Pixelflut

Fast pixelflut server written in C. It is a collaborative coding game. See https://cccgoe.de/wiki/Pixelflut for details about the game itself. In short: project the pixelflut server output onto a wall where many people can see it. Connected clients can then set single pixels by sending a string like "PX [x] [y] [color]\n" (e.g. "PX 100 300 00FF42\n") to its TCP socket. Use netcat, python or whatever you want.

Hardware requirements

Every cpu with a little bit of power (for 2D SDL) should work. On an Core i3-4010U you can easily utilize a 1 GBit Nic. On Raspberry Pi 4B you get around 30 megabytes/sek. On large events, 10 GBit fiber and a few more CPU-Cores are even more fun. On real server hardware you want to add a graphics card. One thread per CPU-Core seems to be a good rule of thumb.

Features

• Multithreaded
• Can display an overlay with some statistics
• Webinterface serves real-time WebGL histogram and help text (same TCP port)
• Optional fade to black for old pixels to encourage pixel refreshes
• Take manual and periodic screenshots
• Supported commands:
  • send pixel: 'PX {x} {y} {GG or RRGGBB or RRGGBBAA as HEX}\n'
  • set offset for future pixels: 'OFFSET {x} {y}\n'
  • request pixel color: 'PX {x} {y}\n'
  • request output resolution: 'SIZE\n'
  • request client connection count: 'CONNECTIONS\n'
  • request help message with all commands: 'HELP\n'

Build

On a clean Debian installation with the "SSH server" and "standard system utilities" selected during setup. A system with desktop should also work.

apt update
apt install xorg git build-essential pkg-config libsdl2-dev libpng-dev -y
git clone https://github.com/larsmm/pixelflut.git
cd pixelflut
make

Keys

• q or ctrl+c: quit
• f: toggle fullscreen
• s: take screenshot (png file in pixelflut dir)

Options

./pixelflut --help

usage: ./pixelflut [OPTION]...
options:
        --width <pixels>                Framebuffer width. Default: Screen width.
        --height <pixels>               Framebuffer width. Default: Screen height.
        --port <port>                   TCP port. Default: 1234.
        --connection_timeout <seconds>  Connection timeout on idle. Default: 5s.
        --connections_max <n>           Maximum number of open connections. [1-60000] Default: 1000.
        --threads <n>                   Number of connection handler threads. Default: 4.
        --no-histogram                  Disable calculating and serving the histogram over HTTP.
        --window                        Start in window mode.
        --fade_out                      Enable fading out the framebuffer contents.
        --fade_interval <frames>        Interval for fading out the framebuffer as number of displayed frames. Default: 4.
        --hide_text                     Hide the overlay text.
        --show_ip_instead_of_hostname   Show IPv4 of interface with default-gateway on overlay.
        --show_custom_ip <IP>           Show specific IP instead of hostname.
        --screenshot_interval <seconds> Time between screenshots. (png files in pixelflut dir) Default: disabled.
        --screenshot_use_bmp            Use bmp instead of png because of speed.

Start Server locally

If you are using linux without a desktop environment, start x server first:

startx &  # start in background

Start pixelflut server:

./pixelflut

Start Server over ssh

If you are using linux with a desktop environment it has to run on the same user as you are using to connect over ssh. List availible displays:

ps e | grep -Po " DISPLAY=[\.0-9A-Za-z:]* " | sort -u

Start pixelflut on local display (x forwarding), e.g.:

DISPLAY=localhost:10.0 ./pixelflut --window

Start pixelflut on server display, e.g.:

DISPLAY=:0.0 ./pixelflut

Connection limit

Best practise: set overall limit of the pixelflut-server high (--connections_max 1000) and limit max connections to the pixelflut port (default: 1234) per IP via iptables to 10-20:

nano iptables.save

paste (set limit in --connlimit-above):

*filter
:INPUT ACCEPT [0:0]
:FORWARD ACCEPT [0:0]
:OUTPUT ACCEPT [0:0]
-A INPUT -p tcp -m tcp --dport 1234 --tcp-flags FIN,SYN,RST,ACK SYN -m connlimit --connlimit-above 20 --connlimit-mask 32 --connlimit-saddr -j REJECT
COMMIT

Ctrl+x, y, return to save. Activate:

iptables-restore < iptables.save

Display driver

Sometimes the free NVidia driver has problems on multiple displays. So install the proprietary driver:

1. detect the chip and find the right driver:

nano /etc/sources/sources.list

Add: "contrib non-free" after each source

apt install nvidia-detect
nvidia-detect

2. install driver:

apt install for example: nvidia-legacy-340xx-driver
reboot

3. configure your displays:

startx

run the following inside the x-session and setup your displays:

nvidia-settings

or via SSH:

DISPLAY=:0 nvidia-settings

maybe restart x server

Prevent standby

If you are using a notebook and want to close the lit. To disable all standby stuff:

systemctl mask sleep.target suspend.target hibernate.target hybrid-sleep.target

TODO

• Use epoll() to check multiple sockets for I/O events at once
• better network-statistics
• ipv6