Re: hardware initialization

On Wed, Nov 23, 2005 at 11:42:58AM +0100, Andreas V. Meier wrote:
> On Tue, Nov 22, 2005 at 06:32:34PM +0100, Benjamin Schieder wrote:
> > Andreas V. Meier wrote:
> > > Hi,
> > > 
> > > I have thought a bit about our scheme of hardware initialization, and suggest
> > > the follwoing changes to enhance our udev support:
>  
> > Can you elaborate a bit more on how hardware detection will work then? Will
> > udev send hotplug events to the hotplug scripts or how is that going to work?
> > Mind you, I'm no expert in that field :)
> I am neither, and unfortunately my research was interrupted by that conference
> call. But it should go the following:
> 
> First: Devices already present:
>  - Through the call of all hotplug scripts from hotplug.init the present
>  hardware is evaluated and respective drivers are loaded.
>  - Through kernel.init which calls /etc/conf/kernel the custom modprobe commands
>  are executed.

Ah, okay, so hotplug still loads the modules even if udevsend is in
proc/kernel/hotplug.

> Second: plugging devices:
>  - kernel calls udevsend, which forwards the event to udevd
>  - udevd reorders the events in appropriate manner and forwards them to hotplug
>    the reordering is used to prevent race conditions
>  - hotplug loads the respective drivers in accordance to the received events. 
>  - udevd also calls hal, which can be used with dbus and ivman/pmount to
>  automount plugged devices or just to execute arbitrary scripts for plug events
>  or other hardware changes.

Does this require any udev rules or is this handled by udev automatically?

> This already works nicely on my notebook. I decided to use the recent versions
> for hal and dbus, because I had some issues with the old versions, which were
> required for kde 3.4. But the upcoming kde 3.5 has support for those recent
> versions, so until we have decided on this topic we will probably also have kde
> 3.5 then ;)

I sure hope we could get this over with before christmas :)

Greetings,
	Benjamin

-- 
Today, memory either forgets things when you don't want it to,
or remembers things long after they're better forgotten.

_______________________________________________
rock-devel mailing list
[email protected]
http://www.rocklinux.net/mailman/listinfo/rock-devel

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)

iD8DBQFDhF+zr0OTeImXvg8RAhM+AJ9LiraKJIctNvakMKXtwPbwOhUJZwCfT3uB
ynb9SbtGEY4WZRzHyKXfj4I=
=PYBN
-----END PGP SIGNATURE-----