Siduction Forum

Siduction Forum => Experimental / Experimentelles => Experimental (EN) => Topic started by: devil on 2015/08/22, 11:33:00

Title: Test Kdbus
Post by: devil on 2015/08/22, 11:33:00
For all users that due to the current state of Sid are depressive and close to jumping off something high, here is a little treat:

The (hopefuly) upcoming kdbus can now be tested in Sid and Wily Werewolf. Martin Pitt has written a dkms-module ( to make this as easy as possible. I have written a little news item ( about this, it is in German, but shows the neccessary steps. Have fun.

Title: Re: Test Kdbus
Post by: ralul on 2015/08/23, 01:03:24
3 weeks ago Kde-4 did not run using kdbus. I did not look up why. That same kernel for my Gentoo kde-4 works without issues since then. Although this kernel is not modular I could run it with Debian by adding kdbus=0 to the grub kernel command line.
Using kdbus with the Gentoo partition is a no brainer since systemD recognizes this facility in the kernel and a handful of dbus1 proxy processes are started, which are waiting for traditional dbus1 calls then. Look  at a few:
Code: [Select]
8976 systemd-b   1   0 1469M  3152  2368 S  0.0  0.1  0:00.03 /usr/lib/systemd/systemd-bus-proxyd --address=kernel:path=/sys/fs/kdbus/0-system/bus
 9009 systemd-b   1   0 1469M  3152  2368 S  0.0  0.1  0:00.13 /usr/lib/systemd/systemd-bus-proxyd --address=kernel:path=/sys/fs/kdbus/0-system/bus
  255 systemd-b   1   0 1469M  3152  2368 S  0.0  0.1  0:00.94 /usr/lib/systemd/systemd-bus-proxyd --address=kernel:path=/sys/fs/kdbus/0-system/bus
  936 ral         1   0 1749M  3456  2248 S  0.0  0.1  0:00.21 /usr/lib/systemd/systemd-bus-proxyd --address=kernel:path=/sys/fs/kdbus/1000-user/bus
  939 ral         1   0 1749M  3456  2248 S  0.0  0.1  0:00.35 /usr/lib/systemd/systemd-bus-proxyd --address=kernel:path=/sys/fs/kdbus/1000-user/bus
  988 ral         1   0 1749M  3456  2248 S  0.0  0.1  0:00.00 /usr/lib/systemd/systemd-bus-proxyd --address=kernel:path=/sys/fs/kdbus/1000-user/bus

You will not feel any difference. Which does not wonder since dbus1 - as a slow compangion - is used mostly to communicate short messages. If you look at Gnome-3, there probably is more to expect, because it is far more cut into pieces than kde-4.

The problem the kernel guys discuss most these days is about what to expect from kernel dbus code if there is a user space program with ugly intentions. I don't think traditional dbus1 code ever was secure in this case, but never was kernel code running at pid1!

For me the interesting about kdbus is:  systemD could get rid of all the hidden special services which are run automatically at the beginning of the boot process. That way I could experiment with the very first boot sequences ...