[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#621803: mountkernfs.sh unconditionally mounts /run



* rleigh <rleigh@codelibre.net> [110510 23:55]:
> On Tue, May 10, 2011 at 11:31:23PM +0200, chris h wrote:
> > * rleigh <rleigh@codelibre.net> [110510 20:43]:
> > > On Tue, May 10, 2011 at 11:55:48AM +0200, Michael Biebl wrote:
[..]
> > > 
> > > Could you retry with
> > > 
> > > http://people.debian.org/~rleigh/sysvinit_2.88dsf-13.6.dsc
> > > http://people.debian.org/~rleigh/initscripts_2.88dsf-13.6_amd64.deb
> > > 
> > > Works for me, but since I don't use startpar it would be good to
> > > have that confirmed to work as well.  Chris, could you also try
> > > this out please?
> > 
> > Tried with latest initramfs-tools maks/run plus initscripts_2.88dsf-13.6
> > and the /run warning is indeed gone.
> 
> Great, thanks for testing!

Thank you for your fast response.

> > > [..]
> > > It would be ideal if we could upgrade to a current version of mount,
> > > such as 2.19 (2.19.1 is due out soon).
> > > http://www.kernel.org/pub/linux/utils/util-linux/v2.19/util-linux-2.19.tar.bz2
> > > This will work much better with /proc/mounts when built with libmount
> > > support enabled, and it's now actively used with /etc/mtab being a
> > > symlink (with Fedora/systemd) so it's likely that this sort of niggle
> > > when using /proc/mounts will have been fixed.
> > > Given that util-linux is not exactly kept up-to-date in Debian, if
> > > anyone wanted to package and NMU it, that would be another big step
> > > to getting systemd and read-only root working in Debian.  With this
> > > version in Debian, we can finally make the switch to having /etc/mtab
> > > as a symlink.
> > 
> > On a related note: we're building a live CD with an aufs-based
> > overlay /-fs.
> > As an artifact of this, during initramfs time we have to loop-mount
> > a squashfs to /image.squashfs. This mount point is never visible to 
> > the user, and mtab.sh complains that the mount point doesn't exist
> > when it tries to populate /etc/mtab.
> > 
> > I guess you don't want to have a general exception for
> > mounts of type 'squashfs' or something similar broad in mtab.sh?
> 
> Probably not.  The main reason being, /etc/mtab will be replaced by
> a symlink to /proc/mounts fairly soon.  With the new mount; test
> packages at
>   http://people.debian.org/~rleigh/util-linux_2.19.1-1.dsc
> and #603858 fixed (just sent a patch)
> and /run available
> mount will use /proc/mounts and /run/mount/utab (additional info
> about mounts not storable in /proc/mounts) and mtab will never
> be updated again.

FTR, util-linux_2.19.1-1 doesn't seem to build for me, on amd64:
if [ -d debian/cfdisk-udeb ]; then \
                cd debian/util-linux-locales && find usr/share/locale -type f | while read x; do ln $x ../cfdisk-udeb/$x; done \
        fi
ln: creating hard link `../cfdisk-udeb/usr/share/locale/pl/LC_MESSAGES/util-linux.mo' => `usr/share/locale/pl/LC_MESSAGES/util-linux.mo': No such file or directory
ln: creating hard link `../cfdisk-udeb/usr/share/locale/zh_CN/LC_MESSAGES/util-linux.mo' => `usr/share/locale/zh_CN/LC_MESSAGES/util-linux.mo': No such file or directory
ln: creating hard link `../cfdisk-udeb/usr/share/locale/zh_TW/LC_MESSAGES/util-linux.mo' => `usr/share/locale/zh_TW/LC_MESSAGES/util-linux.mo': No such file or directory
ln: creating hard link `../cfdisk-udeb/usr/share/locale/eu/LC_MESSAGES/util-linux.mo' => `usr/share/locale/eu/LC_MESSAGES/util-linux.mo': No such file or directory
ln: creating hard link `../cfdisk-udeb/usr/share/locale/hu/LC_MESSAGES/util-linux.mo' => `usr/share/locale/hu/LC_MESSAGES/util-linux.mo': No such file or directory
ln: creating hard link `../cfdisk-udeb/usr/share/locale/gl/LC_MESSAGES/util-linux.mo' => `usr/share/locale/gl/LC_MESSAGES/util-linux.mo': No such file or directory
make: *** [install] Error 1


> So it's probably not worth adding any special casing at this point
> given that the majority of the mtab.sh script will be deleted when
> this is done.

Yep.
For the time being, I will probably introduce a workaround on our side.

Thanks,
  -ch

Attachment: signature.asc
Description: Digital signature


Reply to: