From mboxrd@z Thu Jan 1 00:00:00 1970 From: Colin Guthrie Subject: Re: [ANNOUNCE] dracut 034 Date: Mon, 28 Oct 2013 15:23:50 +0000 Message-ID: References: <5253DD0C.7050506@redhat.com> <526A7412.8030804@redhat.com> <20131028134216.049c6a01@work.puleglot> Mime-Version: 1.0 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: <20131028134216.049c6a01-d/9N6kaejYfpx9ONITL1FA@public.gmane.org> Sender: initramfs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="utf-8" To: initramfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org 'Twas brillig, and Alexander Tsoy at 28/10/13 09:42 did gyre and gimble= : > =D0=92 Fri, 25 Oct 2013 15:37:22 +0200 > Harald Hoyer =D0=BF=D0=B8=D1=88=D0=B5=D1=82: >=20 >> On 10/25/2013 01:22 PM, EatDirt wrote: >>> On 08/10/13 12:23, Harald Hoyer wrote: >>>> Here comes a new dracut version. Thanks to all who have contribute= d to this >>>> release. >>> >>> Hi guys, >>> sorry if this is not the right place to post, but we have some prob= lems with mga >>> distributed dracut-034 >>> >>> For users having two partitions / and /usr; specified in their /etc= /fstab, >>> without including systemd module, dracut does not automatically inc= ludes the >>> necessary usrmount module. >>> >>> Without usrmount, the initrd yields a kernel panics at boot time. I= naively >>> though that dracut would automatically take care of the partitions = specified in >>> /etc/fstab and therefore "sees" that without systemd added, usrmoun= t must be >>> present. >>> >>> Do you think it is a dracut bug, or we should take care of this at = the "distro" >>> packaging level? >>> >>> Thanks for your help! >>> >>> cheers, >>> chris. >>> >>> >>> PS: the mga full bug report is there (check the last posts): >>> >>> https://bugs.mageia.org/show_bug.cgi?id=3D11292 >>> >> >> Hmmm, the usrmount module is automatically included, if /sbin/init i= s on /usr. >> If it is not on /usr, normally /sbin/init can mount /usr via fstab. >> >=20 > But this logic breaks systems with separate (not merged) /usr. Accord= ing > to [1], /usr should be always mounted inside initramfs. For example, > openrc in Gentoo launches udev before mounting filesystems via fstab. I think you're ultimately correct, but keep in mind that if you do have a separate /usr with a non-merged system, you need to have udev and all it's components in / anyway (i.e. /lib, /sbin etc) such that udev could be started prior to mounting /usr. Of course when this happens you run into a whole manner of knock on problems such as helper binaries from udev rules not yet being available, missing firmware and similar - i.e. a big pile of problems. This was indeed one of the issues pointed out during the eudev talk at =46OSDEM by the likes of Grek KH, Kay and co. There was some vague talk about developing a mechanism to delay the actual handling of such rules when /usr is not mounted but the resounding suggestion from the audienc= e was to "just let the initramfs handle /usr mounting!!" So I think you're probably reading a bit too much into Harald's statement above. In theory it's possible to have /sbin/init mount /usr = - but it's something I think everyone versed in the issues would highly discourage! That being the case, I would propose that the usrmount module be automatically included when /usr is a mount point regardless of on whic= h partition init lives? Col --=20 Colin Guthrie gmane(at)colin.guthr.ie http://colin.guthr.ie/ Day Job: Tribalogic Limited http://www.tribalogic.net/ Open Source: Mageia Contributor http://www.mageia.org/ PulseAudio Hacker http://www.pulseaudio.org/ Trac Hacker http://trac.edgewall.org/