[3283] in linux-net channel archive
Re: Binary Driver Issues
daemon@ATHENA.MIT.EDU (Mike Kilburn)
Tue Jun 18 06:01:34 1996
To: "Leonard N. Zubkoff" <lnz@dandelion.com>
cc: linux-net@vger.rutgers.edu
In-reply-to: Your message of "Sun, 16 Jun 1996 14:21:36 MST."
<199606162121.OAA00382@dandelion.com>
Date: Tue, 18 Jun 1996 07:11:21 +0200
From: Mike Kilburn <mike@lserv.conexio.co.za>
Binary only drivers need to somehow be classified into those which are
true hardware drivers and those which are "value-added" software extentions
to the Linux kernel which in my view violate the GPL. I know this is tough
and can be subjective and I really dont know how it could be done. Take
the following example:
If a company were to add say frame-relay with bandwith management to the
kernel networking code and then sell these Linux based routers without
releasing the source this would be a violation of the GPL. So instead
the company includes this with the hardware driver module and since
binary only modules are OK, they get away with it.
I strongly belive companies which use Linux for a profit should give
something back, this is one of the reasons for the GPL - prevent free software
rape. In the example above the company has been able to use the kernel to
save lots of money on RAM and ROM. This is not so bad but when they sell
it as a router product in competetion with, say, Cisco they have also gained
all the routing code without spending a dime on development. They have
effectively 'stole' the software. This company would probaly turn around and
say they dont release there source so people wont steal it - how ironic.
The GPL should prevent this but the binary only driver hook in the kernel is a
loophole for companies like this. Now this is different than a binary only
driver which only contains code to control the hardware is supposes to - no
value-added features to the kernel.
