From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:44833) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1d8BwP-0005AR-Hf for guix-patches@gnu.org; Tue, 09 May 2017 16:41:06 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1d8BwM-000755-5y for guix-patches@gnu.org; Tue, 09 May 2017 16:41:05 -0400 Received: from debbugs.gnu.org ([208.118.235.43]:60113) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1d8BwL-00074x-Tl for guix-patches@gnu.org; Tue, 09 May 2017 16:41:02 -0400 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1d8BwL-0007FB-KX for guix-patches@gnu.org; Tue, 09 May 2017 16:41:01 -0400 Subject: bug#26339: [PATCH 02/18] system: Add extlinux support. Resent-Message-ID: From: ludo@gnu.org (Ludovic =?UTF-8?Q?Court=C3=A8s?=) References: <20170402135242.2958-1-m.othacehe@gmail.com> <20170402135242.2958-2-m.othacehe@gmail.com> <87mvanjepx.fsf@gnu.org> <871sryqy3g.fsf@gmail.com> <87bmr29x4q.fsf@gnu.org> <864lwuje63.fsf@gmail.com> Date: Tue, 09 May 2017 22:40:23 +0200 In-Reply-To: <864lwuje63.fsf@gmail.com> (Mathieu Othacehe's message of "Tue, 09 May 2017 16:30:28 +0200") Message-ID: <87inl9zruw.fsf@gnu.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-patches-bounces+kyle=kyleam.com@gnu.org Sender: "Guix-patches" To: Mathieu Othacehe Cc: 26339@debbugs.gnu.org, David Craven Hello! Mathieu Othacehe skribis: >> Woow, that was fast! :-) > > I had time to kill during a train trip :) I like train trips too. :-) >> Good question! I guess we need to store the name of the bootloader type >> in the =E2=80=9Cparameters=E2=80=9D file, and then look it up upon switc= h-generation? >> But maybe we also need to store a reference to the bootloader package so >> that it isn=E2=80=99t GC=E2=80=99d? > > We have three options for switch-generation : > > 1. Do not reinstall bootloader, only update config (current behaviour). > > -> It was ok with one bootloader, it's more problematic with multiple > bootloaders. For example switching between two generations with two > different bootloaders. OTOH, if it turns out to be safer to keep the current bootloader, which is likely I think, then perhaps it=E2=80=99s best to not try to change. I mean, I prefer a rollback that doesn=E2=80=99t roll back the bootloader o= ver a rollback that tries to do something smart and makes the system unbootable. Of course, if we can make it safe to reinstall the previous bootloader, that=E2=80=99s optimal. > 2. Save the bootloader type in "parameters" and reinstall bootloader > (current bootloader build, not necessarily the one used when creating > this generation) > > -> It's the option I choose in my series. OK. I guess I should read that part of the series. > 3. Same as 2. but also save bootloader used (store reference) and add a > gc-root to keep it around. > > I can try to implement option 3 if you're ok ? Hmm maybe, I need to investigate. >> I think the most important is to make sure that our current use case, >> where people always use GRUB, doesn=E2=80=99t break. > > With my basic-one-unencrypted-partition setup it work fine, but it would > be great if people with more complex setup could try this serie :) Did you try something like: make check-system TESTS=3D"encrypted-root-os btrfs-root-os" ? It takes quite a bit of time to run, but it=E2=80=99s a good starting po= int to test these things. Thanks! Ludo=E2=80=99.