1 .\" Copyright (c) 2002-2004 Networks Associates Technology, Inc.
2 .\" All rights reserved.
4 .\" This software was developed for the FreeBSD Project by Chris Costello
5 .\" at Safeport Network Services and Network Associates Laboratories, the
6 .\" Security Research Division of Network Associates, Inc. under
7 .\" DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the
8 .\" DARPA CHATS research program.
10 .\" Redistribution and use in source and binary forms, with or without
11 .\" modification, are permitted provided that the following conditions
13 .\" 1. Redistributions of source code must retain the above copyright
14 .\" notice, this list of conditions and the following disclaimer.
15 .\" 2. Redistributions in binary form must reproduce the above copyright
16 .\" notice, this list of conditions and the following disclaimer in the
17 .\" documentation and/or other materials provided with the distribution.
19 .\" THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
20 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
23 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 .Nd "Multi-Level Security confidentiality policy"
40 To compile MLS into your kernel, place the following lines in your kernel
42 .Bd -ragged -offset indent
47 Alternately, to load the MLS module at boot time, place the following line
48 in your kernel configuration file:
49 .Bd -ragged -offset indent
55 .Bd -literal -offset indent
61 policy module implements the Multi-Level Security, or MLS model,
62 which controls access between subjects and objects based on their
63 confidentiality by means of a strict information flow policy.
64 Each subject and object in the system has an MLS label associated with it;
65 each subject's MLS label contains information on its clearance level,
66 and each object's MLS label contains information on its classification.
68 In MLS, all system subjects and objects are assigned confidentiality labels,
69 made up of a sensitivity level and zero or more compartments.
70 Together, these label elements permit all labels to be placed in a partial
71 order, with confidentiality protections based on a dominance operator
73 The sensitivity level is expressed as a value between 0 and
74 65535, with higher values reflecting higher sensitivity levels.
75 The compartment field is expressed as a set of up to 256 components,
76 numbered from 1 to 256.
77 A complete label consists of both sensitivity and compartment
80 With normal labels, dominance is defined as a label having a higher
81 or equal active sensitivity level, and having at least
82 all of the same compartments as the label to which it is being compared.
83 With respect to label comparisons,
85 is defined as being dominated by the label to which it is being compared,
88 is defined as dominating the label to which it is being compared,
91 is defined as both labels being able to satisfy the dominance requirements
94 Three special label values exist:
95 .Bl -column -offset indent ".Li mls/equal" "dominated by all other labels"
96 .It Sy Label Ta Sy Comparison
97 .It Li mls/low Ta "dominated by all other labels"
98 .It Li mls/equal Ta "equal to all other labels"
99 .It Li mls/high Ta "dominates all other labels"
104 label may be applied to subjects and objects for which no enforcement of the
105 MLS security policy is desired.
107 The MLS model enforces the following basic restrictions:
110 Subjects may not observe the processes of another subject if its
111 clearance level is lower than the clearance level of the object it is
112 attempting to observe.
114 Subjects may not read, write, or otherwise observe objects without proper
115 clearance (e.g.\& subjects may not observe objects whose classification label
116 dominates its own clearance label)
118 Subjects may not write to objects with a lower classification level than
119 its own clearance level.
121 A subject may read and write to an object if its clearance level is equal
122 to the object's classification level as though MLS protections were not in
126 These rules prevent subjects of lower clearance from gaining access
127 information classified beyond its clearance level in order to protect the
128 confidentiality of classified information, subjects of higher clearance
129 from writing to objects of lower classification in order to prevent the
130 accidental or malicious leaking of information, and subjects of lower
131 clearance from observing subjects of higher clearance altogether.
132 In traditional trusted operating systems, the MLS confidentiality model is
133 used in concert with the Biba integrity model
135 in order to protect the Trusted Code Base (TCB).
137 Almost all system objects are tagged with an effective, active label element,
138 reflecting the classification of the object, or classification of the data
139 contained in the object.
140 In general, object labels are represented in the following form:
143 .D1 Li mls / Ar grade : compartments
147 .Bd -literal -offset indent
152 Subject labels consist of three label elements: an effective (active) label,
153 as well as a range of available labels.
154 This range is represented using two ordered MLS label elements, and when set
155 on a process, permits the process to change its active label to any label of
156 greater or equal integrity to the low end of the range, and lesser or equal
157 integrity to the high end of the range.
158 In general, subject labels are represented in the following form:
161 .D1 Li mls / Ar effectivegrade : effectivecompartments ( lograde : locompartments No -
162 .D1 Ar higrade : hicompartments )
166 .Bd -literal -offset indent
167 mls/10:2+3+6(5:2+3-20:2+3+4+5+6)
171 Valid ranged labels must meet the following requirement regarding their
174 .D1 Ar rangehigh No \[>=] Ar effective No \[>=] Ar rangelow
176 One class of objects with ranges currently exists, the network interface.
177 In the case of the network interface, the effective label element references
178 the default label for packets received over the interface, and the range
179 represents the range of acceptable labels of packets to be transmitted over
181 .Ss Runtime Configuration
184 MIBs are available for fine-tuning the enforcement of this MAC policy.
185 .Bl -tag -width ".Va security.mac.mls.ptys_equal"
186 .It Va security.mac.mls.enabled
187 Enables the enforcement of the MLS confidentiality policy.
189 .It Va security.mac.mls.ptys_equal
196 .It Va security.mac.mls.revocation_enabled
197 Revoke access to objects if the label is changed to a more sensitive
198 level than the subject.
201 .Sh IMPLEMENTATION NOTES
204 policy relies on superuser status
206 in order to change network interface MLS labels.
207 This will eventually go away, but it is currently a liability and may
208 allow the superuser to bypass MLS protections.
212 .Xr mac_bsdextended 4 ,
217 .Xr mac_partition 4 ,
219 .Xr mac_seeotheruids 4 ,
226 policy module first appeared in
228 and was developed by the
232 This software was contributed to the
234 Project by Network Associates Laboratories,
235 the Security Research Division of Network Associates
236 Inc.\& under DARPA/SPAWAR contract N66001-01-C-8035
238 as part of the DARPA CHATS research program.
242 concerning appropriateness for production use.
245 MAC Framework is considered experimental in
248 While the MAC Framework design is intended to support the containment of
249 the root user, not all attack channels are currently protected by entry
251 As such, MAC Framework policies should not be relied on, in isolation,
252 to protect against a malicious privileged user.