Re: Binary-only firmware covered by the GPL?
From: Chris Cheney
Date: Thu Mar 25 2004 - 22:02:35 EST
On Thu, Mar 25, 2004 at 06:06:37PM -0800, David Schwartz wrote:
> > At Fri, 26 Mar 2004 00:33:39 +0000,
> > Matthew Wilcox wrote:
> > > I realise there's a grey area between "magic data you write to a device"
> > > and "a program that is executed on a different processor". For example,
> > > palette data for a frame buffer. But nobody's arguing for that grey
> > > area here -- it's clearly a program without source code that Debian
> > > can't distribute.
> > Well, I also think this is grey area.
> On what basis? How is this file different from an executable?
> The gray area cases are where the code, as orginally written, is obscure.
> Perhaps because it uses 'magic numbers' from data sheets rather than
> symbolic constants. However, the GPL doesn't require you to add comments or
> to write clear code. It simply prohibits deliberate obfuscation by one
> particular means, namely having two forms of the code, one that you
> distribute and one that you use to make modifications. (Other forms of
> obfuscation are the gray areas.)
> But this case is squarely where the GPL says "no". In this case, there is
> one form of the firmware for the purposes of making modifications to it and
> there is another form that's distributed, ostensibly under the GPL.
So is a reverse engineered driver where there is a binary blob the
preferred form of source? Otherwise the case isn't that binary blobs are
against the GPL, just that if the author knows what generates the binary
blob and doesn't disclose it then it is against the GPL. Of course
reverse engineered drivers with binary blobs in them are probably
copyright infringements anyway...
Description: Digital signature