Synaptics MIT license approved

Matthew Garrett mjg59 at
Thu May 17 05:14:04 PDT 2007

On Thu, May 17, 2007 at 04:46:23AM -0400, James Cloos wrote:

> The FSF holds that linking includes loading dynamic modules, as the X
> server does with drivers.  If you accept that definition, that any
> vendor who ships the X server with the synaptics driver is bound by the
> GPL for the server as well as the driver, by virtue of shipping the two
> products linked together.  This means that no non-GPL-compatible
> drivers could also be included.  Or visa-versa; if they include a
> proprietary X driver they cannot also include a GPL X driver.

As Alan noted, the issue is derivation rather than linking. For this to 
actually be a problem, we would need to hold that the X server is a 
derived work of the GPLed Synaptics driver (thereby falling under the 
GPL), and that the shipped proprietary X driver is a derived work of the 
GPLed X server rather than the byte-for-byte identical MIT/X11 one 
without the Synaptics driver. This clearly isn't true at the source 
level, so the issue is whether it's true when distributing binaries. 

I think there's a fairly strong case to be made that there's no 
derivation there whatsoever, unless you're talking about the runtime 
case - but the GPL covers distribution and not use. Note that this is 
different to the Linux/nvidia case, as it's fairly easy to make a claim 
that the nvidia module is a derived work of the GPLed kernel.

But really, this is only a problem if you think that someone can argue 
that the X server is a derived work of the Synaptics driver, even though 
the Synaptics driver is built out of tree. That argument would imply 
that shipping a GPLed binary alongside an LGPLed library means that the 
library is now under the GPL, and therefore you couldn't also ship a 
proprietary binary. The result would be that libc is effectively GPLed 
(not LGPLed) and that no Linux distribution could include any 
proprietary userspace applications. I think this is clearly an 
unreasoanble interpretation.

Matthew Garrett | mjg59 at

More information about the xorg mailing list