On Wed, 2008-08-27 at 15:49 +1000, Murray McAllister wrote: > Hi, > > The following is a draft of the Introduction sections for the SELinux > User Guide. Any comments and corrections are appreciated. > > Thanks. > > On Linux® operating systems, files, directories, sockets, devices, and > so on, are called objects, and processes, such as a user running a > command, the Mozilla® Firefox® application, and the Apache HTTP Server, > are called subjects. Linux operating systems use a Discretionary Access > Control (DAC) system that controls how subjects interact and access > objects, and how subjects interact with each other. The first two statements are actually true of most operating systems. > On systems using > DAC, users control the permissions of objects (files and directories) > that they own. They could, for example, make their home directories > world-readable, giving users and subjects (processes) access to > potentially sensitive information. > > DAC mechanisms are fundamentally inadequate for strong system security. > DAC access decisions are only based on user identity and ownership, > ignoring other security-relevant information such as the role of the > user, the function and trustworthiness of the program, and the > sensitivity and integrity of the data. Each user has complete discretion > over his objects, making it impossible to enforce a system-wide security > policy. Furthermore, every program run by a user inherits all of the > permissions granted to the user and is free to change access to the > user's objects, so no protection is provided against malicious software. > Typically, only two major categories of users are supported by DAC > mechanisms: completely trusted administrators and completely untrusted > ordinary users. Many system services and privileged programs must run > with coarse-grained privileges that far exceed their requirements, so > that a flaw in any one of these programs can be exploited to obtain > complete system access[1]. > > The following is an example of permissions used on Linux operating > systems that do not run Security-Enhanced Linux (SELinux). Use the ls -l > command to view object (such as a file) permissions: > > -rwxrw-r-- 1 user1 group1 0 Aug 18 10:08 file1 > > The first three permission bits, rwx, control the access the Linux > user1 user (in this case, the owner) has to the file1 object. The next > three permission bits, rw-, control the access the Linux group1 group > has to the file1 object. The last three permission bits, r--, control > the access everyone else has to the file1 object. This includes all > subjects (such as users and processes). By default, when a new object > (such as a file) is created, everyone has read permissions. Depends on the umask, and can be customized in /etc/profile or ~/.bashrc or similar files. It isn't really the permissiveness of the default umask that is the key issue here, btw. > Security-Enhanced Linux (SELinux) adds Mandatory Access Control (MAC) to > the Linux kernel, and is enabled by default in Fedora. A general purpose > MAC architecture needs the ability to enforce an administratively-set > security policy over all subjects and objects in the system, basing > decisions on labels containing a variety of security-relevant > information. When properly implemented, it enables a system to > adequately defend itself and offers critical support for application > security by protecting against the tampering with, and bypassing of, > secured applications. It allows critical processing pipelines to be > established and guaranteed. MAC provides strong separation of > applications that permits the safe execution of untrustworthy > applications. Its ability to limit the privileges associated with > executing processes limits the scope of potential damage that can result > from the exploitation of vulnerabilities in applications and system > services. MAC enables information to be protected from legitimate users > with limited authorization as well as from authorized users who have > unwittingly executed malicious applications. The ability for the system > to do these types of things is necessary before the construction of > secure systems will be possible[2]. > > The following is an example of the labels containing security-relevant > information that are used on subjects and objects on Linux operating > systems that run SELinux. This information is called the SELinux > context, and is viewed using the ls -Z command: > > -rwxrw-r-- user1 group1 unconfined_u:object_r:user_home_t:s0 file1 > > In this example, SELinux provides a user (unconfined_u), a role > (object_r), a type (user_home_t), and a category (s0). This information > is used to make access control decisions. On DAC systems, access is > controlled based on Linux user and group IDs. SELinux allow rules are > checked after DAC rules. SELinux allow rules are not used if DAC rules > deny access first. > > Linux and SELinux users > > On Linux operating systems that run SELinux, there are Linux users as > well as SELinux users. SELinux users are part of SELinux policy. Linux > users are mapped to SElinux users. To avoid confusion, this guide uses > "Linux user" and "SELinux user" to differentiate between the two. > > Benefits of running SELinux > > SELinux provides: > > * An additional layer of security: a system where subjects (processes) > are separated from each other by running in their own domains, and rules > define how subjects interact and access objects (such as files), as well > as how subjects interact with each other. Actions, such as a subject > opening an object, are only allowed if a rule exists that specifically > allows it. > > * Fine-grained access control. Stepping beyond traditional UNIX® > permissions that are controlled at user discretion and based on Linux > user and group IDs, SELinux access decisions are based on all available > information, such as an SELinux user, role, and type. SELinux policy is > administratively-defined, and is not set at user discretion. SELinux > allow rules (the default action is deny) are not used if DAC rules deny > access first. This bullet mixes two different concepts: fine-grained access control and administratively-defined policy. Both are important, but they are distinct. > * Type Enforcement®. All subjects and objects are labeled with a type. > Rules define how types interact and access each other. Access is only > allowed if a rule exists that specifically allows it. This bullet seems redundant with the first one - it merely refines the general notions of subject and object to types. > * Prevention against privilege escalation. Since subjects run in > domains, and are therefore separated from each other, and rules > determine how subjects access objects and other subjects, if a service > is compromised, the attacker only has access to the normal functions of > that service, and to files that the service has been configured to have > access to. For example, if the Apache HTTP Server is compromised, an > attacker is unable to read files in user home directories, unless a > specific rule was added or configured to allow such access. This is one way in which SELinux can be applied. Other goals to which SELinux can be applied include enforcing data confidentiality and integrity, protecting privileged subjects from untrustworthy inputs, and ensuring the invocation of critical processing. > SELinux is not: > > * Antivirus software. > > * A replacement for passwords, firewalls, or other security systems. > > * An all-in-one security solution. > > SELinux is an addition to DAC rules. This doesn't quite sound right, as it makes it sound as though SELinux is an extension to DAC vs. an orthogonal access control scheme (MAC). > It is designed to enhance existing > security solutions, not replace them. Even when running SELinux, > continue to follow good security practices, such as keeping software > up-to-date, using hard-to-guess passwords, firewalls, and so on. > > Examples > > The following examples demonstrate how SELinux increases security: > > * The default action is deny. If a rule does not exist to allow a > subject access to an object, or a subject access to another subject, > access is denied. > > * Confining users: SELinux can confine Linux users. A number of > restricted SELinux users exist. Linux users can be mapped to restricted > SELinux users to take advantage of confined SELinux users. For example, > mapping a Linux user account to the SELinux user_u user, results in a > Linux user that is not able to run (unless configured otherwise) set > user ID (setuid) applications, such as /usr/bin/sudo and su. Also, you > can disable the execution of objects in user home directories for Linux > users that are mapped to the SELinux user_u user. If configured, this > prevents users from executing malicious files, that they may have > downloaded from the Internet, from their home directories. > > * Subject (process) separation. Subjects run in their own domains. This > prevents other subjects from accessing objects used by other subjects, > as well as subjects accessing other subjects. For example, when running > SELinux, unless otherwise configured, an attacker can not compromise a > Samba server, and then use that Samba server to read and write to > objects used by other subjects, such as files comprising a website that > is read by the Apache HTTP server. Here you can point specifically to past examples of flaws that would be mitigated by SELinux. See for example: http://www.tresys.com/innovation.php (SELinux Mitigation News, on the right) http://www.redhatmagazine.com/2008/02/26/risk-report-three-years-of-red-hat-enterprise-linux-4/ http://www.linuxworld.com/news/2008/022408-selinux.html?page=1 http://james-morris.livejournal.com/25421.html > * Help limit the damage done by configuration mistakes. An administrator > may forget to limit zone transfers when running ISC BIND; however, the > default SELinux context for zone files does not allow them to be updated > by zone transfers, or written to by named, the ISC BIND daemon, and > other subjects. > > SELinux Architecture and Performance > > SELinux is a Linux kernel module. Actually, to be precise, SELinux is a Linux kernel subsystem or a Linux security module. It has to be built-in; you cannot build and load it as a loadable kernel module due to its tight integration with core kernel structures and interfaces and its need to start handling kernel objects and operations from the beginning. > Part of this module is the SELinux > security server. The security server contains policy rules that define > what access is allowed. I'd tend to say "is driven by loadable policy rules" or "is driven by a loadable policy configuration" rather than "contains policy rules" as the latter suggests that the rules are fixed/hardcoded. > When a subject attempts to interact with an > object, for example, a process opening a file, a hook in the Linux > kernel intercepts the system call the process makes to open the file. > The hook communicates with the security server to check if access should > be allowed or denied. I'd change "communicates" to "calls" since it is a function call rather than an IPC in Linux. > Decisions made by the security server are cached > by an Access Vector Cache (AVC). This decreases how often SELinux rules > in the security server need to be checked, which increases performance. > SELinux has no effect if DAC rules deny access first. > > [performance details to be added later] > > > > [1] "Integrating Flexible Support for Security Policies into the Linux > Operating System", by Peter Loscocco and Stephen Smalley. This paper was > originally prepared for the National Security Agency and is, > consequently, in the public domain. Refer to the > [http://www.nsa.gov/selinux/papers/freenix01/freenix01.html original > paper] for details and the document as it was first released. Any edits > and changes were done by Murray McAllister. > > [2] "Meeting Critical Security Objectives with Security-Enhanced Linux", > by Peter Loscocco and Stephen Smalley. This paper was originally > prepared for the National Security Agency and is, consequently, in the > public domain. Refer to the > [http://www.nsa.gov/selinux/papers/ottawa01/index.html original paper] > for details and the document as it was first released. Any edits and > changes were done by Murray McAllister. > > -- > This message was distributed to subscribers of the selinux mailing list. > If you no longer wish to subscribe, send mail to majordomo@xxxxxxxxxxxxx with > the words "unsubscribe selinux" without quotes as the message. -- Stephen Smalley National Security Agency -- This message was distributed to subscribers of the selinux mailing list. If you no longer wish to subscribe, send mail to majordomo@xxxxxxxxxxxxx with the words "unsubscribe selinux" without quotes as the message.
