so the kernel can send d-bus messages

[Robert Love]

Yo,

FYI.  I finally proposed the Kernel Events Layer patch at last week's
Kernel Summit.  If you were at OLS, I also gave a talk on it.

I submitted the patch to Mr. Andrew Morton yesterday.  The discussion is
ongoing on lkml.  If anyone has something to say, please chime in.

The patch is at

	http://www.kernel.org/pub/linux/kernel/people/rml/events/v2.6/kernel-events-rml-2.6.8-rc2-5.patch

What exactly is this?  It let's the kernel send asynchronous events to
user-space via a netlink socket (read: simple and fast).  The events
resemble D-BUS signals (they have an object name, signal name, and
payload) but are agnostic to whatever user-space does and policy free.

My idea, though, is to hook this in to D-BUS so as to allow the kernel
to effectively send D-BUS signals.  I, personally, find this really
fucking cool.

Mr. Kay Sievers has a lot of test code and examples available at

	http://vrfy.org/projects/kdbusd/

kdbusd.c is a simple daemon to push the kernel events onto the system
message bus.  This can also be done directly in the D-BUS daemon (my
personal preference).

The idea is to get the kernel in on the D-BUS fun, furthering the
unification of the Linux system.  We have plenty of uses for it in
Project Utopia.  Driver error logging folks should find this useful,
too.

Much love and little war,

	Robert Love