Date: Mon, 4 Jan 2016 13:21:16 -0700 From: Warner Losh <imp@bsdimp.com> To: Colin Percival <cperciva@tarsnap.com> Cc: Warner Losh <imp@freebsd.org>, src-committers <src-committers@freebsd.org>, "svn-src-all@freebsd.org" <svn-src-all@freebsd.org>, "svn-src-head@freebsd.org" <svn-src-head@freebsd.org> Subject: Re: svn commit: r293115 - head/etc Message-ID: <CANCZdfqvsTsAPVDCd4sdBaN1QsUPeb6ZRXCG50cAuC9A-Y7fMA@mail.gmail.com> In-Reply-To: <000001520dbaad1a-55cdc7aa-a009-4672-a226-cb140785c55c-000000@email.amazonses.com> References: <201601031918.u03JImBs012182@repo.freebsd.org> <000001520d9553a7-b3ef495a-89d6-44ec-91c1-c4f9afc2c55b-000000@email.amazonses.com> <CANCZdfr=Cffp5N-Q1o2UtNhVvLtOVLZnqh3yzBh3=mX=rzsu0w@mail.gmail.com> <000001520dbaad1a-55cdc7aa-a009-4672-a226-cb140785c55c-000000@email.amazonses.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jan 4, 2016 at 10:41 AM, Colin Percival <cperciva@tarsnap.com> wrote: > On 01/04/16 09:09, Warner Losh wrote: > > On Mon, Jan 4, 2016 at 10:00 AM, Colin Percival <cperciva@tarsnap.com > > <mailto:cperciva@tarsnap.com>> wrote: > > On 01/03/16 11:18, Warner Losh wrote: > > > Fix the read-only > > > root case with horrible kludge of mounting rw removing the > files, then > > > mounting ro. > > > > The solution I intended when I introduced this (and used elsewhere) > was to > > set $firstboot_sentinel in /etc(/defaults)?/rc.conf. This case is > precisely > > why it's a shell variable, in fact. > > > > Except that's not exactly useful. NanoBSD boots with no filesystems > writable > > that are permanent. So I could set it to /var/firstboot or something > like that, > > and the error would go away. However, that wouldn't solve the problem > > because /var is repopulated from base seed files every boot with NanoBSD > > so we'd get firstboot behavior on every single boot. Or, we could remount > > / rw and remove the file and remount it ro when a read-only root was > > requested. > > Huh, ok. I assumed that you had a /conf/ or something like that for > storing > persistent configuration data. > It does. But all that's mirrored in a MFS and it takes an explicit command to write it all back. This is poorly matched with where the firstboot files are actually removed. > > I wondered to myself why we didn't use the same mechanism as nextboot > > for this feature. Do you know? > > Doesn't that still write to the filesystem? > So it does. I thought it mucked about with flags and such. Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfqvsTsAPVDCd4sdBaN1QsUPeb6ZRXCG50cAuC9A-Y7fMA>