guek / autorandr Goto Github PK

*** This guide applies for Linux using the LXDE windows manager ***

Automatic monitor switch is a very useful utility when you often move your laptop. At home you have one or more external monitors, at the office you have another one or you code while commuting from A to B.

However, on my LXDE installation, apparently there's not a tool that automatically provides such a feature, so switching can be accomplished the usual Linux way: a bit of scripting and a lot of googling. Namely you have to intercept the UDEV event, if the graphic driver in the kernel supports your chipset.

I wanted a solution that does not sit on the background polling on kernel events. I wanted something triggered by kernel events, that's what UDEV is there for.

# for xrandr
$ sudo apt-get install x11-xserver-utils

Check whether the kernel is aware of the plug/unplug events. Tail the kernel events with udevadm while you plug/unplug the external monitor:

$ udevadm monitor --property

if you see something like the following, you're good to go:

KERNEL[28.029974] change   /devices/pci0000:00/0000:00:02.0/drm/card0 (drm)
ACTION=change
DEVNAME=/dev/dri/card0
DEVPATH=/devices/pci0000:00/0000:00:02.0/drm/card0
DEVTYPE=drm_mino
HOTPLUG=1
MAJOR=226
MINOR=0
SEQNUM=2959
SUBSYSTEM=drm

UDEV  [28.038149] change   /devices/pci0000:00/0000:00:02.0/drm/card0 (drm)
ACTION=change
DEVNAME=/dev/dri/card0
DEVPATH=/devices/pci0000:00/0000:00:02.0/drm/card0
DEVTYPE=drm_minor
HOTPLUG=1
ID_FOR_SEAT=drm-pci-0000_00_02_0
ID_PATH=pci-0000:00:02.0
ID_PATH_TAG=pci-0000_00_02_0
MAJOR=226
MINOR=0
SEQNUM=2959
SUBSYSTEM=drm
TAGS=:seat:uaccess:master-of-seat:
USEC_INITIALIZED=4743165

The useful info here are:

• action=change: type of action triggered
• devname=/dev/dri/card0: name of the graphic card device
• devpath=/devices/pci0000:00/...: path of the graphic card device

Write down the DEVPATH and DEVNAME from the previous output, that's where UDEV sees your graphic cards. To double check the path/name of your device you can query it with:

$ udevadm info -n /dev/dri/card0
$ udevadm info -q path -n /dev/dri/card0

Create a new UDEV rule /etc/udev/95-monitor-switch.rules, replace user with your Linux user:

ACTION=="change", KERNEL=="card0", SUBSYSTEM=="drm", ENV{DISPLAY}=":0", ENV{XAUTHORITY}="/home/user/.Xauthority", RUN+="/path/to/autorandr.sh"

This rule can probably be refined and YMMV, but it works for me (:copyright: 2016).

Test the syntax of your rule executing a dry-run on your graphic card path:

udevadm test /devices/pci0000:00/0000:00:02.0/drm/card0 2>&1 | less

You should see your new rule appearing and no errors.

You need to create two xrandr commands: one for the single monitor and one with the extended desktop on the external monitor. In order to do so, use arandr and note down the configuration commands created.

Purely as an example, here's mines; I have a Dell XPS13 laptop with a FullHD screen (the eDPI1 primary device) and a Dell U2414 external monitor (the DPI secondary device):

• Switch to single monitor: /usr/bin/xrandr --output DP1 --off --output eDP1 --mode 1920x1080 --pos 0x0 --rotate normal
• Switch to double monitor: /usr/bin/xrandr --output DP1 --mode 1920x1080 --pos 0x0 --output eDP1 --primary --mode 1920x1080 --pos 0x1080 (notice how DPI1 becomes the primary monitor and eDPI1 is stacked under DPI1).

Check how many connected monitors you have with xrandr --listmonitors after running alternatively these two commands. You will notice the monitor count flips from 1 to 2 and viceversa.

Also you can check their status by peeking at the status attribute of the kernel device. In my case these two devices translates to the following paths:

$ cat /sys/class/drm/card0-DP-1/status
$ cat /sys/class/drm/card0-eDP-1/status

Note down these paths, we'll use them later.

Put the xrandr commands in the if-then in the /path/to/autorandr.sh file you chose in the UDEV rule. Set the EXTERNAL_MONITOR_STATUS variable to read whether the external monitor is connected and execute the proper xrandr command.

#!/bin/sh

set -e

# Is the external monitor connected?
EXTERNAL_MONITOR_STATUS=$( /sys/class/drm/card0-DP-1/status )
if [ $EXTERNAL_MONITOR_STATUS == "connected" ]; then
    TYPE="double"
    /usr/bin/xrandr ... (command for single+external monitors)
else
    TYPE="single"
    /usr/bin/xrandr ... (command for single monitor)
fi

# log to syslog (not necessary)
logger -t autorandr "Switched to $TYPE monitor mode"

# show a popup
/usr/bin/notify-send --urgency=low -t 500 "Switched to $TYPE monitor configuration"

exit 0

Test the bash script with the external monitor plugged or unplugged. You should see the monitor switch happening.

Of course things might not be so simple. For example I've encountered an issue running the xrandr command to attach the external monitor: the monitor did not show up with the available resolutions, therefore the xrandr command failed.

With some more google-fu, turns out the solution is to set Option "HotPlug" "false" in your Xorg config file, wherever it is located (usually in /usr/share/X11/xorg.conf.d/*). If you don't have a config file with a Device directive, you need to manually create it.

For example a minimal /usr/share/X11/xorg.conf.d/20-intel.conf did the trick for me (I have an Intel i915 chipset):

Section "Device"
    Identifier "Intel Graphics"
    Driver "intel"
    Option "HotPlug" "false"
    EndSection

Restart X (or reboot) for this change to take effect. However after this change Xorg error reports started popping out upon startup, so I must have overriden some default settings of Xorg; well... what the hell, I'll fix that another time.

I've also experienced that xrandr is sometimes slow to detect the new monitor resolutions (although the command shows them all correctly). As a workaround you can have xrandr poll the device configuration and exit when the switch is done.

I've put a sample script autorandr.sh as a customizable template. Device ids are mines, YMMV.

Linux on desktop sucks. Why did I have to spend an evening copying and pasting stuff from the Interwebz? Why did I have to figure out how things worked until I managed to hack together a BASH script triggered by a kernel event? OSX users have this utility out of the box.

You can of course customize this script based on your configurations, this is just a succint guide on the steps to follow.

If after reading this guide you think that I just wasted my time because there's a tool that does this automagically, please let me know, I'll be super-happy to try it! :-)

• The thread on HN that urged me to itch this scratch
• The always GREAT Arch wiki
• This thread on StackExchange
• This outdated but still useful guide for Ubuntu (the UDEV rule they describe didn't work for me, but it was a start on what I needed)
• man udevadm