From mboxrd@z Thu Jan  1 00:00:00 1970
From: Lee Jones <lee.jones@canonical.com>
Date: Tue, 06 Jul 2010 11:28:51 +0000
Subject: Re: UDEV - DRIVER parameter
Message-Id: <4C331373.1080702@canonical.com>
List-Id: <linux-hotplug.vger.kernel.org>
References: <4C33057C.10805@canonical.com>
In-Reply-To: <4C33057C.10805@canonical.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: linux-hotplug@vger.kernel.org

> Only some devices have a driver bound. Only "bus" devices, "class"
> devices never have a driver.
> 
> It's just the symlink "driver" in /sys in the device directory.
> 
> Your platform device may just be a plain registered device, and not
> probed and bound by a driver in that sense. Usually only enumeratable
> buses have drivers that create devices when they are discovered.

I guess I'd better tell you the full story then.

I am working on a driver (module) that provides a direct link "CPU <-> Video Codec", which I believe is non-discoverable. 

I figured that I'd be able to send the uevent by registering the device as a platform one in architecture specific code.

A uevent is sent, but it is missing the DRIVER parameter.

This is the generic parameter which is used in the udev rules.

/lib/udev/rules.d/80-drivers.rules, DRIVER!="?*", ENV{MODALIAS}="?*", RUN+="/sbin/modprobe -b $env{MODALIAS}"

I want to shy away from sending another rule upstream.

Are there any other sensible alternatives? 

Kind regards,
Lee