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
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
|
# gmid
**the `master` branch is WIP: it's what gmid 2.0 will be, with breaking
changes et al. Please use the latest release from the 1.8 branch for a
stable and documented experience, thank you.**
gmid is a fast Gemini server written with security in mind. I
initially wrote it to serve static files, but it has grown into a
featureful server.
## Features
(random order)
- sandboxed by default on OpenBSD, Linux and FreeBSD
- reconfiguration: reload the running configuration without
interruption
- automatic redirect/error pages (see `block return`)
- IRI support (RFC3987)
- automatic certificate generation for config-less mode
- reverse proxying
- FastCGI support
- virtual hosts
- location rules
- event-based asynchronous I/O model
- low memory footprint
- small codebase, easily hackable
## Internationalisation (IRIs, UNICODE, punycode, all that stuff)
Even thought the current Gemini specification doesn't mention anything
in this regard, I do think these are important things and so I tried
to implement them in the most user-friendly way I could think of.
For starters, gmid has full support for IRI (RFC3987 —
Internationalized Resource Identifiers). IRIs are a superset of URIs,
so there aren't incompatibilities with URI-only clients.
There is full support also for punycode. In theory, the user doesn't
even need to know that punycode is a thing. The hostname in the
configuration file can (and must be) in the decoded form (e.g. `naïve`
and not `xn--nave-6pa`), gmid will do the rest.
The only missing piece is UNICODE normalisation of the IRI path: gmid
doesn't do that (yet).
## Configuration
gmid has a rich configuration file, heavily inspired by OpenBSD'
httpd, with every detail carefully documented in the manpage. Here's
a minimal example of a config file:
```conf
server "example.com" {
cert "/path/to/cert.pem"
key "/path/to/key.pem"
root "/var/gemini/example.com"
}
```
and a slightly more complex one
```conf
ipv6 on # enable ipv6
# define a macro
cert_root = "/path/to/keys"
server "example.com" {
alias "foobar.com"
cert $cert_root "/example.com.crt"
key $cert_root "/example.com.pem"
root "/var/gemini/example.com"
# lang for text/gemini files
lang "en"
# only for locations that matches /files/*
location "/files/*" {
# generate directory listings
auto index on
}
location "/repo/*" {
# change the index file name
index "README.gmi"
lang "it"
}
}
```
## Building
gmid depends on libevent2, OpenSSL/LibreSSL and libtls (provided
either by LibreSSL or libretls). At build time, yacc (or GNU bison)
is also needed.
The build is as simple as
./configure
make
or `make static` to build a statically-linked executable.
If the configure scripts fails to pick up something, please open an
issue or notify me via email.
To install execute:
make install
Please keep in mind that the master branch, from time to time, may be
accidentally broken on some platforms. gmid is developed primarily on
OpenBSD/amd64 and commits on the master branch don't get always tested
in other OSes. Before tagging a release however, a comprehensive
testing on various platform is done to ensure that everything is
working as intended.
### Testing
Execute
make regress
to start the suite. Keep in mind that the regression tests needs to
create files inside the `regress` directory and bind the 10965 port.
## Contributing
Any form of contribution is welcome, not only patches or bug reports.
If you have a sample configuration for some specific use-case, a
script or anything that could be useful to others, consider adding it
to the `contrib` directory.
## Architecture/Security considerations
**outdated: revisit for gmid 2.0**
gmid is composed by four processes: the parent process, the logger,
the listener and the executor. The parent process is the only one
that doesn't drop privileges, but all it does is to wait for a SIGHUP
to reload the configuration and spawn a new generation of children
process. The logger process gathers the logs and prints 'em to
stderr or syslog (for the time being.) The listener process is the
only one that needs internet access and is sandboxed by default. The
executor process exists only to fork and execute CGI scripts, and
optionally to connect to FastCGI applications.
On OpenBSD the processes are all `pledge(2)`d and `unveil(2)`ed.
On FreeBSD, the listener and logger process are sandboxed with `capsicum(4)`.
On Linux, a `seccomp(2)` filter is installed in the listener to allow
only certain syscalls, see [sandbox.c](sandbox.c) for more information
about the BPF program. If available, landlock is used to limit the
portion of the file system gmid can access (requires linux 5.13+.)
In any case, it's advisable to run gmid inside some sort of
container/jail/chroot.
|