Transcript Security Policies C. Edward Chow – Chapter 5.2/5.4 of Security in Computing CS591

Security Policies
C. Edward Chow
CS591 – Chapter 5.2/5.4 of Security in Computing
cs591
1
chow
Goals of Security Policies

Security policy is a statement of the security we expect the system to
enforce.

Security policy also called information flow policy, prevents unauthorized
disclosure of information.

Example: Privacy Act requires that certain personal data be kept
confidential. E.g., income tax return info only available to IRS and legal
authority with court order. It limits the distribution of documents/info.

Example: Military security policy is based on protecting classified
information.

Each piece of info is ranked at a particular sensitivity level such as
unclassified, restricted, confidential, secret, or top secret.

Users are classified with the same classification level.

Policy guards against higher sensitive document to be leaked to lower
ranked users.
cs591
2
chow
Hierarchy of Sensitivities
cs591
3
chow
Discretionary Access Control (DAC)



DAC: Mechanism where a user can set access control to allow or
deny access to an object (Section 5.4)
Also called Identity-based access control (IBAC).
It is a traditional access control techniques implemented by
traditional operating system such as Unix.
 Based on user identity and ownership
 Programs run by a user inherits all privileges granted to the
user.
 Programs is free to change access to the user’s objects
 Support only two major categories of users:
– Completely trusted admins
– Completely untrusted ordinary users
cs591
4
chow
Problems with DAC





Each users has complete discretion over his objects.
 What is wrong with that?
 Difficult to enforce a system-wide security policy, e.g.
– A user can leak classified documents to a unclassified users.
– Other examples?
Only based user’s identity and ownership, Ignoring security relevant info such as
 User’s role
 Function of the program
 Trustworthiness of the program
– Compromised program can change access to the user’s objects
– Compromised program inherit all the permissions granted to the users
(especially the root user)
 Sensitivity of the data
 Integrity of the data
Only support coarse-grained privileges
Unbounded privilege escalation
Too simple classification of users (How about more than two categories of users?)
cs591
5
chow
Mandatory Access Control (MAC)

MAC: Mechanism where system control access to an
object and a user cannot alter that access.
 Occasionally called rule-based access control?
 Defined by three major properties:
 Administratively-defined security policy
 Control over all subjects (process) and objects (files,
sockets, network interfaces)
 Decisions based on all security-relevant info
 MAC access decisions are based on labels that
contains security-relevant info.
cs591
6
chow
What Can MAC Offer?






Supports a wide variety of categories of users in system.
 For example, Users with labels: (secret, {EUR, US}) (top secret,
{NUC, US}).
 Here security level is specified by the two-tuple: (clearance,
category)
Strong separation of security domains
System, application, and data integrity
Ability to limit program privileges
 Confine the damage caused by flowed or malicious software
Processing pipeline guarantees
Authorization limits for legitimate users
cs591
7
chow
Mandatory and Discretionary
Access Control


Bell-LaPadula model combines Mandatory and Discretionary
Access Controls.
“S has discretionary read (write) access to O”
means that the access control matrix entry for S and O
corresponding to the discretionary access control component
contains a read (write) right.
A

B
C
D
O
Q
S
read(D)
T
If the mandatory controls not present, S would be able to read
(write) O.
cs591
8
chow
Bell-LaPadula Model






Also called the multi-level model,
Was proposed by Bell and LaPadula of MITRE for enforcing access
control in government and military applications.
It corresponds to military-style classifications.
In such applications, subjects and objects are often partitioned into
different security levels.
A subject can only access objects at certain levels determined by
his security level.
For instance, the following are two typical access specifications:
“Unclassified personnel cannot read data at confidential levels” and
“Top-Secret data cannot be written into the files at unclassified
levels”
cs591
9
chow
Informal Description




Simplest type of confidentiality classification is a set of
security clearances arranged in a linear (total) ordering.
Clearances represent the security levels.
The higher the clearance, the more sensitive the info.
Basic confidential classification system:
individuals
documents
Top Secret (TS) Tamara, Thomas
Personnel Files
Secret (S)
Sally, Samuel
Electronic Mails
Confidential (C) Claire, Clarence
Activity Log Files
Restricted
Unclassified (UC)Ulaley, Ursula
Telephone Lists
cs591
10
chow
Star Property (Preliminary Version)








Let L(S)=ls be the security clearance of subject S.
Let L(O)=lo be the security classification of object ).
For all security classification li, i=0,…, k-1, li<li+1
Simple Security Condition (Read Down):
S can read O if and only if lo<=ls and
S has discretionary read access to O.
*-Property (Star property) (Write Up):
S can write O if and only if ls<=lo and
S has discretionary write access to O.
TS guy can not write documents lower than TS.  Prevent
classified information leak.
No Read UP; No Write Down!
But how can different groups communicate?
Trusted Solaris Example.
cs591
11
chow
Secure Flow of Information
cs591
12
chow
Basic Security Theorem



Let  be a system with secure initial state 0
Let T be the set of state transformations.
If every element of T preserves the simple security
condition, preliminary version, and the *-property,
preliminary version,
Then every state i, i≥0, is secure.
cs591
13
chow
Categories and Need to Know Principle

Expand the model by adding a set of categories.
 Each category describe a kind of information.
 These categories arise from the “need to know” principle
 no subject should be able to read objects unless
reading them is necessary for that subject to perform its
function.
 Example: three categories: NUC, EUR, US.
 Each security level and category form a security level or
compartment.
 Subjects have clearance at (are cleared into, or are in) a
security level.
 Objects are at the level of (or are in) a security level.
cs591
14
chow
Security Lattice
{NUC, EUR, US}
{NUC, EUR}
{NUC}
{NUC, US}
{EUR}
{EUR, US}
{US}


William may be cleared into level (SECRET, {EUR})
 George into level (TS, {NUC, US}).
 A document may be classified as (C, {EUR})
 Someone with clearance at (TS, {NUC, US}) will be
denied access to document with category EUR.
cs591
15
chow
Dominate (dom) Relation
The security level (L, C) dominates the security level (L’,
C’) if and only if L’  L and C’  C
 Dom  dominate relation is false.
 Geroge is cleared into security level (S, {NUC, EUR})
 DocA is classified as (C, {NUC})
 DocB is classified as (S, {EUR, US})
 DocC is classified as (S, {EUR})
 George dom DocA
 George  dom DocB
 George dom DocC

cs591
16
chow
New Security Condition and *-Property





Let C(S) be the category set of subject S.
Let C(O) be the category set of object O.
Simple Security Condition (no read up):
S can read O if and only if S dom O and
S has discretionary read access to O.
*-Property (no write down):
S can write to O if and only if O dom S and
S has discretionary write access to O.
Basic Security Theorem:
Let  be a system with secure initial state 0
Let T be the set of state transformations.
If every element of T preserves the simple security condition,
preliminary version, and the *-property, preliminary version,
Then every state i, i≥0, is secure.
cs591
17
chow
Allow Write Down?

Bell-LaPadula allows higher-level subject to write into
lower level object that low level subject can read.
 A subject has a maximum security level and a current
security level. maximum security level must dominate
current security level.
 A subject may (effectively) decrease its security level
from the maximum in order to communicate with entities
at lower security levels.
 Colonel’s maximum security level is (S, {NUC, EUR}).
She changes her current security level to (S, {EUR}).
Now she can create document at Major is clearance
level (S, {EUR}).
cs591
18
chow
Data General B2 Unix System









Data General B2 Unix (DG/UX) provides mandatory access
controls (MAC).
The MAC label is a label identifying a particular compartment.
The initial label (assigned at login time) is the label assigned to the
user in a database called Authorization and Authentication (A&A)
Database.
When a process begins, it is assigned to MAC label of its parent
(whoever creates it).
Objects are assigned labels at creation. The labels can be explicit
or implicit.
The explicit label is stored as parts of the object’s attributes.
The implicit label derives from the parent directory of the object.
IMPL_HI: the least upper bound of all components in DG/UX lattice
has IMPL_HI as label.
IMPL_LO: the greatest lower bound of all components in DG/UX
lattice has IMPL_LO as the label
cs591
19
chow
Three MAC Regions in
DG/UX MAC Lattice
Figure 5-3 The three MAC regions in the MAC lattice (modified from the DG/UX Security
Manual [257], p. 4-7, Figure 4-4). TCB stands for "trusted computing base.“
cs591
20
chow
Accesses with MAC Labels
• Read up and write up from users to Admin Region
not allowed.
• Admin processes sanitize data sent to user
processes with MAC Labels in the user region.
• System programs are in the lowest region.
• No user can write to or alter them.
• Only programs with the same label as the directory
can create files in that directory.
• The above restriction will prevent
• compiling (need to access /tmp)
• mail delivery
(need to access mail spool
directory)
• Solution multilevel directory.
cs591
21
chow
Multilevel Directory








A directory with a set of subdirectories, one for each label.
These hidden directories normally invisible to the user.
When a process with label MAC_A creates a file in /tmp, it actually
create a file in hidden directory under /tmp with label MAC_A
The parent directory of a file in /tmp is the hidden directory.
A reference to the parent directory goes to the hidden directory.
Process A with MAC_A creates /tmp/a. Process B with MAC_B
creates /tmp/a. Each of them performs
“cd /tmp/a; cd ..”
The system call stat(“.”, &stat_buffer) returns different inode
number for each process. It returns the inode number of the
respective hidden directory.
Try “stat” command to display file and related status.
DG/UX provides dg_mstat(“.”, &stat_buffer) to translate the current
working directory to the multilevel directory
cs591
22
chow
Mounting Unlabeled File System



1.
2.
3.
4.
5.
All files in that file system need to be labeled.
Symbolic links aggravate this problem. Does the MAC label the target of the link
control, or does the MAC label the link itself? DG/UX uses a notion of inherited
labels (called implicit labels) to solve this problem.
The following rules control the way objects are labeled.
Roots of file systems have explicit MAC labels. If a file system without labels is
mounted on a labeled file system, the root directory of the mounted file system
receives an explicit label equal to that of the mount point. However, the label of the
mount point, and of the underlying tree, is no longer visible, and so its label is
unchanged (and will become visible again when the file system is unmounted).
An object with an implicit MAC label inherits the label of its parent.
When a hard link to an object is created, that object must have an explicit label; if it
does not, the object's implicit label is converted to an explicit label. A corollary is that
moving a file to a different directory makes its label explicit.
If the label of a directory changes, any immediate children with implicit labels have
those labels converted to explicit labels before the parent directory's label is
changed.
When the system resolves a symbolic link, the label of the object is the label of the
target of the symbolic link. However, to resolve the link, the process needs access to
the symbolic link itself.
cs591
23
chow
Interesting Case with Hard Links




Let /x/y/z: and /x/a/b be hard links to the same object. Suppose y has an explicit
label IMPL_HI and a an explicit label IMPL_B. Then the file object can be accessed
by a process at IMPL_HI as /x/y/z and by a process at IMPL_B as /x/alb.
Which label is correct? Two cases arise.
Suppose the hard link is created while the file system is on a DG/UX B2 system.
Then the DG/UX system converts the target's implicit label to an explicit one (rule 3).
Thus, regardless of the path used to refer to the object, the label of the object will be
the same.
Suppose the hard link exists when the file system is mounted on the DG/UX B2
system. In this case, the target had no file label when it was created, and one must
be added. If no objects on the paths to the target have explicit labels, the target will
have the same (implicit) label regardless of the path being used. But if any object on
any path to the target of the link acquires an explicit label, the target's label may
depend on which path is taken. To avoid this, the implicit labels of a directory's
children must be preserved when the directory's label is made explicit. Rule 4 does
this.
Because symbolic links interpolate path names of files, rather than store Mode
numbers, computing the label of symbolic links is straightforward. If /x/y/z is a symbolic link to /a/b/c, then the MAC label of c is computed in the usual way. However,
the symbolic link itself is a file, and so the process must also have access to the link
file z.
cs591
24
chow
Enable Flexible Write in DG/UX







Provide a range of labels called MAC tuple.
A range is a set of labels expressed by a lower bound and an upper
hound. A MAC tuple consists of up to three ranges (one for each of
the regions in Figure 5-3).
Example: A system has two security levels. TS and S, the former
dominating the latter. The categories are COMP. NUC, and ASIA.
Examples of ranges are:
[(S, { COMP } ),
(TS, { COMP } )]
[( S,  ),
(TS, { COMP, NUC. ASIA } )]
[( S, { ASIA } ),
( TS, { ASIA, NUC } )]
The label ( TS, { COMP }) is in the first two ranges.
The label ( S, { NUC, ASIA } ) is in the last two ranges. However,
[( S, {ASIA} ), ( TS, { COMP, NUC} )]
is not a valid range because ( TS, {COMP. NUC } ) dom ( S, {
ASIA } ).
cs591
25
chow
Integrity




Problem area: systems require data to be changed accurately and follow the rules.
Disclosure is not a major concern.
Lipner [636] identifies five requirements for preserving data integriy:
1. Users will not write their own programs, but will use existing production programs
and databases.
2. Programmers will develop and test programs on a nonproduction system; if they
need access to actual data, they will be given production data via a special
process, but will use it on their development system.
3. A special process must be followed to install a program from the development
system onto the production system.
4. The special process in requirement 3 must be controlled and audited.
5. The managers and auditors must have access to both the system state and the
system logs that are generated.
Auditing: the process of analyzing systems to determine what actions took place
and who performed them. It uses extensive logging.
These requirement suggest 3 principles of operation:

Separation of duty (two different people? perform two critical steps)

Separation of function (program not developed on production system;
production data for development needs to be sanitized.)

Auditing. (Commercial systems emphasize recovery and accountability.)
cs591
26
chow
Different Needs



Commercial firms grant access based on individual
needs and has a larger categories  large number of
security levels.
In military environment, creation of compartment is
centralized. In commercial firms, it is decentralized.
Aggregating distributed inoncuous info, one can often
deduce sensitive information. The Bell-LaPadula Model
lack capability to track what questions have been asked.
cs591
27
chow
Biba Integrity Model









In 1977, Biba [94] studied the nature of the integrity of systems. He
proposed three policies, one of which was the mathematical dual of the
Bell-LaPadula Model.
A system consists of a set S of subjects, a set 0 of objects, and a set I of
integrity levels. The levels are ordered.
The relation <  I x I holds when the second integrity level dominates the
first.
The relation ≤  I x I holds when the second integrity level either dominates
or is the same as the first.
The function min: I x II gives the lesser of the two integrity levels
The function i:S  O1 returns the integrity level of an object or a subject.
The relation r  S x 0 defines the ability of a subject to read an object;
the relation w  S x 0 defines the ability of a subject to write to an object;
the relation x  S x S defines the ability of a subject to invoke (execute)
another subject.
cs591
28
chow
Intuition Behind Model Construction




The higher the level, the more confidence one has that
a program will execute correctly (or detect problems
with its inputs and stop executing).
Data at a higher level is more accurate, reliable,
trustworthy than data at a lower level.
Integrity labels, in general, are not also security labels.
They are assigned and maintained separately, because
the reasons behind the labels are different. Security
labels primarily limit the flow of information; integrity
labels primarily inhibit the modification of information.
They may overlap, however, with surprising results
cs591
29
chow
Test case: Information Transfer Path



Biba tests his policies against the notion of an
information transfer path:
Definition 6-1. An information transfer path is a
sequence of objects o1, ..., on+1 and a corresponding
sequence of subjects s1, ..., sn such that
si r oi and si w oi+1 for all i,1≤i≤n.
Intuitively, data in the object o1 can be transferred into
the object on+1 along an information flow path by a
succession of reads and writes.
cs591
30
chow
Low-Water-Mark Policy




Whenever a subject accesses an object, the policy changes the
integrity level of the subject to the lower of the subject and the
object. Specifically:
1. s  S can write to o  O if and only if i(o) ≤ i(s).
2. If s  S reads o  O, then i'(s) = rnin(i(s), i(o)), where i'(s) is the
subject's integrity level after the read.
3. s1  S can execute s2  S if and only if i(s2) ≤ i(s1).
Rule 1 prevents writing to higher level (higher trusted). Prevent
implant of incorrect or false data.
Rule 2 assume that the subject will rely on the data with lower
integrity level. Therefore his integrity level should be lowered.
(Contaminating subject and actions)
Rule 3 prevent a less trusted invoker to control the execution of
more truested subjects.
cs591
31
chow
Constrains Information Transfer Path







This low-water-mark policy constrains any information transfer path.
Theorem 6-1. If there is an information transfer path from object o1  O to
object on+1  O, then enforcement of the low-water-mark policy requires
that
i(on+1) ≤ i(o1) for all n > 1.
Proof: Without loss of generality, assume that each read and write was performed in the order of the indices of the vertices. By induction, for any 1 ≤ k
≤ n,
i(sk) = min { i(oj) I 1 ≤ j ≤ k } after k reads. As the nth write succeeds, by rule
1, i(on+1) < i(sn ). Thus, by transitivity. i(on+1) < i(o1).
This policy prevents direct modifications that would lower integrity labels. It
also prevents indirect modification by lowering the integrity label of a
subject that reads from an object with a lower integrity level.
The problem with this policy is that, in practice. the subjects change
integrity levels. In particular, the level of a subject is nonincreasing. which
means that it will soon be unable to access objects at a high integrity level.
How about decrease object integrity level rather than subject integrity
level?
cs591
32
chow
Ring Policy


The ring policy ignores the issue of indirect modification
and focuses on direct modification only. This solves the
problems described above. The rules are as follows.
1. Any subject may read any object, regardless of
integrity levels.
2. s  S can write to o  O if and only if i(o) ≤ i(s).
3. s1  S can execute s2  S if and only if i(s2) -< i(s1)
The difference between this policy and the low-watermark policy is simply that any subject can read any
object. Hence, Theorem 6-1 holds for this model, too.
cs591
33
chow
Biba Model (Strict Integrity Policy)




This model is the dual of the Bell-LaPadula Model, and is most
commonly called "Biba's model."
Its rules are as follows.
1. s  S can read o  O if and only if i(s) ≤ i(o).
2. s  S can write to o  O if and only if i(o) ≤ i(s).
3. s1  S can execute s2  S if and only if i(s2) ≤ i(s1).
Given these rules, Theorem 6-1 still holds, but its proof changes
(see Exercise 1). Note that rules I and 2 imply that if both read and
write are allowed, i(s) = i(o).
Like the low-water-mark policy, this policy prevents indirect as well
as direct modification of entities without authorization. By replacing
the notion of "integrity level" with "integrity compartments," and
adding the notion of discretionary controls, one obtains the full dual
of Bell-LaPadula.
cs591
34
chow
Example: LOCUS Distributed OS









Pozzo and Gray [817, 818] implemented Biba's strict integrity model on the
distributed operating, system LOCUS [811 ].
Goal: limit execution domains for each program to prevent untrusted
software from altering data or other software.
Approach: make the level of trust in software and data explicit.
They have different classes of executable programs.
Their credibility ratings (Biba's integrity levels) assign a measure of
trustworthiness on a scale from 0 (untrusted) to n (highly trusted),
depending on the source of the software.
Trusted file systems contain only executable files with the same credibility
level.
Associated with each user (process) is a risk level that starts out set to the
highest credibility level at which that user can execute.
Users may execute programs with credibility levels at least as great as the
user's risk level.
To execute programs at a lower credibility level, a user must use the rununtrusted command. This acknowledges the risk that the user is taking.
cs591
35
chow
Chinese Wall Model






It describes policies that prevent conflict of interest.
Examples
 in British Law, provide defense against criminal charges.
 Stock Exchange and Investment house. Prevent traders
represents clients with conflict interest.
Definition 7-1. The objects of the database are items of information
related to a company.
Definition 7-2. A company dataset (CD) contains objects related to
a single company.
Definition 7-3. A conflict of interest (COI) class contains the
datasets of companies in competition.
Let COI(O) represent the COI class that contains object 0, and let
CD(O) be the company dataset that contains object 0. The model
assumes that each object belongs to exactly one COI class.
cs591
36
chow
CD and COI
cs591
37
chow
CW-Simple Security Condition

Consider temporal element. After accessing Bank of
America, Anthony should not transfer to work on
Cityband’s profolio.
 PR(S) is the set of objects that S has read.
 CW-Simple Security Condition, Preliminary Version:
S can read 0 if and only if either of the following is true.
1. There is an object O' such that S has accessed O'
and CD(O') = CD(O).
2. For all objects O’, O’ PR(S)  COI(O') COI(O).
 Initially, PR(S) = 0, and the initial read request is
assumed to be granted.
cs591
38
chow
Consider Sanitized Data


In practice, companies have information they can release publicly,
such as annual stockholders' reports and filings before government
commissions. The Chinese Wall model should not consider this
information restricted, because it is available to all. Hence, the
model distinguishes between sanitized data and unsanitized data;
the latter falls under the CW-simple security condition, preliminary
version, whereas the former does not. The CW-simple security
condition can be reformulated to include this notion.
CW-Simple Security Condition: S can read 0 if and only if any of
the following holds.
1. There is an object O' such that S has accessed O' and CD(O')
= CD(O).
2. For all objects O', O'  PR(S)  COI(O')  COI(O).
3. O is a sanitized object.
cs591
39
chow
CW-*-Property

Suppose Anthony and Susan work in the same trading house. Anthony can
read objects in Bank of America's CD, and Susan can read objects in
Citibank's CD. Both can read objects in ARCO's CD. If Anthony can also
write to objects in ARCO's CD, then he can read information from objects in
Bank of America's CD and write to objects in ARCO's CD, and then Susan
can read that information; so, Susan can indirectly obtain information from
Bank of America's CD, causing a conflict of interest. The CW-simple
security condition must be augmented to prevent this.
 CW-*-Property: A subject S may write to an object 0 if and only if both of
the following conditions hold.
1. The CW-simple security condition permits S to read O.
2. For all unsanitized objects O S can read 0'  CD(O') = CD(O).
 In the example above, Anthony can read objects in both Bank of America's
CD and ARCO's CD. Thus, condition 1 is met. However, assuming that
Bank of America's CD contains unsanitized objects (a reasonable
assumption), then because Anthony can read those objects, condition 2 is
false. Hence, Anthony cannot write to objects in ARCO's CD.
cs591
40
chow
Take-Grant System

Introduce by Jone[Jo78] Expanded by Lipton and
Snyder[Lip77][SNY81].
cs591
41
chow
Trusted Solaris



It implements MLS.
Here is the link to
trusted solaris user
guide.
To the right is a
label builder.
cs591
42
chow
Workspace and related Menu


Bottom panel shows the current security label
Menu for changing roles
cs591
43
chow
Warning when
Copying Files
cs591
44
chow
History Security-Enhanced Linux
(SELinux)





National Security Agency (NSA) and Secure Computing Corporation (SCC)
provide strong MAC.
 Flexible support for security policies (no single MAC policy can satisfy
everyone’s security requirements)
 Cleanly separate the security policy logic from enforcing mechanism
 Developed DTMach, DTOS (Mach-based prototype)
 Apply formal method to validate the security properties of the
architecture (High Assurance)
Work with Univ. Utah Flux Research Group
 integrate the architecture to Fluke research operating system
 Result: Flask architecture support dynamic security policies.
NSA create SELinux integrate Flash architecture to Linux OS.
NAI implements control on procfs and devpts fiel ssytems
MITRE/SCC contribute application security policies, modified utility
programs
cs591
45
chow
SELinux

Support
 Separation policies:
– Enforce legal restriction on data
– Establish well-defined user roles
– Restrict access to classified data
 Containment policies for
– Restrict web server access to only authorized data
– Minimize damage caused by virues/malicious code
 Integrity policies that protect unauthorized modifications to data
and applications
 Invocation policies that guarantee data is processed as required.
cs591
46
chow