From: "Jeremy Blosser (emrys)" <jblosser-smgl AT firinn.org>
To: "Sergey A. Lipnevich" <sergey AT sourcemage.org>, Arwed von Merkatz <v.merkatz AT gmx.net>, duane_malcolm <d.malcolm AT auckland.ac.nz>, Unet <unet AT sourcemage.org>, Gareth Clay <gareth AT caffeinefuelled.co.uk>, Eric Sandall <eric AT sandall.us>, Pol Vinogradov <vin.public AT gmail.com>, SM-Commit Daemon <sm-commit AT lists.ibiblio.org>, Andrew Stitt <a AT t.armory.com>, Arjan Bouter <abouter AT sourcemage.org>, vladimir_marek <vlmarek AT volny.cz>, Ladislav Hagara <ladislav.hagara AT unob.cz>, Maurizio Boriani <baux AT member.fsf.org>, Ethan Grammatikidis <eekee AT eekee.is-a-geek.org>
Subject: Re: [SM-Commit] PERFORCE change 78002 by Sergey A. Lipnevich for review
Date: Wed, 19 Apr 2006 18:56:29 -0500
On Apr 19, Jeremy Blosser (emrys) [jblosser-smgl AT firinn.org] wrote:
> On Apr 19, Sergey A. Lipnevich [sergey AT sourcemage.org] wrote:
> > Quoting Arwed von Merkatz <v.merkatz AT gmx.net>:
> > > All special filesystem like proc, tmpfs, devfs, sysfs, ... can be
> > > mounted fine before any explicit mounting (rw or ro) of /, udev systems
> > > already mount several of them in the DEV runlevel because udev requires
> > > them to work.
> >
> > Can we safely move mounting /proc to that code? Or maybe I should
> > simply try to
> > redistribute things between mountroot and mountall to mount /proc before
> > RAID
> > and LVM are initialized?
>
> proc should probably just get moved to mountroot. I'll look at this all
> and get it figured out probably tonight.
I think I know what all is needed and it's somewhat contradictory depending
on how the system is set up. Other distros I think avoid this by just
making some choices for the users but we of course don't want to needlessly
do that.
I'll get some fix in there soon but given how long it was there til it bit
Sergey it's not that likely many people have lvm+initrd+test grimoire I
guess.