Return-path: Received: from shadbolt.e.decadent.org.uk ([88.96.1.126]:39343 "EHLO shadbolt.e.decadent.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750738AbcEYOvZ (ORCPT ); Wed, 25 May 2016 10:51:25 -0400 Message-ID: <1464187874.2762.18.camel@decadent.org.uk> (sfid-20160525_165200_281145_B1BC1A93) Subject: Re: [RESEND] Re: updating carl9170-1.fw in linux-firmware.git From: Ben Hutchings To: Christian Lamparter , Kalle Valo Cc: Lauri Kasanen , linux-wireless@vger.kernel.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org, Linux Firmware Maintainers , Xose Vazquez Perez Date: Wed, 25 May 2016 15:51:14 +0100 In-Reply-To: <2185358.G319l0XIJC@debian64> References: <20160416113312.8de2266358165182bb4b4a15@gmx.com> <11418613.XH7h8m2RjK@debian64> <87a8koiubz.fsf@kamboji.qca.qualcomm.com> <2185358.G319l0XIJC@debian64> Content-Type: multipart/signed; micalg="pgp-sha512"; protocol="application/pgp-signature"; boundary="=-aVguvMqGs0TR7FTSzLqi" Mime-Version: 1.0 Sender: linux-wireless-owner@vger.kernel.org List-ID: --=-aVguvMqGs0TR7FTSzLqi Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Wed, 2016-04-20 at 23:11 +0200, Christian Lamparter wrote: > On Wednesday, April 20, 2016 10:59:44 AM Kalle Valo wrote: > > Christian Lamparter writes: > >=20 > > > On Monday, April 18, 2016 07:42:05 PM Kalle Valo wrote: > > > > Christian Lamparter writes: > > > >=20 > > > > > On Monday, April 18, 2016 06:45:09 PM Kalle Valo wrote: > > > > >=20 > > > > > > Why even mention anything about a "special firmware" as the fir= mware is > > > > > > already available from linux-firmware.git?=C2=A0 > > > > >=20 > > > > > Yes and no. 1.9.6 is in linux-firmware.git. I've tried to add 1.9= .9 too > > > > > but that failed. > > > > >=20 > > > >=20 > > > > Rick's comment makes sense to me, better just to provide the latest > > > > version. No need to unnecessary confuse the users. And if someone r= eally > > > > wants to use an older version that she can retrieve it from the git > > > > history. > > >=20 > > > Part of the fun here is that firmware is GPLv2. The linux-firmware.gi= t has > > > to point to or add the firmware source to their tree. They have added= every > > > single source file to it.... instead of "packaging" it in a tar.bz2/g= z/xz > > > like you normally do for release sources. > > >=20 > > > If you want to read more about it: > > > > >=20 > > Yeah, that's more work. I get that. But I'm still not understanding > > what's the actual problem which prevents us from updating carl9170 > > firmware in linux-firmware. > I'm not sure, but why not ask? I've added the cc'ed Linux Firmware > Maintainers. So for those people reading the fw list: >=20 > What would it take to update the carl9170-1.fw firmware file in your > repository to the latest version? >=20 > Who has to sent the firmware update. Does it have to be the person who > sent the first request? (Xose)? The maintainer of the firmware (me)? > someone from Qualcomm Atheros? Or someone else (specific)? (the=C2=A0 > firmware is licensed as GPLv2 - in theory anyone should be able to > do that) Given the licence, I don't particularly care. > How should the firmware source update be handled? Currently the latest > .tar.xz of the firmware has ~130kb. The formated patches from 1.9.6 to > latest are about ~100kb (182 individual patches). Either patches that 'git am' can handle, or a git branch. > How does linux-firmware handle new binary firmware images and new=C2=A0 > sources? What if carl9170fw-2.bin is added. Do we need another > source directory for this in the current tree then? Because=C2=A0 > carl9170fw-1.bin will still be needed for backwards compatibility > so we basically need to duplicate parts of the source? We still need to include the old binary for compatibility, and the old source for GPL compliance. =C2=A0(If there was a source version that could build firmware for both ABI versions, then we could update both binaries and have one set of source files. =C2=A0But it doesn't sound like that's the case.) > Also, how's the situation with ath9k_htc? The 1.4.0 image contains > some GPLv2 code as well? I didn't realise that. > So, why is there no source in the tree, but=C2=A0just the link to it? An oversight which we need to fix. > Because, I would like to do basically the same > for carl9170fw and just add a link to the carl9170fw repository and > save everyone this source update "song and dance". Merely linking to upstream source doesn't satisfy GPLv2 source requirements, at least not in case of commercial distribution. =C2=A0Linux distributors should be able to use a snapshot of linux-firmware as the upstream source for a package, without worrying about whether there are extra sources they need to include. (I'm aware that there are several files that don't actually have clear licences for. =C2=A0But those are at least called out in WHENCE, and were previously distributed as part of the kernel sources for years.) Ben. --=20 Ben Hutchings Time is nature's way of making sure that everything doesn't happen at once. --=-aVguvMqGs0TR7FTSzLqi Content-Type: application/pgp-signature; name="signature.asc" Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAABCgAGBQJXRbviAAoJEOe/yOyVhhEJKU0P/0KwlFF1GiMaJM03AnQxhJnY Rf2GbVfIwy1T6vNF+iL+XOnNVI6lpxoyIi5kwa9lMW6IK9eDsD4PckpYIWvCgPSk Q96YuOpsF5G0MUIkJSJQeyY4E29+RS3uh9Nh9mF2Bjm1nbZdHVlY+ZU4sdCrVfLu lc2t+xhGw+97Mmk6w8jnvdaWNH9nBT6OcIBZ9ZUJUYmNysee4B5NzCRuzUq2fDBI AjcQqgHPJo95AAbLv56xny79Ecxv+p1tSIvQ5CFevYude7frbDBH/+QaIxgnrJtW otAFuNoJ090qYF54mOUgCMwqft0Ji2qlBxKiapIUIzZ+2T/M6n3e3n7r2Q00RaKf UCRtJ0B9HujSDeGNftzq+V4YDsTNG5WQ6K0rCiqZTYc81xi3bMKszzhT7ZDbSsHF 9c0KFP6d6ZvOKl4olsWANKqBLuu7gTOi/BkZQbqNKKS7VcveyPq9XfJwNTB/9F6/ tS+Eux0nsvh77J6ahI/1FwstPUDu7ssFFzWdPhytWDrcankM/V+J1XqPom3j/xUv A2cXW67Lm3FqKzcmbNvAMRevV9HkhoDRcx9fFvb0ca7nYTGcWNM55qV9FUqgYC4V 9/rGdkIg0gzH1g0RB7I+UzBQXPmvQGPIAZWF/t5r8Sh3/wiJEvjHsrlNAcIK7BnI gCsIDmJhDzi7dj84HsD9 =lKd7 -----END PGP SIGNATURE----- --=-aVguvMqGs0TR7FTSzLqi--