From mboxrd@z Thu Jan 1 00:00:00 1970 From: Marc Dietrich Subject: Re: autobuild of tegrarcm Date: Mon, 28 Jul 2014 10:03:39 +0200 Message-ID: <15635619.1sXO2FqnMG@fb07-iapwap2> References: <18236773.n1AHpLtjGA@fb07-iapwap2.physik.uni-giessen.de> <20140727182320.GA14656@simplex.0x539.de> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: <20140727182320.GA14656-yKPz972ugOjHWlwuStCS9A@public.gmane.org> Sender: linux-tegra-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Philipp Kern Cc: nonfree-sMDJvqjagnvx+JS5GvM4/R2eb7JE58TQ@public.gmane.org, Julian Andres Klode , dktrkranz-8fiUuRrzOP0dnm+yROfE0A@public.gmane.org, ftpmaster-8fiUuRrzOP0dnm+yROfE0A@public.gmane.org, 'Stephen Warren' , linux-tegra-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Allen Martin List-Id: linux-tegra@vger.kernel.org added tegra ml, tegra maintainer, and tegrarcm maintainer. I think the license isn't appropriete for this software at all. See my = notes=20 regarding this here [1]. So lets take the chance and ping nvidia again = on this=20 issue. [1] http://www.spinics.net/lists/linux-tegra/msg11945.html=20 Am Sonntag, 27. Juli 2014, 20:23:20 schrieb Philipp Kern: > Hi, >=20 > On Sun, May 12, 2013 at 09:55:20PM +0200, Marc Dietrich wrote: > > On Sun, 12 May 2013, Philipp Kern wrote: > > >Hi, > > > > > >On Tue, Apr 23, 2013 at 11:31:21AM +0200, Marc Dietrich wrote: > > >>The proprietary license only applies to the binaries included in = some > > >>header> > > > >>files. FYI, here is the license: > > >it seems like an oversight: LICENSE says "The files tegra20-minilo= ader.h > > >and tegra30-miniloader.h are provided pursuant to the following li= cense > > >agreement". But there is also tegra114-miniloader.h, which bears t= he > > >same "not usable without a specific agreement" header. > >=20 > > thanks for finding this. I think this can be fixed easily. I'll ask= my > > NVIDIA contact to update the LICENSE file regarding this. > >=20 > > >Interestingly NVIDIA tells us in the license that it is "revocable= ", > > >which > > >is not the case for the graphics driver one. I presume we're alrea= dy > > >talking about the binary code form here. It does not really make m= e happy > > >that we can only distribute this to sublicensees that agree to be = bound > > >by the license and to owners of NVIDIA hardware. Both is untrue wh= en > >=20 > > >looking at buildds. ): > > Does the term "revocable" cause problems with the distribution? Doe= s this > > mean that users downloading the code (or the binary) must delete it= if > > NVIDIA tells debian to do so? Of course this is not possbile becaus= e we > > don't know the identity of the users downloading this code. Also we= don't > > ask the users to aggree with the license before downloading/running= the > > program. > >=20 > > The code in the miniloader files is not run on the machine which ru= ns > > tegrarcm. It is run on the target SoC, which is likely Tegra, becau= se all > > other SoCs won't run this bootloader. So I think this is more or le= ss > > harmless. >=20 > whenever I go back to this license, I feel bad about it. Adding Luca = who > accepted it. >=20 > The graphic driver license clearly exempts Linux from the problematic > provisions. The tegra one does not even try. Here's 2.1: >=20 > 2.1 Rights and Limitations of Grant. NVIDIA hereby grants to You t= he > following non-exclusive, non-transferable, revocable right under > NVIDIA=E2=80=99s copyrights to use, copy, distribute and sublicense = the > SOFTWARE (solely in binary code form) to Your sublicensees (a) solel= y > for use in connection with NVIDIA hardware or software products; and > (b) pursuant to the terms and conditions of any form of end-user > software license agreement; provided, that Your sublicensees agree t= o > be bound by this LICENSE or terms and conditions that are as > protective of NVIDIA=E2=80=99s Intellectual Property Rights in the S= OFTWARE as > this LICENSE >=20 > I'm unsure if the latter part means that the users need to consent be= fore we > can offer it (e.g. click-through). Unless we do not actually sublicen= se. > OTOH if we'd sublicense, then the users would not need to destroy cop= ies > when we need to. >=20 > And then there's clause 3: >=20 > 3. TERM AND TERMINATION > . > This LICENSE and the licenses granted hereunder shall be effective a= s > of the date You download the applicable SOFTWARE (=E2=80=9CEffective= Date=E2=80=9D) > and continue for a period of one (1) year (=E2=80=9CInitial Term=E2=80= =9D) > respectively, unless terminated earlier in accordance with the > =E2=80=9CTermination=E2=80=9D provision of this LICENSE. Unless eit= her party notifies > the other party of its intent to terminate this LICENSE at least thr= ee > (3) months prior to the end of the Initial Term or the applicable > renewal period, this License will be automatically renewed for one (= 1) > year renewal periods thereafter, unless terminated in accordance wit= h > the =E2=80=9CTermination=E2=80=9D provision of this LICENSE > . > NVIDIA may terminate this LICENSE at any time if You violate its > terms. Upon termination, You will immediately destroy the SOFTWARE o= r > return all copies of the SOFTWARE to NVIDIA, and certify to NVIDIA i= n > writing that such actions have been completed. Upon termination or > expiration of this LICENSE the license grants to Licensee shall > terminate, except that sublicenses rightfully granted by Licensee > under this LICENSE in connection with Paragraph 2 of this LICENSE > provided by Licensee prior to the termination or expiration of this > LICENSE shall survive in accordance with their respective form of > license terms and conditions. > [...] >=20 > Do we have a precedent for such a clause? The whole language about on= e year > terms and announcements of termination worries me. That'd mean that > someone'd need to monitor NVIDIA's announcements and then request rem= oval > from a stable release in time? Or will that only happen towards singl= e > users? (I.e. must the notification be direct?) >=20 > I guess the language about NVIDIA's products can be ok, given that th= e > compiled product is only useful on NVIDIA hardware anyway and will no= t be > executed on buildds=E2=80=A6 >=20 > Kind regards > Philipp Kern