1 <?xml version="1.0" encoding="UTF-8"?>
2 <!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"
3 "http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd" []>
5 <article class="whitepaper" id="LinuxSecurityModule" lang="en">
7 <title>Linux Security Modules: General Security Hooks for Linux</title>
10 <firstname>Stephen</firstname>
11 <surname>Smalley</surname>
13 <orgname>NAI Labs</orgname>
14 <address><email>ssmalley@nai.com</email></address>
18 <firstname>Timothy</firstname>
19 <surname>Fraser</surname>
21 <orgname>NAI Labs</orgname>
22 <address><email>tfraser@nai.com</email></address>
26 <firstname>Chris</firstname>
27 <surname>Vance</surname>
29 <orgname>NAI Labs</orgname>
30 <address><email>cvance@nai.com</email></address>
36 <sect1 id="Introduction"><title>Introduction</title>
39 In March 2001, the National Security Agency (NSA) gave a presentation
40 about Security-Enhanced Linux (SELinux) at the 2.5 Linux Kernel
41 Summit. SELinux is an implementation of flexible and fine-grained
42 nondiscretionary access controls in the Linux kernel, originally
43 implemented as its own particular kernel patch. Several other
44 security projects (e.g. RSBAC, Medusa) have also developed flexible
45 access control architectures for the Linux kernel, and various
46 projects have developed particular access control models for Linux
47 (e.g. LIDS, DTE, SubDomain). Each project has developed and
48 maintained its own kernel patch to support its security needs.
52 In response to the NSA presentation, Linus Torvalds made a set of
53 remarks that described a security framework he would be willing to
54 consider for inclusion in the mainstream Linux kernel. He described a
55 general framework that would provide a set of security hooks to
56 control operations on kernel objects and a set of opaque security
57 fields in kernel data structures for maintaining security attributes.
58 This framework could then be used by loadable kernel modules to
59 implement any desired model of security. Linus also suggested the
60 possibility of migrating the Linux capabilities code into such a
65 The Linux Security Modules (LSM) project was started by WireX to
66 develop such a framework. LSM is a joint development effort by
67 several security projects, including Immunix, SELinux, SGI and Janus,
68 and several individuals, including Greg Kroah-Hartman and James
69 Morris, to develop a Linux kernel patch that implements this
70 framework. The patch is currently tracking the 2.4 series and is
71 targeted for integration into the 2.5 development series. This
72 technical report provides an overview of the framework and the example
73 capabilities security module provided by the LSM kernel patch.
78 <sect1 id="framework"><title>LSM Framework</title>
81 The LSM kernel patch provides a general kernel framework to support
82 security modules. In particular, the LSM framework is primarily
83 focused on supporting access control modules, although future
84 development is likely to address other security needs such as
85 auditing. By itself, the framework does not provide any additional
86 security; it merely provides the infrastructure to support security
87 modules. The LSM kernel patch also moves most of the capabilities
88 logic into an optional security module, with the system defaulting
89 to the traditional superuser logic. This capabilities module
90 is discussed further in <xref linkend="cap"/>.
94 The LSM kernel patch adds security fields to kernel data structures
95 and inserts calls to hook functions at critical points in the kernel
96 code to manage the security fields and to perform access control. It
97 also adds functions for registering and unregistering security
98 modules, and adds a general <function>security</function> system call
99 to support new system calls for security-aware applications.
103 The LSM security fields are simply <type>void*</type> pointers. For
104 process and program execution security information, security fields
105 were added to <structname>struct task_struct</structname> and
106 <structname>struct linux_binprm</structname>. For filesystem security
107 information, a security field was added to
108 <structname>struct super_block</structname>. For pipe, file, and socket
109 security information, security fields were added to
110 <structname>struct inode</structname> and
111 <structname>struct file</structname>. For packet and network device security
112 information, security fields were added to
113 <structname>struct sk_buff</structname> and
114 <structname>struct net_device</structname>. For System V IPC security
115 information, security fields were added to
116 <structname>struct kern_ipc_perm</structname> and
117 <structname>struct msg_msg</structname>; additionally, the definitions
118 for <structname>struct msg_msg</structname>, <structname>struct
119 msg_queue</structname>, and <structname>struct
120 shmid_kernel</structname> were moved to header files
121 (<filename>include/linux/msg.h</filename> and
122 <filename>include/linux/shm.h</filename> as appropriate) to allow
123 the security modules to use these definitions.
127 Each LSM hook is a function pointer in a global table,
128 security_ops. This table is a
129 <structname>security_operations</structname> structure as defined by
130 <filename>include/linux/security.h</filename>. Detailed documentation
131 for each hook is included in this header file. At present, this
132 structure consists of a collection of substructures that group related
133 hooks based on the kernel object (e.g. task, inode, file, sk_buff,
134 etc) as well as some top-level hook function pointers for system
135 operations. This structure is likely to be flattened in the future
136 for performance. The placement of the hook calls in the kernel code
137 is described by the "called:" lines in the per-hook documentation in
138 the header file. The hook calls can also be easily found in the
139 kernel code by looking for the string "security_ops->".
144 Linus mentioned per-process security hooks in his original remarks as a
145 possible alternative to global security hooks. However, if LSM were
146 to start from the perspective of per-process hooks, then the base
147 framework would have to deal with how to handle operations that
148 involve multiple processes (e.g. kill), since each process might have
149 its own hook for controlling the operation. This would require a
150 general mechanism for composing hooks in the base framework.
151 Additionally, LSM would still need global hooks for operations that
152 have no process context (e.g. network input operations).
153 Consequently, LSM provides global security hooks, but a security
154 module is free to implement per-process hooks (where that makes sense)
155 by storing a security_ops table in each process' security field and
156 then invoking these per-process hooks from the global hooks.
157 The problem of composition is thus deferred to the module.
161 The global security_ops table is initialized to a set of hook
162 functions provided by a dummy security module that provides
163 traditional superuser logic. A <function>register_security</function>
164 function (in <filename>security/security.c</filename>) is provided to
165 allow a security module to set security_ops to refer to its own hook
166 functions, and an <function>unregister_security</function> function is
167 provided to revert security_ops to the dummy module hooks. This
168 mechanism is used to set the primary security module, which is
169 responsible for making the final decision for each hook.
173 LSM also provides a simple mechanism for stacking additional security
174 modules with the primary security module. It defines
175 <function>register_security</function> and
176 <function>unregister_security</function> hooks in the
177 <structname>security_operations</structname> structure and provides
178 <function>mod_reg_security</function> and
179 <function>mod_unreg_security</function> functions that invoke these
180 hooks after performing some sanity checking. A security module can
181 call these functions in order to stack with other modules. However,
182 the actual details of how this stacking is handled are deferred to the
183 module, which can implement these hooks in any way it wishes
184 (including always returning an error if it does not wish to support
185 stacking). In this manner, LSM again defers the problem of
186 composition to the module.
190 Although the LSM hooks are organized into substructures based on
191 kernel object, all of the hooks can be viewed as falling into two
192 major categories: hooks that are used to manage the security fields
193 and hooks that are used to perform access control. Examples of the
194 first category of hooks include the
195 <function>alloc_security</function> and
196 <function>free_security</function> hooks defined for each kernel data
197 structure that has a security field. These hooks are used to allocate
198 and free security structures for kernel objects. The first category
199 of hooks also includes hooks that set information in the security
200 field after allocation, such as the <function>post_lookup</function>
201 hook in <structname>struct inode_security_ops</structname>. This hook
202 is used to set security information for inodes after successful lookup
203 operations. An example of the second category of hooks is the
204 <function>permission</function> hook in
205 <structname>struct inode_security_ops</structname>. This hook checks
206 permission when accessing an inode.
211 <sect1 id="cap"><title>LSM Capabilities Module</title>
214 The LSM kernel patch moves most of the existing POSIX.1e capabilities
215 logic into an optional security module stored in the file
216 <filename>security/capability.c</filename>. This change allows
217 users who do not want to use capabilities to omit this code entirely
218 from their kernel, instead using the dummy module for traditional
219 superuser logic or any other module that they desire. This change
220 also allows the developers of the capabilities logic to maintain and
221 enhance their code more freely, without needing to integrate patches
222 back into the base kernel.
226 In addition to moving the capabilities logic, the LSM kernel patch
227 could move the capability-related fields from the kernel data
228 structures into the new security fields managed by the security
229 modules. However, at present, the LSM kernel patch leaves the
230 capability fields in the kernel data structures. In his original
231 remarks, Linus suggested that this might be preferable so that other
232 security modules can be easily stacked with the capabilities module
233 without needing to chain multiple security structures on the security field.
234 It also avoids imposing extra overhead on the capabilities module
235 to manage the security fields. However, the LSM framework could
236 certainly support such a move if it is determined to be desirable,
237 with only a few additional changes described below.
241 At present, the capabilities logic for computing process capabilities
242 on <function>execve</function> and <function>set*uid</function>,
243 checking capabilities for a particular process, saving and checking
244 capabilities for netlink messages, and handling the
245 <function>capget</function> and <function>capset</function> system
246 calls have been moved into the capabilities module. There are still a
247 few locations in the base kernel where capability-related fields are
248 directly examined or modified, but the current version of the LSM
249 patch does allow a security module to completely replace the
250 assignment and testing of capabilities. These few locations would
251 need to be changed if the capability-related fields were moved into
252 the security field. The following is a list of known locations that
253 still perform such direct examination or modification of
254 capability-related fields:
256 <listitem><para><filename>fs/open.c</filename>:<function>sys_access</function></para></listitem>
257 <listitem><para><filename>fs/lockd/host.c</filename>:<function>nlm_bind_host</function></para></listitem>
258 <listitem><para><filename>fs/nfsd/auth.c</filename>:<function>nfsd_setuser</function></para></listitem>
259 <listitem><para><filename>fs/proc/array.c</filename>:<function>task_cap</function></para></listitem>
projects / linux-2.6 / blob