Comment 10 for bug 175743

I have backported the (temporary?) fix from bluez CVS to Ubuntu's bluez-utils 3.26-0ubuntu6 package that makes the IO timeout for input devices configurable via a conf file, including the default option of having no IO timeout. Attached is the diff of the package source.

I have tested it (using Gnome) and having no IO timeout for input devices appears to have solved the problem for me. However, I do not know what long-term effects, if any, this will have on things like mouse battery life, etc. Removal of IO timeouts for input devices may have other effects that I have not forseen. (I don't really understand why there was an IO timeout for input devices to begin with...) Also, I use a desktop computer, so I don't know what effects, if any, it has for laptop and other battery-powered computers.

Assuming you know how to build Debian/Ubuntu packages from source, you should only have to get the build-dep and source for the bluez-utils package and use the following commands:

patch -p0 <bluez-utils-3.26__input__.diff

The patch should apply with no errors or warnings. Then you need to:

cd bluez-utils-3.26
aclocal
automake

All should then be ready to build the package.

To experiment with the IO timeout, copy input/input.conf to /etc/bluetooth/input.conf and change the IdleTimeout parameter to however many minutes you want. No conf file, or a commented-out setting, means no IO timeout for input devices.

Based on the link in comment 9, it seems that bluez developers might work to find a better long-term solution that correctly disconnects all input devices when they timeout. However, this patch is here right now, and it seems to fix this problem without breaking anything else.