On Sat, Jan 04, 2014 at 11:36:13PM +0100, David Härdeman wrote: >Add socket activation and lazy opening of /dev/uinput which makes >acpi_fakekeyd both more robust and helps speed up boot (by avoiding >sleeping in the init script). > >An added benefit of the systemd socket activation is that the daemon >is never actually started on systems (like my laptop) where the "extra" >keys don't actually generate ACPI events. >--- > debian/acpi-fakekey.init | 18 ++--- > debian/acpi-fakekey.install | 3 + > debian/acpi-fakekey.modules.conf | 3 + > debian/acpi-fakekey.service | 7 ++ > debian/acpi-fakekey.socket | 9 ++ > debian/control | 3 + > debian/patches/acpi_fakekey.diff | 148 +++++++++++++++++++++++--------------- > debian/rules | 2 - > 8 files changed, 121 insertions(+), 72 deletions(-) > create mode 100644 debian/acpi-fakekey.modules.conf
After a second look, I realize that this file (acpi-fakekey.modules.conf which was installed to /etc/modules-load.d/) is actually not necessary. udev in unstable already has the functionality to pre-create dev nodes such as /dev/uinput which, when opened, will trigger the kernel module autoloading functionality, triggering the modprobe of uinput. That means the module doesn't need to be explicitly loaded in the systemd *or* initscript scenario (unless the autoloading functionality is broken of course). I'd be happy to provide an updated patch of the acpi-support maintainer so wishes. //David -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
