Bug#492140: xserver-xorg: xorg.conf generated with no ServerLayout section
Hi Carl,
On Thu, Jul 24, 2008 at 07:49:04 -0700, Carl Worth wrote:
> But I *think* that with no ServerLayout section the server adds default
> input devices rather than hunting them out in the config file, (but I
> could be wrong about that part).
>
when AllowEmptyInput is off, the server calls checkCoreInputDevices(),
which does:
* Locate the core input devices. These can be specified/located in
* the following ways, in order of priority:
*
* 1. The InputDevices named by the -pointer and -keyboard command line
* options.
* 2. The "CorePointer" and "CoreKeyboard" InputDevices referred to by
* the active ServerLayout.
* 3. The first InputDevices marked as "CorePointer" and "CoreKeyboard".
* 4. The first InputDevices that use the 'mouse' and 'keyboard' or 'kbd'
* drivers.
* 5. Default devices with an empty (default) configuration. These defaults
* will reference the 'mouse' and 'keyboard' drivers.
The xorg.conf generated by the xserver-xorg package relies on 4 above,
which breaks when AllowEmptyInput is turned on.
I think the first four points should still be tried when no ServerLayout
section is present, even with AllowEmptyInput.
> Meanwhile, I agree that it's a bug that the server doesn't respect the
> InputDevice sections without a ServerLayout section, (it does respect a
> Screen section without a ServerLayout section). I even looked a bit at
> xf86Config.c to see how easy that might be to fix.
>
Yeah, I looked too, and will try to fix that upstream, time permitting.
> But regardless, that's not fixed now, and the configuration file that
> "dpkg-reconfigure xserver-xorg" generates results in a non-functional
> setup with xserver >= 1.5. So that's something that should be fixed
> still. There are no downsides to actually generating the ServerLayout
> section that the server wants, right?
>
It makes xorg.conf longer :)
Seriously though, it probably doesn't hurt, so I'll try to make that
change soonish.
Cheers,
Julien
Reply to: