The text of and illustrations in this document are licensed by Red Hat under a Creative Commons Attribution–Share Alike 3.0 Unported license ("CC-BY-SA"). An explanation of CC-BY-SA is available at http://creativecommons.org/licenses/by-sa/3.0/. In accordance with CC-BY-SA, if you distribute this document or an adaptation of it, you must provide the URL for the original version.
Red Hat, as the licensor of this document, waives the right to enforce, and agrees not to assert, Section 4d of CC-BY-SA to the fullest extent permitted by applicable law.
Red Hat, Red Hat Enterprise Linux, the Shadowman logo, JBoss, MetaMatrix, Fedora, the Infinity Logo, and RHCE are trademarks of Red Hat, Inc., registered in the United States and other countries.
Linux® is the registered trademark of Linus Torvalds in the United States and other countries.
Java® is a registered trademark of Oracle and/or its affiliates.
XFS® is a trademark of Silicon Graphics International Corp. or its subsidiaries in the United States and/or other countries.
MySQL® is a registered trademark of MySQL AB in the United States, the European Union and other countries.
All other trademarks are the property of their respective owners.
The Red Hat Enterprise Linux 6 SELinux User Guide is for people with minimal or no experience with SELinux. Although system administration experience is not necessary, content in this guide is written for system administration tasks. This guide provides an introduction to fundamental concepts and practical applications of SELinux. After reading this guide you should have an intermediate understanding of SELinux.
Thank you to everyone who offered encouragement, help, and testing - it is most appreciated. Very special thanks to:
Dominick Grift, Stephen Smalley, and Russell Coker for their contributions, help, and patience.
1. Document Conventions
This manual uses several conventions to highlight certain words and phrases and draw attention to specific pieces of information.
In PDF and paper editions, this manual uses typefaces drawn from the Liberation Fonts set. The Liberation Fonts set is also used in HTML editions if the set is installed on your system. If not, alternative but equivalent typefaces are displayed. Note: Red Hat Enterprise Linux 5 and later includes the Liberation Fonts set by default.
1.1. Typographic Conventions
Four typographic conventions are used to call attention to specific words and phrases. These conventions, and the circumstances they apply to, are as follows.
Mono-spaced Bold
Used to highlight system input, including shell commands, file names and paths. Also used to highlight keycaps and key combinations. For example:
To see the contents of the file my_next_bestselling_novel in your current working directory, enter the cat my_next_bestselling_novel command at the shell prompt and press Enter to execute the command.
The above includes a file name, a shell command and a keycap, all presented in mono-spaced bold and all distinguishable thanks to context.
Key combinations can be distinguished from keycaps by the hyphen connecting each part of a key combination. For example:
Press Enter to execute the command.
Press Ctrl+Alt+F2 to switch to the first virtual terminal. Press Ctrl+Alt+F1 to return to your X-Windows session.
The first paragraph highlights the particular keycap to press. The second highlights two key combinations (each a set of three keycaps with each set pressed simultaneously).
If source code is discussed, class names, methods, functions, variable names and returned values mentioned within a paragraph will be presented as above, in mono-spaced bold. For example:
File-related classes include filesystem for file systems, file for files, and dir for directories. Each class has its own associated set of permissions.
Proportional Bold
This denotes words or phrases encountered on a system, including application names; dialog box text; labeled buttons; check-box and radio button labels; menu titles and sub-menu titles. For example:
Choose System → Preferences → Mouse from the main menu bar to launch Mouse Preferences. In the Buttons tab, click the Left-handed mouse check box and click Close to switch the primary mouse button from the left to the right (making the mouse suitable for use in the left hand).
To insert a special character into a gedit file, choose Applications → Accessories → Character Map from the main menu bar. Next, choose Search → Find… from the Character Map menu bar, type the name of the character in the Search field and click Next. The character you sought will be highlighted in the Character Table. Double-click this highlighted character to place it in the Text to copy field and then click the Copy button. Now switch back to your document and choose Edit → Paste from the gedit menu bar.
The above text includes application names; system-wide menu names and items; application-specific menu names; and buttons and text found within a GUI interface, all presented in proportional bold and all distinguishable by context.
Mono-spaced Bold Italic or Proportional Bold Italic
Whether mono-spaced bold or proportional bold, the addition of italics indicates replaceable or variable text. Italics denotes text you do not input literally or displayed text that changes depending on circumstance. For example:
To connect to a remote machine using ssh, type ssh username@domain.name at a shell prompt. If the remote machine is example.com and your username on that machine is john, type ssh john@example.com.
The mount -o remount file-system command remounts the named file system. For example, to remount the /home file system, the command is mount -o remount /home.
To see the version of a currently installed package, use the rpm -q package command. It will return a result as follows: package-version-release.
Note the words in bold italics above — username, domain.name, file-system, package, version and release. Each word is a placeholder, either for text you enter when issuing a command or for text displayed by the system.
Aside from standard usage for presenting the title of a work, italics denotes the first use of a new and important term. For example:
Publican is a DocBook publishing system.
1.2. Pull-quote Conventions
Terminal output and source code listings are set off visually from the surrounding text.
Output sent to a terminal is set in mono-spaced roman and presented thus:
Finally, we use three visual styles to draw attention to information that might otherwise be overlooked.
Note
Notes are tips, shortcuts or alternative approaches to the task at hand. Ignoring a note should have no negative consequences, but you might miss out on a trick that makes your life easier.
Important
Important boxes detail things that are easily missed: configuration changes that only apply to the current session, or services that need restarting before an update will apply. Ignoring a box labeled 'Important' will not cause data loss but may cause irritation and frustration.
Warning
Warnings should not be ignored. Ignoring warnings will most likely cause data loss.
2. We Need Feedback!
If you find a typographical error in this manual, or if you have thought of a way to make this manual better, we would love to hear from you! Please submit a report in Bugzilla: http://bugzilla.redhat.com/ against the product Red Hat Enterprise Linux.
When submitting a bug report, be sure to mention the manual's identifier: doc-SELinux_User_Guide and version number: 6.
If you have a suggestion for improving the documentation, try to be as specific as possible when describing it. If you have found an error, please include the section number and some of the surrounding text so we can find it easily.
Chapter 1. Trademark Information
Linux® is the registered trademark of Linus Torvalds in the U.S. and other countries.
UNIX is a registered trademark of The Open Group.
Type Enforcement is a trademark of Secure Computing, LLC, a wholly owned subsidiary of McAfee, Inc., registered in the U.S. and in other countries. Neither McAfee nor Secure Computing, LLC, has consented to the use or reference to this trademark by the author outside of this guide.
Apache is a trademark of The Apache Software Foundation.
MySQL is a trademark or registered trademark of MySQL AB in the U.S. and other countries.
Other products mentioned may be trademarks of their respective corporations.
Security-Enhanced Linux (SELinux) is an implementation of a mandatory access control mechanism in the Linux kernel, checking for allowed operations after standard discretionary access controls are checked. It was created by the National Security Agency and can enforce rules on files and processes in a Linux system, and on their actions, based on defined policy.
When using SELinux, files, including directories and devices, are referred to as objects. Processes, such as a user running a command or the Mozilla® Firefox® application, are referred to as subjects. Most operating systems use a Discretionary Access Control (DAC) system that controls how subjects interact with objects, and how subjects interact with each other. On operating systems using DAC, users control the permissions of files (objects) that they own. For example, on Linux® operating systems, users could make their home directories world-readable, giving users and processes (subjects) access to potentially sensitive information, with no further protection over this unwanted action.
Relying on DAC mechanisms alone is 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 typically has complete discretion over their files, making it difficult 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 files, so minimal protection is provided against malicious software. Many system services and privileged programs run with coarse-grained privileges that far exceed their requirements, so that a flaw in any one of these programs could be exploited to obtain further system access.[1]
The following is an example of permissions used on Linux operating systems that do not run Security-Enhanced Linux (SELinux). The permissions and output in these examples may differ slightly from your system. Use the ls -l command to view file permissions:
In this example, the first three permission bits, rwx, control the access the Linux user1 user (in this case, the owner) has to file1. The next three permission bits, rw-, control the access the Linux group1 group has to file1. The last three permission bits, r--, control the access everyone else has to file1, which includes all users and processes.
Security-Enhanced Linux (SELinux) adds Mandatory Access Control (MAC) to the Linux kernel, and is enabled by default in Red Hat Enterprise Linux. A general purpose MAC architecture needs the ability to enforce an administratively-set security policy over all processes and files 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. 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.[2]
The following is an example of the labels containing security-relevant information that are used on processes, Linux users, and files, on Linux operating systems that run SELinux. This information is called the SELinux context, and is viewed using the ls -Z command:
$ ls -Z file1
-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 level (s0). This information is used to make access control decisions. With DAC, access is controlled based only on Linux user and group IDs. It is important to remember that SELinux policy rules are checked after DAC rules. SELinux policy 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.
2.1. Benefits of running SELinux
All processes and files are labeled with a type. A type defines a domain for processes, and a type for files. Processes are separated from each other by running in their own domains, and SELinux policy rules define how processes interact with files, as well as how processes interact with each other. Access is only allowed if an SELinux policy 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, type, and, optionally, a level.
SELinux policy is administratively-defined, enforced system-wide, and is not set at user discretion.
Reduced vulnerability to privilege escalation attacks. One example: since processes run in domains, and are therefore separated from each other, and because SELinux policy rules define how processes access files and other processes, if a process is compromised, the attacker only has access to the normal functions of that process, and to files the process has been configured to have access to. For example, if the Apache HTTP Server is compromised, an attacker can not use that process to read files in user home directories, unless a specific SELinux policy rule was added or configured to allow such access.
SELinux can be used to enforce data confidentiality and integrity, as well as protecting processes from untrusted inputs.
However, SELinux is not:
antivirus software,
a replacement for passwords, firewalls, or other security systems,
an all-in-one security solution.
SELinux is designed to enhance existing security solutions, not replace them. Even when running SELinux, it is important to continue to follow good security practices, such as keeping software up-to-date, using hard-to-guess passwords, firewalls, and so on.
2.2. Examples
The following examples demonstrate how SELinux increases security:
The default action is deny. If an SELinux policy rule does not exist to allow access, such as for a process opening a file, access is denied.
SELinux can confine Linux users. A number of confined SELinux users exist in SELinux policy. Linux users can be mapped to confined SELinux users to take advantage of the security rules and mechanisms applied to them. For example, mapping a Linux user 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 sudo and su, as well as preventing them from executing files and applications in their home directory - if configured, this prevents users from executing malicious files from their home directories.
Process separation is used. Processes run in their own domains, preventing processes from accessing files used by other processes, as well as preventing processes from accessing other processes. For example, when running SELinux, unless otherwise configured, an attacker can not compromise a Samba server, and then use that Samba server as an attack vector to read and write to files used by other processes, such as databases used by MySQL®.
SELinux helps limit the damage made by configuration mistakes. Domain Name System (DNS) servers often replicate information between each other in what is known as a zone transfer. Attackers can use zone transfers to update DNS servers with false information. When running the Berkeley Internet Name Domain (BIND) as a DNS server in Red Hat Enterprise Linux, even if an administrator forgets to limit which servers can perform a zone transfer, the default SELinux policy prevents zone files [3] from being updated via zone transfers, by the BIND named daemon itself, and by other processes.
SELinux is a Linux security module that is built into the Linux kernel. SELinux is driven by loadable policy rules. When security-relevant access is taking place, such as when a process attempts to open a file, the operation is intercepted in the kernel by SELinux. If an SELinux policy rule allows the operation, it continues, otherwise, the operation is blocked and the process receives an error.
SELinux decisions, such as allowing or disallowing access, are cached. This cache is known as the Access Vector Cache (AVC). Caching decisions decrease how often SELinux policy rules need to be checked, which increases performance. Remember that SELinux policy rules have no effect if DAC rules deny access first.
2.4. SELinux on Other Operating Systems
Refer to the following for information about running SELinux on other Linux distributions:
[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 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 original paper for details and the document as it was first released. Any edits and changes were done by Murray McAllister.
[3]
Text files that include information, such as hostname to IP address mappings, that are used by DNS servers.
Processes and files are labeled with an SELinux context that contains additional information, such as an SELinux user, role, type, and, optionally, a level. When running SELinux, all of this information is used to make access control decisions. In Red Hat Enterprise Linux, SELinux provides a combination of Role-Based Access Control (RBAC), Type Enforcement® (TE), and, optionally, Multi-Level Security (MLS).
The following is an example showing SELinux context. SELinux contexts are used on processes, Linux users, and files, on Linux operating systems that run SELinux. Use the ls -Z command to view the SELinux context of files and directories:
$ ls -Z file1
-rwxrw-r-- user1 group1 unconfined_u:object_r:user_home_t:s0 file1
SELinux contexts follow the SELinux user:role:type:level syntax. The fields are as follows:
SELinux user
The SELinux user identity is an identity known to the policy that is authorized for a specific set of roles, and for a specific MLS range. Each Linux user is mapped to an SELinux user via SELinux policy. This allows Linux users to inherit the restrictions placed on SELinux users. The mapped SELinux user identity is used in the SELinux context for processes in that session, in order to define what roles and levels they can enter. Run the semanage login -l command as the Linux root user to view a list of mappings between SELinux and Linux user accounts (you need to have the policycoreutils-python package installed):
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
Output may differ slightly from system to system. The Login Name column lists Linux users, and the SELinux User column lists which SELinux user the Linux user is mapped to. For processes, the SELinux user limits which roles and levels are accessible. The last column, MLS/MCS Range, is the level used by Multi-Level Security (MLS) and Multi-Category Security (MCS). Levels are briefly discussed later.
role
Part of SELinux is the Role-Based Access Control (RBAC) security model. The role is an attribute of RBAC. SELinux users are authorized for roles, and roles are authorized for domains. The role serves as an intermediary between domains and SELinux users. The roles that can be entered determine which domains can be entered - ultimately, this controls which object types can be accessed. This helps reduce vulnerability to privilege escalation attacks.
type
The type is an attribute of Type Enforcement. The type defines a domain for processes, and a type for files. SELinux policy rules define how types can access each other, whether it be a domain accessing a type, or a domain accessing another domain. Access is only allowed if a specific SELinux policy rule exists that allows it.
level
The level is an attribute of MLS and Multi-Category Security (MCS). An MLS range is a pair of levels, written as lowlevel-highlevel if the levels differ, or lowlevel if the levels are identical (s0-s0 is the same as s0). Each level is a sensitivity-category pair, with categories being optional. If there are categories, the level is written as sensitivity:category-set. If there are no categories, it is written as sensitivity.
If the category set is a contiguous series, it can be abbreviated. For example, c0.c3 is the same as c0,c1,c2,c3. The /etc/selinux/targeted/setrans.conf file maps levels (s0:c0) to human-readable form (ie. CompanyConfidential). Do not edit setrans.conf with a text editor: use semanage to make changes. Refer to the semanage(8) manual page for further information. In Red Hat Enterprise Linux, targeted policy enforces MCS, and in MCS, there is just one sensitivity, s0. MCS in Red Hat Enterprise Linux supports 1024 different categories: c0 through to c1023. s0-s0:c0.c1023 is sensitivity s0 and authorized for all categories.
MLS enforces the Bell-La Padula Mandatory Access Model, and is used in Labeled Security Protection Profile (LSPP) environments. To use MLS restrictions, install the selinux-policy-mls package, and configure MLS to be the default SELinux policy. The MLS policy shipped with Red Hat Enterprise Linux omits many program domains that were not part of the evaluated configuration, and therefore, MLS on a desktop workstation is unusable (no support for the X Window System); however, an MLS policy from the upstream SELinux Reference Policy can be built that includes all program domains.
3.1. Domain Transitions
A process in one domain transitions to another domain by executing an application that has the entrypoint type for the new domain. The entrypoint permission is used in SELinux policy, and controls which applications can be used to enter a domain. The following example demonstrates a domain transition:
A user wants to change their password. To do this, they run the passwd application. The /usr/bin/passwd executable is labeled with the passwd_exec_t type:
$ ls -Z /usr/bin/passwd
-rwsr-xr-x root root system_u:object_r:passwd_exec_t:s0 /usr/bin/passwd
The passwd application accesses /etc/shadow, which is labeled with the shadow_t type:
$ ls -Z /etc/shadow
-r--------. root root system_u:object_r:shadow_t:s0 /etc/shadow
An SELinux policy rule states that processes running in the passwd_t domain are allowed to read and write to files labeled with the shadow_t type. The shadow_t type is only applied to files that are required for a password change. This includes /etc/gshadow, /etc/shadow, and their backup files.
An SELinux policy rule states that the passwd_t domain has entrypoint permission to the passwd_exec_t type.
When a user runs the /usr/bin/passwd application, the user's shell process transitions to the passwd_t domain. With SELinux, since the default action is to deny, and a rule exists that allows (among other things) applications running in the passwd_t domain to access files labeled with the shadow_t type, the passwd application is allowed to access /etc/shadow, and update the user's password.
This example is not exhaustive, and is used as a basic example to explain domain transition. Although there is an actual rule that allows subjects running in the passwd_t domain to access objects labeled with the shadow_t file type, other SELinux policy rules must be met before the subject can transition to a new domain. In this example, Type Enforcement ensures:
The passwd_t domain can only be entered by executing an application labeled with the passwd_exec_t type; can only execute from authorized shared libraries, such as the lib_t type; and can not execute any other applications.
Only authorized domains, such as passwd_t, can write to files labeled with the shadow_t type. Even if other processes are running with superuser privileges, those processes can not write to files labeled with the shadow_t type, as they are not running in the passwd_t domain.
Only authorized domains can transition to the passwd_t domain. For example, the sendmail process running in the sendmail_t domain does not have a legitimate reason to execute passwd; therefore, it can never transition to the passwd_t domain.
Processes running in the passwd_t domain can only read and write to authorized types, such as files labeled with the etc_t or shadow_t types. This prevents the passwd application from being tricked into reading or writing arbitrary files.
3.2. SELinux Contexts for Processes
Use the ps -eZ command to view the SELinux context for processes. For example:
Open a terminal, such as Applications → System Tools → Terminal.
Run the /usr/bin/passwd command. Do not enter a new password.
Open a new tab, or another terminal, and run the ps -eZ | grep passwd command. The output is similar to the following:
In the first tab/terminal, press Ctrl+C to cancel the passwd application.
In this example, when the /usr/bin/passwd application (labeled with the passwd_exec_t type) is executed, the user's shell process transitions to the passwd_t domain. Remember: the type defines a domain for processes, and a type for files.
Use the ps -eZ command to view the SELinux contexts for running processes. The following is a truncated example of the output, and may differ on your system:
In Red Hat Enterprise Linux, Linux users run unconfined by default. This SELinux context shows that the Linux user is mapped to the SELinux unconfined_u user, running as the unconfined_r role, and is running in the unconfined_t domain. s0-s0 is an MLS range, which in this case, is the same as just s0. The categories the user has access to is defined by c0.c1023, which is all categories (c0 through to c1023).
Targeted policy is the default SELinux policy used in Red Hat Enterprise Linux. When using targeted policy, processes that are targeted run in a confined domain, and processes that are not targeted run in an unconfined domain. For example, by default, logged in users run in the unconfined_t domain, and system processes started by init run in the initrc_t domain - both of these domains are unconfined.
Unconfined domains (as well as confined domains) are subject to executable and writeable memory checks. By default, subjects running in an unconfined domain can not allocate writeable memory and execute it. This reduces vulnerability to buffer overflow attacks. These memory checks are disabled by setting Booleans, which allow the SELinux policy to be modified at runtime. Boolean configuration is discussed later.
4.1. Confined Processes
Almost every service that listens on a network, such as sshd or httpd, is confined in Red Hat Enterprise Linux. Also, most processes that run as the Linux root user and perform tasks for users, such as the passwd application, are confined. When a process is confined, it runs in its own domain, such as the httpd process running in the httpd_t domain. If a confined process is compromised by an attacker, depending on SELinux policy configuration, an attacker's access to resources and the possible damage they can do is limited.
The following example demonstrates how SELinux prevents the Apache HTTP Server (httpd) from reading files that are not correctly labeled, such as files intended for use by Samba. This is an example, and should not be used in production. It assumes that the httpd and wget packages are installed, the SELinux targeted policy is used, and that SELinux is running in enforcing mode:
Run the sestatus command to confirm that SELinux is enabled, is running in enforcing mode, and that targeted policy is being used:
$ /usr/sbin/sestatus
SELinux status: enabled
SELinuxfs mount: /selinux
Current mode: enforcing
Mode from config file: enforcing
Policy version: 24
Policy from config file: targeted
SELinux status: enabled is returned when SELinux is enabled. Current mode: enforcing is returned when SELinux is running in enforcing mode. Policy from config file: targeted is returned when the SELinux targeted policy is used.
As the Linux root user, run the touch /var/www/html/testfile command to create a file.
Run the ls -Z /var/www/html/testfile command to view the SELinux context:
By default, Linux users run unconfined in Red Hat Enterprise Linux, which is why the testfile file is labeled with the SELinux unconfined_u user. RBAC is used for processes, not files. Roles do not have a meaning for files - the object_r role is a generic role used for files (on persistent storage and network file systems). Under the /proc/ directory, files related to processes may use the system_r role.[6] The httpd_sys_content_t type allows the httpd process to access this file.
As the Linux root user, run the service httpd start command to start the httpd process. The output is as follows if httpd starts successfully:
# /sbin/service httpd start
Starting httpd: [ OK ]
Change into a directory where your Linux user has write access to, and run the wget http://localhost/testfile command. Unless there are changes to the default configuration, this command succeeds:
--2009-11-06 17:43:01-- http://localhost/testfile
Resolving localhost... 127.0.0.1
Connecting to localhost|127.0.0.1|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 0 [text/plain]
Saving to: `testfile'
[ <=> ] 0 --.-K/s in 0s
2009-11-06 17:43:01 (0.00 B/s) - `testfile' saved [0/0]
The chcon command relabels files; however, such label changes do not survive when the file system is relabeled. For permanent changes that survive a file system relabel, use the semanage command, which is discussed later. As the Linux root user, run the following command to change the type to a type used by Samba:
# chcon -t samba_share_t /var/www/html/testfile
Run the ls -Z /var/www/html/testfile command to view the changes:
Note: the current DAC permissions allow the httpd process access to testfile. Change into a directory where your Linux user has write access to, and run the wget http://localhost/testfile command. Unless there are changes to the default configuration, this command fails:
As the Linux root user, run the rm -i /var/www/html/testfile command to remove testfile.
If you do not require httpd to be running, as the Linux root user, run the service httpd stop command to stop httpd:
# /sbin/service httpd stop
Stopping httpd: [ OK ]
This example demonstrates the additional security added by SELinux. Although DAC rules allowed the httpd process access to testfile in step 7, because the file was labeled with a type that the httpd process does not have access to, SELinux denied access.
An error similar to the following is logged to /var/log/audit/audit.log:
Also, an error similar to the following is logged to /var/log/httpd/error_log:
[Wed May 06 23:00:54 2009] [error] [client 127.0.0.1] (13)Permission denied: access to /testfile denied
4.2. Unconfined Processes
Unconfined processes run in unconfined domains, for example, init programs run in the unconfined initrc_t domain, unconfined kernel processes run in the kernel_t domain, and unconfined Linux users run in the unconfined_t domain. For unconfined processes, SELinux policy rules are applied, but policy rules exist that allow processes running in unconfined domains almost all access. Processes running in unconfined domains fall back to using DAC rules exclusively. If an unconfined process is compromised, SELinux does not prevent an attacker from gaining access to system resources and data, but of course, DAC rules are still used. SELinux is a security enhancement on top of DAC rules - it does not replace them.
The following example demonstrates how the Apache HTTP Server (httpd) can access data intended for use by Samba, when running unconfined. Note: in Red Hat Enterprise Linux, the httpd process runs in the confined httpd_t domain by default. This is an example, and should not be used in production. It assumes that the httpd, wget, dbus and audit packages are installed, that the SELinux targeted policy is used, and that SELinux is running in enforcing mode:
Run the sestatus command to confirm that SELinux is enabled, is running in enforcing mode, and that targeted policy is being used:
$ /usr/sbin/sestatus
SELinux status: enabled
SELinuxfs mount: /selinux
Current mode: enforcing
Mode from config file: enforcing
Policy version: 24
Policy from config file: targeted
SELinux status: enabled is returned when SELinux is enabled. Current mode: enforcing is returned when SELinux is running in enforcing mode. Policy from config file: targeted is returned when the SELinux targeted policy is used.
As the Linux root user, run the touch /var/www/html/test2file command to create a file.
Run the ls -Z /var/www/html/test2file command to view the SELinux context:
$ ls -Z /var/www/html/test2file
-rw-r--r-- root root unconfined_u:object_r:httpd_sys_content_t:s0 /var/www/html/test2file
By default, Linux users run unconfined in Red Hat Enterprise Linux, which is why the test2file file is labeled with the SELinux unconfined_u user. RBAC is used for processes, not files. Roles do not have a meaning for files - the object_r role is a generic role used for files (on persistent storage and network file systems). Under the /proc/ directory, files related to processes may use the system_r role.[7] The httpd_sys_content_t type allows the httpd process to access this file.
The chcon command relabels files; however, such label changes do not survive when the file system is relabeled. For permanent changes that survive a file system relabel, use the semanage command, which is discussed later. As the Linux root user, run the following command to change the type to a type used by Samba:
# chcon -t samba_share_t /var/www/html/test2file
Run the ls -Z /var/www/html/test2file command to view the changes:
$ ls -Z /var/www/html/test2file
-rw-r--r-- root root unconfined_u:object_r:samba_share_t:s0 /var/www/html/test2file
Run the service httpd status command to confirm that the httpd process is not running:
$ /sbin/service httpd status
httpd is stopped
If the output differs, run the service httpd stop command as the Linux root user to stop the httpd process:
# /sbin/service httpd stop
Stopping httpd: [ OK ]
To make the httpd process run unconfined, run the following command as the Linux root user to change the type of /usr/sbin/httpd, to a type that does not transition to a confined domain:
# chcon -t unconfined_exec_t /usr/sbin/httpd
Run the ls -Z /usr/sbin/httpd command to confirm that /usr/sbin/httpd is labeled with the unconfined_exec_t type:
$ ls -Z /usr/sbin/httpd
-rwxr-xr-x root root system_u:object_r:unconfined_exec_t:s0 /usr/sbin/httpd
As the Linux root user, run the service httpd start command to start the httpd process. The output is as follows if httpd starts successfully:
# /sbin/service httpd start
Starting httpd: [ OK ]
Run the ps -eZ | grep httpd command to view the httpd running in the unconfined_t domain:
Change into a directory where your Linux user has write access to, and run the wget http://localhost/test2file command. Unless there are changes to the default configuration, this command succeeds:
$ wget http://localhost/test2file
--2009-05-07 01:41:10-- http://localhost/test2file
Resolving localhost... 127.0.0.1
Connecting to localhost|127.0.0.1|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 0 [text/plain]
Saving to: `test2file.1'
[ <=> ]--.-K/s in 0s
2009-05-07 01:41:10 (0.00 B/s) - `test2file.1' saved [0/0]
Although the httpd process does not have access to files labeled with the samba_share_t type, httpd is running in the unconfined unconfined_t domain, and falls back to using DAC rules, and as such, the wget command succeeds. Had httpd been running in the confined httpd_t domain, the wget command would have failed.
The restorecon command restores the default SELinux context for files. As the Linux root user, run the restorecon -v /usr/sbin/httpd command to restore the default SELinux context for /usr/sbin/httpd:
Run the ls -Z /usr/sbin/httpd command to confirm that /usr/sbin/httpd is labeled with the httpd_exec_t type:
$ ls -Z /usr/sbin/httpd
-rwxr-xr-x root root system_u:object_r:httpd_exec_t:s0 /usr/sbin/httpd
As the Linux root user, run the /sbin/service httpd restart command to restart httpd. After restarting, run the ps -eZ | grep httpd to confirm that httpd is running in the confined httpd_t domain:
If you do not require httpd to be running, as the Linux root user, run the service httpd stop command to stop httpd:
# /sbin/service httpd stop
Stopping httpd: [ OK ]
The examples in these sections demonstrate how data can be protected from a compromised confined-process (protected by SELinux), as well as how data is more accessible to an attacker from a compromised unconfined-process (not protected by SELinux).
4.3. Confined and Unconfined Users
Each Linux user is mapped to an SELinux user via SELinux policy. This allows Linux users to inherit the restrictions on SELinux users. This Linux user mapping is seen by running the semanage login -l command as the Linux root user:
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
In Red Hat Enterprise Linux 6, Linux users are mapped to the SELinux __default__ login by default (which is mapped to the SELinux unconfined_u user). The following defines the default-mapping:
__default__ unconfined_u s0-s0:c0.c1023
The following example demonstrates adding a new Linux user, and that Linux user being mapped to the SELinux unconfined_u user. It assumes that the Linux root user is running unconfined, as it does by default in Red Hat Enterprise Linux 6:
As the Linux root user, run the /usr/sbin/useradd newuser command to create a new Linux user named newuser.
As the Linux root user, run the passwd newuser command to assign a password to the Linux newuser user:
# passwd newuser
Changing password for user newuser.
New UNIX password: Enter a password
Retype new UNIX password: Enter the same password again
passwd: all authentication tokens updated successfully.
Log out of your current session, and log in as the Linux newuser user. When you log in, pam_selinux maps the Linux user to an SELinux user (in this case, unconfined_u), and sets up the resulting SELinux context. The Linux user's shell is then launched with this context. Run the id -Z command to view the context of a Linux user:
[newuser@localhost ~]$ id -Z
unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
Log out of the Linux newuser's session, and log in with your account. If you do not want the Linux newuser user, run the /usr/sbin/userdel -r newuser command as the Linux root user to remove it, along with the Linux newuser's home directory.
Confined and unconfined Linux users are subject to executable and writeable memory checks, and are also restricted by MCS (and MLS, if the MLS policy is used). If unconfined Linux users execute an application that SELinux policy defines can transition from the unconfined_t domain to its own confined domain, unconfined Linux users are still subject to the restrictions of that confined domain. The security benefit of this is that, even though a Linux user is running unconfined, the application remains confined, and therefore, the exploitation of a flaw in the application can be limited by policy. Note: this does not protect the system from the user. Instead, the user and the system are being protected from possible damage caused by a flaw in the application.
The following confined SELinux users are available in Red Hat Enterprise Linux 6:
Table 4.1. SELinux User Capabilities
User
Domain
X Window System
su and sudo
Execute in home directory and /tmp/
Networking
guest_u
guest_t
no
no
optional
no
xguest_u
xguest_t
yes
no
optional
only Firefox
user_u
user_t
yes
no
optional
yes
staff_u
staff_t
yes
only sudo
optional
yes
Linux users in the guest_t, xguest_t, and user_t domains can only run set user ID (setuid) applications if SELinux policy permits it (such as passwd). They can not run the su and /usr/bin/sudo setuid applications, and therefore, can not use these applications to become the Linux root user.
Linux users in the guest_t domain have no network access, and can only log in via a terminal (including ssh; they can log in via ssh, but can not use ssh to connect to another system).
The only network access Linux users in the xguest_t domain have is Firefox connecting to web pages.
Linux users in the xguest_t, user_t and staff_t domains can log in via the X Window System and a terminal.
By default, Linux users in the staff_t domain do not have permissions to execute applications with /usr/bin/sudo. These permissions must be configured by an administrator.
By default, Linux users in the guest_t and xguest_t domains can not execute applications in their home directories or /tmp/, preventing them from executing applications (which inherit users' permissions) in directories they have write access to. This helps prevent flawed or malicious applications from modifying files users' own.
By default, Linux users in the user_t and staff_t domains can execute applications in their home directories and /tmp/. Refer to Section 6.6, “Booleans for Users Executing Applications” for information about allowing and preventing users from executing applications in their home directories and /tmp/.
[6]
When using other policies, such as MLS, other roles may be used, for example, secadm_r.
[7]
When using other policies, such as MLS, other roles may also be used, for example, secadm_r.
The following sections give a brief overview of the main SELinux packages in Red Hat Enterprise Linux; installing and updating packages; which log files are used; the main SELinux configuration file; enabling and disabling SELinux; SELinux modes; configuring Booleans; temporarily and persistently changing file and directory labels; overriding file system labels with the mount command; mounting NFS file systems; and how to preserve SELinux contexts when copying and archiving files and directories.
5.1. SELinux Packages
In Red Hat Enterprise Linux, the SELinux packages are installed by default, in a full installation, unless they are manually excluded during installation. If performing a minimal installation in text mode, the policycoreutils-python package will not be installed by default. Also, by default, SELinux targeted policy is used, and SELinux runs in enforcing mode. The following is a brief description of the main SELinux packages:
policycoreutils-python — provides utilities such as semanage, audit2allow, audit2why and chcat, for operating and managing SELinux.
policycoreutils — provides utilities such as restorecon, secon, setfiles, semodule, load_policy, and setsebool, for operating and managing SELinux.
policycoreutils-gui — provides system-config-selinux, a graphical tool for managing SELinux.
selinux-policy — provides the SELinux Reference Policy. The SELinux Reference Policy is a complete SELinux policy, and is used as a basis for other policies, such as the SELinux targeted policy; refer to the Tresys Technology SELinux Reference Policy page for further information. This package also provides the /usr/share/selinux/devel/policygentool development utility, as well as example policy files.
selinux-policy-policy — provides SELinux policies. For targeted policy, install selinux-policy-targeted. For MLS, install selinux-policy-mls.
setroubleshoot-server — translates denial messages, produced when access is denied by SELinux, into detailed descriptions that are viewed with sealert (which is provided by this package).
setools-console — this package provides the Tresys Technology SETools distribution, a number of tools and libraries for analyzing and querying policy, audit log monitoring and reporting, and file context management[8]. The setools package is a meta-package for SETools. The setools-gui package provides the apol, seaudit, and sediffx tools. The setools-console package provides the seaudit-report, sechecker, sediff, seinfo, sesearch, findcon, replcon, and indexcon command line tools. Refer to the Tresys Technology SETools page for information about these tools.
mcstrans — translates levels, such as s0-s0:c0.c1023, to an easier to read form, such as SystemLow-SystemHigh. This package is not installed by default.
To install packages in Red Hat Enterprise Linux, as the Linux root user, run the yum install package-name command. For example, to install the mcstrans package, run the yum install mcstrans command. To upgrade all installed packages in Red Hat Enterprise Linux, run the yum update command.
5.2. Which Log File is Used
In Red Hat Enterprise Linux 6, the dbus, setroubleshoot-server and audit packages are installed if packages are not removed from the default package selection.
SELinux denial messages, such as the following, are written to /var/log/audit/audit.log by default:
May 7 18:55:56 localhost setroubleshoot: SELinux is preventing httpd (httpd_t) "getattr" to /var/www/html/file1 (samba_share_t). For complete SELinux messages. run sealert -l de7e30d6-5488-466d-a606-92c9f40d316d
In Red Hat Enterprise Linux 6, setroubleshootd no longer constantly runs as a service, however it is still used to analyze the AVC messages. Two new programs act as a method to start setroubleshoot when needed: sedispatch and seapplet. sedispatch runs as part of the audit subsystem, and via dbus, sends a message when an AVC denial occurs, which will go straight to setroubleshootd if it is already running, or it will start setroubleshootd if it is not running. seapplet is a tool which runs in the system's toolbar, waiting for dbus messages in setroubleshootd, and will launch the notification bubble, allowing the user to review the denial.
Starting Daemons Automatically
To configure the auditd and rsyslogd daemons to automatically start at boot, run the following commands as the Linux root user:
/sbin/chkconfig --levels 2345 auditd on
/sbin/chkconfig --levels 2345 rsyslog on
Use the service service-name status command to check if these services are running, for example:
$ /sbin/service auditd status
auditd (pid 1318) is running...
If the above services are not running (service-name is stopped), use the service service-name start command as the Linux root user to start them. For example:
# /sbin/service auditd start
Starting auditd: [ OK ]
5.3. Main Configuration File
The /etc/selinux/config file is the main SELinux configuration file. It controls the SELinux mode and the SELinux policy to use:
# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
# enforcing - SELinux security policy is enforced.
# permissive - SELinux prints warnings instead of enforcing.
# disabled - No SELinux policy is loaded.
SELINUX=enforcing
# SELINUXTYPE= can take one of these two values:
# targeted - Targeted processes are protected,
# mls - Multi Level Security protection.
SELINUXTYPE=targeted
SELINUX=enforcing
The SELINUX option sets the mode SELinux runs in. SELinux has three modes: enforcing, permissive, and disabled. When using enforcing mode, SELinux policy is enforced, and SELinux denies access based on SELinux policy rules. Denial messages are logged. When using permissive mode, SELinux policy is not enforced. SELinux does not deny access, but denials are logged for actions that would have been denied if running SELinux in enforcing mode. When using disabled mode, SELinux is disabled (the SELinux module is not registered with the Linux kernel), and only DAC rules are used.
SELINUXTYPE=targeted
The SELINUXTYPE option sets the SELinux policy to use. Targeted policy is the default policy. Only change this option if you want to use the MLS policy. To use the MLS policy, install the selinux-policy-mls package; configure SELINUXTYPE=mls in /etc/selinux/config; and reboot your system.
Important
When systems run with SELinux in permissive or disabled mode, users have permission to label files incorrectly. Also, files created while SELinux is disabled are not labeled. This causes problems when changing to enforcing mode. To prevent incorrectly labeled and unlabeled files from causing problems, file systems are automatically relabeled when changing from disabled mode to permissive or enforcing mode.
5.4. Enabling and Disabling SELinux
Use the /usr/sbin/getenforce or /usr/sbin/sestatus commands to check the status of SELinux. The getenforce command returns Enforcing, Permissive, or Disabled. The getenforce command returns Enforcing when SELinux is enabled (SELinux policy rules are enforced):
$ /usr/sbin/getenforce
Enforcing
The getenforce command returns Permissive when SELinux is enabled, but SELinux policy rules are not enforced, and only DAC rules are used. The getenforce command returns Disabled if SELinux is disabled.
The sestatus command returns the SELinux status and the SELinux policy being used:
$ /usr/sbin/sestatus
SELinux status: enabled
SELinuxfs mount: /selinux
Current mode: enforcing
Mode from config file: enforcing
Policy version: 24
Policy from config file: targeted
SELinux status: enabled is returned when SELinux is enabled. Current mode: enforcing is returned when SELinux is running in enforcing mode. Policy from config file: targeted is returned when the SELinux targeted policy is used.
5.4.1. Enabling SELinux
On systems with SELinux disabled, the SELINUX=disabled option is configured in /etc/selinux/config:
# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
# enforcing - SELinux security policy is enforced.
# permissive - SELinux prints warnings instead of enforcing.
# disabled - No SELinux policy is loaded.
SELINUX=disabled
# SELINUXTYPE= can take one of these two values:
# targeted - Targeted processes are protected,
# mls - Multi Level Security protection.
SELINUXTYPE=targeted
Also, the getenforce command returns Disabled:
$ /usr/sbin/getenforce
Disabled
To enable SELinux:
Use the rpm -qa | grep selinux, rpm -q policycoreutils, and rpm -qa | grep setroubleshoot commands to confirm that the SELinux packages are installed. This guide assumes the following packages are installed: selinux-policy-targeted, selinux-policy, libselinux, libselinux-python, libselinux-utils, policycoreutils, policycoreutils-python, setroubleshoot, setroubleshoot-server, setroubleshoot-plugins. If these packages are not installed, as the Linux root user, install them via the yum install package-name command. The following packages are optional: policycoreutils-gui, setroubleshoot, selinux-policy-devel, and mcstrans.
Before SELinux is enabled, each file on the file system must be labeled with an SELinux context. Before this happens, confined domains may be denied access, preventing your system from booting correctly. To prevent this, configure SELINUX=permissive in /etc/selinux/config:
# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
# enforcing - SELinux security policy is enforced.
# permissive - SELinux prints warnings instead of enforcing.
# disabled - No SELinux policy is loaded.
SELINUX=permissive
# SELINUXTYPE= can take one of these two values:
# targeted - Targeted processes are protected,
# mls - Multi Level Security protection.
SELINUXTYPE=targeted
As the Linux root user, run the reboot command to restart the system. During the next boot, file systems are labeled. The label process labels all files with an SELinux context:
*** Warning -- SELinux targeted policy relabel is required.
*** Relabeling could take a very long time, depending on file
*** system size and speed of hard drives.
****
Each * character on the bottom line represents 1000 files that have been labeled. In the above example, four * characters represent 4000 files have been labeled. The time it takes to label all files depends upon the number of files on the system, and the speed of the hard disk drives. On modern systems, this process can take as little as 10 minutes.
In permissive mode, SELinux policy is not enforced, but denials are still logged for actions that would have been denied if running in enforcing mode. Before changing to enforcing mode, as the Linux root user, run the grep "SELinux is preventing" /var/log/messages command as the Linux root user to confirm that SELinux did not deny actions during the last boot. If SELinux did not deny actions during the last boot, this command does not return any output. Refer to Chapter 8, Troubleshooting for troubleshooting information if SELinux denied access during boot.
If there were no denial messages in /var/log/messages, configure SELINUX=enforcing in /etc/selinux/config:
# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
# enforcing - SELinux security policy is enforced.
# permissive - SELinux prints warnings instead of enforcing.
# disabled - No SELinux policy is loaded.
SELINUX=enforcing
# SELINUXTYPE= can take one of these two values:
# targeted - Targeted processes are protected,
# mls - Multi Level Security protection.
SELINUXTYPE=targeted
Reboot your system. After reboot, confirm that the getenforce command returns Enforcing:
$ /usr/sbin/getenforce
Enforcing
As the Linux root user, run the /usr/sbin/semanage login -l command to view the mapping between SELinux and Linux users. The output should be as follows:
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
If this is not the case, run the following commands as the Linux root user to fix the user mappings. It is safe to ignore the SELinux-user username is already defined warnings if they occur, where username can be unconfined_u, guest_u, or xguest_u:
/usr/sbin/semanage user -a -S targeted -P user -R "unconfined_r system_r" -r s0-s0:c0.c1023 unconfined_u
/usr/sbin/semanage user -a -S targeted -P user -R guest_r guest_u
/usr/sbin/semanage user -a -S targeted -P user -R xguest_r xguest_u
Important
When systems run with SELinux in permissive or disabled mode, users have permission to label files incorrectly. Also, files created while SELinux is disabled are not labeled. This causes problems when changing to enforcing mode. To prevent incorrectly labeled and unlabeled files from causing problems, file systems are automatically relabeled when changing from disabled mode to permissive or enforcing mode.
5.4.2. Disabling SELinux
To disable SELinux, configure SELINUX=disabled in /etc/selinux/config:
# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
# enforcing - SELinux security policy is enforced.
# permissive - SELinux prints warnings instead of enforcing.
# disabled - No SELinux policy is loaded.
SELINUX=disabled
# SELINUXTYPE= can take one of these two values:
# targeted - Targeted processes are protected,
# mls - Multi Level Security protection.
SELINUXTYPE=targeted
Reboot your system. After reboot, confirm that the getenforce command returns Disabled:
$ /usr/sbin/getenforce
Disabled
5.5. SELinux Modes
SELinux has three modes:
Enforcing: SELinux policy is enforced. SELinux denies access based on SELinux policy rules.
Permissive: SELinux policy is not enforced. SELinux does not deny access, but denials are logged for actions that would have been denied if running in enforcing mode.
Disabled: SELinux is disabled. Only DAC rules are used.
Use the /usr/sbin/setenforce command to change between enforcing and permissive mode. Changes made with /usr/sbin/setenforce do not persist across reboots. To change to enforcing mode, as the Linux root user, run the /usr/sbin/setenforce 1 command. To change to permissive mode, run the /usr/sbin/setenforce 0 command. Use the /usr/sbin/getenforce command to view the current SELinux mode.
Booleans allow parts of SELinux policy to be changed at runtime, without any knowledge of SELinux policy writing. This allows changes, such as allowing services access to NFS file systems, without reloading or recompiling SELinux policy.
5.6.1. Listing Booleans
For a list of Booleans, an explanation of what each one is, and whether they are on or off, run the semanage boolean -l command as the Linux root user. The following example does not list all Booleans:
# /usr/sbin/semanage boolean -l
SELinux boolean Description
ftp_home_dir -> off Allow ftp to read and write files in the user home directories
xen_use_nfs -> off Allow xen to manage nfs files
xguest_connect_network -> on Allow xguest to configure Network Manager
The SELinux boolean column lists Boolean names. The Description column lists whether the Booleans are on or off, and what they do.
In the following example, the ftp_home_dir Boolean is off, preventing the FTP daemon (vsftpd) from reading and writing to files in user home directories:
ftp_home_dir -> off Allow ftp to read and write files in the user home directories
The getsebool -a command lists Booleans, whether they are on or off, but does not give a description of each one. The following example does not list all Booleans:
$ /usr/sbin/getsebool -a
allow_console_login --> off
allow_cvs_read_shadow --> off
allow_daemons_dump_core --> on
Run the getsebool boolean-name command to only list the status of the boolean-name Boolean:
$ /usr/sbin/getsebool allow_console_login
allow_console_login --> off
Use a space-separated list to list multiple Booleans:
$ /usr/sbin/getsebool allow_console_login allow_cvs_read_shadow allow_daemons_dump_core
allow_console_login --> off
allow_cvs_read_shadow --> off
allow_daemons_dump_core --> on
5.6.2. Configuring Booleans
The setsebool boolean-namex command turns Booleans on or off, where boolean-name is a Boolean name, and x is either on to turn the Boolean on, or off to turn it off.
The following example demonstrates configuring the httpd_can_network_connect_db Boolean:
By default, the httpd_can_network_connect_db Boolean is off, preventing Apache HTTP Server scripts and modules from connecting to database servers:
$ /usr/sbin/getsebool httpd_can_network_connect_db
httpd_can_network_connect_db --> off
To temporarily enable Apache HTTP Server scripts and modules to connect to database servers, run the setsebool httpd_can_network_connect_db on command as the Linux root user.
Use the getsebool httpd_can_network_connect_db command to verify the Boolean is turned on:
$ /usr/sbin/getsebool httpd_can_network_connect_db
httpd_can_network_connect_db --> on
This allows Apache HTTP Server scripts and modules to connect to database servers.
This change is not persistent across reboots. To make changes persistent across reboots, run the setsebool -P boolean-name on command as the Linux root user:
# /usr/sbin/setsebool -P httpd_can_network_connect_db on
To temporarily revert to the default behavior, as the Linux root user, run the setsebool httpd_can_network_connect_db off command. For changes that persist across reboots, run the setsebool -P httpd_can_network_connect_db off command.
5.6.3. Booleans for NFS and CIFS
By default, NFS mounts on the client side are labeled with a default context defined by policy for NFS file systems. In common policies, this default context uses the nfs_t type. Also, by default, Samba shares mounted on the client side are labeled with a default context defined by policy. In common policies, this default context uses the cifs_t type.
Depending on policy configuration, services may not be able to read files labeled with the nfs_t or cifs_t types. This may prevent file systems labeled with these types from being mounted and then read or exported by other services. Booleans can be turned on or off to control which services are allowed to access the nfs_t and cifs_t types.
The setsebool and semanage commands must be run as the Linux root user. The setsebool -P command makes persistent changes. Do not use the -P option if you do not want changes to persist across reboots:
Apache HTTP Server
To allow access to NFS file systems (files labeled with the nfs_t type):
/usr/sbin/setsebool -P httpd_use_nfs on
To allow access to Samba file systems (files labeled with the cifs_t type):
/usr/sbin/setsebool -P httpd_use_cifs on
Samba
To export NFS file systems:
/usr/sbin/setsebool -P samba_share_nfs on
FTP (vsftpd)
To allow access to NFS file systems:
/usr/sbin/setsebool -P allow_ftpd_use_nfs on
To allow access to Samba file systems:
/usr/sbin/setsebool -P allow_ftpd_use_cifs on
Other Services
For a list of NFS related Booleans for other services:
/usr/sbin/semanage boolean -l | grep nfs
For a list of Samba related Booleans for other services:
/usr/sbin/semanage boolean -l | grep cifs
Note
These Booleans exist in SELinux policy as shipped with Red Hat Enterprise Linux 6. They may not exist in policy shipped with other versions of Red Hat Enterprise Linux or other operating systems.
5.7. SELinux Contexts - Labeling Files
On systems running SELinux, all processes and files are labeled in a way that represents security-relevant information. This information is called the SELinux context. For files, this is viewed using the ls -Z command:
$ ls -Z file1
-rw-rw-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 level (s0). This information is used to make access control decisions. On DAC systems, access is controlled based on Linux user and group IDs. SELinux policy rules are checked after DAC rules. SELinux policy rules are not used if DAC rules deny access first.
There are multiple commands for managing the SELinux context for files, such as chcon, semanage fcontext, and restorecon.
5.7.1. Temporary Changes: chcon
The chcon command changes the SELinux context for files. However, changes made with the chcon command do not survive a file system relabel, or the execution of the /sbin/restorecon command. SELinux policy controls whether users are able to modify the SELinux context for any given file. When using chcon, users provide all or part of the SELinux context to change. An incorrect file type is a common cause of SELinux denying access.
Quick Reference
Run the chcon -t typefile-name command to change the file type, where type is a type, such as httpd_sys_content_t, and file-name is a file or directory name.
Run the chcon -R -t typedirectory-name command to change the type of the directory and its contents, where type is a type, such as httpd_sys_content_t, and directory-name is a directory name.
Changing a File's or Directory's Type
The following example demonstrates changing the type, and no other attributes of the SELinux context:
Run the cd command without arguments to change into your home directory.
Run the touch file1 command to create a new file. Use the ls -Z file1 command to view the SELinux context for file1:
$ ls -Z file1
-rw-rw-r-- user1 group1 unconfined_u:object_r:user_home_t:s0 file1
In this example, the SELinux context for file1 includes the SELinux unconfined_u user, object_r role, user_home_t type, and the s0 level. For a description of each part of the SELinux context, refer to Chapter 3, SELinux Contexts.
Run the chcon -t samba_share_t file1 command to change the type to samba_share_t. The -t option only changes the type. View the change with ls -Z file1:
$ ls -Z file1
-rw-rw-r-- user1 group1 unconfined_u:object_r:samba_share_t:s0 file1
Use the /sbin/restorecon -v file1 command to restore the SELinux context for the file1 file. Use the -v option to view what changes:
In this example, the previous type, samba_share_t, is restored to the correct, user_home_t type. When using targeted policy (the default SELinux policy in Red Hat Enterprise Linux 6), the /sbin/restorecon command reads the files in the /etc/selinux/targeted/contexts/files/ directory, to see which SELinux context files should have.
The example in this section works the same for directories, for example, if file1 was a directory.
Changing a Directory and its Contents Types
The following example demonstrates creating a new directory, and changing the directory's file type (along with its contents) to a type used by the Apache HTTP Server. The configuration in this example is used if you want Apache HTTP Server to use a different document root (instead of /var/www/html/):
As the Linux root user, run the mkdir /web command to create a new directory, and then the touch /web/file{1,2,3} command to create 3 empty files (file1, file2, and file3). The /web/ directory and files in it are labeled with the default_t type:
As the Linux root user, run the chcon -R -t httpd_sys_content_t /web/ command to change the type of the /web/ directory (and its contents) to httpd_sys_content_t:
Refer to the chcon(1) manual page for further information about chcon.
Note
Type Enforcement is the main permission control used in SELinux targeted policy. For the most part, SELinux users and roles can be ignored.
5.7.2. Persistent Changes: semanage fcontext
The /usr/sbin/semanage fcontext command changes the SELinux context for files. When using targeted policy, changes made with this command are added to the /etc/selinux/targeted/contexts/files/file_contexts file if the changes are to files that exists in file_contexts, or are added to file_contexts.local for new files and directories, such as creating a /web/ directory. setfiles, which is used when a file system is relabeled, and /sbin/restorecon, which restores the default SELinux contexts, read these files. This means that changes made by /usr/sbin/semanage fcontext are persistent, even if the file system is relabeled. SELinux policy controls whether users are able to modify the SELinux context for any given file.
Quick Reference
To make SELinux context changes that survive a file system relabel:
Run the /usr/sbin/semanage fcontext -a optionsfile-name|directory-name command, remembering to use the full path to the file or directory.
Run the /sbin/restorecon -v file-name|directory-name command to apply the context changes.
Changing a File's Type
The following example demonstrates changing a file's type, and no other attributes of the SELinux context:
As the Linux root user, run the touch /etc/file1 command to create a new file. By default, newly-created files in the /etc/ directory are labeled with the etc_t type:
# ls -Z /etc/file1
-rw-r--r-- root root unconfined_u:object_r:etc_t:s0 /etc/file1
As the Linux root user, run the /usr/sbin/semanage fcontext -a -t samba_share_t /etc/file1 command to change the file1 type to samba_share_t. The -a option adds a new record, and the -t option defines a type (samba_share_t). Note: running this command does not directly change the type - file1 is still labeled with the etc_t type:
# /usr/sbin/semanage fcontext -a -t samba_share_t /etc/file1
# ls -Z /etc/file1
-rw-r--r-- root root unconfined_u:object_r:etc_t:s0 /etc/file1
The /usr/sbin/semanage fcontext -a -t samba_share_t /etc/file1 command adds the following entry to /etc/selinux/targeted/contexts/files/file_contexts.local:
/etc/file1 unconfined_u:object_r:samba_share_t:s0
As the Linux root user, run the /sbin/restorecon -v /etc/file1 command to change the type. Since the semanage command added an entry to file.contexts.local for /etc/file1, the /sbin/restorecon command changes the type to samba_share_t:
As the Linux root user, run the rm -i /etc/file1 command to remove file1.
As the Linux root user, run the /usr/sbin/semanage fcontext -d /etc/file1 command to remove the context added for /etc/file1. When the context is removed, running restorecon changes the type to etc_t, rather than samba_share_t.
Changing a Directory's Type
The following example demonstrates creating a new directory and changing that directory's file type, to a type used by Apache HTTP Server:
As the Linux root user, run the mkdir /web command to create a new directory. This directory is labeled with the default_t type:
# ls -dZ /web
drwxr-xr-x root root unconfined_u:object_r:default_t:s0 /web
The ls-d option makes ls list information about a directory, rather than its contents, and the -Z option makes ls display the SELinux context (in this example, unconfined_u:object_r:default_t:s0).
As the Linux root user, run the /usr/sbin/semanage fcontext -a -t httpd_sys_content_t /web command to change the /web/ type to httpd_sys_content_t. The -a option adds a new record, and the -t option defines a type (httpd_sys_content_t). Note: running this command does not directly change the type - /web/ is still labeled with the default_t type:
# /usr/sbin/semanage fcontext -a -t httpd_sys_content_t /web
# ls -dZ /web
drwxr-xr-x root root unconfined_u:object_r:default_t:s0 /web
The /usr/sbin/semanage fcontext -a -t httpd_sys_content_t /web command adds the following entry to /etc/selinux/targeted/contexts/files/file_contexts.local:
/web unconfined_u:object_r:httpd_sys_content_t:s0
As the Linux root user, run the /sbin/restorecon -v /web command to change the type. Since the semanage command added an entry to file.contexts.local for /web, the /sbin/restorecon command changes the type to httpd_sys_content_t:
By default, newly-created files and directories inherit the SELinux type of their parent folders. When using this example, and before removing the SELinux context added for /web/, files and directories created in the /web/ directory are labeled with the httpd_sys_content_t type.
As the Linux root user, run the /usr/sbin/semanage fcontext -d /web command to remove the context added for /web/.
As the Linux root user, run the /sbin/restorecon -v /web command to restore the default SELinux context.
Changing a Directory and its Contents Types
The following example demonstrates creating a new directory, and changing the directory's file type (along with its contents) to a type used by Apache HTTP Server. The configuration in this example is used if you want Apache HTTP Server to use a different document root (instead of /var/www/html/):
As the Linux root user, run the mkdir /web command to create a new directory, and then the touch /web/file{1,2,3} command to create 3 empty files (file1, file2, and file3). The /web/ directory and files in it are labeled with the default_t type:
As the Linux root user, run the /usr/sbin/semanage fcontext -a -t httpd_sys_content_t "/web(/.*)?" command to change the type of the /web/ directory and the files in it, to httpd_sys_content_t. The -a option adds a new record, and the -t option defines a type (httpd_sys_content_t). The "/web(/.*)?" regular expression causes the semanage command to apply changes to the /web/ directory, as well as the files in it. Note: running this command does not directly change the type - /web/ and files in it are still labeled with the default_t type:
The /usr/sbin/semanage fcontext -a -t httpd_sys_content_t "/web(/.*)?" command adds the following entry to /etc/selinux/targeted/contexts/files/file_contexts.local:
As the Linux root user, run the /sbin/restorecon -R -v /web command to change the type of the /web/ directory, as well as all files in it. The -R is for recursive, which means all files and directories under the /web/ directory are labeled with the httpd_sys_content_t type. Since the semanage command added an entry to file.contexts.local for /web(/.*)?, the /sbin/restorecon command changes the types to httpd_sys_content_t:
By default, newly-created files and directories inherit the SELinux type of their parents. In this example, files and directories created in the /web/ directory will be labeled with the httpd_sys_content_t type.
As the Linux root user, run the /usr/sbin/semanage fcontext -d "/web(/.*)?" command to remove the context added for "/web(/.*)?".
As the Linux root user, run the /sbin/restorecon -R -v /web command to restore the default SELinux contexts.
Deleting an added Context
The following example demonstrates adding and removing an SELinux context:
As the Linux root user, run the /usr/sbin/semanage fcontext -a -t httpd_sys_content_t /test command. The /test/ directory does not have to exist. This command adds the following context to /etc/selinux/targeted/contexts/files/file_contexts.local:
/test system_u:object_r:httpd_sys_content_t:s0
To remove the context, as the Linux root user, run the /usr/sbin/semanage fcontext -d file-name|directory-name command, where file-name|directory-name is the first part in file_contexts.local. The following is an example of a context in file_contexts.local:
/test system_u:object_r:httpd_sys_content_t:s0
With the first part being /test. To prevent the /test/ directory from being labeled with the httpd_sys_content_t after running /sbin/restorecon, or after a file system relabel, run the following command as the Linux root user to delete the context from file_contexts.local:
/usr/sbin/semanage fcontext -d /test
If the context is part of a regular expression, for example, /web(/.*)?, use quotation marks around the regular expression:
/usr/sbin/semanage fcontext -d "/web(/.*)?"
Refer to the semanage(8) manual page for further information about /usr/sbin/semanage.
Important
When changing the SELinux context with /usr/sbin/semanage fcontext -a, use the full path to the file or directory to avoid files being mislabeled after a file system relabel, or after the /sbin/restorecon command is run.
5.8. The file_t and default_t Types
For file systems that support extended attributes, when a file that lacks an SELinux context on disk is accessed, it is treated as if it had a default context as defined by SELinux policy. In common policies, this default context uses the file_t type. This should be the only use of this type, so that files without a context on disk can be distinguished in policy, and generally kept inaccessible to confined domains. The file_t type should not exist on correctly-labeled file systems, because all files on a system running SELinux should have an SELinux context, and the file_t type is never used in file-context configuration[9].
The default_t type is used on files that do not match any other pattern in file-context configuration, so that such files can be distinguished from files that do not have a context on disk, and generally kept inaccessible to confined domains. If you create a new top-level directory, such as /mydirectory/, this directory may be labeled with the default_t type. If services need access to such a directory, update the file-contexts configuration for this location. Refer to Section 5.7.2, “Persistent Changes: semanage fcontext” for details on adding a context to the file-context configuration.
5.9. Mounting File Systems
By default, when a file system that supports extended attributes is mounted, the security context for each file is obtained from the security.selinux extended attribute of the file. Files in file systems that do not support extended attributes are assigned a single, default security context from the policy configuration, based on file system type.
Use the mount -o context command to override existing extended attributes, or to specify a different, default context for file systems that do not support extended attributes. This is useful if you do not trust a file system to supply the correct attributes, for example, removable media used in multiple systems. The mount -o context command can also be used to support labeling for file systems that do not support extended attributes, such as File Allocation Table (FAT) or NFS file systems. The context specified with the context is not written to disk: the original contexts are preserved, and are seen when mounting without a context option (if the file system had extended attributes in the first place).
To mount a file system with the specified context, overriding existing contexts if they exist, or to specify a different, default context for a file system that does not support extended attributes, as the Linux root user, use the mount -o context=SELinux_user:role:type:level command when mounting the desired file system. Context changes are not written to disk. By default, NFS mounts on the client side are labeled with a default context defined by policy for NFS file systems. In common policies, this default context uses the nfs_t type. Without additional mount options, this may prevent sharing NFS file systems via other services, such as the Apache HTTP Server. The following example mounts an NFS file system so that it can be shared via the Apache HTTP Server:
# mount server:/export /local/mount/point -o \
context="system_u:object_r:httpd_sys_content_t:s0"
Newly-created files and directories on this file system appear to have the SELinux context specified with -o context; however, since context changes are not written to disk for these situations, the context specified with the context option is only retained if the context option is used on the next mount, and if the same context is specified.
Type Enforcement is the main permission control used in SELinux targeted policy. For the most part, SELinux users and roles can be ignored, so, when overriding the SELinux context with -o context, use the SELinux system_u user and object_r role, and concentrate on the type. If you are not using the MLS policy or multi-category security, use the s0 level.
Note
When a file system is mounted with a context option, context changes (by users and processes) are prohibited. For example, running chcon on a file system mounted with a context option results in a Operation not supported error.
5.9.2. Changing the Default Context
As mentioned in Section 5.8, “The file_t and default_t Types”, on file systems that support extended attributes, when a file that lacks an SELinux context on disk is accessed, it is treated as if it had a default context as defined by SELinux policy. In common policies, this default context uses the file_t type. If it is desirable to use a different default context, mount the file system with the defcontext option.
The following example mounts a newly-created file system (on /dev/sda2) to the newly-created /test/ directory. It assumes that there are no rules in /etc/selinux/targeted/contexts/files/ that define a context for the /test/ directory:
# mount /dev/sda2 /test/ -o defcontext="system_u:object_r:samba_share_t:s0"
In this example:
the defcontext option defines that system_u:object_r:samba_share_t:s0 is "the default security context for unlabeled files"[10].
when mounted, the root directory (/test/) of the file system is treated as if it is labeled with the context specified by defcontext (this label is not stored on disk). This affects the labeling for files created under /test/: new files inherit the samba_share_t type, and these labels are stored on disk.
files created under /test/ while the file system was mounted with a defcontext option retain their labels.
5.9.3. Mounting an NFS File System
By default, NFS mounts on the client side are labeled with a default context defined by policy for NFS file systems. In common policies, this default context uses the nfs_t type. Depending on policy configuration, services, such as Apache HTTP Server and MySQL, may not be able to read files labeled with the nfs_t type. This may prevent file systems labeled with this type from being mounted and then read or exported by other services.
If you would like to mount an NFS file system and read or export that file system with another service, use the context option when mounting to override the nfs_t type. Use the following context option to mount NFS file systems so that they can be shared via the Apache HTTP Server:
mount server:/export /local/mount/point -o \
context="system_u:object_r:httpd_sys_content_t:s0"
Since context changes are not written to disk for these situations, the context specified with the context option is only retained if the context option is used on the next mount, and if the same context is specified.
As an alternative to mounting file systems with context options, Booleans can be turned on to allow services access to file systems labeled with the nfs_t type. Refer to Section 5.6.3, “Booleans for NFS and CIFS” for instructions on configuring Booleans to allow services access to the nfs_t type.
5.9.4. Multiple NFS Mounts
When mounting multiple mounts from the same NFS export, attempting to override the SELinux context of each mount with a different context, results in subsequent mount commands failing. In the following example, the NFS server has a single export, /export, which has two subdirectories, web/ and database/. The following commands attempt two mounts from a single NFS export, and try to override the context for each one:
# mount server:/export/web /local/web -o \
context="system_u:object_r:httpd_sys_content_t:s0"
# mount server:/export/database /local/database -o \
context="system_u:object_r:mysqld_db_t:s0"
The second mount command fails, and the following is logged to /var/log/messages:
kernel: SELinux: mount invalid. Same superblock, different security settings for (dev 0:15, type nfs)
To mount multiple mounts from a single NFS export, with each mount having a different context, use the -o nosharecache,context options. The following example mounts multiple mounts from a single NFS export, with a different context for each mount (allowing a single service access to each one):
# mount server:/export/web /local/web -o \
nosharecache,context="system_u:object_r:httpd_sys_content_t:s0"
# mount server:/export/database /local/database -o \
nosharecache,context="system_u:object_r:mysqld_db_t:s0"
In this example, server:/export/web is mounted locally to /local/web/, with all files being labeled with the httpd_sys_content_t type, allowing Apache HTTP Server access. server:/export/database is mounted locally to /local/database, with all files being labeled with the mysqld_db_t type, allowing MySQL access. These type changes are not written to disk.
Important
The nosharecache options allows you to mount the same subdirectory of an export multiple times with different contexts (for example, mounting /export/web multiple times). Do not mount the same subdirectory from an export multiple times with different contexts, as this creates an overlapping mount, where files are accessible under two different contexts.
5.9.5. Making Context Mounts Persistent
To make context mounts persistent across remounting and reboots, add entries for the file systems in /etc/fstab or an automounter map, and use the desired context as a mount option. The following example adds an entry to /etc/fstab for an NFS context mount:
These sections describe what happens to SELinux contexts when copying, moving, and archiving files and directories. Also, it explains how to preserve contexts when copying and archiving.
5.10.1. Copying Files and Directories
When a file or directory is copied, a new file or directory is created if it does not exist. That new file or directory's context is based on default-labeling rules, not the original file or directory's context (unless options were used to preserve the original context). For example, files created in user home directories are labeled with the user_home_t type:
If such a file is copied to another directory, such as /etc/, the new file is created in accordance to default-labeling rules for the /etc/ directory. Copying a file (without additional options) may not preserve the original context:
When file1 is copied to /etc/, if /etc/file1 does not exist, /etc/file1 is created as a new file. As shown in the example above, /etc/file1 is labeled with the etc_t type, in accordance to default-labeling rules.
When a file is copied over an existing file, the existing file's context is preserved, unless the user specified cp options to preserve the context of the original file, such as --preserve=context. SELinux policy may prevent contexts from being preserved during copies.
Copying Without Preserving SELinux Contexts
When copying a file with the cp command, if no options are given, the type is inherited from the targeted, parent directory:
In this example, file1 is created in a user's home directory, and is labeled with the user_home_t type. The /var/www/html/ directory is labeled with the httpd_sys_content_t type, as shown with the ls -dZ /var/www/html/ command. When file1 is copied to /var/www/html/, it inherits the httpd_sys_content_t type, as shown with the ls -Z /var/www/html/file1 command.
Preserving SELinux Contexts When Copying
Use the cp --preserve=context command to preserve contexts when copying:
In this example, file1 is created in a user's home directory, and is labeled with the user_home_t type. The /var/www/html/ directory is labeled with the httpd_sys_content_t type, as shown with the ls -dZ /var/www/html/ command. Using the --preserve=context option preserves SELinux contexts during copy operations. As shown with the ls -Z /var/www/html/file1 command, the file1user_home_t type was preserved when the file was copied to /var/www/html/.
Copying and Changing the Context
Use the cp -Z command to change the destination copy's context. The following example was performed in the user's home directory:
In this example, the context is defined with the -Z option. Without the -Z option, file2 would be labeled with the unconfined_u:object_r:user_home_t context.
Copying a File Over an Existing File
When a file is copied over an existing file, the existing file's context is preserved (unless an option is used to preserve contexts). For example:
In this example, two files are created: /etc/file1, labeled with the etc_t type, and /tmp/file2, labeled with the user_tmp_t type. The cp /tmp/file2 /etc/file1 command overwrites file1 with file2. After copying, the ls -Z /etc/file1 command shows file1 labeled with the etc_t type, not the user_tmp_t type from /tmp/file2 that replaced /etc/file1.
Important
Copy files and directories, rather than moving them. This helps ensure they are labeled with the correct SELinux contexts. Incorrect SELinux contexts can prevent processes from accessing such files and directories.
5.10.2. Moving Files and Directories
File and directories keep their current SELinux context when they are moved. In many cases, this is incorrect for the location they are being moved to. The following example demonstrates moving a file from a user's home directory to /var/www/html/, which is used by the Apache HTTP Server. Since the file is moved, it does not inherit the correct SELinux context:
Run the cd command without any arguments to change into your home directory. Once in your home directory, run the touch file1 command to create a file. This file is labeled with the user_home_t type:
$ ls -Z file1
-rw-rw-r-- user1 group1 unconfined_u:object_r:user_home_t:s0 file1
Run the ls -dZ /var/www/html/ command to view the SELinux context of the /var/www/html/ directory:
$ ls -dZ /var/www/html/
drwxr-xr-x root root system_u:object_r:httpd_sys_content_t:s0 /var/www/html/
By default, the /var/www/html/ directory is labeled with the httpd_sys_content_t type. Files and directories created under the /var/www/html/ directory inherit this type, and as such, they are labeled with this type.
As the Linux root user, run the mv file1 /var/www/html/ command to move file1 to the /var/www/html/ directory. Since this file is moved, it keeps its current user_home_t type:
By default, the Apache HTTP Server can not read files that are labeled with the user_home_t type. If all files comprising a web page are labeled with the user_home_t type, or another type that the Apache HTTP Server can not read, permission is denied when attempting to access them via Firefox or text-based Web browsers.
Important
Moving files and directories with the mv command may result in the wrong SELinux context, preventing processes, such as the Apache HTTP Server and Samba, from accessing such files and directories.
5.10.3. Checking the Default SELinux Context
Use the /usr/sbin/matchpathcon command to check if files and directories have the correct SELinux context. From the matchpathcon(8) manual page: "matchpathcon queries the system policy and outputs the default security context associated with the file path."[11]. The following example demonstrates using the /usr/sbin/matchpathcon command to verify that files in /var/www/html/ directory are labeled correctly:
As the Linux root user, run the touch /var/www/html/file{1,2,3} command to create three files (file1, file2, and file3). These files inherit the httpd_sys_content_t type from the /var/www/html/ directory:
As the Linux root user, run the chcon -t samba_share_t /var/www/html/file1 command to change the file1 type to samba_share_t. Note: the Apache HTTP Server can not read files or directories labeled with the samba_share_t type.
The /usr/sbin/matchpathcon-V option compares the current SELinux context to the correct, default context in SELinux policy. Run the /usr/sbin/matchpathcon -V /var/www/html/* command to check all files in the /var/www/html/ directory:
$ /usr/sbin/matchpathcon -V /var/www/html/*
/var/www/html/file1 has context unconfined_u:object_r:samba_share_t:s0, should be system_u:object_r:httpd_sys_content_t:s0
/var/www/html/file2 verified.
/var/www/html/file3 verified.
The following output from the /usr/sbin/matchpathcon command explains that file1 is labeled with the samba_share_t type, but should be labeled with the httpd_sys_content_t type:
/var/www/html/file1 has context unconfined_u:object_r:samba_share_t:s0, should be system_u:object_r:httpd_sys_content_t:s0
To resolve the label problem and allow the Apache HTTP Server access to file1, as the Linux root user, run the /sbin/restorecon -v /var/www/html/file1 command:
tar does not retain extended attributes by default. Since SELinux contexts are stored in extended attributes, contexts can be lost when archiving files. Use tar --selinux to create archives that retain contexts. If a Tar archive contains files without extended attributes, or if you want the extended attributes to match the system defaults, run the archive through /sbin/restorecon:
$ tar -xvf archive.tar | /sbin/restorecon -f -
Note: depending on the directory, you may need to be the Linux root user to run the /sbin/restorecon command.
The following example demonstrates creating a Tar archive that retains SELinux contexts:
As the Linux root user, run the touch /var/www/html/file{1,2,3} command to create three files (file1, file2, and file3). These files inherit the httpd_sys_content_t type from the /var/www/html/ directory:
Run the cd /var/www/html/ command to change into the /var/www/html/ directory. Once in this directory, as the Linux root user, run the tar --selinux -cf test.tar file{1,2,3} command to create a Tar archive named test.tar.
As the Linux root user, run the mkdir /test command to create a new directory, and then, run the chmod 777 /test/ command to allow all users full-access to the /test/ directory.
Run the cp /var/www/html/test.tar /test/ command to copy the test.tar file in to the /test/ directory.
Run the cd /test/ command to change into the /test/ directory. Once in this directory, run the tar -xvf test.tar command to extract the Tar archive.
Run the ls -lZ /test/ command to view the SELinux contexts. The httpd_sys_content_t type has been retained, rather than being changed to default_t, which would have happened had the --selinux not been used:
If the /test/ directory is no longer required, as the Linux root user, run the rm -ri /test/ command to remove it, as well as all files in it.
Refer to the tar(1) manual page for further information about tar, such as the --xattrs option that retains all extended attributes.
5.10.5. Archiving Files with star
star does not retain extended attributes by default. Since SELinux contexts are stored in extended attributes, contexts can be lost when archiving files. Use star -xattr -H=exustar to create archives that retain contexts. The star package is not installed by default. To install star, run the yum install star command as the Linux root user.
The following example demonstrates creating a Star archive that retains SELinux contexts:
As the Linux root user, run the touch /var/www/html/file{1,2,3} command to create three files (file1, file2, and file3). These files inherit the httpd_sys_content_t type from the /var/www/html/ directory:
Run the cd /var/www/html/ command to change into the /var/www/html/ directory. Once in this directory, as the Linux root user, run the star -xattr -H=exustar -c -f=test.star file{1,2,3} command to create a Star archive named test.star:
# star -xattr -H=exustar -c -f=test.star file{1,2,3}
star: 1 blocks + 0 bytes (total of 10240 bytes = 10.00k).
As the Linux root user, run the mkdir /test command to create a new directory, and then, run the chmod 777 /test/ command to allow all users full-access to the /test/ directory.
Run the cp /var/www/html/test.star /test/ command to copy the test.star file in to the /test/ directory.
Run the cd /test/ command to change into the /test/ directory. Once in this directory, run the star -x -f=test.star command to extract the Star archive:
$ star -x -f=test.star
star: 1 blocks + 0 bytes (total of 10240 bytes = 10.00k).
Run the ls -lZ /test/ command to view the SELinux contexts. The httpd_sys_content_t type has been retained, rather than being changed to default_t, which would have happened had the --selinux not been used:
If the /test/ directory is no longer required, as the Linux root user, run the rm -ri /test/ command to remove it, as well as all files in it.
If star is no longer required, as the Linux root user, run the yum remove star command to remove the package.
Refer to the star(1) manual page for further information about star.
5.11. Information Gathering Tools
These tools are command line tools, providing formatted output. They are harder to use as part of command line piping, but they provide gathered and well formatted information quickly.
avcstat
This command provides a short output of the access vector cache statistics since boot. You can watch the statistics in real time by specifying a time interval in seconds. This provides updated statistics since the initial output. The statistics file used is /selinux/avc/cache_stats, and you can specify a different cache file with the -f /path/to/file option.
This utility is useful in describing the break-down of a policy, such as the number of classes, types, Booleans, allow rules, etc. seinfo is a command line utility that uses either the policy.conf file or a binary policy file as input.
The output of seinfo will vary between binary and source files. For example, the policy source file uses the { } brackets to group multiple rule elements onto a single line. A similar effect happens with attributes, where a single attribute expands into one or many types. Because these are expanded and no longer relevant in the binary policy file, they have a return value of zero in the search results. However, the number of rules greatly increases as each formerly one line rule using brackets is now a number of individual lines.
Some items are not present in the binary policy. For example, neverallow rules are only checked during policy compile, not during runtime, and initial SIDs are not part of the binary policy since they are required prior to the policy being loaded by the kernel during boot.
The seinfo command can also list the number of types with the domain attribute, giving an estimate of the number of different confined processes:
# seinfo -adomain -x | wc -l
550
Not all domain types are confined. To look at the number of unconfined domains, use the unconfined_domain attribute:
# seinfo -aunconfined_domain_type -x | wc -l
52
Permissive domains can be counted with the --permissive option.
# seinfo --permissive -x | wc -l
31
Remove the | wc -l option in the above commands to see the full lists.
sesearch
You can use the sesearch command to search for a particular type in the policy. You can search either policy source files or the binary file. For example:
The sesearch command can provide the number of allow rules:
# sesearch --allow | wc -l
262798
And the number of dontaudit rules:
# sesearch --dontaudit | wc -l
156712
[8]
Brindle, Joshua. "Re: blurb for fedora setools packages" Email to Murray McAllister. 1 November 2008. Any edits or changes in this version were done by Murray McAllister.
[9]
Files in /etc/selinux/targeted/contexts/files/ define contexts for files and directories. Files in this directory are read by restorecon and setfiles to restore files and directories to their default contexts.
[11]
The matchpathcon(8) manual page, as shipped with the libselinux-utils package in Red Hat Enterprise Linux, is written by Daniel Walsh. Any edits or changes in this version were done by Murray McAllister.
A number of confined SELinux users are available in Red Hat Enterprise Linux 6. Each Linux user is mapped to an SELinux user via SELinux policy, allowing Linux users to inherit the restrictions placed on SELinux users, for example (depending on the user), not being able to: run the X Window System; use networking; run setuid applications (unless SELinux policy permits it); or run the su and sudo commands. This helps protect the system from the user. Refer to Section 4.3, “Confined and Unconfined Users” for further information about confined users.
6.1. Linux and SELinux User Mappings
As the Linux root user, run the semanage login -l command to view the mapping between Linux users and SELinux users:
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
In Red Hat Enterprise Linux 6, Linux users are mapped to the SELinux __default__ login by default (which is in turn mapped to the SELinux unconfined_u user). When a Linux user is created with the useradd command, if no options are specified, they are mapped to the SELinux unconfined_u user. The following defines the default-mapping:
__default__ unconfined_u s0-s0:c0.c1023
6.2. Confining New Linux Users: useradd
Linux users mapped to the SELinux unconfined_u user run in the unconfined_t domain. This is seen by running the id -Z command while logged-in as a Linux user mapped to unconfined_u:
$ id -Z
unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
When Linux users run in the unconfined_t domain, SELinux policy rules are applied, but policy rules exist that allow Linux users running in the unconfined_t domain almost all access. If unconfined Linux users execute an application that SELinux policy defines can transition from the unconfined_t domain to its own confined domain, unconfined Linux users are still subject to the restrictions of that confined domain. The security benefit of this is that, even though a Linux user is running unconfined, the application remains confined, and therefore, the exploitation of a flaw in the application can be limited by policy. Note: this does not protect the system from the user. Instead, the user and the system are being protected from possible damage caused by a flaw in the application.
When creating Linux users with useradd, use the -Z option to specify which SELinux user they are mapped to. The following example creates a new Linux user, useruuser, and maps that user to the SELinux user_u user. Linux users mapped to the SELinux user_u user run in the user_t domain. In this domain, Linux users are unable to run setuid applications unless SELinux policy permits it (such as passwd), and can not run su or sudo, preventing them from becoming the Linux root user with these commands.
As the Linux root user, run the /usr/sbin/useradd -Z user_u useruuser command to create a new Linux user (useruuser) that is mapped to the SELinux user_u user.
As the Linux root user, run the semanage login -l command to view the mapping between the Linux useruuser user and user_u:
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
useruuser user_u s0
As the Linux root user, run the passwd useruuser command to assign a password to the Linux useruuser user:
# passwd useruuser
Changing password for user useruuser.
New UNIX password: Enter a password
Retype new UNIX password: Enter the same password again
passwd: all authentication tokens updated successfully.
Log out of your current session, and log in as the Linux useruuser user. When you log in, pam_selinux maps the Linux user to an SELinux user (in this case, user_u), and sets up the resulting SELinux context. The Linux user's shell is then launched with this context. Run the id -Z command to view the context of a Linux user:
[useruuser@localhost ~]$ id -Z
user_u:user_r:user_t:s0
Log out of the Linux useruuser's session, and log back in with your account. If you do not want the Linux useruuser user, run the /usr/sbin/userdel -r useruuser command as the Linux root user to remove it, along with its home directory.
6.3. Confining Existing Linux Users: semanage login
If a Linux user is mapped to the SELinux unconfined_u user (the default behavior), and you would like to change which SELinux user they are mapped to, use the semanage login command. The following example creates a new Linux user named newuser, then maps that Linux user to the SELinux user_u user:
As the Linux root user, run the useradd newuser command to create a new Linux user (newuser). Since this user uses the default mapping, it does not appear in the semanage login -l output:
# /usr/sbin/useradd newuser
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
To map the Linux newuser user to the SELinux user_u user, run the following command as the Linux root user:
# /usr/sbin/semanage login -a -s user_u newuser
The -a option adds a new record, and the -s option specifies the SELinux user to map a Linux user to. The last argument, newuser, is the Linux user you want mapped to the specified SELinux user.
To view the mapping between the Linux newuser user and user_u, run the semanage login -l command as the Linux root user:
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
newuser user_u s0
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
As the Linux root user, run the passwd newuser command to assign a password to the Linux newuser user:
# passwd newuser
Changing password for user newuser.
New password: Enter a password
Retype new password: Enter the same password again
passwd: all authentication tokens updated successfully.
Log out of your current session, and log in as the Linux newuser user. Run the id -Z command to view the newuser's SELinux context:
[newuser@rlocalhost ~]$ id -Z
user_u:user_r:user_t:s0
Log out of the Linux newuser's session, and log back in with your account. If you do not want the Linux newuser user, run the userdel -r newuser command as the Linux root user to remove it, along with its home directory. Also, run the semanage login -d newuser command to remove the mapping between the Linux newuser user and user_u:
# /usr/sbin/userdel -r newuser
# /usr/sbin/semanage login -d newuser
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ unconfined_u s0-s0:c0.c1023
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
6.4. Changing the Default Mapping
In Red Hat Enterprise Linux 6, Linux users are mapped to the SELinux __default__ login by default (which is in turn mapped to the SELinux unconfined_u user). If you would like new Linux users, and Linux users not specifically mapped to an SELinux user to be confined by default, change the default mapping with the semanage login command.
For example, run the following command as the Linux root user to change the default mapping from unconfined_u to user_u:
Run the semanage login -l command as the Linux root user to verify the __default__ login is mapped to user_u:
# /usr/sbin/semanage login -l
Login Name SELinux User MLS/MCS Range
__default__ user_u s0
root unconfined_u s0-s0:c0.c1023
system_u system_u s0-s0:c0.c1023
If a new Linux user is created and an SELinux user is not specified, or if an existing Linux user logs in and does not match a specific entry from the semanage login -l output, they are mapped to user_u, as per the __default__ login.
To change back to the default behavior, run the following command as the Linux root user to map the __default__ login to the SELinux unconfined_u user:
The xguest package provides a kiosk user account. This account is used to secure machines that people walk up to and use, such as those at libraries, banks, airports, information kiosks, and coffee shops. The kiosk user account is very limited: essentially, it only allows users to log in and use Firefox to browse Internet websites. Any changes made while logged in with his account, such as creating files or changing settings, are lost when you log out.
To set up the kiosk account:
As the Linux root user, run yum install xguest command to install the xguest package. Install dependencies as required.
In order to allow the kiosk account to be used by a variety of people, the account is not password-protected, and as such, the account can only be protected if SELinux is running in enforcing mode. Before logging in with this account, use the getenforce command to confirm that SELinux is running in enforcing mode:
$ /usr/sbin/getenforce
Enforcing
If this is not the case, refer to Section 5.5, “SELinux Modes” for information about changing to enforcing mode. It is not possible to log in with this account if SELinux is in permissive mode or disabled.
You can only log in to this account via the GNOME Display Manager (GDM). Once the xguest package is installed, a Guest account is added to the GDM login screen.
6.6. Booleans for Users Executing Applications
Not allowing Linux users to execute applications (which inherit users' permissions) in their home directories and /tmp/, which they have write access to, helps prevent flawed or malicious applications from modifying files that users own. In Red Hat Enterprise Linux 6, by default, Linux users in the guest_t and xguest_t domains can not execute applications in their home directories or /tmp/; however, by default, Linux users in the user_t and staff_t domains can.
Booleans are available to change this behavior, and are configured with the setsebool command. The setsebool command must be run as the Linux root user. The setsebool -P command makes persistent changes. Do not use the -P option if you do not want changes to persist across reboots:
guest_t
To allow Linux users in the guest_t domain to execute applications in their home directories and /tmp/:
/usr/sbin/setsebool -P allow_guest_exec_content on
xguest_t
To allow Linux users in the xguest_t domain to execute applications in their home directories and /tmp/:
/usr/sbin/setsebool -P allow_xguest_exec_content on
user_t
To prevent Linux users in the user_t domain from executing applications in their home directories and /tmp/:
/usr/sbin/setsebool -P allow_user_exec_content off
staff_t
To prevent Linux users in the staff_t domain from executing applications in their home directories and /tmp/:
/usr/sbin/setsebool -P allow_staff_exec_content off
sVirt is a technology included in Red Hat Enterprise Linux 6 that integrates SELinux and virtualization. sVirt applies Mandatory Access Control (MAC) to improve security when using virtual machines. The main reasons for integrating these technologies are to improve security and harden the system against bugs in the hypervisor that might be used as an attack vector aimed toward the host or to another virtual machine.
This chapter describes how sVirt integrates with virtualization technologies in Red Hat Enterprise Linux 6.
Non-Virtualized Environment
In a non-virtualized environment, hosts are separated from each other physically and each host has a self-contained environment, consisting of services such as a Web server, or a DNS server. These services communicate directly to their own user space, host kernel and physical host, offering their services directly to the network. The following image represents a non-virtualized environment:
Virtualized Environment
In a virtualized environment, several operating systems can be housed (as "guests") within a single host kernel and physical host. The following image represents a virtualized environment:
7.1. Security and Virtualization
When services are not virtualized, machines are physically separated. Any exploit is usually contained to the affected machine, with the obvious exception of network attacks. When services are grouped together in a virtualized environment, extra vulnerabilities emerge in the system. If there is a security flaw in the hypervisor that can be exploited by a guest instance, this guest may be able to not only attack the host, but also other guests running on that host. This is not theoretical; attacks already exist on hypervisors. These attacks can extend beyond the guest instance and could expose other guests to attack.
sVirt is an effort to isolate guests and limit their ability to launch further attacks if exploited. This is demonstrated in the following image, where an attack can not break out of the virtual machine and extend to another host instance:
SELinux introduces a pluggable security framework for virtualized instances in its implementation of Mandatory Access Control (MAC). The sVirt framework allows guests and their resources to be uniquely labelled. Once labelled, rules can be applied which can reject access between different guests.
7.2. sVirt Labelling
Like other services under the protection of SELinux, sVirt uses process-based mechanisms and restrictions to provide an extra layer of security over guest instances. Under typical use, you should not even notice that sVirt is working in the background. This section describes the labelling features of sVirt.
As shown in the following output, when using sVirt, each Virtual Machine (VM) process is labelled and runs with a dynamically generated level. Each process is isolated from other VMs with different levels:
The actual disk images are automatically labelled to match the processes, as shown in the following output:
# ls -lZ /var/lib/libvirt/images/*
system_u:object_r:svirt_image_t:s0:c87,c520 image1
The following table outlines the different labels that can be assigned when using sVirt:
Table 7.1. sVirt Labels
Type
SELinux Context
Description
Virtual Machine Processes
system_u:system_r:svirt_t:MCS1
MCS1 is a randomly selected MCS field. Currently approximately 500,000 labels are supported.
Virtual Machine Image
system_u:object_r:svirt_image_t:MCS1
Only svirt_t processes with the same MCS fields are able to read/write these image files and devices.
Virtual Machine Shared Read/Write Content
system_u:object_r:svirt_image_t:s0
All svirt_t processes are allowed to write to the svirt_image_t:s0 files and devices.
Virtual Machine Shared Shared Read Only content
system_u:object_r:svirt_content_t:s0
All svirt_t processes are able to read files/devices with this label.
Virtual Machine Image
system_u:object_r:virt_content_t:s0
System default label used when an image exits. No svirt_t virtual processes are allowed to read files/devices with this label.
It is also possible to perform static labeling when using sVirt. Static labels allow the administrator to select a specific label, including the MCS/MLS field, for a virtual machine. Administrators who run statically-labeled virtual machines are responsible for setting the correct label on the image files. The virtual machine will always be started with that label, and the sVirt system will never modify the label of a statically-labelled virtual machine's content. This allows the sVirt component to run in an MLS environment. You can also run multiple virtual machines with different sensitivity levels on a system, depending on your requirements.
The following chapter describes what happens when SELinux denies access; the top three causes of problems; where to find information about correct labeling; analyzing SELinux denials; and creating custom policy modules with audit2allow.
8.1. What Happens when Access is Denied
SELinux decisions, such as allowing or disallowing access, are cached. This cache is known as the Access Vector Cache (AVC). Denial messages are logged when SELinux denies access. These denials are also known as "AVC denials", and are logged to a different location, depending on which daemons are running:
Daemon
Log Location
auditd on
/var/log/audit/audit.log
auditd off; rsyslogd on
/var/log/messages
setroubleshootd, rsyslogd, and auditd on
/var/log/audit/audit.log. Easier-to-read denial messages also sent to /var/log/messages
If you are running the X Window System, have the setroubleshoot and setroubleshoot-server packages installed, and the setroubleshootd and auditd daemons are running, a warning is displayed when access is denied by SELinux:
Clicking on 'Show' presents a detailed analysis of why SELinux denied access, and a possible solution for allowing access. If you are not running the X Window System, it is less obvious when access is denied by SELinux. For example, users browsing your website may receive an error similar to the following:
Forbidden
You don't have permission to access file name on this server
For these situations, if DAC rules (standard Linux permissions) allow access, check /var/log/messages and /var/log/audit/audit.log for "SELinux is preventing" and "denied" errors respectively. This can be done by running the following commands as the Linux root user:
grep "SELinux is preventing" /var/log/messages
grep "denied" /var/log/audit/audit.log
8.2. Top Three Causes of Problems
The following sections describe the top three causes of problems: labeling problems, configuring Booleans and ports for services, and evolving SELinux rules.
8.2.1. Labeling Problems
On systems running SELinux, all processes and files are labeled with a label that contains security-relevant information. This information is called the SELinux context. If these labels are wrong, access may be denied. If an application is labeled incorrectly, the process it transitions to may not have the correct label, possibly causing SELinux to deny access, and the process being able to create mislabeled files.
A common cause of labeling problems is when a non-standard directory is used for a service. For example, instead of using /var/www/html/ for a website, an administrator wants to use /srv/myweb/. On Red Hat Enterprise Linux 6, the /srv/ directory is labeled with the var_t type. Files and directories created and /srv/ inherit this type. Also, newly-created top-level directories (such as /myserver/) may be labeled with the default_t type. SELinux prevents the Apache HTTP Server (httpd) from accessing both of these types. To allow access, SELinux must know that the files in /srv/myweb/ are to be accessible to httpd:
# /usr/sbin/semanage fcontext -a -t httpd_sys_content_t \
"/srv/myweb(/.*)?"
This semanage command adds the context for the /srv/myweb/ directory (and all files and directories under it) to the SELinux file-context configuration[12]. The semanage command does not change the context. As the Linux root user, run the restorecon command to apply the changes:
The matchpathcon command checks the context of a file path and compares it to the default label for that path. The following example demonstrates using matchpathcon on a directory that contains incorrectly labeled files:
$ /usr/sbin/matchpathcon -V /var/www/html/*
/var/www/html/index.html has context unconfined_u:object_r:user_home_t:s0, should be system_u:object_r:httpd_sys_content_t:s0
/var/www/html/page1.html has context unconfined_u:object_r:user_home_t:s0, should be system_u:object_r:httpd_sys_content_t:s0
In this example, the index.html and page1.html files are labeled with the user_home_t type. This type is used for files in user home directories. Using the mv command to move files from your home directory may result in files being labeled with the user_home_t type. This type should not exist outside of home directories. Use the restorecon command to restore such files to their correct type:
Services can be run in a variety of ways. To cater for this, you must tell SELinux how you are running services. This can be achieved via Booleans that allow parts of SELinux policy to be changed at runtime, without any knowledge of SELinux policy writing. This allows changes, such as allowing services access to NFS file systems, without reloading or recompiling SELinux policy. Also, running services on non-default port numbers requires policy configuration to be updated via the semanage command.
For example, to allow the Apache HTTP Server to communicate with MySQL, turn the httpd_can_network_connect_db Boolean on:
# /usr/sbin/setsebool -P httpd_can_network_connect_db on
If access is denied for a particular service, use the getsebool and grep commands to see if any Booleans are available to allow access. For example, use the getsebool -a | grep ftp command to search for FTP related Booleans:
$ /usr/sbin/getsebool -a | grep ftp
allow_ftpd_anon_write --> off
allow_ftpd_full_access --> off
allow_ftpd_use_cifs --> off
allow_ftpd_use_nfs --> off
ftp_home_dir --> off
ftpd_connect_db --> off
httpd_enable_ftp_server --> off
tftp_anon_write --> off
For a list of Booleans and whether they are on or off, run the /usr/sbin/getsebool -a command. For a list of Booleans, an explanation of what each one is, and whether they are on or off, run the /usr/sbin/semanage boolean -l command as the Linux root user. Refer to Section 5.6, “Booleans” for information about listing and configuring Booleans.
Port Numbers
Depending on policy configuration, services may only be allowed to run on certain port numbers. Attempting to change the port a service runs on without changing policy may result in the service failing to start. For example, run the semanage port -l | grep http command as the Linux root user to list http related ports:
The http_port_t port type defines the ports Apache HTTP Server can listen on, which in this case, are TCP ports 80, 443, 488, 8008, 8009, and 8443. If an administrator configures httpd.conf so that httpd listens on port 9876 (Listen 9876), but policy is not updated to reflect this, the service httpd start command fails:
# /sbin/service httpd start
Starting httpd: (13)Permission denied: make_sock: could not bind to address [::]:9876
(13)Permission denied: make_sock: could not bind to address 0.0.0.0:9876
no listening sockets available, shutting down
Unable to open logs
[FAILED]
An SELinux denial similar to the following is logged to /var/log/audit/audit.log:
To allow httpd to listen on a port that is not listed for the http_port_t port type, run the semanage port command to add a port to policy configuration[13]:
# /usr/sbin/semanage port -a -t http_port_t -p tcp 9876
The -a option adds a new record; the -t option defines a type; and the -p option defines a protocol. The last argument is the port number to add.
8.2.3. Evolving Rules and Broken Applications
Applications may be broken, causing SELinux to deny access. Also, SELinux rules are evolving - SELinux may not have seen an application running in a certain way, possibly causing it to deny access, even though the application is working as expected. For example, if a new version of PostgreSQL is released, it may perform actions the current policy has not seen before, causing access to be denied, even though access should be allowed.
For these situations, after access is denied, use audit2allow to create a custom policy module to allow access. Refer to Section 8.3.8, “Allowing Access: audit2allow” for information about using audit2allow.
8.3. Fixing Problems
The following sections help troubleshoot issues. They go over: checking Linux permissions, which are checked before SELinux rules; possible causes of SELinux denying access, but no denials being logged; manual pages for services, which contain information about labeling and Booleans; permissive domains, for allowing one process to run permissive, rather than the whole system; how to search for and view denial messages; analyzing denials; and creating custom policy modules with audit2allow.
8.3.1. Linux Permissions
When access is denied, check standard Linux permissions. As mentioned in Chapter 2, Introduction, most operating systems use a Discretionary Access Control (DAC) system to control access, allowing users to control the permissions of files that they own. SELinux policy rules are checked after DAC rules. SELinux policy rules are not used if DAC rules deny access first.
If access is denied and no SELinux denials are logged, use the ls -l command to view the standard Linux permissions:
In this example, index.html is owned by the root user and group. The root user has read and write permissions (-rw), and members of the root group have read permissions (-r-). Everyone else has no access (---). By default, such permissions do not allow httpd to read this file. To resolve this issue, use the chown command to change the owner and group. This command must be run as the Linux root user:
# chown apache:apache /var/www/html/index.html
This assumes the default configuration, in which httpd runs as the Linux apache user. If you run httpd with a different user, replace apache:apache with that user.
In certain situations, AVC denials may not be logged when SELinux denies access. Applications and system library functions often probe for more access than required to perform their tasks. To maintain least privilege without filling audit logs with AVC denials for harmless application probing, the policy can silence AVC denials without allowing a permission by using dontaudit rules. These rules are common in standard policy. The downside of dontaudit is that, although SELinux denies access, denial messages are not logged, making troubleshooting hard.
To temporarily disable dontaudit rules, allowing all denials to be logged, run the following command as the Linux root user:
/usr/sbin/semodule -DB
The -D option disables dontaudit rules; the -B option rebuilds policy. After running semodule -DB, try exercising the application that was encountering permission problems, and see if SELinux denials — relevant to the application — are now being logged. Take care in deciding which denials should be allowed, as some should be ignored and handled via dontaudit rules. If in doubt, or in search of guidance, contact other SELinux users and developers on an SELinux list, such as fedora-selinux-list.
To rebuild policy and enable dontaudit rules, run the following command as the Linux root user:
/usr/sbin/semodule -B
This restores the policy to its original state. For a full list of dontaudit rules, run the sesearch --dontaudit command. Narrow down searches using the -s domain option and the grep command. For example:
Manual pages for services contain valuable information, such as what file type to use for a given situation, and Booleans to change the access a service has (such as httpd accessing NFS file systems). This information may be in the standard manual page, or a manual page with selinux prepended or appended.
For example, the httpd_selinux(8) manual page has information about what file type to use for a given situation, as well as Booleans to allow scripts, sharing files, accessing directories inside user home directories, and so on. Other manual pages with SELinux information for services include:
Samba: the samba_selinux(8) manual page describes that files and directories to be exported via Samba must be labeled with the samba_share_t type, as well as Booleans to allow files labeled with types other than samba_share_t to be exported via Samba.
NFS: the nfs_selinux(8) manual page describes that, by default, file systems can not be exported via NFS, and that to allow file systems to be exported, Booleans such as nfs_export_all_ro or nfs_export_all_rw must be turned on.
Berkeley Internet Name Domain (BIND): the named(8) manual page describes what file type to use for a given situation (see the Red Hat SELinux BIND Security Profile section). The named_selinux(8) manual page describes that, by default, named can not write to master zone files, and to allow such access, the named_write_master_zones Boolean must be turned on.
The information in manual pages helps you configure the correct file types and Booleans, helping to prevent SELinux from denying access.
8.3.4. Permissive Domains
When SELinux is running in permissive mode, SELinux does not deny access, but denials are logged for actions that would have been denied if running in enforcing mode. Previously, it was not possible to make a single domain permissive (remember: processes run in domains). In certain situations, this led to making the whole system permissive to troubleshoot issues.
In Red Hat Enterprise Linux 4 and 5, domain_disable_trans Booleans are available to prevent an application from transitioning to a confined domain, and therefore, the process runs in an unconfined domain, such as initrc_t. Consequently, turning such Booleans on can cause major problems. For example, the following happens if the httpd_disable_trans Boolean is turned on:
The httpd service runs in the unconfined initrc_t domain. Files created by processes running in the initrc_t domain may not have the same labeling rules applied as files created by a process running in the httpd_t domain, potentially allowing processes to create mislabeled files. This causes access problems later on.
Confined domains that are allowed to communicate with httpd_t can not communicate with initrc_t, possibly causing additional failures.
To address these issues, Red Hat Enterprise Linux 6 introduces permissive domains. Permissive domains allow an administrator to configure a single process (domain) to run permissive, rather than making the whole system permissive. SELinux checks are still performed for permissive domains; however, the kernel allows access and reports an AVC denial for situations where SELinux would have denied access.
Permissive domains have the following uses:
They can be used for making a single process (domain) run permissive to troubleshoot an issue without putting the entire system at risk by making it permissive.
They allow an administrator to create policies for new applications. Previously, it was recommended that a minimal policy be created, and then the entire machine put into permissive mode, so that the application could run, but SELinux denials still logged. audit2allow could then be used to help write the policy. This put the whole system at risk. With permissive domains, only the domain in the new policy can be marked permissive, without putting the whole system at risk.
8.3.4.1. Making a Domain Permissive
To make a domain permissive, run the semanage permissive -a domain command, where domain is the domain you want to make permissive. For example, run the following command as the Linux root user to make the httpd_t domain (the domain the Apache HTTP Server runs in) permissive:
/usr/sbin/semanage permissive -a httpd_t
To view a list of domains you have made permissive, run the semodule -l | grep permissive command as the Linux root user. For example:
If you no longer want a domain to be permissive, run the semanage permissive -d domain command as the Linux root user. For example:
/usr/sbin/semanage permissive -d httpd_t
8.3.4.2. Denials for Permissive Domains
The SYSCALL message is different for permissive domains. The following is an example AVC denial (and the associated system call) from the Apache HTTP Server:
By default, the httpd_t domain is not permissive, and as such, the action is denied, and the SYSCALL message contains success=no. The following is an example AVC denial for the same situation, except the semanage permissive -a httpd_t command has been run to make the httpd_t domain permissive:
In this case, although an AVC denial was logged, access was not denied, as shown by success=yes in the SYSCALL message.
Refer to Dan Walsh's "Permissive Domains" blog entry for further information about permissive domains.
8.3.5. Searching For and Viewing Denials
This section assumes the setroubleshoot, setroubleshoot-server, dbus and audit packages are installed, and that the auditd, rsyslogd, and setroubleshootd daemons are running. Refer to Section 5.2, “Which Log File is Used” for information about starting these daemons. A number of tools are available for searching for and viewing SELinux denials, such as ausearch, aureport, and sealert.
ausearch
The audit package provides ausearch. From the ausearch(8) manual page: "ausearch is a tool that can query the audit daemon logs based for events based on different search criteria"[14]. The ausearch tool accesses /var/log/audit/audit.log, and as such, must be run as the Linux root user:
Searching For
Command
all denials
/sbin/ausearch -m avc
denials for that today
/sbin/ausearch -m avc -ts today
denials from the last 10 minutes
/sbin/ausearch -m avc -ts recent
To search for SELinux denials for a particular service, use the -c comm-name option, where comm-name "is the executable’s name"[15], for example, httpd for the Apache HTTP Server, and smbd for Samba:
/sbin/ausearch -m avc -c httpd
/sbin/ausearch -m avc -c smbd
Refer to the ausearch(8) manual page for further ausearch options.
aureport
The audit package provides aureport. From the aureport(8) manual page: "aureport is a tool that produces summary reports of the audit system logs"[16]. The aureport tool accesses /var/log/audit/audit.log, and as such, must be run as the Linux root user. To view a list of SELinux denials and how often each one occurred, run the aureport -a command. The following is example output that includes two denials:
# /sbin/aureport -a
AVC Report
========================================================
# date time comm subj syscall class permission obj event
========================================================
1. 05/01/2009 21:41:39 httpd unconfined_u:system_r:httpd_t:s0 195 file getattr system_u:object_r:samba_share_t:s0 denied 2
2. 05/03/2009 22:00:25 vsftpd unconfined_u:system_r:ftpd_t:s0 5 file read unconfined_u:object_r:cifs_t:s0 denied 4
Refer to the aureport(8) manual page for further aureport options.
sealert
The setroubleshoot-server package provides sealert, which reads denial messages translated by setroubleshoot-server. Denials are assigned IDs, as seen in /var/log/messages. The following is an example denial from messages:
setroubleshoot: SELinux is preventing httpd (httpd_t) "getattr" to /var/www/html/file1 (samba_share_t). For complete SELinux messages. run sealert -l 84e0b04d-d0ad-4347-8317-22e74f6cd020
In this example, the denial ID is 84e0b04d-d0ad-4347-8317-22e74f6cd020. The -l option takes an ID as an argument. Running the sealert -l 84e0b04d-d0ad-4347-8317-22e74f6cd020 command presents a detailed analysis of why SELinux denied access, and a possible solution for allowing access.
If you are running the X Window System, have the setroubleshoot and setroubleshoot-server packages installed, and the setroubleshootd, dbus and auditd daemons are running, a warning is displayed when access is denied by SELinux. Clicking on 'Show' launches the sealert GUI, and displays denials in HTML output:
Run the sealert -b command to launch the sealert GUI.
Run the sealert -l \* command to view a detailed analysis of all denials.
As the Linux root user, run the sealert -a /var/log/audit/audit.log -H > audit.html command to create a HTML version of the sealert analysis, as seen with the sealert GUI.
Refer to the sealert(8) manual page for further sealert options.
8.3.6. Raw Audit Messages
Raw audit messages are logged to /var/log/audit/audit.log. The following is an example AVC denial (and the associated system call) that occurred when the Apache HTTP Server (running in the httpd_t domain) attempted to access the /var/www/html/file1 file (labeled with the samba_share_t type):
The item in braces indicates the permission that was denied. getattr indicates the source process was trying to read the target file's status information. This occurs before reading files. This action is denied due to the file being accessed having the wrong label. Commonly seen permissions include getattr, read, and write.
comm="httpd"
The executable that launched the process. The full path of the executable is found in the exe= section of the system call (SYSCALL) message, which in this case, is exe="/usr/sbin/httpd".
path="/var/www/html/file1"
The path to the object (target) the process attempted to access.
scontext="unconfined_u:system_r:httpd_t:s0"
The SELinux context of the process that attempted the denied action. In this case, it is the SELinux context of the Apache HTTP Server, which is running in the httpd_t domain.
tcontext="unconfined_u:object_r:samba_share_t:s0"
The SELinux context of the object (target) the process attempted to access. In this case, it is the SELinux context of file1. Note: the samba_share_t type is not accessible to processes running in the httpd_t domain.
In certain situations, the tcontext may match the scontext, for example, when a process attempts to execute a system service that will change characteristics of that running process, such as the user ID. Also, the tcontext may match the scontext when a process tries to use more resources (such as memory) than normal limits allow, resulting in a security check to see if that process is allowed to break those limits.
From the system call (SYSCALL) message, two items are of interest:
success=no: indicates whether the denial (AVC) was enforced or not. success=no indicates the system call was not successful (SELinux denied access). success=yes indicates the system call was successful - this can be seen for permissive domains or unconfined domains, such as initrc_t and kernel_t.
exe="/usr/sbin/httpd": the full path to the executable that launched the process, which in this case, is exe="/usr/sbin/httpd".
An incorrect file type is a common cause for SELinux denying access. To start troubleshooting, compare the source context (scontext) with the target context (tcontext). Should the process (scontext) be accessing such an object (tcontext)? For example, the Apache HTTP Server (httpd_t) should only be accessing types specified in the httpd_selinux(8) manual page, such as httpd_sys_content_t, public_content_t, and so on, unless configured otherwise.
8.3.7. sealert Messages
Denials are assigned IDs, as seen in /var/log/messages. The following is an example AVC denial (logged to messages) that occurred when the Apache HTTP Server (running in the httpd_t domain) attempted to access the /var/www/html/file1 file (labeled with the samba_share_t type):
hostname setroubleshoot: SELinux is preventing httpd (httpd_t) "getattr" to /var/www/html/file1 (samba_share_t). For complete SELinux messages. run sealert -l 84e0b04d-d0ad-4347-8317-22e74f6cd020
As suggested, run the sealert -l 84e0b04d-d0ad-4347-8317-22e74f6cd020 command to view the complete message. This command only works on the local machine, and presents the same information as the sealert GUI:
$ sealert -l 84e0b04d-d0ad-4347-8317-22e74f6cd020
Summary:
SELinux is preventing httpd (httpd_t) "getattr" to /var/www/html/file1
(samba_share_t).
Detailed Description:
SELinux denied access to /var/www/html/file1 requested by httpd.
/var/www/html/file1 has a context used for sharing by different program. If you
would like to share /var/www/html/file1 from httpd also, you need to change its
file context to public_content_t. If you did not intend to this access, this
could signal a intrusion attempt.
Allowing Access:
You can alter the file context by executing chcon -t public_content_t
'/var/www/html/file1'
Fix Command:
chcon -t public_content_t '/var/www/html/file1'
Additional Information:
Source Context unconfined_u:system_r:httpd_t:s0
Target Context unconfined_u:object_r:samba_share_t:s0
Target Objects /var/www/html/file1 [ file ]
Source httpd
Source Path /usr/sbin/httpd
Port <Unknown>
Host hostname
Source RPM Packages httpd-2.2.10-2
Target RPM Packages
Policy RPM selinux-policy-3.5.13-11.fc12
Selinux Enabled True
Policy Type targeted
MLS Enabled True
Enforcing Mode Enforcing
Plugin Name public_content
Host Name hostname
Platform Linux hostname 2.6.27.4-68.fc12.i686 #1 SMP Thu Oct
30 00:49:42 EDT 2008 i686 i686
Alert Count 4
First Seen Wed Nov 5 18:53:05 2008
Last Seen Wed Nov 5 01:22:58 2008
Local ID 84e0b04d-d0ad-4347-8317-22e74f6cd020
Line Numbers
Raw Audit Messages
node=hostname type=AVC msg=audit(1225812178.788:101): avc: denied { getattr } for pid=2441 comm="httpd" path="/var/www/html/file1" dev=dm-0 ino=284916 scontext=unconfined_u:system_r:httpd_t:s0 tcontext=unconfined_u:object_r:samba_share_t:s0 tclass=file
node=hostname type=SYSCALL msg=audit(1225812178.788:101): arch=40000003 syscall=196 success=no exit=-13 a0=b8e97188 a1=bf87aaac a2=54dff4 a3=2008171 items=0 ppid=2439 pid=2441 auid=502 uid=48 gid=48 euid=48 suid=48 fsuid=48 egid=48 sgid=48 fsgid=48 tty=(none) ses=3 comm="httpd" exe="/usr/sbin/httpd" subj=unconfined_u:system_r:httpd_t:s0 key=(null)
Summary
A brief summary of the denied action. This is the same as the denial in /var/log/messages. In this example, the httpd process was denied access to a file (file1), which is labeled with the samba_share_t type.
Detailed Description
A more verbose description. In this example, file1 is labeled with the samba_share_t type. This type is used for files and directories that you want to export via Samba. The description suggests changing the type to a type that can be accessed by the Apache HTTP Server and Samba, if such access is desired.
Allowing Access
A suggestion for how to allow access. This may be relabeling files, turning a Boolean on, or making a local policy module. In this case, the suggestion is to label the file with a type accessible to both the Apache HTTP Server and Samba.
Fix Command
A suggested command to allow access and resolve the denial. In this example, it gives the command to change the file1 type to public_content_t, which is accessible to the Apache HTTP Server and Samba.
Additional Information
Information that is useful in bug reports, such as the policy package name and version (selinux-policy-3.5.13-11.fc12), but may not help towards solving why the denial occurred.
Raw Audit Messages
The raw audit messages from /var/log/audit/audit.log that are associated with the denial. Refer to Section 8.3.6, “Raw Audit Messages” for information about each item in the AVC denial.
8.3.8. Allowing Access: audit2allow
Do not use the example in this section in production. It is used only to demonstrate the use of audit2allow.
From the audit2allow(1) manual page: "audit2allow - generate SELinux policy allow rules from logs of denied operations"[17]. After analyzing denials as per Section 8.3.7, “sealert Messages”, and if no label changes or Booleans allowed access, use audit2allow to create a local policy module. After access is denied by SELinux, running the audit2allow command presents Type Enforcement rules that allow the previously denied access.
The following example demonstrates using audit2allow to create a policy module:
A denial and the associated system call are logged to /var/log/audit/audit.log:
In this example, certwatch (comm="certwatch") was denied write access ({ write }) to a directory labeled with the var_t type (tcontext=system_u:object_r:var_t:s0). Analyze the denial as per Section 8.3.7, “sealert Messages”. If no label changes or Booleans allowed access, use audit2allow to create a local policy module.
With a denial logged, such as the certwatch denial in step 1, run the audit2allow -w -a command to produce a human-readable description of why access was denied. The -a option causes all audit logs to be read. The -w option produces the human-readable description. The audit2allow tool accesses /var/log/audit/audit.log, and as such, must be run as the Linux root user:
# audit2allow -w -a
type=AVC msg=audit(1226270358.848:238): avc: denied { write } for pid=13349 comm="certwatch" name="cache" dev=dm-0 ino=218171 scontext=system_u:system_r:certwatch_t:s0 tcontext=system_u:object_r:var_t:s0 tclass=dir
Was caused by:
Missing type enforcement (TE) allow rule.
You can use audit2allow to generate a loadable module to allow this access.
As shown, access was denied due to a missing Type Enforcement rule.
Run the audit2allow -a command to view the Type Enforcement rule that allows the denied access:
# audit2allow -a
#============= certwatch_t ==============
allow certwatch_t var_t:dir write;
Important
Missing Type Enforcement rules are usually caused by bugs in SELinux policy, and should be reported in Red Hat Bugzilla. For Red Hat Enterprise Linux, create bugs against the Red Hat Enterprise Linux product, and select the selinux-policy component. Include the output of the audit2allow -w -a and audit2allow -a commands in such bug reports.
To use the rule displayed by audit2allow -a, run the audit2allow -a -M mycertwatch command as the Linux root user to create custom module. The -M option creates a Type Enforcement file (.te) with the name specified with -M, in your current working directory:
# audit2allow -a -M mycertwatch
******************** IMPORTANT ***********************
To make this policy package active, execute:
semodule -i mycertwatch.pp
# ls
mycertwatch.pp mycertwatch.te
Also, audit2allow compiles the Type Enforcement rule into a policy package (.pp). To install the module, run the /usr/sbin/semodule -i mycertwatch.pp command as the Linux root user.
Important
Modules created with audit2allow may allow more access than required. It is recommended that policy created with audit2allow be posted to an SELinux list, such as fedora-selinux-list, for review. If you believe their is a bug in policy, create a bug in Red Hat Bugzilla.
If you have multiple denials from multiple processes, but only want to create a custom policy for a single process, use the grep command to narrow down the input for audit2allow. The following example demonstrates using grep to only send denials related to certwatch through audit2allow:
# grep certwatch /var/log/audit/audit.log | audit2allow -M mycertwatch2
******************** IMPORTANT ***********************
To make this policy package active, execute:
# /usr/sbin/semodule -i mycertwatch2.pp
[12]
Files in /etc/selinux/targeted/contexts/files/ define contexts for files and directories. Files in this directory are read by restorecon and setfiles to restore files and directories to their default contexts.
[13]
The semanage port -a command adds an entry to the /etc/selinux/targeted/modules/active/ports.local file. Note: by default, this file can only be viewed by the Linux root user.
[14]
From the ausearch(8) manual page, as shipped with the audit package in Red Hat Enterprise Linux 6.
[15]
From the ausearch(8) manual page, as shipped with the audit package in Red Hat Enterprise Linux 6.
[16]
From the aureport(8) manual page, as shipped with the audit package in Red Hat Enterprise Linux 6.
[17]
From the audit2allow(1) manual page, which is available when the policycoreutils-sandbox package in Red Hat Enterprise Linux 6 is installed.
Researchers in NSA's National Information Assurance Research Laboratory (NIARL) designed and implemented flexible mandatory access controls in the major subsystems of the Linux kernel and implemented the new operating system components provided by the Flask architecture, namely the security server and the access vector cache. The NSA researchers reworked the LSM-based SELinux for inclusion in Linux 2.6. NSA has also led the development of similar controls for the X Window System (XACE/XSELinux) and for Xen (XSM/Flask).
