1 b9220ca4 2021-01-11 op # gmid
  2 3e4749f7 2020-10-02 op 
  3 b9231167 2021-02-08 op gmid is a fast Gemini server written with security in mind.  I
  4 b9231167 2021-02-08 op initially wrote it to serve static files, but it has grown into a
  5 747b35d0 2021-02-08 op featureful server.
  6 3e4749f7 2020-10-02 op 
  7 3e4749f7 2020-10-02 op 
  8 b9220ca4 2021-01-11 op ## Features
  9 3e4749f7 2020-10-02 op 
 10 601bc1cc 2021-01-30 op (random order)
 11 601bc1cc 2021-01-30 op 
 12 747b35d0 2021-02-08 op  - reconfiguration: reload the running configuration without
 13 747b35d0 2021-02-08 op    interruption
 14 86edc4f4 2021-02-04 op  - sandboxed by default on OpenBSD, Linux and FreeBSD
 15 4d2ec6d7 2021-01-13 op  - IRI support (RFC3987)
 16 286c4f40 2021-01-27 op  - punycode support
 17 601bc1cc 2021-01-30 op  - dual stack (IPv4 and IPv6)
 18 601bc1cc 2021-01-30 op  - automatic certificate generation for config-less mode
 19 b9220ca4 2021-01-11 op  - CGI scripts
 20 b9220ca4 2021-01-11 op  - (very) low memory footprint
 21 b9231167 2021-02-08 op  - event-based asynchronous I/O model
 22 b9220ca4 2021-01-11 op  - small codebase, easily hackable
 23 b9220ca4 2021-01-11 op  - virtual hosts
 24 286c4f40 2021-01-27 op  - per-location rules
 25 286c4f40 2021-01-27 op  - optional directory listings
 26 286c4f40 2021-01-27 op  - configurable mime types
 27 0b00962d 2021-01-25 op  - chroot support
 28 3e4749f7 2020-10-02 op 
 29 fab952e1 2020-10-03 op 
 30 286c4f40 2021-01-27 op ## Internationalisation (IRIs, UNICODE, punycode, all that stuff)
 31 286c4f40 2021-01-27 op 
 32 286c4f40 2021-01-27 op Even thought the current Gemini specification doesn't mention anything
 33 601bc1cc 2021-01-30 op in this regard, I do think these are important things and so I tried
 34 601bc1cc 2021-01-30 op to implement them in the most user-friendly way I could think of.
 35 286c4f40 2021-01-27 op 
 36 601bc1cc 2021-01-30 op For starters, gmid has full support for IRI (RFC3987 —
 37 22a742e4 2021-01-29 op Internationalized Resource Identifiers).  IRIs are a superset of URIs,
 38 286c4f40 2021-01-27 op so there aren't incompatibilities with URI-only clients.
 39 286c4f40 2021-01-27 op 
 40 601bc1cc 2021-01-30 op There is full support also for punycode.  In theory, the user doesn't
 41 286c4f40 2021-01-27 op even need to know that punycode is a thing.  The hostname in the
 42 601bc1cc 2021-01-30 op configuration file can (and must be) in the decoded form (e.g. `naïve`
 43 601bc1cc 2021-01-30 op and not `xn--nave-6pa`), gmid will do the rest.
 44 286c4f40 2021-01-27 op 
 45 601bc1cc 2021-01-30 op The only missing piece is UNICODE normalisation of the IRI path: gmid
 46 601bc1cc 2021-01-30 op doesn't do that (yet).
 47 286c4f40 2021-01-27 op 
 48 286c4f40 2021-01-27 op 
 49 1487e11e 2021-02-06 op ## Configuration
 50 1487e11e 2021-02-06 op 
 51 1487e11e 2021-02-06 op gmid has a rich configuration file, heavily inspired by OpenBSD'
 52 1487e11e 2021-02-06 op httpd.  While you should definitely check the manpage because it
 53 1487e11e 2021-02-06 op documents every option in depth, here's an example of what gmid can
 54 1487e11e 2021-02-06 op do.
 55 1487e11e 2021-02-06 op 
 56 1487e11e 2021-02-06 op ```conf
 57 1487e11e 2021-02-06 op ipv6 on     # enable ipv6
 58 1487e11e 2021-02-06 op 
 59 1487e11e 2021-02-06 op server "example.com" {
 60 1487e11e 2021-02-06 op     cert "/path/to/cert.pem"
 61 1487e11e 2021-02-06 op     key  "/path/to/key.pem"
 62 1487e11e 2021-02-06 op     root "/var/gemini/example.com"
 63 1487e11e 2021-02-06 op     lang "it"
 64 1487e11e 2021-02-06 op     cgi  "/cgi/*"
 65 1487e11e 2021-02-06 op 
 66 1487e11e 2021-02-06 op     location "/files/*" {
 67 1487e11e 2021-02-06 op         auto index on
 68 1487e11e 2021-02-06 op     }
 69 1487e11e 2021-02-06 op 
 70 1487e11e 2021-02-06 op     location "/repo/*" {
 71 1487e11e 2021-02-06 op         # change the index file name
 72 1487e11e 2021-02-06 op         index "README.gmi"
 73 1487e11e 2021-02-06 op     }
 74 1487e11e 2021-02-06 op 
 75 1487e11e 2021-02-06 op     # redirect /cgi/man/... to man.example.com/...
 76 1487e11e 2021-02-06 op     location "/cgi/man*" {
 77 1487e11e 2021-02-06 op         strip 2
 78 1487e11e 2021-02-06 op         block return 31 "gemini://man.example.com%p"
 79 1487e11e 2021-02-06 op     }
 80 1487e11e 2021-02-06 op }
 81 1487e11e 2021-02-06 op 
 82 1487e11e 2021-02-06 op server "man.example.com" {
 83 1487e11e 2021-02-06 op     cert "..."
 84 1487e11e 2021-02-06 op     key  "..."
 85 1487e11e 2021-02-06 op     root "/var/gemini/man.example.com"
 86 1487e11e 2021-02-06 op 
 87 1487e11e 2021-02-06 op     # handle every request with the CGI script `man'
 88 1487e11e 2021-02-06 op     entrypoint "man"
 89 1487e11e 2021-02-06 op }
 90 1487e11e 2021-02-06 op ```
 91 1487e11e 2021-02-06 op 
 92 1487e11e 2021-02-06 op 
 93 b9220ca4 2021-01-11 op ## Building
 94 3e4749f7 2020-10-02 op 
 95 b9231167 2021-02-08 op gmid depends on a POSIX libc, libevent2, OpenSSL/LibreSSL and libtls
 96 b9231167 2021-02-08 op (provided either by LibreSSL or libretls).  At build time, flex and
 97 b9231167 2021-02-08 op yacc (or GNU bison) are also needed.
 98 3e4749f7 2020-10-02 op 
 99 771d8f28 2021-01-17 op The build is as simple as
100 3e4749f7 2020-10-02 op 
101 601bc1cc 2021-01-30 op     ./configure
102 b9220ca4 2021-01-11 op     make
103 3e4749f7 2020-10-02 op 
104 8f0da580 2021-01-21 op If the configure scripts fails to pick up something, please open an
105 f980545b 2021-01-21 op issue or notify me via email.
106 8f0da580 2021-01-21 op 
107 771d8f28 2021-01-17 op To install execute:
108 881a9dd9 2021-01-16 op 
109 771d8f28 2021-01-17 op     make install
110 771d8f28 2021-01-17 op 
111 601bc1cc 2021-01-30 op ### Docker
112 771d8f28 2021-01-17 op 
113 601bc1cc 2021-01-30 op If you have trouble installing LibreSSL or libretls, you can use
114 601bc1cc 2021-01-30 op Docker to build a `gmid` image with:
115 601bc1cc 2021-01-30 op 
116 17b09e3c 2021-01-18 op     docker build -t gmid .
117 17b09e3c 2021-01-18 op 
118 17b09e3c 2021-01-18 op and then run it with something along the lines of
119 17b09e3c 2021-01-18 op 
120 17b09e3c 2021-01-18 op     docker run --rm -it -p 1965:1965 \
121 6c117838 2021-01-25 op         -v /path/to/gmid.conf:...:ro \
122 17b09e3c 2021-01-18 op         -v /path/to/docs:/var/gemini \
123 6c117838 2021-01-25 op         gmid -c .../gmid.conf
124 17b09e3c 2021-01-18 op 
125 601bc1cc 2021-01-30 op (ellipses used for brevity)
126 17b09e3c 2021-01-18 op 
127 dd8cc7d3 2021-01-22 op ### Local libretls
128 17b09e3c 2021-01-18 op 
129 8f0da580 2021-01-21 op This is **NOT** recommended, please try to port LibreSSL/LibreTLS to
130 8f0da580 2021-01-21 op your distribution of choice or use docker instead.
131 8f0da580 2021-01-21 op 
132 601bc1cc 2021-01-30 op However, it's possible to statically-link `gmid` to locally-installed
133 601bc1cc 2021-01-30 op libretls quite easily.  (It's how I test gmid on Fedora, for instance)
134 8f0da580 2021-01-21 op 
135 04397b32 2021-01-21 op Let's say you have compiled and installed libretls in `$LIBRETLS`,
136 8f0da580 2021-01-21 op then you can build `gmid` with
137 8f0da580 2021-01-21 op 
138 04397b32 2021-01-21 op     ./configure CFLAGS="-I$LIBRETLS/include" \
139 1606927e 2021-02-11 op                 LDFLAGS="$LIBRETLS/lib/libtls.a -lssl -lcrypto -lpthread -levent"
140 8f0da580 2021-01-21 op     make
141 8f0da580 2021-01-21 op 
142 dd8cc7d3 2021-01-22 op ### Testing
143 dd8cc7d3 2021-01-22 op 
144 31a4993a 2021-01-23 op Execute
145 dd8cc7d3 2021-01-22 op 
146 dd8cc7d3 2021-01-22 op     make regress
147 dd8cc7d3 2021-01-22 op 
148 601bc1cc 2021-01-30 op to start the suite.  Keep in mind that the regression tests will
149 601bc1cc 2021-01-30 op create files inside the `regress` directory and bind the 10965 port.
150 dd8cc7d3 2021-01-22 op 
151 dd8cc7d3 2021-01-22 op 
152 881a9dd9 2021-01-16 op ## Architecture/Security considerations
153 881a9dd9 2021-01-16 op 
154 881a9dd9 2021-01-16 op gmid is composed by two processes: a listener and an executor.  The
155 881a9dd9 2021-01-16 op listener process is the only one that needs internet access and is
156 881a9dd9 2021-01-16 op sandboxed.  When a CGI script needs to be executed, the executor
157 881a9dd9 2021-01-16 op (outside of the sandbox) sets up a pipe and gives one end to the
158 881a9dd9 2021-01-16 op listener, while the other is bound to the CGI script standard output.
159 909ea500 2021-01-27 op This way, is still possible to execute CGI scripts without
160 601bc1cc 2021-01-30 op restrictions even in the presence of a sandboxed network process.
161 881a9dd9 2021-01-16 op 
162 601bc1cc 2021-01-30 op On OpenBSD, the listener runs with the `stdio recvfd rpath inet`
163 601bc1cc 2021-01-30 op pledges, while the executor has `stdio sendfd proc exec`; both have
164 601bc1cc 2021-01-30 op unveiled only the served directories.
165 881a9dd9 2021-01-16 op 
166 ea58dab1 2021-01-17 op On FreeBSD, the executor process is sandboxed with `capsicum(4)`.
167 d939d0f0 2021-01-17 op 
168 601bc1cc 2021-01-30 op On Linux, a `seccomp(2)` filter is installed in the listener to allow
169 601bc1cc 2021-01-30 op only certain syscalls, see [sandbox.c](sandbox.c) for more information
170 601bc1cc 2021-01-30 op on the BPF program.
171 ea58dab1 2021-01-17 op 
172 ea58dab1 2021-01-17 op In any case, you are invited to run gmid inside some sort of
173 0b00962d 2021-01-25 op container/jail/chroot.