about summary refs log tree commit diff
path: root/doc/overview.html
blob: 6c58a37a246ad0190a0b837c2d64d6fc2b607658 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
<html>
  <head>
    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
    <meta http-equiv="Content-Language" content="en" />
    <title>utmps: an overview</title>
    <meta name="Description" content="utmps: an overview" />
    <meta name="Keywords" content="utmps overview utmp wtmp utmpx login user accounting unix" />
    <!-- <link rel="stylesheet" type="text/css" href="//skarnet.org/default.css" /> -->
  </head>
<body>

<p>
<a href="index.html">utmps</a><br />
<a href="//skarnet.org/software/">Software</a><br />
<a href="//skarnet.org/">skarnet.org</a>
</p>

<h1> An overview of utmps </h1>

<p>
 utmps is a secure implementation of the <em>utmp</em> functionality, i.e.
user accounting on Unix systems. It includes full POSIX
<a href="http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/utmpx.h.html">utmpx.h</a>
functionality, a few extensions created by GNU, and an underlying
<a href="libutmps/">C client library</a> with better error reporting
than the POSIX interface specifies.
</p>

<h2> The issues with traditional utmp </h2>

<p>
 Traditional <em>utmp</em> implementations, as performed by most Unix
libcs, are woefully insecure. The fundamental issue with <em>utmp</em>
is that it requires user programs to write to files (the utmp or wtmp
databases) owned by either root or a specific system user. That means
having the suid bit set on programs using it.
</p>

<h2> The utmps solution </h2>

<p>
 utmps uses the age-old Unix client-server model, following the
adage "one resource &rarr; one daemon". It provides two daemons,
<a href="utmps-utmpd.html">utmps-utmpd</a> and
<a href="utmps-wtmpd.html">utmps-wtmpd</a>, which should be the only
programs allowed to access the utmp and wtmp databases respectively.
It provides the
<a href="libutmps/">utmps client library</a> to communicate with
those daemons; and it implements the
<a href="http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/utmpx.h.html">utmpx.h</a>
interfaces, and the extensions, as wrappers for this client library.
</p>

<h2> Authentication, local services and superservers </h2>

<p>
<a href="utmps-utmpd.html">utmps-utmpd</a> and
<a href="utmps-wtmpd.html">utmps-wtmpd</a> do not listen to the
network themselves. They are designed to serve only one client
connection, following the
<a href="http://cr.yp.to/proto/ucspi.txt">UCSPI model</a> -
also known as the inetd model. To implement the utmpd and wtmpd
<a href="//skarnet.org/software/s6/localservice.html">local services</a>,
a Unix domain superserver such as
<a href="//skarnet.org/software/s6/s6-ipcserver.html">s6-ipcserver</a>
is required. s6-ipcserver listens to a socket, and spawns a
<a href="utmps-utmpd.html">utmps-utmpd</a> process when a client
calls <a href="http://pubs.opengroup.org/onlinepubs/9699919799/functions/setutxent.html">setutxent()</a>
for instance.
</p>

<p>
 The utmpd and wtmpd services must be started at boot time in
order for utmp calls to succeed. The <tt>examples/</tt> subdirectory
of the utmps package has examples on how to start those services
when using the
<a href="//skarnet.org/software/s6/">s6</a> supervision suite,
the <a href="//skarnet.org/software/s6-rc/">s6-rc</a> service manager, or
the <a href="https://wiki.gentoo.org/wiki/OpenRC">OpenRC</a> service manager.
</p>

<p>
 This model has advantages and drawbacks. The main drawback is that it requires
a daemon to be running in order for the system to provide full POSIX
functionality. The main advantage, on the other hand, is that no program
needs to be suid or sgid, and permissions can actually be quite fine-grained.
</p>

<ul>
 <li> <a href="utmps-utmpd.html">utmps-utmpd</a> will allow any user to
read from the utmp database, but will only allow root, or members of the
same group utmps-utmpd runs as, to write to it. </li>
 <li> <a href="utmps-wtmpd.html">utmps-wtmpd</a> will only allow a user
to add an entry to the wtmp database if the user is root, or if the
<tt>ut_user</tt> field of the added entry resolves to the user's effective
uid. </li>
  <li> The <a href="//skarnet.org/software/s6/s6-ipcserver.html">s6-ipcserver</a>
superserver, which is recommended to implement the utmpd and wtmpd services,
allows fine-tuning the permissions: it is possible to deny users from
connecting to the service, or to only allow certain groups, etc. </li>
</ul>

<p>
 All in all, I believe the flexibility it offers overweighs the inconvenience
of having to run services before providing utmp/wtmp.
</p>

</body>
</html>