[Pkg-sysvinit-devel] Bug#757083: Bug#757083: initscripts: please treat /usr (if separate) the same as /

[Henrique de Moraes Holschuh]

(Copied also Debian bug #697002).

On Fri, 08 Aug 2014, Simon McVittie wrote:
> Vaguely relatedly, I see you're a sysvinit/initscripts maintainer. On
> #697002 I suggested that maybe initscripts' checkfs.sh should use fsck
> -M (ignore mounted filesystems), like systemd-fsck does, instead of fsck
> -R (ignore root) as it currently does. That would make it unnecessary to
> modify util-linux for the "provide /usr early" feature, which could be
> done entirely via changes to initramfs-tools and initscripts. Any
> opinion on that?

Other than /etc separate from / being really fragile, I have no opinion on
that.  But I do agree with the others that it would make more sense to do
the more important /usr work first, and tackle /etc later.  IMO, separate
/usr correctly handled in initramfs is important for jessie, it shouldn't be
delayed by attempts to support a separate /etc.

I also agree that we are not able to support properly the MU-- case anymore.
While this makes me sad, it is how things are.  If you want a separate /usr,
you will need an initramfs for Debian jessie.

As for using -M in the checkfs initscript, it looks like it can be done as
long as the initramfs will make sure to fsck anything it will mount.  With
that in mind, I will second this change in checkfs.

Note that someone else will have to do the work and testing, and come up
with a tested patch, and attach it to #757083.

Also, redundant fsck is not supposed to be a problem: it should skip the
clean filesystem.  It is a non-problem when compared with mounting a dirty
filesystem (even if it would happen only on rare use cases), which causes
data loss.  If this adds some boot latency, so be it. REALLY.

Anyway, if we switch checkfs to fsck -M, there would be no need to change
the fsck wrapper -R behaviour, would there be?  So maybe #697002 can be
dismissed?

BTW: whether the kernel uses modules or not is orthogonal to whether it has
an initramfs or not.  It took me some time to notice you meant "no
initramfs" when you wrote about monolithic (module-less) kernels in #757083
and #697002.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh