Copyright (c) 2001 Ilmar S. Habibulin
* Copyright (c) 2001, 2002 Networks Associates Technology, Inc.
* All rights reserved.
*
* This software was developed by Robert Watson and Ilmar Habibulin for the
* TrustedBSD Project.
*
* This software was developed for the FreeBSD Project in part by NAI Labs,
* the Security Research Division of Network Associates, Inc. under
* DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA
* CHATS research program.
*
* 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.
* 3. The names of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* 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.
*
* $FreeBSD$
*/
/*
* Developed by the TrustedBSD Project.
*
* Framework for extensible kernel access control. Kernel and userland
* interface to the framework, policy registration and composition.
*/
#include "opt_mac.h"
Remember to include "opt_devfs.h" so we get any relevant changes to NDEVFSINO before we include devfs.h. Spotted by: FlexeLint
2002-10-01 15:24:35 +00:00
#include "opt_devfs.h"
Include file cleanup; mac.h and malloc.h at one point had ordering relationship requirements, and no longer do. Reminded by: bde
2002-08-01 17:47:56 +00:00
Stubs for the TrustedBSD MAC system calls to permit TrustedBSD MAC userland code to operate on kernel's from the main tree. Not much in this file yet. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 02:04:05 +00:00
#include <sys/param.h>
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#include <sys/extattr.h>
#include <sys/kernel.h>
#include <sys/lock.h>
Include <sys/malloc.h> instead of depending on namespace pollution 2 layers deep in <sys/proc.h> or <sys/vnode.h>. Removed unused includes. Fixed some printf format errors (1 fatal on i386's; 1 fatal on alphas; 1 not fatal on any supported machine).
2002-09-05 07:02:43 +00:00
#include <sys/malloc.h>
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#include <sys/mutex.h>
#include <sys/mac.h>
Declare a module service "kernel_mac_support" when MAC support is enabled and the kernel provides the MAC registration and entry point service. Declare a dependency on that module service for any MAC module registered using mac_policy.h. For now, hard code the version as 1, but once we've come up with a versioning policy, we'll move to a #define of some sort. In the mean time, this will prevent loading a MAC module when 'options MAC' isn't present, which (due to a bug in the kernel linker) can result if the MAC module is preloaded via loader.conf. This particular evil recommended by: peter Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI LAbs
2002-08-12 02:00:21 +00:00
#include <sys/module.h>
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#include <sys/proc.h>
#include <sys/systm.h>
Stubs for the TrustedBSD MAC system calls to permit TrustedBSD MAC userland code to operate on kernel's from the main tree. Not much in this file yet. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 02:04:05 +00:00
#include <sys/sysproto.h>
#include <sys/sysent.h>
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#include <sys/vnode.h>
#include <sys/mount.h>
#include <sys/file.h>
#include <sys/namei.h>
#include <sys/socket.h>
#include <sys/pipe.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <sys/mac_policy.h>
#include <fs/devfs/devfs.h>
#include <net/bpfdesc.h>
#include <net/if.h>
#include <net/if_var.h>
#include <netinet/in.h>
#include <netinet/ip_var.h>
#ifdef MAC
Declare a module service "kernel_mac_support" when MAC support is enabled and the kernel provides the MAC registration and entry point service. Declare a dependency on that module service for any MAC module registered using mac_policy.h. For now, hard code the version as 1, but once we've come up with a versioning policy, we'll move to a #define of some sort. In the mean time, this will prevent loading a MAC module when 'options MAC' isn't present, which (due to a bug in the kernel linker) can result if the MAC module is preloaded via loader.conf. This particular evil recommended by: peter Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI LAbs
2002-08-12 02:00:21 +00:00
/*
* Declare that the kernel provides MAC support, version 1. This permits
* modules to refuse to be loaded if the necessary support isn't present,
* even if it's pre-boot.
*/
MODULE_VERSION(kernel_mac_support, 1);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
SYSCTL_DECL(_security);
SYSCTL_NODE(_security, OID_AUTO, mac, CTLFLAG_RW, 0,
"TrustedBSD MAC policy controls");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#if MAC_MAX_POLICIES > 32
#error "MAC_MAX_POLICIES too large"
#endif
If MAC_MAX_POLICIES isn't defined, don't try to define it, just let the compile fail. MAC_MAX_POLICIES should always be defined, or we have bigger problems at hand. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-20 03:41:09 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static unsigned int mac_max_policies = MAC_MAX_POLICIES;
static unsigned int mac_policy_offsets_free = (1 << MAC_MAX_POLICIES) - 1;
SYSCTL_UINT(_security_mac, OID_AUTO, max_policies, CTLFLAG_RD,
&mac_max_policies, 0, "");
Comment describing the semantics of mac_late. Trim trailing whitespace. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-25 20:45:27 +00:00
/*
* Has the kernel started generating labeled objects yet? All read/write
* access to this variable is serialized during the boot process. Following
* the end of serialization, we don't update this flag; no locking.
*/
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static int mac_late = 0;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
/*
* Warn about EA transactions only the first time they happen.
* Weak coherency, no locking.
*/
static int ea_warn_once = 0;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static int mac_enforce_fs = 1;
SYSCTL_INT(_security_mac, OID_AUTO, enforce_fs, CTLFLAG_RW,
&mac_enforce_fs, 0, "Enforce MAC policy on file system objects");
TUNABLE_INT("security.mac.enforce_fs", &mac_enforce_fs);
static int mac_enforce_network = 1;
SYSCTL_INT(_security_mac, OID_AUTO, enforce_network, CTLFLAG_RW,
&mac_enforce_network, 0, "Enforce MAC policy on network packets");
TUNABLE_INT("security.mac.enforce_network", &mac_enforce_network);
At the cost of seeming a little gauche, make use of more traditional alphabetization for mac_enforce_pipe sysctl. Obtained from: TrustedBSD Project Sponsored by: DAPRA, NAI Labs
2002-09-18 02:00:19 +00:00
static int mac_enforce_pipe = 1;
SYSCTL_INT(_security_mac, OID_AUTO, enforce_pipe, CTLFLAG_RW,
&mac_enforce_pipe, 0, "Enforce MAC policy on pipe operations");
Add tunables for the existing sysctl twiddles for pipe and vm enforcement so they can be disabled prior to kernel start. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-09-30 20:50:00 +00:00
TUNABLE_INT("security.mac.enforce_pipe", &mac_enforce_pipe);
At the cost of seeming a little gauche, make use of more traditional alphabetization for mac_enforce_pipe sysctl. Obtained from: TrustedBSD Project Sponsored by: DAPRA, NAI Labs
2002-09-18 02:00:19 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static int mac_enforce_process = 1;
SYSCTL_INT(_security_mac, OID_AUTO, enforce_process, CTLFLAG_RW,
&mac_enforce_process, 0, "Enforce MAC policy on inter-process operations");
TUNABLE_INT("security.mac.enforce_process", &mac_enforce_process);
static int mac_enforce_socket = 1;
SYSCTL_INT(_security_mac, OID_AUTO, enforce_socket, CTLFLAG_RW,
&mac_enforce_socket, 0, "Enforce MAC policy on socket operations");
TUNABLE_INT("security.mac.enforce_socket", &mac_enforce_socket);
Add a toggle to disable VM enforcement. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-09-18 02:02:08 +00:00
static int mac_enforce_vm = 1;
SYSCTL_INT(_security_mac, OID_AUTO, enforce_vm, CTLFLAG_RW,
&mac_enforce_vm, 0, "Enforce MAC policy on vm operations");
Add tunables for the existing sysctl twiddles for pipe and vm enforcement so they can be disabled prior to kernel start. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-09-30 20:50:00 +00:00
TUNABLE_INT("security.mac.enforce_vm", &mac_enforce_vm);
Add a toggle to disable VM enforcement. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-09-18 02:02:08 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static int mac_cache_fslabel_in_vnode = 1;
SYSCTL_INT(_security_mac, OID_AUTO, cache_fslabel_in_vnode, CTLFLAG_RW,
&mac_cache_fslabel_in_vnode, 0, "Cache mount fslabel in vnode");
TUNABLE_INT("security.mac.cache_fslabel_in_vnode",
&mac_cache_fslabel_in_vnode);
Add security.mac.mmap_revocation, a flag indicating whether we should revoke access to memory maps on a process label change. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-09-09 17:12:24 +00:00
static int mac_mmap_revocation = 1;
SYSCTL_INT(_security_mac, OID_AUTO, mmap_revocation, CTLFLAG_RW,
&mac_mmap_revocation, 0, "Revoke mmap access to files on subject "
"relabel");
Sync to trustedbsd_mac tree: default to sigsegv rather than copy-on-write during a label change resulting in an mmap removal. This is "fail stop" behavior, which is preferred, although it offers slightly less transparency. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-15 02:28:32 +00:00
static int mac_mmap_revocation_via_cow = 0;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
SYSCTL_INT(_security_mac, OID_AUTO, mmap_revocation_via_cow, CTLFLAG_RW,
&mac_mmap_revocation_via_cow, 0, "Revoke mmap access to files via "
"copy-on-write semantics, or by removing all write access");
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Push 'security.mac.debug_label_fallback' behind options MAC_DEBUG. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-10-01 03:24:20 +00:00
SYSCTL_NODE(_security_mac, OID_AUTO, debug, CTLFLAG_RW, 0,
"TrustedBSD MAC debug info");
static int mac_debug_label_fallback = 0;
SYSCTL_INT(_security_mac_debug, OID_AUTO, label_fallback, CTLFLAG_RW,
&mac_debug_label_fallback, 0, "Filesystems should fall back to fs label"
"when label is corrupted.");
TUNABLE_INT("security.mac.debug_label_fallback",
&mac_debug_label_fallback);
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_NODE(_security_mac_debug, OID_AUTO, counters, CTLFLAG_RW, 0,
"TrustedBSD MAC object counters");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static unsigned int nmacmbufs, nmaccreds, nmacifnets, nmacbpfdescs,
nmacsockets, nmacmounts, nmactemp, nmacvnodes, nmacdevfsdirents,
nmacipqs, nmacpipes;
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, mbufs, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacmbufs, 0, "number of mbufs in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, creds, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmaccreds, 0, "number of ucreds in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, ifnets, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacifnets, 0, "number of ifnets in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, ipqs, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacipqs, 0, "number of ipqs in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, bpfdescs, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacbpfdescs, 0, "number of bpfdescs in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, sockets, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacsockets, 0, "number of sockets in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, pipes, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacpipes, 0, "number of pipes in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, mounts, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacmounts, 0, "number of mounts in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, temp, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmactemp, 0, "number of temporary labels in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, vnodes, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacvnodes, 0, "number of vnodes in use");
Push the debugging obect label counters into security.mac.debug.counters rather than directly under security.mac.debug. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:30:53 +00:00
SYSCTL_UINT(_security_mac_debug_counters, OID_AUTO, devfsdirents, CTLFLAG_RD,
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
&nmacdevfsdirents, 0, "number of devfs dirents inuse");
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static int error_select(int error1, int error2);
static int mac_policy_register(struct mac_policy_conf *mpc);
static int mac_policy_unregister(struct mac_policy_conf *mpc);
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
static void mac_check_vnode_mmap_downgrade(struct ucred *cred,
struct vnode *vp, int *prot);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static void mac_cred_mmapped_drop_perms_recurse(struct thread *td,
struct ucred *cred, struct vm_map *map);
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
static void mac_destroy_socket_label(struct label *label);
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
static int mac_setlabel_vnode_extattr(struct ucred *cred,
struct vnode *vp, struct label *intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MALLOC_DEFINE(M_MACOPVEC, "macopvec", "MAC policy operation vector");
MALLOC_DEFINE(M_MACPIPELABEL, "macpipelabel", "MAC labels for pipes");
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MALLOC_DEFINE(M_MACTEMP, "mactemp", "MAC temporary label storage");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
/*
* mac_policy_list_lock protects the consistency of 'mac_policy_list',
* the linked list of attached policy modules. Read-only consumers of
* the list must acquire a shared lock for the duration of their use;
* writers must acquire an exclusive lock. Note that for compound
* operations, locks should be held for the entire compound operation,
* and that this is not yet done for relabel requests.
*/
static struct mtx mac_policy_list_lock;
static LIST_HEAD(, mac_policy_conf) mac_policy_list;
static int mac_policy_list_busy;
#define MAC_POLICY_LIST_LOCKINIT() mtx_init(&mac_policy_list_lock, \
"mac_policy_list_lock", NULL, MTX_DEF);
#define MAC_POLICY_LIST_LOCK() mtx_lock(&mac_policy_list_lock);
#define MAC_POLICY_LIST_UNLOCK() mtx_unlock(&mac_policy_list_lock);
#define MAC_POLICY_LIST_BUSY() do { \
MAC_POLICY_LIST_LOCK(); \
mac_policy_list_busy++; \
MAC_POLICY_LIST_UNLOCK(); \
} while (0)
#define MAC_POLICY_LIST_UNBUSY() do { \
MAC_POLICY_LIST_LOCK(); \
mac_policy_list_busy--; \
if (mac_policy_list_busy < 0) \
panic("Extra mac_policy_list_busy--"); \
MAC_POLICY_LIST_UNLOCK(); \
} while (0)
/*
* MAC_CHECK performs the designated check by walking the policy
* module list and checking with each as to how it feels about the
* request. Note that it returns its value via 'error' in the scope
* of the caller.
*/
#define MAC_CHECK(check, args...) do { \
struct mac_policy_conf *mpc; \
\
error = 0; \
MAC_POLICY_LIST_BUSY(); \
LIST_FOREACH(mpc, &mac_policy_list, mpc_list) { \
if (mpc->mpc_ops->mpo_ ## check != NULL) \
error = error_select( \
mpc->mpc_ops->mpo_ ## check (args), \
error); \
} \
MAC_POLICY_LIST_UNBUSY(); \
} while (0)
/*
* MAC_BOOLEAN performs the designated boolean composition by walking
* the module list, invoking each instance of the operation, and
* combining the results using the passed C operator. Note that it
* returns its value via 'result' in the scope of the caller, which
* should be initialized by the caller in a meaningful way to get
* a meaningful result.
*/
#define MAC_BOOLEAN(operation, composition, args...) do { \
struct mac_policy_conf *mpc; \
\
MAC_POLICY_LIST_BUSY(); \
LIST_FOREACH(mpc, &mac_policy_list, mpc_list) { \
if (mpc->mpc_ops->mpo_ ## operation != NULL) \
result = result composition \
mpc->mpc_ops->mpo_ ## operation (args); \
} \
MAC_POLICY_LIST_UNBUSY(); \
} while (0)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
#define MAC_EXTERNALIZE(type, label, elementlist, outbuf, \
outbuflen) do { \
char *curptr, *curptr_start, *element_name, *element_temp; \
size_t left, left_start, len; \
int claimed, first, first_start, ignorenotfound; \
\
error = 0; \
element_temp = elementlist; \
curptr = outbuf; \
curptr[0] = '\0'; \
left = outbuflen; \
first = 1; \
while ((element_name = strsep(&element_temp, ",")) != NULL) { \
curptr_start = curptr; \
left_start = left; \
first_start = first; \
if (element_name[0] == '?') { \
element_name++; \
ignorenotfound = 1; \
} else \
ignorenotfound = 0; \
claimed = 0; \
if (first) { \
len = snprintf(curptr, left, "%s/", \
element_name); \
first = 0; \
} else \
len = snprintf(curptr, left, ",%s/", \
element_name); \
if (len >= left) { \
error = EINVAL; /* XXXMAC: E2BIG */ \
break; \
} \
curptr += len; \
left -= len; \
\
MAC_CHECK(externalize_ ## type, label, element_name, \
curptr, left, &len, &claimed); \
if (error) \
break; \
if (claimed == 1) { \
if (len >= outbuflen) { \
error = EINVAL; /* XXXMAC: E2BIG */ \
break; \
} \
curptr += len; \
left -= len; \
} else if (claimed == 0 && ignorenotfound) { \
/* \
* Revert addition of the label element \
* name. \
*/ \
curptr = curptr_start; \
*curptr = '\0'; \
left = left_start; \
first = first_start; \
} else { \
error = EINVAL; /* XXXMAC: ENOLABEL */ \
break; \
} \
} \
} while (0)
#define MAC_INTERNALIZE(type, label, instring) do { \
char *element, *element_name, *element_data; \
int claimed; \
\
error = 0; \
element = instring; \
while ((element_name = strsep(&element, ",")) != NULL) { \
element_data = element_name; \
element_name = strsep(&element_data, "/"); \
if (element_data == NULL) { \
error = EINVAL; \
break; \
} \
claimed = 0; \
MAC_CHECK(internalize_ ## type, label, element_name, \
element_data, &claimed); \
if (error) \
break; \
if (claimed != 1) { \
/* XXXMAC: Another error here? */ \
error = EINVAL; \
break; \
} \
} \
} while (0)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
/*
* MAC_PERFORM performs the designated operation by walking the policy
* module list and invoking that operation for each policy.
*/
#define MAC_PERFORM(operation, args...) do { \
struct mac_policy_conf *mpc; \
\
MAC_POLICY_LIST_BUSY(); \
LIST_FOREACH(mpc, &mac_policy_list, mpc_list) { \
if (mpc->mpc_ops->mpo_ ## operation != NULL) \
mpc->mpc_ops->mpo_ ## operation (args); \
} \
MAC_POLICY_LIST_UNBUSY(); \
} while (0)
/*
* Initialize the MAC subsystem, including appropriate SMP locks.
*/
static void
mac_init(void)
{
LIST_INIT(&mac_policy_list);
MAC_POLICY_LIST_LOCKINIT();
}
/*
* For the purposes of modules that want to know if they were loaded
* "early", set the mac_late flag once we've processed modules either
* linked into the kernel, or loaded before the kernel startup.
*/
static void
mac_late_init(void)
{
mac_late = 1;
}
/*
* Allow MAC policy modules to register during boot, etc.
*/
int
mac_policy_modevent(module_t mod, int type, void *data)
{
struct mac_policy_conf *mpc;
int error;
error = 0;
mpc = (struct mac_policy_conf *) data;
switch (type) {
case MOD_LOAD:
if (mpc->mpc_loadtime_flags & MPC_LOADTIME_FLAG_NOTLATE &&
mac_late) {
printf("mac_policy_modevent: can't load %s policy "
"after booting\n", mpc->mpc_name);
error = EBUSY;
break;
}
error = mac_policy_register(mpc);
break;
case MOD_UNLOAD:
/* Don't unregister the module if it was never registered. */
if ((mpc->mpc_runtime_flags & MPC_RUNTIME_FLAG_REGISTERED)
!= 0)
error = mac_policy_unregister(mpc);
else
error = 0;
break;
default:
break;
}
return (error);
}
static int
mac_policy_register(struct mac_policy_conf *mpc)
{
struct mac_policy_conf *tmpc;
struct mac_policy_op_entry *mpe;
int slot;
Remove un-needed stack variable 'ops'. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-09-18 22:35:02 +00:00
MALLOC(mpc->mpc_ops, struct mac_policy_ops *, sizeof(*mpc->mpc_ops),
M_MACOPVEC, M_WAITOK | M_ZERO);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
for (mpe = mpc->mpc_entries; mpe->mpe_constant != MAC_OP_LAST; mpe++) {
switch (mpe->mpe_constant) {
case MAC_OP_LAST:
/*
* Doesn't actually happen, but this allows checking
* that all enumerated values are handled.
*/
break;
case MAC_DESTROY:
mpc->mpc_ops->mpo_destroy =
mpe->mpe_function;
break;
case MAC_INIT:
mpc->mpc_ops->mpo_init =
mpe->mpe_function;
break;
Provide an implementation of mac_syscall() so that security modules can offer new services without reserving system call numbers, or augmented versions of existing services. User code requests a target policy by name, and specifies the policy-specific API plus target. This is required in particular for our port of SELinux/FLASK to the MAC framework since it offers additional security services. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 17:59:48 +00:00
case MAC_SYSCALL:
mpc->mpc_ops->mpo_syscall =
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_BPFDESC_LABEL:
mpc->mpc_ops->mpo_init_bpfdesc_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_CRED_LABEL:
mpc->mpc_ops->mpo_init_cred_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_DEVFSDIRENT_LABEL:
mpc->mpc_ops->mpo_init_devfsdirent_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_IFNET_LABEL:
mpc->mpc_ops->mpo_init_ifnet_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_IPQ_LABEL:
mpc->mpc_ops->mpo_init_ipq_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_MBUF_LABEL:
mpc->mpc_ops->mpo_init_mbuf_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_MOUNT_LABEL:
mpc->mpc_ops->mpo_init_mount_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_MOUNT_FS_LABEL:
mpc->mpc_ops->mpo_init_mount_fs_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_PIPE_LABEL:
mpc->mpc_ops->mpo_init_pipe_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_SOCKET_LABEL:
mpc->mpc_ops->mpo_init_socket_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_SOCKET_PEER_LABEL:
mpc->mpc_ops->mpo_init_socket_peer_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_INIT_VNODE_LABEL:
mpc->mpc_ops->mpo_init_vnode_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_BPFDESC_LABEL:
mpc->mpc_ops->mpo_destroy_bpfdesc_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_CRED_LABEL:
mpc->mpc_ops->mpo_destroy_cred_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_DEVFSDIRENT_LABEL:
mpc->mpc_ops->mpo_destroy_devfsdirent_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_IFNET_LABEL:
mpc->mpc_ops->mpo_destroy_ifnet_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_IPQ_LABEL:
mpc->mpc_ops->mpo_destroy_ipq_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_MBUF_LABEL:
mpc->mpc_ops->mpo_destroy_mbuf_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_MOUNT_LABEL:
mpc->mpc_ops->mpo_destroy_mount_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_MOUNT_FS_LABEL:
mpc->mpc_ops->mpo_destroy_mount_fs_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Begin another merge from the TrustedBSD MAC branch: - Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy entry points to mpo_init_foo_label(label) and mpo_destroy_foo_label(label). This will permit the use of the same entry points for holding temporary type-specific label during internalization and externalization, as well as for caching purposes. - Because of this, break out mpo_{init,destroy}_socket() and mpo_{init,destroy}_mount() into seperate entry points for socket main/peer labels and mount main/fs labels. - Since the prototype for label initialization is the same across almost all entry points, implement these entry points using common implementations for Biba, MLS, and Test, reducing the number of almost identical looking functions. This simplifies policy implementation, as well as preparing us for the merge of the new flexible userland API for managing labels on objects. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 15:10:00 +00:00
case MAC_DESTROY_PIPE_LABEL:
mpc->mpc_ops->mpo_destroy_pipe_label =
mpe->mpe_function;
break;
case MAC_DESTROY_SOCKET_LABEL:
mpc->mpc_ops->mpo_destroy_socket_label =
mpe->mpe_function;
break;
case MAC_DESTROY_SOCKET_PEER_LABEL:
mpc->mpc_ops->mpo_destroy_socket_peer_label =
mpe->mpe_function;
break;
case MAC_DESTROY_VNODE_LABEL:
mpc->mpc_ops->mpo_destroy_vnode_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
case MAC_COPY_PIPE_LABEL:
mpc->mpc_ops->mpo_copy_pipe_label =
mpe->mpe_function;
break;
case MAC_COPY_VNODE_LABEL:
mpc->mpc_ops->mpo_copy_vnode_label =
mpe->mpe_function;
break;
case MAC_EXTERNALIZE_CRED_LABEL:
mpc->mpc_ops->mpo_externalize_cred_label =
mpe->mpe_function;
break;
case MAC_EXTERNALIZE_IFNET_LABEL:
mpc->mpc_ops->mpo_externalize_ifnet_label =
mpe->mpe_function;
break;
case MAC_EXTERNALIZE_PIPE_LABEL:
mpc->mpc_ops->mpo_externalize_pipe_label =
mpe->mpe_function;
break;
case MAC_EXTERNALIZE_SOCKET_LABEL:
mpc->mpc_ops->mpo_externalize_socket_label =
mpe->mpe_function;
break;
case MAC_EXTERNALIZE_SOCKET_PEER_LABEL:
mpc->mpc_ops->mpo_externalize_socket_peer_label =
mpe->mpe_function;
break;
case MAC_EXTERNALIZE_VNODE_LABEL:
mpc->mpc_ops->mpo_externalize_vnode_label =
mpe->mpe_function;
break;
case MAC_INTERNALIZE_CRED_LABEL:
mpc->mpc_ops->mpo_internalize_cred_label =
mpe->mpe_function;
break;
case MAC_INTERNALIZE_IFNET_LABEL:
mpc->mpc_ops->mpo_internalize_ifnet_label =
mpe->mpe_function;
break;
case MAC_INTERNALIZE_PIPE_LABEL:
mpc->mpc_ops->mpo_internalize_pipe_label =
mpe->mpe_function;
break;
case MAC_INTERNALIZE_SOCKET_LABEL:
mpc->mpc_ops->mpo_internalize_socket_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
case MAC_INTERNALIZE_VNODE_LABEL:
mpc->mpc_ops->mpo_internalize_vnode_label =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
case MAC_CREATE_DEVFS_DEVICE:
mpc->mpc_ops->mpo_create_devfs_device =
mpe->mpe_function;
break;
case MAC_CREATE_DEVFS_DIRECTORY:
mpc->mpc_ops->mpo_create_devfs_directory =
mpe->mpe_function;
break;
Integrate a devfs/MAC fix from the MAC tree: avoid a race condition during devfs VOP symlink creation by introducing a new entry point to determine the label of the devfs_dirent prior to allocation of a vnode for the symlink. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 18:40:10 +00:00
case MAC_CREATE_DEVFS_SYMLINK:
mpc->mpc_ops->mpo_create_devfs_symlink =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CREATE_DEVFS_VNODE:
mpc->mpc_ops->mpo_create_devfs_vnode =
mpe->mpe_function;
break;
case MAC_CREATE_MOUNT:
mpc->mpc_ops->mpo_create_mount =
mpe->mpe_function;
break;
case MAC_CREATE_ROOT_MOUNT:
mpc->mpc_ops->mpo_create_root_mount =
mpe->mpe_function;
break;
case MAC_RELABEL_VNODE:
mpc->mpc_ops->mpo_relabel_vnode =
mpe->mpe_function;
break;
case MAC_UPDATE_DEVFSDIRENT:
mpc->mpc_ops->mpo_update_devfsdirent =
mpe->mpe_function;
break;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
case MAC_ASSOCIATE_VNODE_DEVFS:
mpc->mpc_ops->mpo_associate_vnode_devfs =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
case MAC_ASSOCIATE_VNODE_EXTATTR:
mpc->mpc_ops->mpo_associate_vnode_extattr =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
case MAC_ASSOCIATE_VNODE_SINGLELABEL:
mpc->mpc_ops->mpo_associate_vnode_singlelabel =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
case MAC_CREATE_VNODE_EXTATTR:
mpc->mpc_ops->mpo_create_vnode_extattr =
mpe->mpe_function;
break;
case MAC_SETLABEL_VNODE_EXTATTR:
mpc->mpc_ops->mpo_setlabel_vnode_extattr =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_FROM_SOCKET:
mpc->mpc_ops->mpo_create_mbuf_from_socket =
mpe->mpe_function;
break;
case MAC_CREATE_PIPE:
mpc->mpc_ops->mpo_create_pipe =
mpe->mpe_function;
break;
case MAC_CREATE_SOCKET:
mpc->mpc_ops->mpo_create_socket =
mpe->mpe_function;
break;
case MAC_CREATE_SOCKET_FROM_SOCKET:
mpc->mpc_ops->mpo_create_socket_from_socket =
mpe->mpe_function;
break;
case MAC_RELABEL_PIPE:
mpc->mpc_ops->mpo_relabel_pipe =
mpe->mpe_function;
break;
case MAC_RELABEL_SOCKET:
mpc->mpc_ops->mpo_relabel_socket =
mpe->mpe_function;
break;
case MAC_SET_SOCKET_PEER_FROM_MBUF:
mpc->mpc_ops->mpo_set_socket_peer_from_mbuf =
mpe->mpe_function;
break;
case MAC_SET_SOCKET_PEER_FROM_SOCKET:
mpc->mpc_ops->mpo_set_socket_peer_from_socket =
mpe->mpe_function;
break;
case MAC_CREATE_BPFDESC:
mpc->mpc_ops->mpo_create_bpfdesc =
mpe->mpe_function;
break;
case MAC_CREATE_DATAGRAM_FROM_IPQ:
mpc->mpc_ops->mpo_create_datagram_from_ipq =
mpe->mpe_function;
break;
case MAC_CREATE_FRAGMENT:
mpc->mpc_ops->mpo_create_fragment =
mpe->mpe_function;
break;
case MAC_CREATE_IFNET:
mpc->mpc_ops->mpo_create_ifnet =
mpe->mpe_function;
break;
case MAC_CREATE_IPQ:
mpc->mpc_ops->mpo_create_ipq =
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_FROM_MBUF:
mpc->mpc_ops->mpo_create_mbuf_from_mbuf =
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_LINKLAYER:
mpc->mpc_ops->mpo_create_mbuf_linklayer =
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_FROM_BPFDESC:
mpc->mpc_ops->mpo_create_mbuf_from_bpfdesc =
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_FROM_IFNET:
mpc->mpc_ops->mpo_create_mbuf_from_ifnet =
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_MULTICAST_ENCAP:
mpc->mpc_ops->mpo_create_mbuf_multicast_encap =
mpe->mpe_function;
break;
case MAC_CREATE_MBUF_NETLAYER:
mpc->mpc_ops->mpo_create_mbuf_netlayer =
mpe->mpe_function;
break;
case MAC_FRAGMENT_MATCH:
mpc->mpc_ops->mpo_fragment_match =
mpe->mpe_function;
break;
case MAC_RELABEL_IFNET:
mpc->mpc_ops->mpo_relabel_ifnet =
mpe->mpe_function;
break;
case MAC_UPDATE_IPQ:
mpc->mpc_ops->mpo_update_ipq =
mpe->mpe_function;
break;
case MAC_CREATE_CRED:
mpc->mpc_ops->mpo_create_cred =
mpe->mpe_function;
break;
case MAC_EXECVE_TRANSITION:
mpc->mpc_ops->mpo_execve_transition =
mpe->mpe_function;
break;
case MAC_EXECVE_WILL_TRANSITION:
mpc->mpc_ops->mpo_execve_will_transition =
mpe->mpe_function;
break;
case MAC_CREATE_PROC0:
Cosmetic line wrap synchronization.
2002-10-05 16:33:46 +00:00
mpc->mpc_ops->mpo_create_proc0 =
mpe->mpe_function;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
case MAC_CREATE_PROC1:
Cosmetic line wrap synchronization.
2002-10-05 16:33:46 +00:00
mpc->mpc_ops->mpo_create_proc1 =
mpe->mpe_function;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
case MAC_RELABEL_CRED:
mpc->mpc_ops->mpo_relabel_cred =
mpe->mpe_function;
break;
Add a new MAC entry point, mac_thread_userret(td), which permits policy modules to perform MAC-related events when a thread returns to user space. This is required for policies that have floating process labels, as it's not always possible to acquire the process lock at arbitrary points in the stack during system call processing; process labels might represent traditional authentication data, process history information, or other data. LOMAC will use this entry point to perform the process label update prior to the thread returning to userspace, when plugged into the MAC framework. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-02 02:42:38 +00:00
case MAC_THREAD_USERRET:
mpc->mpc_ops->mpo_thread_userret =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CHECK_BPFDESC_RECEIVE:
mpc->mpc_ops->mpo_check_bpfdesc_receive =
mpe->mpe_function;
break;
case MAC_CHECK_CRED_RELABEL:
mpc->mpc_ops->mpo_check_cred_relabel =
mpe->mpe_function;
break;
case MAC_CHECK_CRED_VISIBLE:
mpc->mpc_ops->mpo_check_cred_visible =
mpe->mpe_function;
break;
case MAC_CHECK_IFNET_RELABEL:
mpc->mpc_ops->mpo_check_ifnet_relabel =
mpe->mpe_function;
break;
case MAC_CHECK_IFNET_TRANSMIT:
mpc->mpc_ops->mpo_check_ifnet_transmit =
mpe->mpe_function;
break;
case MAC_CHECK_MOUNT_STAT:
mpc->mpc_ops->mpo_check_mount_stat =
mpe->mpe_function;
break;
case MAC_CHECK_PIPE_IOCTL:
mpc->mpc_ops->mpo_check_pipe_ioctl =
mpe->mpe_function;
break;
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
case MAC_CHECK_PIPE_POLL:
mpc->mpc_ops->mpo_check_pipe_poll =
mpe->mpe_function;
break;
case MAC_CHECK_PIPE_READ:
mpc->mpc_ops->mpo_check_pipe_read =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
case MAC_CHECK_PIPE_RELABEL:
mpc->mpc_ops->mpo_check_pipe_relabel =
mpe->mpe_function;
break;
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
case MAC_CHECK_PIPE_STAT:
mpc->mpc_ops->mpo_check_pipe_stat =
mpe->mpe_function;
break;
case MAC_CHECK_PIPE_WRITE:
mpc->mpc_ops->mpo_check_pipe_write =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CHECK_PROC_DEBUG:
mpc->mpc_ops->mpo_check_proc_debug =
mpe->mpe_function;
break;
case MAC_CHECK_PROC_SCHED:
mpc->mpc_ops->mpo_check_proc_sched =
mpe->mpe_function;
break;
case MAC_CHECK_PROC_SIGNAL:
mpc->mpc_ops->mpo_check_proc_signal =
mpe->mpe_function;
break;
case MAC_CHECK_SOCKET_BIND:
mpc->mpc_ops->mpo_check_socket_bind =
mpe->mpe_function;
break;
case MAC_CHECK_SOCKET_CONNECT:
mpc->mpc_ops->mpo_check_socket_connect =
mpe->mpe_function;
break;
Rename mac_check_socket_receive() to mac_check_socket_deliver() so that we can use the names _receive() and _send() for the receive() and send() checks. Rename related constants, policy implementations, etc. PR: Submitted by: Reviewed by: Approved by: Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs MFC after:
2002-08-15 18:51:26 +00:00
case MAC_CHECK_SOCKET_DELIVER:
mpc->mpc_ops->mpo_check_socket_deliver =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Rename mac_check_socket_receive() to mac_check_socket_deliver() so that we can use the names _receive() and _send() for the receive() and send() checks. Rename related constants, policy implementations, etc. PR: Submitted by: Reviewed by: Approved by: Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs MFC after:
2002-08-15 18:51:26 +00:00
case MAC_CHECK_SOCKET_LISTEN:
mpc->mpc_ops->mpo_check_socket_listen =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
Integrate mac_check_socket_send() and mac_check_socket_receive() checks from the MAC tree: allow policies to perform access control for the ability of a process to send and receive data via a socket. At some point, we might also pass in additional address information if an explicit address is requested on send. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 14:39:15 +00:00
case MAC_CHECK_SOCKET_RECEIVE:
mpc->mpc_ops->mpo_check_socket_receive =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CHECK_SOCKET_RELABEL:
mpc->mpc_ops->mpo_check_socket_relabel =
mpe->mpe_function;
break;
Integrate mac_check_socket_send() and mac_check_socket_receive() checks from the MAC tree: allow policies to perform access control for the ability of a process to send and receive data via a socket. At some point, we might also pass in additional address information if an explicit address is requested on send. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 14:39:15 +00:00
case MAC_CHECK_SOCKET_SEND:
mpc->mpc_ops->mpo_check_socket_send =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CHECK_SOCKET_VISIBLE:
mpc->mpc_ops->mpo_check_socket_visible =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_ACCESS:
mpc->mpc_ops->mpo_check_vnode_access =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_CHDIR:
mpc->mpc_ops->mpo_check_vnode_chdir =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_CHROOT:
mpc->mpc_ops->mpo_check_vnode_chroot =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_CREATE:
mpc->mpc_ops->mpo_check_vnode_create =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_DELETE:
mpc->mpc_ops->mpo_check_vnode_delete =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_DELETEACL:
mpc->mpc_ops->mpo_check_vnode_deleteacl =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_EXEC:
mpc->mpc_ops->mpo_check_vnode_exec =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_GETACL:
mpc->mpc_ops->mpo_check_vnode_getacl =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_GETEXTATTR:
mpc->mpc_ops->mpo_check_vnode_getextattr =
mpe->mpe_function;
break;
Merge support for mac_check_vnode_link(), a MAC framework/policy entry point that instruments the creation of hard links. Policy implementations to follow. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 18:11:36 +00:00
case MAC_CHECK_VNODE_LINK:
mpc->mpc_ops->mpo_check_vnode_link =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CHECK_VNODE_LOOKUP:
mpc->mpc_ops->mpo_check_vnode_lookup =
mpe->mpe_function;
break;
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
case MAC_CHECK_VNODE_MMAP:
mpc->mpc_ops->mpo_check_vnode_mmap =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_MMAP_DOWNGRADE:
mpc->mpc_ops->mpo_check_vnode_mmap_downgrade =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_MPROTECT:
mpc->mpc_ops->mpo_check_vnode_mprotect =
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_OPEN:
mpc->mpc_ops->mpo_check_vnode_open =
mpe->mpe_function;
break;
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
case MAC_CHECK_VNODE_POLL:
mpc->mpc_ops->mpo_check_vnode_poll =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_READ:
mpc->mpc_ops->mpo_check_vnode_read =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case MAC_CHECK_VNODE_READDIR:
mpc->mpc_ops->mpo_check_vnode_readdir =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_READLINK:
mpc->mpc_ops->mpo_check_vnode_readlink =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_RELABEL:
mpc->mpc_ops->mpo_check_vnode_relabel =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_RENAME_FROM:
mpc->mpc_ops->mpo_check_vnode_rename_from =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_RENAME_TO:
mpc->mpc_ops->mpo_check_vnode_rename_to =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_REVOKE:
mpc->mpc_ops->mpo_check_vnode_revoke =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_SETACL:
mpc->mpc_ops->mpo_check_vnode_setacl =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_SETEXTATTR:
mpc->mpc_ops->mpo_check_vnode_setextattr =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_SETFLAGS:
mpc->mpc_ops->mpo_check_vnode_setflags =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_SETMODE:
mpc->mpc_ops->mpo_check_vnode_setmode =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_SETOWNER:
mpc->mpc_ops->mpo_check_vnode_setowner =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_SETUTIMES:
mpc->mpc_ops->mpo_check_vnode_setutimes =
mpe->mpe_function;
break;
case MAC_CHECK_VNODE_STAT:
mpc->mpc_ops->mpo_check_vnode_stat =
mpe->mpe_function;
break;
Introduce MAC_CHECK_VNODE_SWAPON, which permits MAC policies to perform authorization checks during swapon() events; policies might choose to enforce protections based on the credential requesting the swap configuration, the target of the swap operation, or other factors such as internal policy state. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 15:53:43 +00:00
case MAC_CHECK_VNODE_SWAPON:
mpc->mpc_ops->mpo_check_vnode_swapon =
mpe->mpe_function;
break;
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
case MAC_CHECK_VNODE_WRITE:
mpc->mpc_ops->mpo_check_vnode_write =
mpe->mpe_function;
break;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
/*
default:
printf("MAC policy `%s': unknown operation %d\n",
mpc->mpc_name, mpe->mpe_constant);
return (EINVAL);
*/
}
}
MAC_POLICY_LIST_LOCK();
if (mac_policy_list_busy > 0) {
MAC_POLICY_LIST_UNLOCK();
FREE(mpc->mpc_ops, M_MACOPVEC);
mpc->mpc_ops = NULL;
return (EBUSY);
}
LIST_FOREACH(tmpc, &mac_policy_list, mpc_list) {
if (strcmp(tmpc->mpc_name, mpc->mpc_name) == 0) {
MAC_POLICY_LIST_UNLOCK();
FREE(mpc->mpc_ops, M_MACOPVEC);
mpc->mpc_ops = NULL;
return (EEXIST);
}
}
if (mpc->mpc_field_off != NULL) {
slot = ffs(mac_policy_offsets_free);
if (slot == 0) {
MAC_POLICY_LIST_UNLOCK();
FREE(mpc->mpc_ops, M_MACOPVEC);
mpc->mpc_ops = NULL;
return (ENOMEM);
}
slot--;
mac_policy_offsets_free &= ~(1 << slot);
*mpc->mpc_field_off = slot;
}
mpc->mpc_runtime_flags |= MPC_RUNTIME_FLAG_REGISTERED;
LIST_INSERT_HEAD(&mac_policy_list, mpc, mpc_list);
/* Per-policy initialization. */
if (mpc->mpc_ops->mpo_init != NULL)
(*(mpc->mpc_ops->mpo_init))(mpc);
MAC_POLICY_LIST_UNLOCK();
printf("Security policy loaded: %s (%s)\n", mpc->mpc_fullname,
mpc->mpc_name);
return (0);
}
static int
mac_policy_unregister(struct mac_policy_conf *mpc)
{
Synch from TrustedBSD MAC tree: - If a policy isn't registered when a policy module unloads, silently succeed. - Hold the policy list lock across more of the validity tests to avoid races. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:46:03 +00:00
/*
* If we fail the load, we may get a request to unload. Check
* to see if we did the run-time registration, and if not,
* silently succeed.
*/
MAC_POLICY_LIST_LOCK();
if ((mpc->mpc_runtime_flags & MPC_RUNTIME_FLAG_REGISTERED) == 0) {
MAC_POLICY_LIST_UNLOCK();
return (0);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#if 0
/*
* Don't allow unloading modules with private data.
*/
Synch from TrustedBSD MAC tree: - If a policy isn't registered when a policy module unloads, silently succeed. - Hold the policy list lock across more of the validity tests to avoid races. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:46:03 +00:00
if (mpc->mpc_field_off != NULL) {
MAC_POLICY_LIST_UNLOCK();
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (EBUSY);
Synch from TrustedBSD MAC tree: - If a policy isn't registered when a policy module unloads, silently succeed. - Hold the policy list lock across more of the validity tests to avoid races. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:46:03 +00:00
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
#endif
Synch from TrustedBSD MAC tree: - If a policy isn't registered when a policy module unloads, silently succeed. - Hold the policy list lock across more of the validity tests to avoid races. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:46:03 +00:00
/*
* Only allow the unload to proceed if the module is unloadable
* by its own definition.
*/
if ((mpc->mpc_loadtime_flags & MPC_LOADTIME_FLAG_UNLOADOK) == 0) {
MAC_POLICY_LIST_UNLOCK();
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (EBUSY);
Synch from TrustedBSD MAC tree: - If a policy isn't registered when a policy module unloads, silently succeed. - Hold the policy list lock across more of the validity tests to avoid races. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:46:03 +00:00
}
/*
* Right now, we EBUSY if the list is in use. In the future,
* for reliability reasons, we might want to sleep and wakeup
* later to try again.
*/
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (mac_policy_list_busy > 0) {
MAC_POLICY_LIST_UNLOCK();
return (EBUSY);
}
if (mpc->mpc_ops->mpo_destroy != NULL)
(*(mpc->mpc_ops->mpo_destroy))(mpc);
LIST_REMOVE(mpc, mpc_list);
MAC_POLICY_LIST_UNLOCK();
FREE(mpc->mpc_ops, M_MACOPVEC);
mpc->mpc_ops = NULL;
Make sure to clear the 'registered' flag for MAC policies when they unregister. Under some obscure (perhaps demented) circumstances, this can result in a panic if a policy is unregistered, and then someone foolishly unregisters it again. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-19 20:30:12 +00:00
mpc->mpc_runtime_flags &= ~MPC_RUNTIME_FLAG_REGISTERED;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
printf("Security policy unload: %s (%s)\n", mpc->mpc_fullname,
mpc->mpc_name);
return (0);
}
/*
* Define an error value precedence, and given two arguments, selects the
* value with the higher precedence.
*/
static int
error_select(int error1, int error2)
{
/* Certain decision-making errors take top priority. */
if (error1 == EDEADLK || error2 == EDEADLK)
return (EDEADLK);
/* Invalid arguments should be reported where possible. */
if (error1 == EINVAL || error2 == EINVAL)
return (EINVAL);
/* Precedence goes to "visibility", with both process and file. */
if (error1 == ESRCH || error2 == ESRCH)
return (ESRCH);
if (error1 == ENOENT || error2 == ENOENT)
return (ENOENT);
/* Precedence goes to DAC/MAC protections. */
if (error1 == EACCES || error2 == EACCES)
return (EACCES);
/* Precedence goes to privilege. */
if (error1 == EPERM || error2 == EPERM)
return (EPERM);
/* Precedence goes to error over success; otherwise, arbitrary. */
if (error1 != 0)
return (error1);
return (error2);
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
static void
mac_init_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
bzero(label, sizeof(*label));
label->l_flags = MAC_FLAG_INITIALIZED;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
static void
mac_destroy_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
KASSERT(label->l_flags & MAC_FLAG_INITIALIZED,
("destroying uninitialized label"));
bzero(label, sizeof(*label));
/* implicit: label->l_flags &= ~MAC_FLAG_INITIALIZED; */
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_bpfdesc(struct bpf_d *bpf_d)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_label(&bpf_d->bd_label);
MAC_PERFORM(init_bpfdesc_label, &bpf_d->bd_label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_add_int(&nmacbpfdescs, 1);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
static void
mac_init_cred_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_label(label);
MAC_PERFORM(init_cred_label, label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#ifdef MAC_DEBUG
atomic_add_int(&nmaccreds, 1);
#endif
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
void
mac_init_cred(struct ucred *cred)
{
mac_init_cred_label(&cred->cr_label);
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_devfsdirent(struct devfs_dirent *de)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_label(&de->de_label);
MAC_PERFORM(init_devfsdirent_label, &de->de_label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_add_int(&nmacdevfsdirents, 1);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#endif
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
static void
mac_init_ifnet_label(struct label *label)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_label(label);
MAC_PERFORM(init_ifnet_label, label);
Push 'security.mac.debug_label_fallback' behind options MAC_DEBUG. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-10-01 03:24:20 +00:00
#ifdef MAC_DEBUG
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
atomic_add_int(&nmacifnets, 1);
Push 'security.mac.debug_label_fallback' behind options MAC_DEBUG. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-10-01 03:24:20 +00:00
#endif
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
void
mac_init_ifnet(struct ifnet *ifp)
{
mac_init_ifnet_label(&ifp->if_label);
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_ipq(struct ipq *ipq)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_label(&ipq->ipq_label);
MAC_PERFORM(init_ipq_label, &ipq->ipq_label);
Push 'security.mac.debug_label_fallback' behind options MAC_DEBUG. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-10-01 03:24:20 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_add_int(&nmacipqs, 1);
Push 'security.mac.debug_label_fallback' behind options MAC_DEBUG. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-10-01 03:24:20 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
int
mac_init_mbuf(struct mbuf *m, int flag)
{
While the MAC API has supported the ability to handle M_NOWAIT passed to mbuf label initialization, that functionality was never merged to the main tree. Go ahead and merge that functionality now. Note that this requires policy modules to accept the case where the label element may be destroyed even if init has not succeeded on it (in the event that policy failed the init). This will shortly also apply to sockets. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:44:49 +00:00
int error;
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
KASSERT(m->m_flags & M_PKTHDR, ("mac_init_mbuf on non-header mbuf"));
mac_init_label(&m->m_pkthdr.label);
While the MAC API has supported the ability to handle M_NOWAIT passed to mbuf label initialization, that functionality was never merged to the main tree. Go ahead and merge that functionality now. Note that this requires policy modules to accept the case where the label element may be destroyed even if init has not succeeded on it (in the event that policy failed the init). This will shortly also apply to sockets. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:44:49 +00:00
MAC_CHECK(init_mbuf_label, &m->m_pkthdr.label, flag);
if (error) {
MAC_PERFORM(destroy_mbuf_label, &m->m_pkthdr.label);
mac_destroy_label(&m->m_pkthdr.label);
}
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
#ifdef MAC_DEBUG
While the MAC API has supported the ability to handle M_NOWAIT passed to mbuf label initialization, that functionality was never merged to the main tree. Go ahead and merge that functionality now. Note that this requires policy modules to accept the case where the label element may be destroyed even if init has not succeeded on it (in the event that policy failed the init). This will shortly also apply to sockets. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:44:49 +00:00
if (error == 0)
atomic_add_int(&nmacmbufs, 1);
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
#endif
While the MAC API has supported the ability to handle M_NOWAIT passed to mbuf label initialization, that functionality was never merged to the main tree. Go ahead and merge that functionality now. Note that this requires policy modules to accept the case where the label element may be destroyed even if init has not succeeded on it (in the event that policy failed the init). This will shortly also apply to sockets. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:44:49 +00:00
return (error);
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_mount(struct mount *mp)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_label(&mp->mnt_mntlabel);
mac_init_label(&mp->mnt_fslabel);
MAC_PERFORM(init_mount_label, &mp->mnt_mntlabel);
MAC_PERFORM(init_mount_fs_label, &mp->mnt_fslabel);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_add_int(&nmacmounts, 1);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#endif
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
static void
mac_init_pipe_label(struct label *label)
{
mac_init_label(label);
MAC_PERFORM(init_pipe_label, label);
#ifdef MAC_DEBUG
atomic_add_int(&nmacpipes, 1);
#endif
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_pipe(struct pipe *pipe)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
struct label *label;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
label = malloc(sizeof(struct label), M_MACPIPELABEL, M_ZERO|M_WAITOK);
pipe->pipe_label = label;
pipe->pipe_peer->pipe_label = label;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_pipe_label(label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
static int
mac_init_socket_label(struct label *label, int flag)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
int error;
mac_init_label(label);
MAC_CHECK(init_socket_label, label, flag);
if (error) {
MAC_PERFORM(destroy_socket_label, label);
mac_destroy_label(label);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#ifdef MAC_DEBUG
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
if (error == 0)
atomic_add_int(&nmacsockets, 1);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
#endif
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
return (error);
}
static int
mac_init_socket_peer_label(struct label *label, int flag)
{
int error;
mac_init_label(label);
MAC_CHECK(init_socket_peer_label, label, flag);
if (error) {
MAC_PERFORM(destroy_socket_label, label);
mac_destroy_label(label);
}
return (error);
}
int
mac_init_socket(struct socket *socket, int flag)
{
int error;
error = mac_init_socket_label(&socket->so_label, flag);
if (error)
return (error);
error = mac_init_socket_peer_label(&socket->so_peerlabel, flag);
if (error)
mac_destroy_socket_label(&socket->so_label);
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
void
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_vnode_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_label(label);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_PERFORM(init_vnode_label, label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
atomic_add_int(&nmacvnodes, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_init_vnode(struct vnode *vp)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_vnode_label(&vp->v_label);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_destroy_bpfdesc(struct bpf_d *bpf_d)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
MAC_PERFORM(destroy_bpfdesc_label, &bpf_d->bd_label);
mac_destroy_label(&bpf_d->bd_label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacbpfdescs, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
static void
mac_destroy_cred_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_PERFORM(destroy_cred_label, label);
mac_destroy_label(label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmaccreds, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
void
mac_destroy_cred(struct ucred *cred)
{
mac_destroy_cred_label(&cred->cr_label);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_destroy_devfsdirent(struct devfs_dirent *de)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
MAC_PERFORM(destroy_devfsdirent_label, &de->de_label);
mac_destroy_label(&de->de_label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacdevfsdirents, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
static void
mac_destroy_ifnet_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_PERFORM(destroy_ifnet_label, label);
mac_destroy_label(label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacifnets, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
void
mac_destroy_ifnet(struct ifnet *ifp)
{
mac_destroy_ifnet_label(&ifp->if_label);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_destroy_ipq(struct ipq *ipq)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
MAC_PERFORM(destroy_ipq_label, &ipq->ipq_label);
mac_destroy_label(&ipq->ipq_label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacipqs, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
void
mac_destroy_mbuf(struct mbuf *m)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
MAC_PERFORM(destroy_mbuf_label, &m->m_pkthdr.label);
mac_destroy_label(&m->m_pkthdr.label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacmbufs, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
void
mac_destroy_mount(struct mount *mp)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
MAC_PERFORM(destroy_mount_label, &mp->mnt_mntlabel);
MAC_PERFORM(destroy_mount_fs_label, &mp->mnt_fslabel);
mac_destroy_label(&mp->mnt_fslabel);
mac_destroy_label(&mp->mnt_mntlabel);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacmounts, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
static void
mac_destroy_pipe_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_PERFORM(destroy_pipe_label, label);
mac_destroy_label(label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacpipes, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
void
mac_destroy_pipe(struct pipe *pipe)
{
mac_destroy_pipe_label(pipe->pipe_label);
free(pipe->pipe_label, M_MACPIPELABEL);
}
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
static void
mac_destroy_socket_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
MAC_PERFORM(destroy_socket_label, label);
mac_destroy_label(label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
atomic_subtract_int(&nmacsockets, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Modify label allocation semantics for sockets: pass in soalloc's malloc flags so that we can call malloc with M_NOWAIT if necessary, avoiding potential sleeps while holding mutexes in the TCP syncache code. Similar to the existing support for mbuf label allocation: if we can't allocate all the necessary label store in each policy, we back out the label allocation and fail the socket creation. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 21:23:47 +00:00
static void
mac_destroy_socket_peer_label(struct label *label)
{
MAC_PERFORM(destroy_socket_peer_label, label);
mac_destroy_label(label);
}
void
mac_destroy_socket(struct socket *socket)
{
mac_destroy_socket_label(&socket->so_label);
mac_destroy_socket_peer_label(&socket->so_peerlabel);
}
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
void
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_destroy_vnode_label(struct label *label)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_PERFORM(destroy_vnode_label, label);
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_destroy_label(label);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#ifdef MAC_DEBUG
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
atomic_subtract_int(&nmacvnodes, 1);
Wrap maintenance of varios nmac{objectname} counters in MAC_DEBUG so we can avoid the cost of a large number of atomic operations if we're not interested in the object count statistics. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 14:21:38 +00:00
#endif
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
void
Rearrange object and label init/destroy functions to match the order used in mac_policy.h and elsewhere. Sort order is basically "by operation category", then "alphabetically by object". Sync to MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 17:38:45 +00:00
mac_destroy_vnode(struct vnode *vp)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_destroy_vnode_label(&vp->v_label);
}
static void
mac_copy_pipe_label(struct label *src, struct label *dest)
{
MAC_PERFORM(copy_pipe_label, src, dest);
}
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
void
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_copy_vnode_label(struct label *src, struct label *dest)
{
MAC_PERFORM(copy_vnode_label, src, dest);
}
static int
mac_check_structmac_consistent(struct mac *mac)
{
if (mac->m_buflen > MAC_MAX_LABEL_BUF_LEN)
return (EINVAL);
return (0);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Another big diff, little functional change: move label internalization, externalization, and cred label life cycle events to entirely above devfs and vnode events. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:57:16 +00:00
static int
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_externalize_cred_label(struct label *label, char *elements,
char *outbuf, size_t outbuflen, int flags)
Another big diff, little functional change: move label internalization, externalization, and cred label life cycle events to entirely above devfs and vnode events. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:57:16 +00:00
{
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_EXTERNALIZE(cred_label, label, elements, outbuf, outbuflen);
Another big diff, little functional change: move label internalization, externalization, and cred label life cycle events to entirely above devfs and vnode events. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:57:16 +00:00
return (error);
}
static int
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_externalize_ifnet_label(struct label *label, char *elements,
char *outbuf, size_t outbuflen, int flags)
Another big diff, little functional change: move label internalization, externalization, and cred label life cycle events to entirely above devfs and vnode events. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:57:16 +00:00
{
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
MAC_EXTERNALIZE(ifnet_label, label, elements, outbuf, outbuflen);
return (error);
}
static int
mac_externalize_pipe_label(struct label *label, char *elements,
char *outbuf, size_t outbuflen, int flags)
{
int error;
MAC_EXTERNALIZE(pipe_label, label, elements, outbuf, outbuflen);
return (error);
}
static int
mac_externalize_socket_label(struct label *label, char *elements,
char *outbuf, size_t outbuflen, int flags)
{
int error;
MAC_EXTERNALIZE(socket_label, label, elements, outbuf, outbuflen);
return (error);
}
static int
mac_externalize_socket_peer_label(struct label *label, char *elements,
char *outbuf, size_t outbuflen, int flags)
{
int error;
MAC_EXTERNALIZE(socket_peer_label, label, elements, outbuf, outbuflen);
return (error);
}
static int
mac_externalize_vnode_label(struct label *label, char *elements,
char *outbuf, size_t outbuflen, int flags)
{
int error;
MAC_EXTERNALIZE(vnode_label, label, elements, outbuf, outbuflen);
return (error);
}
static int
mac_internalize_cred_label(struct label *label, char *string)
{
int error;
MAC_INTERNALIZE(cred_label, label, string);
return (error);
}
static int
mac_internalize_ifnet_label(struct label *label, char *string)
{
int error;
MAC_INTERNALIZE(ifnet_label, label, string);
return (error);
}
static int
mac_internalize_pipe_label(struct label *label, char *string)
{
int error;
MAC_INTERNALIZE(pipe_label, label, string);
return (error);
}
static int
mac_internalize_socket_label(struct label *label, char *string)
{
int error;
MAC_INTERNALIZE(socket_label, label, string);
return (error);
}
static int
mac_internalize_vnode_label(struct label *label, char *string)
{
int error;
MAC_INTERNALIZE(vnode_label, label, string);
Another big diff, little functional change: move label internalization, externalization, and cred label life cycle events to entirely above devfs and vnode events. Sync from MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:57:16 +00:00
return (error);
}
/*
* Initialize MAC label for the first kernel process, from which other
* kernel processes and threads are spawned.
*/
void
mac_create_proc0(struct ucred *cred)
{
MAC_PERFORM(create_proc0, cred);
}
/*
* Initialize MAC label for the first userland process, from which other
* userland processes and threads are spawned.
*/
void
mac_create_proc1(struct ucred *cred)
{
MAC_PERFORM(create_proc1, cred);
}
void
mac_thread_userret(struct thread *td)
{
MAC_PERFORM(thread_userret, td);
}
/*
* When a new process is created, its label must be initialized. Generally,
* this involves inheritence from the parent process, modulo possible
* deltas. This function allows that processing to take place.
*/
void
mac_create_cred(struct ucred *parent_cred, struct ucred *child_cred)
{
MAC_PERFORM(create_cred, parent_cred, child_cred);
}
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
void
mac_update_devfsdirent(struct devfs_dirent *de, struct vnode *vp)
{
MAC_PERFORM(update_devfsdirent, de, &de->de_label, vp, &vp->v_label);
}
void
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_associate_vnode_devfs(struct mount *mp, struct devfs_dirent *de,
struct vnode *vp)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
MAC_PERFORM(associate_vnode_devfs, mp, &mp->mnt_fslabel, de,
&de->de_label, vp, &vp->v_label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
}
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
int
mac_associate_vnode_extattr(struct mount *mp, struct vnode *vp)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
int error;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
ASSERT_VOP_LOCKED(vp, "mac_associate_vnode_extattr");
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
MAC_CHECK(associate_vnode_extattr, mp, &mp->mnt_fslabel, vp,
&vp->v_label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
return (error);
}
void
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_associate_vnode_singlelabel(struct mount *mp, struct vnode *vp)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
MAC_PERFORM(associate_vnode_singlelabel, mp, &mp->mnt_fslabel, vp,
&vp->v_label);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
}
int
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_create_vnode_extattr(struct ucred *cred, struct mount *mp,
struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
int error;
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
ASSERT_VOP_LOCKED(dvp, "mac_create_vnode_extattr");
ASSERT_VOP_LOCKED(vp, "mac_create_vnode_extattr");
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
error = VOP_OPENEXTATTR(vp, cred, curthread);
if (error == EOPNOTSUPP) {
/* XXX: Optionally abort if transactions not supported. */
if (ea_warn_once == 0) {
printf("Warning: transactions not supported "
"in EA write.\n");
ea_warn_once = 1;
}
} else if (error)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
return (error);
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
MAC_CHECK(create_vnode_extattr, cred, mp, &mp->mnt_fslabel,
dvp, &dvp->v_label, vp, &vp->v_label, cnp);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
if (error) {
VOP_CLOSEEXTATTR(vp, 0, NOCRED, curthread);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
return (error);
}
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
error = VOP_CLOSEEXTATTR(vp, 1, NOCRED, curthread);
if (error == EOPNOTSUPP)
error = 0; /* XXX */
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
return (error);
}
static int
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_setlabel_vnode_extattr(struct ucred *cred, struct vnode *vp,
struct label *intlabel)
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
{
int error;
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
ASSERT_VOP_LOCKED(vp, "mac_setlabel_vnode_extattr");
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
error = VOP_OPENEXTATTR(vp, cred, curthread);
if (error == EOPNOTSUPP) {
/* XXX: Optionally abort if transactions not supported. */
if (ea_warn_once == 0) {
printf("Warning: transactions not supported "
"in EA write.\n");
ea_warn_once = 1;
}
} else if (error)
return (error);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
MAC_CHECK(setlabel_vnode_extattr, cred, vp, &vp->v_label, intlabel);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
if (error) {
VOP_CLOSEEXTATTR(vp, 0, NOCRED, curthread);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
return (error);
}
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
error = VOP_CLOSEEXTATTR(vp, 1, NOCRED, curthread);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
if (error == EOPNOTSUPP)
error = 0; /* XXX */
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
return (error);
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
}
void
mac_execve_transition(struct ucred *old, struct ucred *new, struct vnode *vp)
{
ASSERT_VOP_LOCKED(vp, "mac_execve_transition");
MAC_PERFORM(execve_transition, old, new, vp, &vp->v_label);
}
int
mac_execve_will_transition(struct ucred *old, struct vnode *vp)
{
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
int result;
Move all object label init/destroy routines to the head of the entry points to better match the entry point ordering in mac_policy.h. Big diff, no functional change; merge from the MAC tree. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 16:54:59 +00:00
result = 0;
MAC_BOOLEAN(execve_will_transition, ||, old, vp, &vp->v_label);
return (result);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int
mac_check_vnode_access(struct ucred *cred, struct vnode *vp, int flags)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_access");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_access, cred, vp, &vp->v_label, flags);
return (error);
}
int
mac_check_vnode_chdir(struct ucred *cred, struct vnode *dvp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_chdir");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_chdir, cred, dvp, &dvp->v_label);
return (error);
}
int
mac_check_vnode_chroot(struct ucred *cred, struct vnode *dvp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_chroot");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_chroot, cred, dvp, &dvp->v_label);
return (error);
}
int
mac_check_vnode_create(struct ucred *cred, struct vnode *dvp,
struct componentname *cnp, struct vattr *vap)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_create");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_create, cred, dvp, &dvp->v_label, cnp, vap);
return (error);
}
int
mac_check_vnode_delete(struct ucred *cred, struct vnode *dvp, struct vnode *vp,
struct componentname *cnp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_delete");
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_delete");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_delete, cred, dvp, &dvp->v_label, vp,
&vp->v_label, cnp);
return (error);
}
int
mac_check_vnode_deleteacl(struct ucred *cred, struct vnode *vp,
acl_type_t type)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_deleteacl");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_deleteacl, cred, vp, &vp->v_label, type);
return (error);
}
int
mac_check_vnode_exec(struct ucred *cred, struct vnode *vp)
{
int error;
Add a missing vnode assertion for the exec() check. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 15:28:39 +00:00
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_exec");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (!mac_enforce_process && !mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_exec, cred, vp, &vp->v_label);
return (error);
}
int
mac_check_vnode_getacl(struct ucred *cred, struct vnode *vp, acl_type_t type)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_getacl");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_getacl, cred, vp, &vp->v_label, type);
return (error);
}
int
mac_check_vnode_getextattr(struct ucred *cred, struct vnode *vp,
int attrnamespace, const char *name, struct uio *uio)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_getextattr");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_getextattr, cred, vp, &vp->v_label,
attrnamespace, name, uio);
return (error);
}
Merge support for mac_check_vnode_link(), a MAC framework/policy entry point that instruments the creation of hard links. Policy implementations to follow. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 18:11:36 +00:00
int
mac_check_vnode_link(struct ucred *cred, struct vnode *dvp,
struct vnode *vp, struct componentname *cnp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_link");
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_link");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_link, cred, dvp, &dvp->v_label, vp,
&vp->v_label, cnp);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int
mac_check_vnode_lookup(struct ucred *cred, struct vnode *dvp,
struct componentname *cnp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_lookup");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_lookup, cred, dvp, &dvp->v_label, cnp);
return (error);
}
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
int
mac_check_vnode_mmap(struct ucred *cred, struct vnode *vp, int prot)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
int error;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_mmap");
Add a toggle to disable VM enforcement. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-09-18 02:02:08 +00:00
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
if (!mac_enforce_fs || !mac_enforce_vm)
return (0);
MAC_CHECK(check_vnode_mmap, cred, vp, &vp->v_label, prot);
return (error);
}
void
mac_check_vnode_mmap_downgrade(struct ucred *cred, struct vnode *vp, int *prot)
{
int result = *prot;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_mmap_downgrade");
if (!mac_enforce_fs || !mac_enforce_vm)
return;
MAC_PERFORM(check_vnode_mmap_downgrade, cred, vp, &vp->v_label,
&result);
*prot = result;
}
int
mac_check_vnode_mprotect(struct ucred *cred, struct vnode *vp, int prot)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_mprotect");
if (!mac_enforce_fs || !mac_enforce_vm)
return (0);
MAC_CHECK(check_vnode_mprotect, cred, vp, &vp->v_label, prot);
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
int
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
mac_check_vnode_open(struct ucred *cred, struct vnode *vp, mode_t acc_mode)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
int error;
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_open");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (!mac_enforce_fs)
return (0);
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
MAC_CHECK(check_vnode_open, cred, vp, &vp->v_label, acc_mode);
return (error);
}
int
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
mac_check_vnode_poll(struct ucred *active_cred, struct ucred *file_cred,
struct vnode *vp)
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_poll");
if (!mac_enforce_fs)
return (0);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
MAC_CHECK(check_vnode_poll, active_cred, file_cred, vp,
&vp->v_label);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
int
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
mac_check_vnode_read(struct ucred *active_cred, struct ucred *file_cred,
struct vnode *vp)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
int error;
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_read");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (!mac_enforce_fs)
return (0);
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
MAC_CHECK(check_vnode_read, active_cred, file_cred, vp,
&vp->v_label);
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
int
mac_check_vnode_readdir(struct ucred *cred, struct vnode *dvp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_readdir");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_readdir, cred, dvp, &dvp->v_label);
return (error);
}
int
mac_check_vnode_readlink(struct ucred *cred, struct vnode *vp)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_readlink");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_readlink, cred, vp, &vp->v_label);
return (error);
}
static int
mac_check_vnode_relabel(struct ucred *cred, struct vnode *vp,
struct label *newlabel)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_relabel");
MAC_CHECK(check_vnode_relabel, cred, vp, &vp->v_label, newlabel);
return (error);
}
int
mac_check_vnode_rename_from(struct ucred *cred, struct vnode *dvp,
struct vnode *vp, struct componentname *cnp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_rename_from");
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_rename_from");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_rename_from, cred, dvp, &dvp->v_label, vp,
&vp->v_label, cnp);
return (error);
}
int
mac_check_vnode_rename_to(struct ucred *cred, struct vnode *dvp,
struct vnode *vp, int samedir, struct componentname *cnp)
{
int error;
ASSERT_VOP_LOCKED(dvp, "mac_check_vnode_rename_to");
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_rename_to");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_rename_to, cred, dvp, &dvp->v_label, vp,
vp != NULL ? &vp->v_label : NULL, samedir, cnp);
return (error);
}
int
mac_check_vnode_revoke(struct ucred *cred, struct vnode *vp)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_revoke");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_revoke, cred, vp, &vp->v_label);
return (error);
}
int
mac_check_vnode_setacl(struct ucred *cred, struct vnode *vp, acl_type_t type,
struct acl *acl)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_setacl");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_setacl, cred, vp, &vp->v_label, type, acl);
return (error);
}
int
mac_check_vnode_setextattr(struct ucred *cred, struct vnode *vp,
int attrnamespace, const char *name, struct uio *uio)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_setextattr");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_setextattr, cred, vp, &vp->v_label,
attrnamespace, name, uio);
return (error);
}
int
mac_check_vnode_setflags(struct ucred *cred, struct vnode *vp, u_long flags)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_setflags");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_setflags, cred, vp, &vp->v_label, flags);
return (error);
}
int
mac_check_vnode_setmode(struct ucred *cred, struct vnode *vp, mode_t mode)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_setmode");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_setmode, cred, vp, &vp->v_label, mode);
return (error);
}
int
mac_check_vnode_setowner(struct ucred *cred, struct vnode *vp, uid_t uid,
gid_t gid)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_setowner");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_setowner, cred, vp, &vp->v_label, uid, gid);
return (error);
}
int
mac_check_vnode_setutimes(struct ucred *cred, struct vnode *vp,
struct timespec atime, struct timespec mtime)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_setutimes");
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_vnode_setutimes, cred, vp, &vp->v_label, atime,
mtime);
return (error);
}
int
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
mac_check_vnode_stat(struct ucred *active_cred, struct ucred *file_cred,
struct vnode *vp)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_stat");
if (!mac_enforce_fs)
return (0);
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
MAC_CHECK(check_vnode_stat, active_cred, file_cred, vp,
&vp->v_label);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
Introduce MAC_CHECK_VNODE_SWAPON, which permits MAC policies to perform authorization checks during swapon() events; policies might choose to enforce protections based on the credential requesting the swap configuration, the target of the swap operation, or other factors such as internal policy state. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 15:53:43 +00:00
int
mac_check_vnode_swapon(struct ucred *cred, struct vnode *vp)
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_swapon");
Comment describing the semantics of mac_late. Trim trailing whitespace. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-25 20:45:27 +00:00
if (!mac_enforce_fs)
Introduce MAC_CHECK_VNODE_SWAPON, which permits MAC policies to perform authorization checks during swapon() events; policies might choose to enforce protections based on the credential requesting the swap configuration, the target of the swap operation, or other factors such as internal policy state. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 15:53:43 +00:00
return (0);
MAC_CHECK(check_vnode_swapon, cred, vp, &vp->v_label);
return (error);
}
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
int
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
mac_check_vnode_write(struct ucred *active_cred, struct ucred *file_cred,
struct vnode *vp)
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
{
int error;
ASSERT_VOP_LOCKED(vp, "mac_check_vnode_write");
if (!mac_enforce_fs)
return (0);
Pass active_cred and file_cred into the MAC framework explicitly for mac_check_vnode_{poll,read,stat,write}(). Pass in fp->f_cred when calling these checks with a struct file available. Otherwise, pass NOCRED. All currently MAC policies use active_cred, but could now offer the cached credential semantic used for the base system security model. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 19:04:53 +00:00
MAC_CHECK(check_vnode_write, active_cred, file_cred, vp,
&vp->v_label);
Break out mac_check_vnode_op() into three seperate checks: mac_check_vnode_poll(), mac_check_vnode_read(), mac_check_vnode_write(). This improves the consistency with other existing vnode checks, and allows policies to avoid implementing switch statements to determine what operations they do and do not want to authorize. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:43:25 +00:00
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
/*
* When relabeling a process, call out to the policies for the maximum
* permission allowed for each object type we know about in its
* memory space, and revoke access (in the least surprising ways we
* know) when necessary. The process lock is not held here.
*/
static void
mac_cred_mmapped_drop_perms(struct thread *td, struct ucred *cred)
{
/* XXX freeze all other threads */
mac_cred_mmapped_drop_perms_recurse(td, cred,
&td->td_proc->p_vmspace->vm_map);
/* XXX allow other threads to continue */
}
static __inline const char *
prot2str(vm_prot_t prot)
{
switch (prot & VM_PROT_ALL) {
case VM_PROT_READ:
return ("r--");
case VM_PROT_READ | VM_PROT_WRITE:
return ("rw-");
case VM_PROT_READ | VM_PROT_EXECUTE:
return ("r-x");
case VM_PROT_READ | VM_PROT_WRITE | VM_PROT_EXECUTE:
return ("rwx");
case VM_PROT_WRITE:
return ("-w-");
case VM_PROT_EXECUTE:
return ("--x");
case VM_PROT_WRITE | VM_PROT_EXECUTE:
return ("-wx");
default:
return ("---");
}
}
static void
mac_cred_mmapped_drop_perms_recurse(struct thread *td, struct ucred *cred,
struct vm_map *map)
{
struct vm_map_entry *vme;
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
int result;
vm_prot_t revokeperms;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
vm_object_t object;
vm_ooffset_t offset;
struct vnode *vp;
Add security.mac.mmap_revocation, a flag indicating whether we should revoke access to memory maps on a process label change. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-09-09 17:12:24 +00:00
if (!mac_mmap_revocation)
return;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
vm_map_lock_read(map);
for (vme = map->header.next; vme != &map->header; vme = vme->next) {
if (vme->eflags & MAP_ENTRY_IS_SUB_MAP) {
mac_cred_mmapped_drop_perms_recurse(td, cred,
vme->object.sub_map);
continue;
}
/*
* Skip over entries that obviously are not shared.
*/
if (vme->eflags & (MAP_ENTRY_COW | MAP_ENTRY_NOSYNC) ||
!vme->max_protection)
continue;
/*
* Drill down to the deepest backing object.
*/
offset = vme->offset;
object = vme->object.vm_object;
if (object == NULL)
continue;
while (object->backing_object != NULL) {
object = object->backing_object;
offset += object->backing_object_offset;
}
/*
* At the moment, vm_maps and objects aren't considered
* by the MAC system, so only things with backing by a
* normal object (read: vnodes) are checked.
*/
if (object->type != OBJT_VNODE)
continue;
vp = (struct vnode *)object->handle;
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
Sync from MAC tree: break out the single mmap entry point into seperate entry points for each occasion: mac_check_vnode_mmap() Check at initial mapping mac_check_vnode_mprotect() Check at mapping protection change mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade should take place following subject relabel. Implement mmap() and mprotect() entry points for labeled vnode policies. These entry points are currently not hooked up to the VM system in the base tree. These changes improve the consistency of the access control interface and offer more flexibility regarding limiting access to vnode mmaping. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 02:46:26 +00:00
result = vme->max_protection;
mac_check_vnode_mmap_downgrade(cred, vp, &result);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
VOP_UNLOCK(vp, 0, td);
/*
* Find out what maximum protection we may be allowing
* now but a policy needs to get removed.
*/
revokeperms = vme->max_protection & ~result;
if (!revokeperms)
continue;
Include <sys/malloc.h> instead of depending on namespace pollution 2 layers deep in <sys/proc.h> or <sys/vnode.h>. Removed unused includes. Fixed some printf format errors (1 fatal on i386's; 1 fatal on alphas; 1 not fatal on any supported machine).
2002-09-05 07:02:43 +00:00
printf("pid %ld: revoking %s perms from %#lx:%ld "
"(max %s/cur %s)\n", (long)td->td_proc->p_pid,
prot2str(revokeperms), (u_long)vme->start,
(long)(vme->end - vme->start),
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
prot2str(vme->max_protection), prot2str(vme->protection));
vm_map_lock_upgrade(map);
/*
* This is the really simple case: if a map has more
* max_protection than is allowed, but it's not being
* actually used (that is, the current protection is
* still allowed), we can just wipe it out and do
* nothing more.
*/
if ((vme->protection & revokeperms) == 0) {
vme->max_protection -= revokeperms;
} else {
if (revokeperms & VM_PROT_WRITE) {
/*
* In the more complicated case, flush out all
* pending changes to the object then turn it
* copy-on-write.
*/
vm_object_reference(object);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
vm_object_page_clean(object,
OFF_TO_IDX(offset),
OFF_TO_IDX(offset + vme->end - vme->start +
PAGE_MASK),
OBJPC_SYNC);
VOP_UNLOCK(vp, 0, td);
vm_object_deallocate(object);
/*
* Why bother if there's no read permissions
* anymore? For the rest, we need to leave
* the write permissions on for COW, or
* remove them entirely if configured to.
*/
if (!mac_mmap_revocation_via_cow) {
vme->max_protection &= ~VM_PROT_WRITE;
vme->protection &= ~VM_PROT_WRITE;
} if ((revokeperms & VM_PROT_READ) == 0)
vme->eflags |= MAP_ENTRY_COW |
MAP_ENTRY_NEEDS_COPY;
}
if (revokeperms & VM_PROT_EXECUTE) {
vme->max_protection &= ~VM_PROT_EXECUTE;
vme->protection &= ~VM_PROT_EXECUTE;
}
if (revokeperms & VM_PROT_READ) {
vme->max_protection = 0;
vme->protection = 0;
}
pmap_protect(map->pmap, vme->start, vme->end,
vme->protection & ~revokeperms);
vm_map_simplify_entry(map, vme);
}
vm_map_lock_downgrade(map);
}
vm_map_unlock_read(map);
}
/*
* When the subject's label changes, it may require revocation of privilege
* to mapped objects. This can't be done on-the-fly later with a unified
* buffer cache.
*/
static void
mac_relabel_cred(struct ucred *cred, struct label *newlabel)
{
MAC_PERFORM(relabel_cred, cred, newlabel);
}
void
mac_relabel_vnode(struct ucred *cred, struct vnode *vp, struct label *newlabel)
{
MAC_PERFORM(relabel_vnode, cred, vp, &vp->v_label, newlabel);
}
void
mac_create_ifnet(struct ifnet *ifnet)
{
MAC_PERFORM(create_ifnet, ifnet, &ifnet->if_label);
}
void
mac_create_bpfdesc(struct ucred *cred, struct bpf_d *bpf_d)
{
MAC_PERFORM(create_bpfdesc, cred, bpf_d, &bpf_d->bd_label);
}
void
mac_create_socket(struct ucred *cred, struct socket *socket)
{
MAC_PERFORM(create_socket, cred, socket, &socket->so_label);
}
void
mac_create_pipe(struct ucred *cred, struct pipe *pipe)
{
MAC_PERFORM(create_pipe, cred, pipe, pipe->pipe_label);
}
void
mac_create_socket_from_socket(struct socket *oldsocket,
struct socket *newsocket)
{
MAC_PERFORM(create_socket_from_socket, oldsocket, &oldsocket->so_label,
newsocket, &newsocket->so_label);
}
static void
mac_relabel_socket(struct ucred *cred, struct socket *socket,
struct label *newlabel)
{
MAC_PERFORM(relabel_socket, cred, socket, &socket->so_label, newlabel);
}
static void
mac_relabel_pipe(struct ucred *cred, struct pipe *pipe, struct label *newlabel)
{
MAC_PERFORM(relabel_pipe, cred, pipe, pipe->pipe_label, newlabel);
}
void
mac_set_socket_peer_from_mbuf(struct mbuf *mbuf, struct socket *socket)
{
MAC_PERFORM(set_socket_peer_from_mbuf, mbuf, &mbuf->m_pkthdr.label,
socket, &socket->so_peerlabel);
}
void
mac_set_socket_peer_from_socket(struct socket *oldsocket,
struct socket *newsocket)
{
MAC_PERFORM(set_socket_peer_from_socket, oldsocket,
&oldsocket->so_label, newsocket, &newsocket->so_peerlabel);
}
void
mac_create_datagram_from_ipq(struct ipq *ipq, struct mbuf *datagram)
{
MAC_PERFORM(create_datagram_from_ipq, ipq, &ipq->ipq_label,
datagram, &datagram->m_pkthdr.label);
}
void
mac_create_fragment(struct mbuf *datagram, struct mbuf *fragment)
{
MAC_PERFORM(create_fragment, datagram, &datagram->m_pkthdr.label,
fragment, &fragment->m_pkthdr.label);
}
void
mac_create_ipq(struct mbuf *fragment, struct ipq *ipq)
{
MAC_PERFORM(create_ipq, fragment, &fragment->m_pkthdr.label, ipq,
&ipq->ipq_label);
}
void
mac_create_mbuf_from_mbuf(struct mbuf *oldmbuf, struct mbuf *newmbuf)
{
MAC_PERFORM(create_mbuf_from_mbuf, oldmbuf, &oldmbuf->m_pkthdr.label,
newmbuf, &newmbuf->m_pkthdr.label);
}
void
mac_create_mbuf_from_bpfdesc(struct bpf_d *bpf_d, struct mbuf *mbuf)
{
MAC_PERFORM(create_mbuf_from_bpfdesc, bpf_d, &bpf_d->bd_label, mbuf,
&mbuf->m_pkthdr.label);
}
void
mac_create_mbuf_linklayer(struct ifnet *ifnet, struct mbuf *mbuf)
{
MAC_PERFORM(create_mbuf_linklayer, ifnet, &ifnet->if_label, mbuf,
&mbuf->m_pkthdr.label);
}
void
mac_create_mbuf_from_ifnet(struct ifnet *ifnet, struct mbuf *mbuf)
{
MAC_PERFORM(create_mbuf_from_ifnet, ifnet, &ifnet->if_label, mbuf,
&mbuf->m_pkthdr.label);
}
void
mac_create_mbuf_multicast_encap(struct mbuf *oldmbuf, struct ifnet *ifnet,
struct mbuf *newmbuf)
{
MAC_PERFORM(create_mbuf_multicast_encap, oldmbuf,
&oldmbuf->m_pkthdr.label, ifnet, &ifnet->if_label, newmbuf,
&newmbuf->m_pkthdr.label);
}
void
mac_create_mbuf_netlayer(struct mbuf *oldmbuf, struct mbuf *newmbuf)
{
MAC_PERFORM(create_mbuf_netlayer, oldmbuf, &oldmbuf->m_pkthdr.label,
newmbuf, &newmbuf->m_pkthdr.label);
}
int
mac_fragment_match(struct mbuf *fragment, struct ipq *ipq)
{
int result;
result = 1;
MAC_BOOLEAN(fragment_match, &&, fragment, &fragment->m_pkthdr.label,
ipq, &ipq->ipq_label);
return (result);
}
void
mac_update_ipq(struct mbuf *fragment, struct ipq *ipq)
{
MAC_PERFORM(update_ipq, fragment, &fragment->m_pkthdr.label, ipq,
&ipq->ipq_label);
}
void
mac_create_mbuf_from_socket(struct socket *socket, struct mbuf *mbuf)
{
MAC_PERFORM(create_mbuf_from_socket, socket, &socket->so_label, mbuf,
&mbuf->m_pkthdr.label);
}
void
mac_create_mount(struct ucred *cred, struct mount *mp)
{
MAC_PERFORM(create_mount, cred, mp, &mp->mnt_mntlabel,
&mp->mnt_fslabel);
}
void
mac_create_root_mount(struct ucred *cred, struct mount *mp)
{
MAC_PERFORM(create_root_mount, cred, mp, &mp->mnt_mntlabel,
&mp->mnt_fslabel);
}
int
mac_check_bpfdesc_receive(struct bpf_d *bpf_d, struct ifnet *ifnet)
{
int error;
if (!mac_enforce_network)
return (0);
MAC_CHECK(check_bpfdesc_receive, bpf_d, &bpf_d->bd_label, ifnet,
&ifnet->if_label);
return (error);
}
static int
mac_check_cred_relabel(struct ucred *cred, struct label *newlabel)
{
int error;
MAC_CHECK(check_cred_relabel, cred, newlabel);
return (error);
}
int
mac_check_cred_visible(struct ucred *u1, struct ucred *u2)
{
int error;
if (!mac_enforce_process)
return (0);
MAC_CHECK(check_cred_visible, u1, u2);
return (error);
}
int
mac_check_ifnet_transmit(struct ifnet *ifnet, struct mbuf *mbuf)
{
int error;
if (!mac_enforce_network)
return (0);
KASSERT(mbuf->m_flags & M_PKTHDR, ("packet has no pkthdr"));
if (!(mbuf->m_pkthdr.label.l_flags & MAC_FLAG_INITIALIZED))
Use if_printf(ifp, "blah") instead of printf("%s%d: blah", ifp->if_name, ifp->if_xname).
2002-10-21 02:51:56 +00:00
if_printf(ifnet, "not initialized\n");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MAC_CHECK(check_ifnet_transmit, ifnet, &ifnet->if_label, mbuf,
&mbuf->m_pkthdr.label);
return (error);
}
int
mac_check_mount_stat(struct ucred *cred, struct mount *mount)
{
int error;
if (!mac_enforce_fs)
return (0);
MAC_CHECK(check_mount_stat, cred, mount, &mount->mnt_mntlabel);
return (error);
}
int
mac_check_pipe_ioctl(struct ucred *cred, struct pipe *pipe, unsigned long cmd,
void *data)
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
if (!mac_enforce_pipe)
return (0);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MAC_CHECK(check_pipe_ioctl, cred, pipe, pipe->pipe_label, cmd, data);
return (error);
}
int
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
mac_check_pipe_poll(struct ucred *cred, struct pipe *pipe)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
if (!mac_enforce_pipe)
return (0);
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
MAC_CHECK(check_pipe_poll, cred, pipe, pipe->pipe_label);
return (error);
}
int
mac_check_pipe_read(struct ucred *cred, struct pipe *pipe)
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
if (!mac_enforce_pipe)
return (0);
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
MAC_CHECK(check_pipe_read, cred, pipe, pipe->pipe_label);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
static int
mac_check_pipe_relabel(struct ucred *cred, struct pipe *pipe,
struct label *newlabel)
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
if (!mac_enforce_pipe)
return (0);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MAC_CHECK(check_pipe_relabel, cred, pipe, pipe->pipe_label, newlabel);
return (error);
}
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
int
mac_check_pipe_stat(struct ucred *cred, struct pipe *pipe)
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
if (!mac_enforce_pipe)
return (0);
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
MAC_CHECK(check_pipe_stat, cred, pipe, pipe->pipe_label);
return (error);
}
int
mac_check_pipe_write(struct ucred *cred, struct pipe *pipe)
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
if (!mac_enforce_pipe)
return (0);
Break out mac_check_pipe_op() into component check entry points: mac_check_pipe_poll(), mac_check_pipe_read(), mac_check_pipe_stat(), and mac_check_pipe_write(). This is improves consistency with other access control entry points and permits security modules to only control the object methods that they are interested in, avoiding switch statements. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 16:59:37 +00:00
MAC_CHECK(check_pipe_write, cred, pipe, pipe->pipe_label);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int
mac_check_proc_debug(struct ucred *cred, struct proc *proc)
{
int error;
Assert process locks in proces-related access control checks. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 15:30:30 +00:00
PROC_LOCK_ASSERT(proc, MA_OWNED);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (!mac_enforce_process)
return (0);
MAC_CHECK(check_proc_debug, cred, proc);
return (error);
}
int
mac_check_proc_sched(struct ucred *cred, struct proc *proc)
{
int error;
Assert process locks in proces-related access control checks. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 15:30:30 +00:00
PROC_LOCK_ASSERT(proc, MA_OWNED);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (!mac_enforce_process)
return (0);
MAC_CHECK(check_proc_sched, cred, proc);
return (error);
}
int
mac_check_proc_signal(struct ucred *cred, struct proc *proc, int signum)
{
int error;
Assert process locks in proces-related access control checks. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 15:30:30 +00:00
PROC_LOCK_ASSERT(proc, MA_OWNED);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (!mac_enforce_process)
return (0);
MAC_CHECK(check_proc_signal, cred, proc, signum);
return (error);
}
int
mac_check_socket_bind(struct ucred *ucred, struct socket *socket,
struct sockaddr *sockaddr)
{
int error;
if (!mac_enforce_socket)
return (0);
MAC_CHECK(check_socket_bind, ucred, socket, &socket->so_label,
sockaddr);
return (error);
}
int
mac_check_socket_connect(struct ucred *cred, struct socket *socket,
struct sockaddr *sockaddr)
{
int error;
if (!mac_enforce_socket)
return (0);
MAC_CHECK(check_socket_connect, cred, socket, &socket->so_label,
sockaddr);
return (error);
}
int
Rename mac_check_socket_receive() to mac_check_socket_deliver() so that we can use the names _receive() and _send() for the receive() and send() checks. Rename related constants, policy implementations, etc. PR: Submitted by: Reviewed by: Approved by: Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs MFC after:
2002-08-15 18:51:26 +00:00
mac_check_socket_deliver(struct socket *socket, struct mbuf *mbuf)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
int error;
if (!mac_enforce_socket)
return (0);
Rename mac_check_socket_receive() to mac_check_socket_deliver() so that we can use the names _receive() and _send() for the receive() and send() checks. Rename related constants, policy implementations, etc. PR: Submitted by: Reviewed by: Approved by: Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs MFC after:
2002-08-15 18:51:26 +00:00
MAC_CHECK(check_socket_deliver, socket, &socket->so_label, mbuf,
&mbuf->m_pkthdr.label);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
int
Rename mac_check_socket_receive() to mac_check_socket_deliver() so that we can use the names _receive() and _send() for the receive() and send() checks. Rename related constants, policy implementations, etc. PR: Submitted by: Reviewed by: Approved by: Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs MFC after:
2002-08-15 18:51:26 +00:00
mac_check_socket_listen(struct ucred *cred, struct socket *socket)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
int error;
if (!mac_enforce_socket)
return (0);
Rename mac_check_socket_receive() to mac_check_socket_deliver() so that we can use the names _receive() and _send() for the receive() and send() checks. Rename related constants, policy implementations, etc. PR: Submitted by: Reviewed by: Approved by: Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs MFC after:
2002-08-15 18:51:26 +00:00
MAC_CHECK(check_socket_listen, cred, socket, &socket->so_label);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
Integrate mac_check_socket_send() and mac_check_socket_receive() checks from the MAC tree: allow policies to perform access control for the ability of a process to send and receive data via a socket. At some point, we might also pass in additional address information if an explicit address is requested on send. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 14:39:15 +00:00
int
mac_check_socket_receive(struct ucred *cred, struct socket *so)
{
int error;
if (!mac_enforce_socket)
return (0);
MAC_CHECK(check_socket_receive, cred, so, &so->so_label);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
static int
mac_check_socket_relabel(struct ucred *cred, struct socket *socket,
struct label *newlabel)
{
int error;
MAC_CHECK(check_socket_relabel, cred, socket, &socket->so_label,
newlabel);
return (error);
}
Integrate mac_check_socket_send() and mac_check_socket_receive() checks from the MAC tree: allow policies to perform access control for the ability of a process to send and receive data via a socket. At some point, we might also pass in additional address information if an explicit address is requested on send. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-06 14:39:15 +00:00
int
mac_check_socket_send(struct ucred *cred, struct socket *so)
{
int error;
if (!mac_enforce_socket)
return (0);
MAC_CHECK(check_socket_send, cred, so, &so->so_label);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int
mac_check_socket_visible(struct ucred *cred, struct socket *socket)
{
int error;
if (!mac_enforce_socket)
return (0);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MAC_CHECK(check_socket_visible, cred, socket, &socket->so_label);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
int
mac_ioctl_ifnet_get(struct ucred *cred, struct ifreq *ifr,
struct ifnet *ifnet)
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *elements, *buffer;
struct mac mac;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(ifr->ifr_ifru.ifru_data, &mac, sizeof(mac));
if (error)
return (error);
error = mac_check_structmac_consistent(&mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
elements = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, elements, mac.m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
error = mac_externalize_ifnet_label(&ifnet->if_label, elements,
buffer, mac.m_buflen, M_WAITOK);
if (error == 0)
error = copyout(buffer, mac.m_string, strlen(buffer)+1);
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
int
mac_ioctl_ifnet_set(struct ucred *cred, struct ifreq *ifr,
struct ifnet *ifnet)
{
struct label intlabel;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac mac;
char *buffer;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(ifr->ifr_ifru.ifru_data, &mac, sizeof(mac));
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = mac_check_structmac_consistent(&mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, buffer, mac.m_buflen, NULL);
if (error) {
free(buffer, M_MACTEMP);
return (error);
}
mac_init_ifnet_label(&intlabel);
error = mac_internalize_ifnet_label(&intlabel, buffer);
free(buffer, M_MACTEMP);
if (error) {
mac_destroy_ifnet_label(&intlabel);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
/*
* XXX: Note that this is a redundant privilege check, since
* policies impose this check themselves if required by the
* policy. Eventually, this should go away.
*/
error = suser_cred(cred, 0);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
if (error) {
mac_destroy_ifnet_label(&intlabel);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MAC_CHECK(check_ifnet_relabel, cred, ifnet, &ifnet->if_label,
&intlabel);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
if (error) {
mac_destroy_ifnet_label(&intlabel);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
MAC_PERFORM(relabel_ifnet, cred, ifnet, &ifnet->if_label, &intlabel);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_destroy_ifnet_label(&intlabel);
return (0);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
void
mac_create_devfs_vnode(struct devfs_dirent *de, struct vnode *vp)
{
MAC_PERFORM(create_devfs_vnode, de, &de->de_label, vp, &vp->v_label);
}
void
mac_create_devfs_device(dev_t dev, struct devfs_dirent *de)
{
MAC_PERFORM(create_devfs_device, dev, de, &de->de_label);
}
Integrate a devfs/MAC fix from the MAC tree: avoid a race condition during devfs VOP symlink creation by introducing a new entry point to determine the label of the devfs_dirent prior to allocation of a vnode for the symlink. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-05 18:40:10 +00:00
void
mac_create_devfs_symlink(struct ucred *cred, struct devfs_dirent *dd,
struct devfs_dirent *de)
{
MAC_PERFORM(create_devfs_symlink, cred, dd, &dd->de_label, de,
&de->de_label);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
void
mac_create_devfs_directory(char *dirname, int dirnamelen,
struct devfs_dirent *de)
{
MAC_PERFORM(create_devfs_directory, dirname, dirnamelen, de,
&de->de_label);
}
int
mac_setsockopt_label_set(struct ucred *cred, struct socket *so,
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac *mac)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
struct label intlabel;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *buffer;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = mac_check_structmac_consistent(mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
buffer = malloc(mac->m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac->m_string, buffer, mac->m_buflen, NULL);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error) {
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
free(buffer, M_MACTEMP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_socket_label(&intlabel, M_WAITOK);
error = mac_internalize_socket_label(&intlabel, buffer);
free(buffer, M_MACTEMP);
if (error) {
mac_destroy_socket_label(&intlabel);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_check_socket_relabel(cred, so, &intlabel);
if (error) {
mac_destroy_socket_label(&intlabel);
return (error);
}
mac_relabel_socket(cred, so, &intlabel);
mac_destroy_socket_label(&intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (0);
}
int
mac_pipe_label_set(struct ucred *cred, struct pipe *pipe, struct label *label)
{
int error;
Improve locking of pipe mutexes in the context of MAC: (1) Where previously the pipe mutex was selectively grabbed during pipe_ioctl(), now always grab it and then release if if not needed. This protects the call to mac_check_pipe_ioctl() to make sure the label remains consistent. (Note: it looks like sigio locking may be incorrect for fgetown() since we call it not-by-reference and sigio locking assumes call by reference). (2) In pipe_stat(), lock the pipe if MAC is compiled in so that the call to mac_check_pipe_stat() gets a locked pipe to protect label consistency. We still release the lock before returning actual stat() data, risking inconsistency, but apparently our pipe locking model accepts that risk. (3) In various pipe MAC authorization checks, assert that the pipe lock is held. (4) Grab the lock when performing a pipe relabel operation, and assert it a little deeper in the stack. Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-01 04:30:19 +00:00
PIPE_LOCK_ASSERT(pipe, MA_OWNED);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
error = mac_check_pipe_relabel(cred, pipe, label);
if (error)
return (error);
mac_relabel_pipe(cred, pipe, label);
return (0);
}
int
mac_getsockopt_label_get(struct ucred *cred, struct socket *so,
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac *mac)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *buffer, *elements;
int error;
error = mac_check_structmac_consistent(mac);
if (error)
return (error);
elements = malloc(mac->m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac->m_string, elements, mac->m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
buffer = malloc(mac->m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
error = mac_externalize_socket_label(&so->so_label, elements,
buffer, mac->m_buflen, M_WAITOK);
if (error == 0)
error = copyout(buffer, mac->m_string, strlen(buffer)+1);
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
int
mac_getsockopt_peerlabel_get(struct ucred *cred, struct socket *so,
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac *mac)
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *elements, *buffer;
int error;
error = mac_check_structmac_consistent(mac);
if (error)
return (error);
elements = malloc(mac->m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac->m_string, elements, mac->m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
buffer = malloc(mac->m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
error = mac_externalize_socket_peer_label(&so->so_peerlabel,
elements, buffer, mac->m_buflen, M_WAITOK);
if (error == 0)
error = copyout(buffer, mac->m_string, strlen(buffer)+1);
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
/*
* Implementation of VOP_SETLABEL() that relies on extended attributes
* to store label data. Can be referenced by filesystems supporting
* extended attributes.
*/
int
vop_stdsetlabel_ea(struct vop_setlabel_args *ap)
{
struct vnode *vp = ap->a_vp;
struct label *intlabel = ap->a_label;
int error;
ASSERT_VOP_LOCKED(vp, "vop_stdsetlabel_ea");
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
if ((vp->v_mount->mnt_flag & MNT_MULTILABEL) == 0)
return (EOPNOTSUPP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
error = mac_setlabel_vnode_extattr(ap->a_cred, vp, intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
mac_relabel_vnode(ap->a_cred, vp, intlabel);
return (0);
}
static int
vn_setlabel(struct vnode *vp, struct label *intlabel, struct ucred *cred)
{
int error;
if (vp->v_mount == NULL) {
/* printf("vn_setlabel: null v_mount\n"); */
Remove all use of vnode->v_tag, replacing with appropriate substitutes. v_tag is now const char * and should only be used for debugging. Additionally: 1. All users of VT_NTS now check vfsconf->vf_type VFCF_NETWORK 2. The user of VT_PROCFS now checks for the new flag VV_PROCDEP, which is propagated by pseudofs to all child vnodes if the fs sets PFS_PROCDEP. Suggested by: phk Reviewed by: bde, rwatson (earlier version)
2002-09-14 09:02:28 +00:00
if (vp->v_type != VNON)
printf("vn_setlabel: null v_mount with non-VNON\n");
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (EBADF);
}
if ((vp->v_mount->mnt_flag & MNT_MULTILABEL) == 0)
return (EOPNOTSUPP);
/*
* Multi-phase commit. First check the policies to confirm the
* change is OK. Then commit via the filesystem. Finally,
* update the actual vnode label. Question: maybe the filesystem
* should update the vnode at the end as part of VOP_SETLABEL()?
*/
error = mac_check_vnode_relabel(cred, vp, intlabel);
if (error)
return (error);
/*
* VADMIN provides the opportunity for the filesystem to make
* decisions about who is and is not able to modify labels
* and protections on files. This might not be right. We can't
* assume VOP_SETLABEL() will do it, because we might implement
* that as part of vop_stdsetlabel_ea().
*/
error = VOP_ACCESS(vp, VADMIN, cred, curthread);
if (error)
return (error);
error = VOP_SETLABEL(vp, intlabel, cred, curthread);
if (error)
return (error);
return (0);
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
int
__mac_get_pid(struct thread *td, struct __mac_get_pid_args *uap)
{
char *elements, *buffer;
struct mac mac;
struct proc *tproc;
struct ucred *tcred;
int error;
error = copyin(SCARG(uap, mac_p), &mac, sizeof(mac));
if (error)
return (error);
error = mac_check_structmac_consistent(&mac);
if (error)
return (error);
tproc = pfind(uap->pid);
if (tproc == NULL)
return (ESRCH);
tcred = NULL; /* Satisfy gcc. */
error = p_cansee(td, tproc);
if (error == 0)
tcred = crhold(tproc->p_ucred);
PROC_UNLOCK(tproc);
if (error)
return (error);
elements = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, elements, mac.m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
crfree(tcred);
return (error);
}
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
error = mac_externalize_cred_label(&tcred->cr_label, elements,
buffer, mac.m_buflen, M_WAITOK);
if (error == 0)
error = copyout(buffer, mac.m_string, strlen(buffer)+1);
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
crfree(tcred);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
/*
* MPSAFE
*/
int
__mac_get_proc(struct thread *td, struct __mac_get_proc_args *uap)
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *elements, *buffer;
struct mac mac;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(uap->mac_p, &mac, sizeof(mac));
if (error)
return (error);
error = mac_check_structmac_consistent(&mac);
if (error)
return (error);
elements = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, elements, mac.m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
error = mac_externalize_cred_label(&td->td_ucred->cr_label,
elements, buffer, mac.m_buflen, M_WAITOK);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error == 0)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyout(buffer, mac.m_string, strlen(buffer)+1);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
/*
* MPSAFE
*/
int
__mac_set_proc(struct thread *td, struct __mac_set_proc_args *uap)
{
struct ucred *newcred, *oldcred;
struct label intlabel;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct proc *p;
struct mac mac;
char *buffer;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(uap->mac_p, &mac, sizeof(mac));
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = mac_check_structmac_consistent(&mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
return (error);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, buffer, mac.m_buflen, NULL);
if (error) {
free(buffer, M_MACTEMP);
return (error);
}
mac_init_cred_label(&intlabel);
error = mac_internalize_cred_label(&intlabel, buffer);
free(buffer, M_MACTEMP);
if (error) {
mac_destroy_cred_label(&intlabel);
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
newcred = crget();
p = td->td_proc;
PROC_LOCK(p);
oldcred = p->p_ucred;
error = mac_check_cred_relabel(oldcred, &intlabel);
if (error) {
PROC_UNLOCK(p);
crfree(newcred);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
goto out;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
setsugid(p);
crcopy(newcred, oldcred);
mac_relabel_cred(newcred, &intlabel);
p->p_ucred = newcred;
Close a race in process label changing opened due to dropping the proc locking when revoking access to mmaps. Instead, perform this later once we've changed the process label (hold onto a reference to the new cred so that we don't lose it when we release the process lock if another thread changes the credential). Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 20:26:32 +00:00
/*
* Grab additional reference for use while revoking mmaps, prior
* to releasing the proc lock and sharing the cred.
*/
crhold(newcred);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
PROC_UNLOCK(p);
Close a race in process label changing opened due to dropping the proc locking when revoking access to mmaps. Instead, perform this later once we've changed the process label (hold onto a reference to the new cred so that we don't lose it when we release the process lock if another thread changes the credential). Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 20:26:32 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
if (mac_enforce_vm) {
mtx_lock(&Giant);
mac_cred_mmapped_drop_perms(td, newcred);
mtx_unlock(&Giant);
}
Close a race in process label changing opened due to dropping the proc locking when revoking access to mmaps. Instead, perform this later once we've changed the process label (hold onto a reference to the new cred so that we don't lose it when we release the process lock if another thread changes the credential). Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 20:26:32 +00:00
crfree(newcred); /* Free revocation reference. */
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
crfree(oldcred);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
out:
mac_destroy_cred_label(&intlabel);
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
/*
* MPSAFE
*/
int
__mac_get_fd(struct thread *td, struct __mac_get_fd_args *uap)
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *elements, *buffer;
struct label intlabel;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
struct file *fp;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac mac;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
struct vnode *vp;
struct pipe *pipe;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
short label_type;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(uap->mac_p, &mac, sizeof(mac));
if (error)
return (error);
error = mac_check_structmac_consistent(&mac);
if (error)
return (error);
elements = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, elements, mac.m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
mtx_lock(&Giant); /* VFS */
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
error = fget(td, SCARG(uap, fd), &fp);
if (error)
goto out;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
label_type = fp->f_type;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
switch (fp->f_type) {
case DTYPE_FIFO:
case DTYPE_VNODE:
vp = (struct vnode *)fp->f_data;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_vnode_label(&intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_copy_vnode_label(&vp->v_label, &intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
VOP_UNLOCK(vp, 0, td);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
case DTYPE_PIPE:
pipe = (struct pipe *)fp->f_data;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_pipe_label(&intlabel);
PIPE_LOCK(pipe);
mac_copy_pipe_label(pipe->pipe_label, &intlabel);
PIPE_UNLOCK(pipe);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
default:
error = EINVAL;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
fdrop(fp, td);
goto out;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
fdrop(fp, td);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
switch (label_type) {
case DTYPE_FIFO:
case DTYPE_VNODE:
if (error == 0)
error = mac_externalize_vnode_label(&intlabel,
elements, buffer, mac.m_buflen, M_WAITOK);
mac_destroy_vnode_label(&intlabel);
break;
case DTYPE_PIPE:
error = mac_externalize_pipe_label(&intlabel, elements,
buffer, mac.m_buflen, M_WAITOK);
mac_destroy_pipe_label(&intlabel);
break;
default:
panic("__mac_get_fd: corrupted label_type");
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
if (error == 0)
error = copyout(buffer, mac.m_string, strlen(buffer)+1);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
out:
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mtx_unlock(&Giant); /* VFS */
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
/*
* MPSAFE
*/
int
__mac_get_file(struct thread *td, struct __mac_get_file_args *uap)
{
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
char *elements, *buffer;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
struct nameidata nd;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct label intlabel;
struct mac mac;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(uap->mac_p, &mac, sizeof(mac));
if (error)
return (error);
error = mac_check_structmac_consistent(&mac);
if (error)
return (error);
elements = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, elements, mac.m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
mtx_lock(&Giant); /* VFS */
NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_USERSPACE, uap->path_p,
td);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
error = namei(&nd);
if (error)
goto out;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_vnode_label(&intlabel);
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_copy_vnode_label(&nd.ni_vp->v_label, &intlabel);
error = mac_externalize_vnode_label(&intlabel, elements, buffer,
mac.m_buflen, M_WAITOK);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
NDFREE(&nd, 0);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_destroy_vnode_label(&intlabel);
if (error == 0)
error = copyout(buffer, mac.m_string, strlen(buffer)+1);
out:
mtx_unlock(&Giant); /* VFS */
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
return (error);
}
/*
* MPSAFE
*/
int
__mac_get_link(struct thread *td, struct __mac_get_link_args *uap)
{
char *elements, *buffer;
struct nameidata nd;
struct label intlabel;
struct mac mac;
int error;
error = copyin(uap->mac_p, &mac, sizeof(mac));
if (error)
return (error);
error = mac_check_structmac_consistent(&mac);
if (error)
return (error);
elements = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, elements, mac.m_buflen, NULL);
if (error) {
free(elements, M_MACTEMP);
return (error);
}
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK | M_ZERO);
mtx_lock(&Giant); /* VFS */
NDINIT(&nd, LOOKUP, LOCKLEAF | NOFOLLOW, UIO_USERSPACE, uap->path_p,
td);
error = namei(&nd);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
goto out;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_vnode_label(&intlabel);
Slightly change the semantics of vnode labels for MAC: rather than "refreshing" the label on the vnode before use, just get the label right from inception. For single-label file systems, set the label in the generic VFS getnewvnode() code; for multi-label file systems, leave the labeling up to the file system. With UFS1/2, this means reading the extended attribute during vfs_vget() as the inode is pulled off disk, rather than hitting the extended attributes frequently during operations later, improving performance. This also corrects sematics for shared vnode locks, which were not previously present in the system. This chances the cache coherrency properties WRT out-of-band access to label data, but in an acceptable form. With UFS1, there is a small race condition during automatic extended attribute start -- this is not present with UFS2, and occurs because EAs aren't available at vnode inception. We'll introduce a work around for this shortly. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-26 14:38:24 +00:00
mac_copy_vnode_label(&nd.ni_vp->v_label, &intlabel);
error = mac_externalize_vnode_label(&intlabel, elements, buffer,
mac.m_buflen, M_WAITOK);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
NDFREE(&nd, 0);
mac_destroy_vnode_label(&intlabel);
if (error == 0)
error = copyout(buffer, mac.m_string, strlen(buffer)+1);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
out:
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mtx_unlock(&Giant); /* VFS */
free(buffer, M_MACTEMP);
free(elements, M_MACTEMP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
/*
* MPSAFE
*/
int
__mac_set_fd(struct thread *td, struct __mac_set_fd_args *uap)
{
struct label intlabel;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct pipe *pipe;
struct file *fp;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
struct mount *mp;
struct vnode *vp;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac mac;
char *buffer;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(uap->mac_p, &mac, sizeof(mac));
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = mac_check_structmac_consistent(&mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, buffer, mac.m_buflen, NULL);
if (error) {
free(buffer, M_MACTEMP);
return (error);
}
mtx_lock(&Giant); /* VFS */
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = fget(td, SCARG(uap, fd), &fp);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
goto out;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
switch (fp->f_type) {
case DTYPE_FIFO:
case DTYPE_VNODE:
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_vnode_label(&intlabel);
error = mac_internalize_vnode_label(&intlabel, buffer);
if (error) {
mac_destroy_vnode_label(&intlabel);
break;
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
vp = (struct vnode *)fp->f_data;
error = vn_start_write(vp, &mp, V_WAIT | PCATCH);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
if (error != 0) {
mac_destroy_vnode_label(&intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
error = vn_setlabel(vp, &intlabel, td->td_ucred);
VOP_UNLOCK(vp, 0, td);
vn_finished_write(mp);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_destroy_vnode_label(&intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
case DTYPE_PIPE:
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mac_init_pipe_label(&intlabel);
error = mac_internalize_pipe_label(&intlabel, buffer);
if (error == 0) {
pipe = (struct pipe *)fp->f_data;
PIPE_LOCK(pipe);
error = mac_pipe_label_set(td->td_ucred, pipe,
&intlabel);
PIPE_UNLOCK(pipe);
}
mac_destroy_pipe_label(&intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
break;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
default:
error = EINVAL;
}
fdrop(fp, td);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
out:
mtx_unlock(&Giant); /* VFS */
free(buffer, M_MACTEMP);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
/*
* MPSAFE
*/
int
__mac_set_file(struct thread *td, struct __mac_set_file_args *uap)
{
struct label intlabel;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct nameidata nd;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
struct mount *mp;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
struct mac mac;
char *buffer;
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
int error;
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = copyin(uap->mac_p, &mac, sizeof(mac));
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
error = mac_check_structmac_consistent(&mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, buffer, mac.m_buflen, NULL);
if (error) {
free(buffer, M_MACTEMP);
return (error);
}
mac_init_vnode_label(&intlabel);
error = mac_internalize_vnode_label(&intlabel, buffer);
free(buffer, M_MACTEMP);
if (error) {
mac_destroy_vnode_label(&intlabel);
return (error);
}
mtx_lock(&Giant); /* VFS */
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_USERSPACE, uap->path_p,
td);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
error = namei(&nd);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
if (error == 0) {
error = vn_start_write(nd.ni_vp, &mp, V_WAIT | PCATCH);
if (error == 0)
error = vn_setlabel(nd.ni_vp, &intlabel,
td->td_ucred);
vn_finished_write(mp);
}
NDFREE(&nd, 0);
mtx_unlock(&Giant); /* VFS */
mac_destroy_vnode_label(&intlabel);
return (error);
}
/*
* MPSAFE
*/
int
__mac_set_link(struct thread *td, struct __mac_set_link_args *uap)
{
struct label intlabel;
struct nameidata nd;
struct mount *mp;
struct mac mac;
char *buffer;
int error;
error = copyin(uap->mac_p, &mac, sizeof(mac));
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
error = mac_check_structmac_consistent(&mac);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
if (error)
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
return (error);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
buffer = malloc(mac.m_buflen, M_MACTEMP, M_WAITOK);
error = copyinstr(mac.m_string, buffer, mac.m_buflen, NULL);
if (error) {
free(buffer, M_MACTEMP);
return (error);
}
mac_init_vnode_label(&intlabel);
error = mac_internalize_vnode_label(&intlabel, buffer);
free(buffer, M_MACTEMP);
if (error) {
mac_destroy_vnode_label(&intlabel);
return (error);
}
mtx_lock(&Giant); /* VFS */
NDINIT(&nd, LOOKUP, LOCKLEAF | NOFOLLOW, UIO_USERSPACE, uap->path_p,
td);
error = namei(&nd);
if (error == 0) {
error = vn_start_write(nd.ni_vp, &mp, V_WAIT | PCATCH);
if (error == 0)
error = vn_setlabel(nd.ni_vp, &intlabel,
td->td_ucred);
vn_finished_write(mp);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
NDFREE(&nd, 0);
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
mtx_unlock(&Giant); /* VFS */
mac_destroy_vnode_label(&intlabel);
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
return (error);
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
/*
* MPSAFE
*/
Provide an implementation of mac_syscall() so that security modules can offer new services without reserving system call numbers, or augmented versions of existing services. User code requests a target policy by name, and specifies the policy-specific API plus target. This is required in particular for our port of SELinux/FLASK to the MAC framework since it offers additional security services. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 17:59:48 +00:00
int
mac_syscall(struct thread *td, struct mac_syscall_args *uap)
{
struct mac_policy_conf *mpc;
char target[MAC_MAX_POLICY_NAME];
int error;
error = copyinstr(SCARG(uap, policy), target, sizeof(target), NULL);
if (error)
return (error);
error = ENOSYS;
MAC_POLICY_LIST_BUSY();
LIST_FOREACH(mpc, &mac_policy_list, mpc_list) {
if (strcmp(mpc->mpc_name, target) == 0 &&
mpc->mpc_ops->mpo_syscall != NULL) {
error = mpc->mpc_ops->mpo_syscall(td,
SCARG(uap, call), SCARG(uap, arg));
goto out;
}
}
out:
MAC_POLICY_LIST_UNBUSY();
return (error);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
SYSINIT(mac, SI_SUB_MAC, SI_ORDER_FIRST, mac_init, NULL);
SYSINIT(mac_late, SI_SUB_MAC_LATE, SI_ORDER_FIRST, mac_late_init, NULL);
#else /* !MAC */
Stubs for the TrustedBSD MAC system calls to permit TrustedBSD MAC userland code to operate on kernel's from the main tree. Not much in this file yet. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 02:04:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
int
__mac_get_pid(struct thread *td, struct __mac_get_pid_args *uap)
{
return (ENOSYS);
}
Stubs for the TrustedBSD MAC system calls to permit TrustedBSD MAC userland code to operate on kernel's from the main tree. Not much in this file yet. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 02:04:05 +00:00
int
__mac_get_proc(struct thread *td, struct __mac_get_proc_args *uap)
{
return (ENOSYS);
}
int
__mac_set_proc(struct thread *td, struct __mac_set_proc_args *uap)
{
return (ENOSYS);
}
int
__mac_get_fd(struct thread *td, struct __mac_get_fd_args *uap)
{
return (ENOSYS);
}
int
__mac_get_file(struct thread *td, struct __mac_get_file_args *uap)
{
return (ENOSYS);
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
int
__mac_get_link(struct thread *td, struct __mac_get_link_args *uap)
{
return (ENOSYS);
}
Stubs for the TrustedBSD MAC system calls to permit TrustedBSD MAC userland code to operate on kernel's from the main tree. Not much in this file yet. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 02:04:05 +00:00
int
__mac_set_fd(struct thread *td, struct __mac_set_fd_args *uap)
{
return (ENOSYS);
}
int
__mac_set_file(struct thread *td, struct __mac_set_file_args *uap)
{
return (ENOSYS);
}
Begin committing support for Mandatory Access Control and extensible kernel access control. The MAC framework permits loadable kernel modules to link to the kernel at compile-time, boot-time, or run-time, and augment the system security policy. This commit includes the initial kernel implementation, although the interface with the userland components of the oeprating system is still under work, and not all kernel subsystems are supported. Later in this commit sequence, documentation of which kernel subsystems will not work correctly with a kernel compiled with MAC support will be added. kern_mac.c contains the body of the MAC framework. Kernel and user APIs defined in mac.h are implemented here, providing a front end to loaded security modules. This code implements a module registration service, state (label) management, security configuration and policy composition. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-07-30 21:36:05 +00:00
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
int
__mac_set_link(struct thread *td, struct __mac_set_link_args *uap)
{
return (ENOSYS);
}
Provide an implementation of mac_syscall() so that security modules can offer new services without reserving system call numbers, or augmented versions of existing services. User code requests a target policy by name, and specifies the policy-specific API plus target. This is required in particular for our port of SELinux/FLASK to the MAC framework since it offers additional security services. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-19 17:59:48 +00:00
int
mac_syscall(struct thread *td, struct mac_syscall_args *uap)
{
return (ENOSYS);
}
Support the new MAC user API in kernel: modify existing system calls to use a modified notion of 'struct mac', and flesh out the new variation system calls (almost identical to existing ones except that they permit a pid to be specified for process label retrieval, and don't follow symlinks). This generalizes the label API so that the framework is now almost entirely policy-agnostic. Approved by: re Obtained from: TrustedBSD Project Sponsored by: DARPA, Network Associates Laboratories
2002-10-22 14:29:47 +00:00
#endif
Reference in New Issue Copy Permalink