doc/mwg/Linux-PAM_MWG.xml - master - githubmirror/linux-pam

doc/mwg/Linux-PAM_MWG.xml
604 строки27 KB
Dmitry V. Levin
Prepare for 1.7.2 release
21 янв 2026, 11:00
21 янв 2026, 11:006247444
<book xmlns="http://docbook.org/ns/docbook" version="5.0" xml:id="mwg">
  <info>
    <title>The Linux-PAM Module Writers' Guide</title>
    <authorgroup>
      <author><personname><firstname>Andrew G.</firstname><surname>Morgan</surname></personname><email>morgan@kernel.org</email></author>
      <author><personname><firstname>Thorsten</firstname><surname>Kukuk</surname></personname><email>kukuk@thkukuk.de</email></author>
    </authorgroup>
    <releaseinfo>Version 1.7.2, January 2026</releaseinfo>
    <abstract>
      <para>
        This manual documents what a programmer needs to know in order
        to write a module that conforms to the
        <emphasis remap="B">Linux-PAM</emphasis> standard.It also
        discusses some security issues from the point of view of the
        module programmer.
      </para>
    </abstract>
  </info>

  <chapter xml:id="mwg-introduction">
    <title>Introduction</title>
    <section xml:id="mwg-introduction-description">
      <title>Description</title>
      <para>
        <emphasis remap="B">Linux-PAM</emphasis> (Pluggable Authentication
        Modules for Linux) is a library that enables the local system
        administrator to choose how individual applications authenticate
        users. For an overview of the
        <emphasis remap="B">Linux-PAM</emphasis> library see the
        <emphasis>Linux-PAM System Administrators' Guide</emphasis>.
      </para>
      <para>
        A <emphasis remap="B">Linux-PAM</emphasis> module is a single
        executable binary file that can be loaded by the
        <emphasis remap="B">Linux-PAM</emphasis> interface library.
        This PAM library is configured locally with a system file,
        <filename>/etc/pam.conf</filename>, to authenticate a user
        request via the locally available authentication modules. The
        modules themselves will usually be located in the directory
        <filename>/lib/security</filename> (or
        <filename>/lib64/security</filename>, depending on the architecture)
        and take the form of dynamically loadable object files (see
        <citerefentry>
          <refentrytitle>dlopen</refentrytitle><manvolnum>3</manvolnum>
        </citerefentry>. Alternatively, the modules can be statically
        linked into the <emphasis remap="B">Linux-PAM</emphasis> library;
        this is mostly to allow <emphasis remap="B">Linux-PAM</emphasis> to
        be used on platforms without dynamic linking available, but this is
        a <emphasis>deprecated</emphasis> functionality. It is the
        <emphasis remap="B">Linux-PAM</emphasis> interface that is called
        by an application and it is the responsibility of the library to
        locate, load and call the appropriate functions in a
        <emphasis remap="B">Linux-PAM</emphasis>-module.
      </para>
      <para>
        Except for the immediate purpose of interacting with the user
        (entering a password etc..) the module should never call the
        application directly. This exception requires a "conversation
        mechanism" which is documented below.
      </para>
    </section>

    <section xml:id="mwg-introduction-synopsis">
      <title>Synopsis</title>
      <programlisting>
#include &lt;security/pam_modules.h&gt;

gcc -fPIC -c pam_module.c
gcc -shared -o pam_module.so pam_module.o -lpam
      </programlisting>
    </section>
  </chapter>

  <chapter xml:id="mwg-expected-by-module">
    <title>What can be expected by the module</title>
    <para>
      Here we list the interface that the conventions that all
      <emphasis remap="B">Linux-PAM</emphasis> modules must adhere to.
    </para>
    <section xml:id="mwg-expected-by-module-item">
      <title>
        Getting and setting <emphasis>PAM_ITEM</emphasis>s and
        <emphasis>data</emphasis>
      </title>
      <para>
        First, we cover what the module should expect from the
        <emphasis remap="B">Linux-PAM</emphasis> library and a
        <emphasis remap="B">Linux-PAM</emphasis> aware application.
        Essentially this is the <filename>libpam.*</filename> library.
      </para>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_set_data.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_get_data.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_set_item.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_get_item.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_get_user.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_conv.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_putenv.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_getenv.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_getenvlist.xml"/>
    </section>
    <section xml:id="mwg-expected-by-module-other">
      <title>
        Other functions provided by <filename>libpam</filename>
      </title>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_strerror.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_fail_delay.xml"/>
    </section>
  </chapter>

  <chapter xml:id="mwg-expected-of-module">
    <title>What is expected of a module</title>
    <para>
      The module must supply a sub-set of the six functions listed
      below. Together they define the function of a
      <emphasis remap="B">Linux-PAM module</emphasis>. Module developers
      are strongly urged to read the comments on security that follow
      this list.
    </para>
    <section xml:id="mwg-expected-of-module-overview">
      <title>Overview</title>
      <para>
        The six module functions are grouped into four independent
        management groups. These groups are as follows:
        <emphasis>authentication</emphasis>, <emphasis>account</emphasis>,
        <emphasis>session</emphasis> and <emphasis>password</emphasis>.
        To be properly defined, a module must define all functions within
        at least one of these groups. A single module may contain the
        necessary functions for <emphasis>all</emphasis> four groups.
      </para>
      <section xml:id="mwg-expected-of-module-overview-1">
        <title>Functional independence</title>
        <para>
          The independence of the four groups of service a module can
          offer means that the module should allow for the possibility
          that any one of these four services may legitimately be called
          in any order. Thus, the module writer should consider the
          appropriateness of performing a service without the prior
          success of some other part of the module.
        </para>
        <para>
          As an informative example, consider the possibility that an
          application applies to change a user's authentication token,
          without having first requested that
          <emphasis remap="B">Linux-PAM</emphasis> authenticate the
          user. In some cases this may be deemed appropriate: when
          <command>root</command> wants to change the authentication
          token of some lesser user. In other cases it may not be
          appropriate: when <command>joe</command> maliciously wants
          to reset <command>alice</command>'s password; or when anyone
          other than the user themself wishes to reset their
          <emphasis>KERBEROS</emphasis> authentication token. A policy
          for this action should be defined by any reasonable
          authentication scheme, the module writer should consider
          this when implementing a given module.
        </para>
      </section>
      <section xml:id="mwg-expected-of-module-overview-2">
        <title>Minimizing administration problems</title>
        <para>
          To avoid system administration problems and the poor
          construction of a <filename>/etc/pam.conf</filename> file,
          the module developer may define all six of the following
          functions. For those functions that would not be called,
          the module should return <errorname>PAM_SERVICE_ERR</errorname>
          and write an appropriate message to the system log. When
          this action is deemed inappropriate, the function would
          simply return <errorname>PAM_IGNORE</errorname>.
        </para>
      </section>
      <section xml:id="mwg-expected-of-module-overview-3">
        <title>Arguments supplied to the module</title>
        <para>
          The <parameter>flags</parameter> argument of each of
          the following functions can be logically OR'd with
          <parameter>PAM_SILENT</parameter>, which is used to inform the
          module to not pass any <emphasis>text</emphasis> (errors or
          warnings) application.
        </para>
        <para>
          The <parameter>argc</parameter> and <parameter>argv</parameter>
          arguments are taken from the line appropriate to this
          module---that is, with the <emphasis>service_name</emphasis>
          matching that of the application---in the configuration file
          (see the <emphasis remap="B">Linux-PAM</emphasis>
          System Administrators' Guide). Together these two parameters
          provide the number of arguments and an array of pointers to
          the individual argument tokens. This will be familiar to C
          programmers as the ubiquitous method of passing command arguments
          to the function <function>main()</function>. Note, however, that
          the first argument (<parameter>argv[0]</parameter>) is a true
          argument and <emphasis>not</emphasis> the name of the module.
        </para>
      </section>
    </section>
    <section xml:id="mwg-expected-of-module-auth">
      <title>Authentication management</title>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_sm_authenticate.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_sm_setcred.xml"/>
    </section>
    <section xml:id="mwg-expected-of-module-acct">
      <title>Account management</title>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_sm_acct_mgmt.xml"/>
    </section>
    <section xml:id="mwg-expected-of-module-session">
      <title>Session management</title>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_sm_open_session.xml"/>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_sm_close_session.xml"/>
    </section>
    <section xml:id="mwg-expected-of-module-chauthtok">
      <title>Authentication token management</title>
      <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="pam_sm_chauthtok.xml"/>
    </section>
  </chapter>

  <chapter xml:id="mwg-see-options">
    <title>Generic optional arguments</title>
    <para>
      Here we list the generic arguments that all modules can expect to
      be passed. They are not mandatory, and their absence should be
      accepted without comment by the module.
    </para>
    <variablelist>
      <varlistentry>
        <term>debug</term>
        <listitem>
           <para>
             Use the <citerefentry>
             <refentrytitle>pam_syslog</refentrytitle><manvolnum>3</manvolnum>
             </citerefentry> call to log debugging information to the system
             log files.
          </para>
        </listitem>
      </varlistentry>
      <varlistentry>
        <term>use_first_pass</term>
        <listitem>
           <para>
             The module should not prompt the user for a password.
             Instead, it should obtain the previously typed password
             (by a call to <function>pam_get_item()</function> for the
             <parameter>PAM_AUTHTOK</parameter> item), and use that. If
             that doesn't work, then the user will not be authenticated.
             (This option is intended for <command>auth</command> and
             <command>passwd</command> modules only).
          </para>
        </listitem>
      </varlistentry>
    </variablelist>
  </chapter>

  <chapter xml:id="mwg-see-programming">
    <title>Programming notes</title>
    <para>
      Here we collect some pointers for the module writer to bear in mind
      when writing/developing a <emphasis remap="B">Linux-PAM</emphasis>
      compatible module.
    </para>

    <section xml:id="mwg-see-programming-sec">
      <title>Security issues for module creation</title>
      <section xml:id="mwg-see-programming-sec-res">
        <title>Sufficient resources</title>
        <para>
          Care should be taken to ensure that the proper execution
          of a module is not compromised by a lack of system resources.
          If a module is unable to open sufficient files to perform its
          task, it should fail gracefully, or request additional resources.
          Specifically, the quantities manipulated by the <citerefentry>
          <refentrytitle>setrlimit</refentrytitle><manvolnum>2</manvolnum>
          </citerefentry> family of commands should be taken into
          consideration.
        </para>
      </section>
      <section xml:id="mwg-see-programming-sec-who">
        <title>Who´s who?</title>
        <para>
          Generally, the module may wish to establish the identity of
          the user requesting a service. This may not be the same as
          the username returned by <function>pam_get_user()</function>.
          Indeed, that is only going to be the name of the user under
          whose identity the service will be given. This is not
          necessarily the user that requests the service.
        </para>
        <para>
          In other words, user X runs a program that is setuid-Y, it
          grants the user to have the permissions of Z. A specific example
          of this sort of service request is the <command>su</command>
          program: user <command>joe</command> executes
          <command>su</command> to become the user <command>jane</command>.
          In this situation X=<command>joe</command>, Y=<command>root</command>
          and Z=<command>jane</command>. Clearly, it is important that
          the module does not confuse these different users and grant an
          inappropriate level of privilege.
        </para>
        <para>
          The following is the convention to be adhered to when juggling
          user-identities.
        </para>
        <itemizedlist>
          <listitem>
            <para>
              X, the identity of the user invoking the service request.
              This is the user identifier; returned by the function
              <citerefentry>
                <refentrytitle>getuid</refentrytitle><manvolnum>2</manvolnum>
              </citerefentry>.
            </para>
          </listitem>
          <listitem>
            <para>
              Y, the privileged identity of the application used to
              grant the requested service. This is the
              <emphasis>effective</emphasis> user identifier;
              returned by the function <citerefentry>
              <refentrytitle>geteuid</refentrytitle><manvolnum>2</manvolnum>
              </citerefentry>.
            </para>
          </listitem>
          <listitem>
            <para>
              Z, the user under whose identity the service will be granted.
              This is the username returned by
              <function>pam_get_user()</function> and also stored in the
              <emphasis remap="B">Linux-PAM</emphasis> item,
              <emphasis>PAM_USER</emphasis>.
            </para>
          </listitem>
          <listitem>
            <para>
              <emphasis remap="B">Linux-PAM</emphasis> has a place for
              an additional user identity that a module may care to make
              use of. This is the <emphasis>PAM_RUSER</emphasis> item.
              Generally, network sensitive modules/applications may wish
              to set/read this item to establish the identity of the user
              requesting a service from a remote location.
            </para>
          </listitem>
        </itemizedlist>
        <para>
          Note, if a module wishes to modify the identity of either the
          <emphasis>uid</emphasis> or <emphasis>euid</emphasis> of the
          running process, it should take care to restore the original
          values prior to returning control to the
          <emphasis remap="B">Linux-PAM</emphasis> library.
        </para>
      </section>
      <section xml:id="mwg-see-programming-sec-conv">
        <title>Using the conversation function</title>
        <para>
          Prior to calling the conversation function, the module should
          reset the contents of the pointer that will return the applications
          response. This is a good idea since the application may fail
          to fill the pointer and the module should be in a position to
          notice!
        </para>
        <para>
          The module should be prepared for a failure from the
          conversation. The generic error would be
          <emphasis>PAM_CONV_ERR</emphasis>, but anything other than
          <emphasis>PAM_SUCCESS</emphasis> should be treated as
          indicating failure.
        </para>
      </section>
      <section xml:id="mwg-see-programming-sec-token">
        <title>Authentication tokens</title>
        <para>
          To ensure that the authentication tokens are not left lying
          around the items, <emphasis>PAM_AUTHTOK</emphasis> and
          <emphasis>PAM_OLDAUTHTOK</emphasis>, are not available to
          the application: they are defined in
          <filename>&lt;security/pam_modules.h&gt;</filename>. This
          is ostensibly for security reasons, but a maliciously
          programmed application will always have access to all memory
          of the process, so it is only superficially enforced. As a
          general rule the module should overwrite authentication tokens
          as soon as they are no longer needed. Especially before
          <function>free()</function>'ing them. The
          <emphasis remap="B">Linux-PAM</emphasis> library is
          required to do this when either of these authentication
          token items are (re)set.
        </para>
        <para>
          Not to dwell too little on this concern; should the module
          store the authentication tokens either as (automatic) function
          variables or using <function>pam_[gs]et_data()</function> the
          associated memory should be over-written explicitly before it
          is released. In the case of the latter storage mechanism, the
          associated <function>cleanup()</function> function should
          explicitly overwrite the <varname>*data</varname> before
          <function>free()</function>'ing it: for example,
          <programlisting>
/*
 * An example cleanup() function for releasing memory that was used to
 * store a password.
 */

int cleanup(pam_handle_t *pamh, void *data, int error_status)
{
    char *xx;

    if ((xx = data)) {
        while (*xx)
            *xx++ = '\0';
        free(data);
    }
    return PAM_SUCCESS;
}
          </programlisting>
        </para>
      </section>
    </section>
    <section xml:id="mwg-see-programming-syslog">
      <title>Use of <citerefentry>
        <refentrytitle>syslog</refentrytitle><manvolnum>3</manvolnum>
      </citerefentry></title>
      <para>
        Only rarely should error information be directed to the user.
        Usually, this is to be limited to
        <quote><emphasis>sorry you cannot login now</emphasis></quote>
        type messages. Information concerning errors in the configuration
        file, <filename>/etc/pam.conf</filename>, or due to some system
        failure encountered by the module, should be written to
        <citerefentry>
        <refentrytitle>syslog</refentrytitle><manvolnum>3</manvolnum>
        </citerefentry> with <emphasis>facility-type</emphasis>
        <emphasis remap="B">LOG_AUTHPRIV</emphasis>.
      </para>
      <para>
        With a few exceptions, the level of logging is, at the discretion
        of the module developer. Here is the recommended usage of different
        logging levels:
      </para>
      <itemizedlist>
        <listitem>
          <para>
            As a general rule, errors encountered by a module should be
            logged at the <emphasis>LOG_ERR</emphasis> level. However,
            information regarding an unrecognized argument, passed to a
            module from an entry in the <filename>/etc/pam.conf</filename>
            file, is <emphasis>required</emphasis> to be logged at the
            <emphasis>LOG_ERR</emphasis> level.
          </para>
        </listitem>
        <listitem>
          <para>
            Debugging information, as activated by the
            <command>debug</command> argument to the module in
            <filename>/etc/pam.conf</filename>, should be logged
            at the <emphasis>LOG_DEBUG</emphasis> level.
          </para>
        </listitem>
        <listitem>
          <para>
            If a module discovers that its personal configuration
            file or some system file it uses for information is
            corrupted or somehow unusable, it should indicate this
            by logging messages at level, <emphasis>LOG_ALERT</emphasis>.
          </para>
        </listitem>
        <listitem>
          <para>
            Shortages of system resources, such as a failure to
            manipulate a file or <function>malloc()</function> failures
            should be logged at level <emphasis>LOG_CRIT</emphasis>.
          </para>
        </listitem>
        <listitem>
          <para>
            Authentication failures, associated with an incorrectly
            typed password should be logged at level,
            <emphasis>LOG_NOTICE</emphasis>.
          </para>
        </listitem>
      </itemizedlist>
    </section>
    <section xml:id="mwg-see-programming-libs">
      <title>Modules that require system libraries</title>
      <para>
        Writing a module is much like writing an application. You
        have to provide the "conventional hooks" for it to work
        correctly, like <function>pam_sm_authenticate()</function>
        etc., which would correspond to the <function>main()</function>
        function in a normal function.
      </para>
      <para>
        Typically, the author may want to link against some standard system
        libraries. As when one compiles a normal program, this can be
        done for modules too: you simply append the
        <parameter>-l</parameter><emphasis>XXX</emphasis> arguments
        for the desired libraries when you create the shared module object.
        To make sure a module is linked to the
        <command>libwhatever.so</command> library
        when it is <function>dlopen()</function>ed, try:
        <programlisting>
% gcc -shared -o pam_module.so pam_module.o -lwhatever
        </programlisting>
      </para>
    </section>
  </chapter>

  <chapter xml:id="mwg-example">
    <title>An example module</title>
    <para>
      At some point, we may include a fully commented example of a module in
      this document. For now, please look at the modules directory of the
      <emphasis remap="B">Linux-PAM</emphasis> sources.
    </para>
  </chapter>

  <chapter xml:id="mwg-see-also">
    <title>See also</title>
    <itemizedlist>
      <listitem>
        <para>
          The Linux-PAM System Administrators' Guide.
        </para>
      </listitem>
      <listitem>
        <para>
          The Linux-PAM Application Developers' Guide.
        </para>
      </listitem>
      <listitem>
        <para>
          The V. Samar and R. Schemers (SunSoft), ``UNIFIED LOGIN WITH
          PLUGGABLE AUTHENTICATION MODULES'', Open Software Foundation
          Request For Comments 86.0, October 1995.
        </para>
      </listitem>
    </itemizedlist>
  </chapter>

  <chapter xml:id="mwg-author">
    <title>Author/acknowledgments</title>
    <para>
      This document was written by Andrew G. Morgan (morgan@kernel.org)
      with many contributions from
      Chris Adams, Peter Allgeyer, Tim Baverstock, Tim Berger, Craig S. Bell,
      Derrick J. Brashear, Ben Buxton, Seth Chaiklin, Oliver Crow, Chris Dent,
      Marc Ewing, Cristian Gafton, Emmanuel Galanos, Brad M. Garcia,
      Eric Hester, Roger Hu, Eric Jacksch, Michael K. Johnson, David Kinchlea,
      Olaf Kirch, Marcin Korzonek, Thorsten Kukuk, Stephen Langasek,
      Nicolai Langfeldt, Elliot Lee, Luke Kenneth Casson Leighton,
      Al Longyear, Ingo Luetkebohle, Marek Michalkiewicz, Robert Milkowski,
      Aleph One, Martin Pool, Sean Reifschneider, Jan Rekorajski, Erik Troan,
      Theodore Ts'o, Jeff Uphoff, Myles Uyema, Savochkin Andrey Vladimirovich,
      Ronald Wahl, David Wood, John Wilmes, Joseph S. D. Yao
      and Alex O. Yuriev.
    </para>
    <para>
      Thanks are also due to Sun Microsystems, especially to Vipin Samar and
      Charlie Lai for their advice. At an early stage in the development of
      <emphasis remap="B">Linux-PAM</emphasis>, Sun graciously made the
      documentation for their implementation of PAM available. This act
      greatly accelerated the development of
      <emphasis remap="B">Linux-PAM</emphasis>.
    </para>
  </chapter>

  <chapter xml:id="mwg-copyright">
    <title>Copyright information for this document</title>
    <programlisting>
Copyright (c) 2006 Thorsten Kukuk &lt;kukuk@thkukuk.de&gt;
Copyright (c) 1996-2002 Andrew G. Morgan &lt;morgan@kernel.org&gt;
    </programlisting>
    <para>
      Redistribution and use in source and binary forms, with or without
      modification, are permitted provided that the following conditions are
      met:
    </para>
    <programlisting>
1. Redistributions of source code must retain the above copyright
   notice, and the entire permission notice in its entirety,
   including the disclaimer of warranties.

2. Redistributions in binary form must reproduce the above copyright
   notice, this list of conditions and the following disclaimer in the
   documentation and/or other materials provided with the distribution.

3. The name of the author may not be used to endorse or promote
   products derived from this software without specific prior
   written permission.
    </programlisting>
    <para>
      Alternatively, this product may be distributed under the terms of
      the GNU General Public License (GPL), in which case the provisions
      of the GNU GPL are required instead of the above restrictions.
      (This clause is necessary due to a potential bad interaction between
      the GNU GPL and the restrictions contained in a BSD-style copyright.)
    </para>
    <programlisting>
THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR
TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
    </programlisting>
  </chapter>
</book>