Timezone changes are not working due to ro /etc and bind mounts

Bug #1227520 reported by Gema Gomez
46
This bug affects 9 people
Affects Status Importance Assigned to Milestone
android (Ubuntu)
Fix Released
Undecided
Oliver Grawert
apparmor (Ubuntu)
Fix Released
High
Martin Pitt
initramfs-tools-ubuntu-touch (Ubuntu)
Fix Released
Undecided
Stéphane Graber
livecd-rootfs (Ubuntu)
Fix Released
Undecided
Martin Pitt
lxc-android-config (Ubuntu)
Fix Released
Undecided
Martin Pitt
systemd (Ubuntu)
Fix Released
High
Martin Pitt
ubuntu-system-settings (Ubuntu)
Invalid
High
Iain Lane

Bug Description

There is no way to set up a different timezone than automatic in the System Settings

ProblemType: Bug
DistroRelease: Ubuntu 13.10
Package: ubuntu-system-settings 0.1+13.10.20130910.3-0ubuntu1
Uname: Linux 3.4.0-3-mako armv7l
ApportVersion: 2.12.2-0ubuntu1
Architecture: armhf
Date: Thu Sep 19 08:41:07 2013
InstallationDate: Installed on 2013-09-18 (0 days ago)
InstallationMedia: Ubuntu Saucy Salamander (development branch) - armhf (20130918)
MarkForUpload: True
ProcEnviron:
 TERM=linux
 PATH=(custom, no user)
SourcePackage: ubuntu-system-settings
UpgradeStatus: No upgrade log present (probably fresh install)

Revision history for this message
Gema Gomez (gema) wrote :
tags: added: avengers qa-touch touch-needs-autopilot
Revision history for this message
Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in ubuntu-system-settings (Ubuntu):
status: New → Confirmed
Revision history for this message
wojox (wojox) wrote :

Also affects nexus 7 grouper.

Julien Funk (jaboing)
tags: added: jfunk
Changed in ubuntu-system-settings (Ubuntu):
importance: Undecided → High
Changed in systemd (Ubuntu):
importance: Undecided → High
status: New → Confirmed
summary: - Timezone cannot be selected on system settings app
+ Timezone changes are not working due to ro /etc and bind mounts
Revision history for this message
Iain Lane (laney) wrote :

If you issue timedated's SetTimezone D-Bus call and strace the daemon, you see that it fails because of this:

  symlink("../usr/share/zoneinfo/Europe/Berlin", "/etc/.localtime4e49606a750b8c2c") = -1 EROFS (Read-only file system)

Which means that timedated is trying, entirely sensibly, to atomically overwrite the /etc/localtime symlink with the new timezone. This doesn't work with our ro-/etc scheme that we have on our device images. There will be a similar problem with updating /etc/timezone atomically too, but that's a file instead of a symlink.

The rough workaround that we have in mind (suggested by cjwatson, thanks) is to create a writable directory and make /etc/localtime /etc/timezone /etc/adjtime be symlinks to files/symlinks as appropriate inside there, which are themselves atomically updated. Do this for the fallback case (EROFS) only.

It does mean that we'll need to play whack-a-mole with anything else which thinks it can update these, sadly. I wonder if it'd be better to revisit making the whole of /etc writable on our images and deal with the conffile issues another way.

Revision history for this message
Sebastien Bacher (seb128) wrote :

(Ccing Steve and Stéphane to get their input.)

Do you have recommendations on what's the best way forward there, to get things working for v1, and to get things properly working on a converged world (without too much workarounds and hacks)?

Revision history for this message
Martin Pitt (pitti) wrote :

Please note that with a readonly /etc and the gross "mount individual files to writable partitions" hack, there is no way to implement that in a race free manner: As that setup makes it impossible to do rename() or use symlinks, the only thing that programs like timedated or tzdata's configure could do is to truncate() the existing /etc/localtime and write the new data into it. That means that during that time (and until the write cache catches up), any program which tries to determine the current time zone will misbehave or even crash, as it can only read junk from /etc/localtime.

I highly suggest to at least change our current hack to have a writable /data/ partition and symlink files from /etc/ to that. This would at least make atomic file changes possible in /data. The best way would still be to either have a writable /etc or overlayfs/aufs on /etc, though. There are certainly other programs which want to change /etc/, and it seems our time would be better spent building overlayfs/aufs for arm than with tracking down an unknown number of programs/applications that change files in /etc?

Revision history for this message
Stéphane Graber (stgraber) wrote :

So I think your best bet here is to change those daemons to fallback to in-place file content changes when atomic edits fail because of writable fs.
If /etc/localtime is already a symlink in the rootfs, you should be able to change its target without much trouble too.

Switching those files to be symlinks in the base rootfs is a bad idea which we initially considered. It's bad because we then have no way to get the original value of those, so we'd need something to create the target with a sensible content at boot time, which isn't terribly optimal. Also symlinks have a tendency to confuse some stat checks in various piece of software (probably not a big concern for timezone/localtime though).

Revision history for this message
Martin Pitt (pitti) wrote : Re: [Bug 1227520] Re: Timezone changes are not working due to ro /etc and bind mounts

Stéphane Graber [2013-09-24 13:57 -0000]:
> So I think your best bet here is to change those daemons to fallback
> to in-place file content changes when atomic edits fail because of
> writable fs.

As I wrote above that can only be a workaround hack. Changing such
kind of files inline is a source of race conditions and program
crashes, not a solution.

> If /etc/localtime is already a symlink in the rootfs, you should be
> able to change its target without much trouble too.

Not if it points to a file in /usr/share/zoneinfo/... You don't want
to change the contents of the target file, but you want to change the
symlink itself to a new target. But the latter again doesn't work on
an r/o file system, so we'd need some intermediate file like
/etc/localtime → /data/localtime → /usr/share/zoneinfo/...

Revision history for this message
Iain Lane (laney) wrote :

We can make writable directories in /etc, so ACK for your setting (it's what cjwatson suggested) but with s!data!etc/.timedated! (or so).

pitti agreed to work on a patch for this in #ubuntu-devel earlier; thanks a lot.

Changed in systemd (Ubuntu):
assignee: nobody → Martin Pitt (pitti)
Revision history for this message
Iain Lane (laney) wrote :

The u-s-s side should be fine (but if there are any unexpected issues here then I'll take care of them).

Changed in ubuntu-system-settings (Ubuntu):
status: Confirmed → Invalid
assignee: nobody → Iain Lane (laney)
Revision history for this message
Martin Pitt (pitti) wrote :

Changing timedated is blocked until we actually get writable directories in /etc/, setting to incomplete until that lands.

Changed in systemd (Ubuntu):
status: Confirmed → Incomplete
Revision history for this message
Iain Lane (laney) wrote :

On Tue, Sep 24, 2013 at 03:50:58PM -0000, Martin Pitt wrote:
> Changing timedated is blocked until we actually get writable directories
> in /etc/, setting to incomplete until that lands.

It works now; try adding it to /etc/system-image/writable-paths (not
sure if you have to make the directory first).

--
Iain Lane [ <email address hidden> ]
Debian Developer [ <email address hidden> ]
Ubuntu Developer [ <email address hidden> ]

Revision history for this message
Stéphane Graber (stgraber) wrote :

On Tue, Sep 24, 2013 at 04:14:41PM -0000, Iain Lane wrote:
> On Tue, Sep 24, 2013 at 03:50:58PM -0000, Martin Pitt wrote:
> > Changing timedated is blocked until we actually get writable directories
> > in /etc/, setting to incomplete until that lands.
>
> It works now; try adding it to /etc/system-image/writable-paths (not
> sure if you have to make the directory first).

The path needs to exist in the rootfs, otherwise we can't bind-mount
stuff onto it. So you need to have livecd-rootfs or some touch package
create the directory, move the files from /etc to that directory and
replace them with symlinks.

I don't believe any of those are conffiles or even shipped within
packages (they're generated at postinst), so there shouldn't be any need
for dpkg-divert or similar magic to keep stuff consistent.

--
Stéphane Graber
Ubuntu developer
http://www.ubuntu.com

Revision history for this message
Steve Langasek (vorlon) wrote :

Stéphane,

I've always been uncomfortable with using per-file bind mounts instead of symlinks for this. Mounts carry a small but real penalty in the kernel; we don't want the kernel to be using memory tracking thousands of mounts on the phone!

Why do you prefer bind mounts instead of symlinks here? I can't see any advantage to this, and while it wouldn't solve the /etc/localtime case, it would help other uses where the handling tools know how to read through symlinks.

Revision history for this message
Stéphane Graber (stgraber) wrote :

My main problem with symlinks are that we wouldn't be able to access the original file as the .tar.xz would contain a symlink instead of the file it's diverting.
Symlinks also come with the problem that some tests may fail (the symlink is typically seen as a file but may also be a directory) and some tools won't recurse through symlinks in the same way they do through directories.

On top of that, outside of the very small performance gain we'd gain, they really wouldn't solve much, this bug highlights two problematic cases:
 - Atomic file modification (by creating another file, then doing the rename+delete) which isn't possible as the directory itself is read-only. This would be no different with symlinks.
 - Symlink target changes. This actually works with bind-mounts, if you change the target of the target file but doesn't with symlinks as it'll instead try to change the original file which is read-only (unless we go through a separate read-write directory and add an extra step as was discussed in this bug).

So all in all, I saw symlinks as adding a lot of complexity (copying any file/directory they replace in some other location of the rootfs, ...) for only a barely measurable performance gain (and arguably a cleaner /proc/mounts, if anyone cares).

Revision history for this message
Martin Pitt (pitti) wrote :

So, to summarize my understanding what should happen (sorry, I'm still quite new to that touch layout, so I rather make sure):

 * livecd-rootfs creates /etc/writable/ (use a generic name as we might need it for other cases, and please no dotfiles)
 * livecd-rootfs moves /etc/localtime to /etc/writable/localtime and adds a corresponding symlink
 * lxc-android-config declares /etc/writable as writable -> this would create a mount point, and hide the original contents of /etc/writable/ as created by livecd-rootfs; how does the mounted file system get /etc/writable/zoneinfo? Is that done magically by the "transition" flag in /etc/system-image/writable-paths?
 * timedated and tzdata's postinst check if /etc/localtime point to /etc/writable (i. e. not to /usr/share/zoneinfo) and do its operation on the symlink target instead as a fallback.

Revision history for this message
Martin Pitt (pitti) wrote :

Oh, and we'd drop /etc/localtime from /etc/system-image/writable-paths instead.

Revision history for this message
Martin Pitt (pitti) wrote :

> how does the mounted file system get /etc/writable/zoneinfo? Is that done magically by the "transition" flag in /etc/system-image/writable-paths?

I just tested that by temporarily making the image writable, replicating the steps that livecd-rootfs and lxc-android-config would do, and that seems to work.

Revision history for this message
Iain Lane (laney) wrote :

Yeah, AFAIK it gets copied according to the transition flag the first time it shows up. Quote from IRC:

24/09 15:42:51 <stgraber> so any file (or directory) listed in writable-paths will be copied to /userdata and bind-mounted on first boot, if the content changes in the rootfs after that, you won't see it until you do a factory reset (which will trigger another copy from the rootfs to /userdata)

Not sure what you mean by /etc/writable/zoneinfo though - a typo?

I think you need to care for /etc/timezone and maybe also /etc/adjtime too.

And the code changes required in timedated too, of course.

Martin Pitt (pitti)
Changed in systemd (Ubuntu):
status: Incomplete → Triaged
Changed in lxc-android-config (Ubuntu):
status: New → Triaged
Changed in livecd-rootfs (Ubuntu):
status: New → In Progress
assignee: nobody → Martin Pitt (pitti)
Changed in systemd (Ubuntu):
milestone: none → ubuntu-13.09
Revision history for this message
Martin Pitt (pitti) wrote :

livecd-rootfs changes pushed to http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-rootfs/trunk/revision/848
writable-paths changes pushed to http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/saucy/lxc-android-config/saucy/revision/102

I tested those by

 * making image writable
 * umounting /etc/{adjtime,localtime,timezone}
 * Running the new live-build/ubuntu-touch/hooks/03-etc-writable.chroot script
 * Replacing /etc/system-image/writable-paths with the new version from above branch
 * making image non-writable again, reboot

Now I have

$ ls -l /etc/{adjtime,localtime,timezone}
lrwxrwxrwx 1 root root 21 Sep 25 09:22 /etc/adjtime -> /etc/writable/adjtime
lrwxrwxrwx 1 root root 23 Sep 25 09:22 /etc/localtime -> /etc/writable/localtime
lrwxrwxrwx 1 root root 22 Sep 25 09:22 /etc/timezone -> /etc/writable/timezone

and I can create new files in /etc/writable/. So I'm fairly sure that this is ok, but peer review highly appreciated before I actually upload this. Also, I believe Oliver needs to add that to some google docs to get a "change request".

Changed in livecd-rootfs (Ubuntu):
assignee: Martin Pitt (pitti) → nobody
status: In Progress → Fix Committed
Changed in lxc-android-config (Ubuntu):
status: Triaged → Fix Committed
assignee: nobody → Martin Pitt (pitti)
Changed in livecd-rootfs (Ubuntu):
assignee: nobody → Martin Pitt (pitti)
Revision history for this message
Martin Pitt (pitti) wrote :

This updates systemd's timedated to get along with those /etc/writable symlinks. Tested with both the normal files as well as an /etc/writable/ structure.

Changed in systemd (Ubuntu):
status: Triaged → Fix Committed
Revision history for this message
Martin Pitt (pitti) wrote :

So as far as I can see, all patches are ready for upload now. I tested them, but I'd appreciate another set of eyes on it. Stephane, are you ok with this approach? I'm ready to upload the three at any time.

tags: added: patch
Revision history for this message
Martin Pitt (pitti) wrote :

Adding initramfs-tools-ubuntu-touch tasks. Stephane says that this needs a new mode for /etc/system-image/writable-paths:

stgraber | pitti: so as a summary, I'd agree with the change if you change from "persistent" to "synced" (better names are welcome) and also propose an initramfs-tools-ubuntu-touch change to introduce that new "synced" mode (the code will need to recursively iterate through the source directory and only copy files and directories that don't already exist in the target)

I'm not sure when I'll have time to work on that; I was originally asked to help out with timedated only, not to competely understand and redesign our approach of handling /etc on the phone.. So help with that from someone who already understands the initramfs magic appreciated.

Revision history for this message
Steve Langasek (vorlon) wrote :

On Wed, Sep 25, 2013 at 08:54:31AM -0000, Iain Lane wrote:

> I think you need to care for /etc/timezone and maybe also /etc/adjtime
> too.

We absolutely do not need to do this for /etc/adjtime. /etc/adjtime is not
used in Ubuntu, and shouldn't even be on the filesystem.

Revision history for this message
Stéphane Graber (stgraber) wrote :

Added a debdiff for initramfs-tools-ubuntu-touch adding the new "synced" mode.

This appears to work as expected on my device. I'll let whoever lands the reset take care of landing that one first.

Revision history for this message
Stéphane Graber (stgraber) wrote :

Added a debdiff for lxc-android-config to drop the existing entries and replace them by a new "synced" entry for /etc/writable.

Note that before this lands, you'll need to make sure initramfs-tools-ubuntu-touch is uploaded and built, that android gets a no change rebuild to pick up initramfs-tools-ubuntu-touch and that the livecd-rootfs change has also landed.

Should any of those be missing, I'd expect the boot to still work but /etc/writable to end up read-only (if it exists).

Revision history for this message
Stéphane Graber (stgraber) wrote :

Since those bits don't have any dependencies between each other, we can't fix the ordering using those, so I'd strongly suggest they all land in the same image so it's actually possible to confirm it all works.

Revision history for this message
Martin Pitt (pitti) wrote :

Thanks Stéphane! Note, the lxc-android-config debdiff needs a bug ref in debian/changelog, otherwise it matches what I previously pushed to the branch (except for the new "synced" bit of course).

The livecd-rootfs change (http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-rootfs/trunk/revision/848) and my systemd patch (attached) are still current.

Changed in initramfs-tools-ubuntu-touch (Ubuntu):
status: New → Fix Committed
assignee: nobody → Stéphane Graber (stgraber)
Revision history for this message
Oliver Grawert (ogra) wrote :

changes to the touch initrd require a no change upload of the android package to pick them up, adding to the tasks

Revision history for this message
Iain Lane (laney) wrote :

On Wed, Sep 25, 2013 at 02:35:46PM -0000, Steve Langasek wrote:
> On Wed, Sep 25, 2013 at 08:54:31AM -0000, Iain Lane wrote:
>
> > I think you need to care for /etc/timezone and maybe also /etc/adjtime
> > too.
>
> We absolutely do not need to do this for /etc/adjtime. /etc/adjtime is not
> used in Ubuntu, and shouldn't even be on the filesystem.

Missed this.

It is on the filesystem, and from reading the code it looks as if
timedated will error out if it is present and can't be written to.

So as long as this is the case, we absolutely do.

--
Iain Lane [ <email address hidden> ]
Debian Developer [ <email address hidden> ]
Ubuntu Developer [ <email address hidden> ]

Revision history for this message
Steve Langasek (vorlon) wrote :

> Missed this.

> It is on the filesystem, and from reading the code it looks as if
> timedated will error out if it is present and can't be written to.

> So as long as this is the case, we absolutely do.

No, absolutely *not*. Do not pile more hacks on top of hacks. It is a bug
that /etc/adjtime is present, and it is a bug that timedated attempts to
manipulate it in Ubuntu.

Revision history for this message
Steve Langasek (vorlon) wrote :

/etc/adjtime was being created by live-build, behavior inherited from Debian where /etc/adjtime is still used. I've uploaded live-build to drop the file.

Revision history for this message
Steve Langasek (vorlon) wrote :

(Note that my system here, which was installed way back from an alternate CD, doesn't have /etc/adjtime and there are no problems with timezone setting - so there should be no risk of regression by having the file removed.)

Revision history for this message
Martin Pitt (pitti) wrote :

I can update the timedatectl patch to get along with a missing file (maybe it does already, need to test).

Revision history for this message
Iain Lane (laney) wrote :

On Fri, Sep 27, 2013 at 05:41:54PM -0000, Steve Langasek wrote:
> > Missed this.
>
> > It is on the filesystem, and from reading the code it looks as if
> > timedated will error out if it is present and can't be written to.
>
> > So as long as this is the case, we absolutely do.
>
> No, absolutely *not*. Do not pile more hacks on top of hacks. It is a bug
> that /etc/adjtime is present, and it is a bug that timedated attempts to
> manipulate it in Ubuntu.

I said "as long as this is the case", which means that it's valid to
make it so that it is not the case. That's what you did, which is fine
(as long as upgrades are cleaned up too). Thanks for that.

--
Iain Lane [ <email address hidden> ]
Debian Developer [ <email address hidden> ]
Ubuntu Developer [ <email address hidden> ]

Revision history for this message
Martin Pitt (pitti) wrote :

http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-rootfs/trunk/revision/848 gets along fine with a nonexistent /etc/adjtime, but will move it to /etc/writable if it is present. That seems right to me, as a non-writable /etc/adjtime is totally useless. If this isn't supposed to be in the phone images, we can throw it out of the patch of course (just drop it from the for loop). Steve, any opinion on that?

> I can update the timedatectl patch to get along with a missing file (maybe it does already, need to test).

I checked this now. timedated does not touch /etc/adjtime for changing timezones, so this isn't a concern for the purposes of this bug. It only creates/updates /etc/adjtime when calling "timedatectl set-local-rtc yes", and removes it again on "no", for storing the LOCAL flag. If we don't want that, I'm happy to take this out, but please let's not block this bug on that. It has already taken way more time than necessary to land this.

Revision history for this message
Steve Langasek (vorlon) wrote :

On Mon, Sep 30, 2013 at 06:44:54AM -0000, Martin Pitt wrote:
> http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-
> rootfs/trunk/revision/848 gets along fine with a nonexistent
> /etc/adjtime, but will move it to /etc/writable if it is present. That
> seems right to me, as a non-writable /etc/adjtime is totally useless. If
> this isn't supposed to be in the phone images, we can throw it out of
> the patch of course (just drop it from the for loop). Steve, any opinion
> on that?

Please drop it from the patch. live-build has already been fixed not to
create it, so there's no reason for livecd-rootfs to include handling.

> I checked this now. timedated does not touch /etc/adjtime for changing
> timezones, so this isn't a concern for the purposes of this bug. It only
> creates/updates /etc/adjtime when calling "timedatectl set-local-rtc
> yes", and removes it again on "no", for storing the LOCAL flag. If we
> don't want that, I'm happy to take this out, but please let's not block
> this bug on that. It has already taken way more time than necessary to
> land this.

Agreed, this should be changed but should not block this bug.

Revision history for this message
Martin Pitt (pitti) wrote :
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package initramfs-tools-ubuntu-touch - 0.50

---------------
initramfs-tools-ubuntu-touch (0.50) saucy; urgency=low

  * Add a new "synced" mode to writable-paths which will automatically copy
    any new file appearing on the filesystem but won't remove any extra file
    from the target (this should be used with care and with full understanding
    of the consequences). (LP: #1227520)
 -- StAphane Graber <email address hidden> Thu, 26 Sep 2013 17:55:27 -0400

Changed in initramfs-tools-ubuntu-touch (Ubuntu):
status: Fix Committed → Fix Released
Revision history for this message
Martin Pitt (pitti) wrote :

I uploaded lxc-android-config with stgraber's patch (plus changelog bug ref), my livecd-rootfs changes, and systemd.

Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package lxc-android-config - 0.104

---------------
lxc-android-config (0.104) saucy; urgency=low

  * Make /etc/writable "synced" and drop the entries that got moved in there.
    (LP: #1227520)
 -- Stephane Graber <email address hidden> Mon, 30 Sep 2013 13:31:37 +0200

Changed in lxc-android-config (Ubuntu):
status: Fix Committed → Fix Released
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package livecd-rootfs - 2.191

---------------
livecd-rootfs (2.191) saucy; urgency=low

  * ubuntu-touch: Make writable symlinks relative.
 -- Martin Pitt <email address hidden> Mon, 30 Sep 2013 13:59:59 +0200

Changed in livecd-rootfs (Ubuntu):
status: Fix Committed → Fix Released
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package systemd - 204-0ubuntu14

---------------
systemd (204-0ubuntu14) saucy; urgency=low

  * Add 0029-Support-phablet-etc-writable.patch: On Ubuntu Phone with readonly
    /etc we symlink /etc/{adjtime,localtime,timezone} to /etc/writable/, so we
    need to update those files instead if the original files are symlinks into
    /etc/writable/. (LP: #1227520)
  * Refresh debian/extra/60-keyboard.hwdb with latest keymaps from upstream.
    Among others, this fixes Samsung Ativ 9 Plus. (LP: #1229936)
  * Rename 0027-CVE-2013-4327.patch and
    0028-fix-reboot-polkit-permissions.patch to 0000-upstream-* as they are in
    upstream git.
  * Ship {ata,cdrom,scsi}_id in udev-udeb. (LP: #1231230)
  * Drop 60-persistent-v4l.rules and 61-accelerometer.rules from udev-udeb, we
    don't need these in the installer.
 -- Martin Pitt <email address hidden> Mon, 30 Sep 2013 13:34:06 +0200

Changed in systemd (Ubuntu):
status: Fix Committed → Fix Released
Oliver Grawert (ogra)
Changed in android (Ubuntu):
status: New → Fix Released
assignee: nobody → Oliver Grawert (ogra)
Revision history for this message
Martin Pitt (pitti) wrote :

Looks like there's some fallout:
[ 21.974853] type=1400 audit(1380573331.014:55): apparmor="DENIED" operation="open" parent=863 profile="/sbin/dhclient" name="/etc/writable/localtime" pid=1369 comm="dhclient" requested_mask="r" denied_mask="r" fsuid=0 ouid=0

Some of the AA profiles need updating for this.

Changed in apparmor (Ubuntu):
assignee: nobody → Martin Pitt (pitti)
importance: Undecided → High
status: New → In Progress
Revision history for this message
Martin Pitt (pitti) wrote :

Due to the last-minute change of making the symlinks relative the systemd patch now stopped working and needs to be adjusted.

Changed in systemd (Ubuntu):
status: Fix Released → In Progress
milestone: ubuntu-13.09 → ubuntu-13.10
Changed in apparmor (Ubuntu):
status: In Progress → Fix Committed
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package apparmor - 2.8.0-0ubuntu29

---------------
apparmor (2.8.0-0ubuntu29) saucy; urgency=low

  * Add 0070-etc-writable.patch: Allow reading time configuration from
    /etc/writable, as we have it on the phone. (LP: #1227520)
 -- Martin Pitt <email address hidden> Tue, 01 Oct 2013 09:55:15 +0200

Changed in apparmor (Ubuntu):
status: Fix Committed → Fix Released
Revision history for this message
Martin Pitt (pitti) wrote :

Updated systemd for the livecd-rootfs change in 2.191, sorry for not spotting it earlier.

Changed in systemd (Ubuntu):
status: In Progress → Fix Committed
Revision history for this message
Martin Pitt (pitti) wrote :

Reopening initramfs-tools-ubuntu-touch. When flashing current devel-proposed builds (74), /etc/writable/ is empty after boot (the symlinks /etc/timezone etc. into it are fine, though). http://cdimage.ubuntu.com/ubuntu-touch/daily-preinstalled/pending/saucy-preinstalled-touch-armhf.tar.gz does have the files, so it seems the "synced" mode doesn't actually copy the files?

Changed in initramfs-tools-ubuntu-touch (Ubuntu):
status: Fix Released → In Progress
Revision history for this message
Martin Pitt (pitti) wrote :

Changing to get along with absolute and relative symlinks.

Changed in systemd (Ubuntu):
status: Fix Committed → In Progress
Revision history for this message
Martin Pitt (pitti) wrote :

systemd uploaded, third time's the charm! I tested this now with relative and absolute symlinks, as well as direct files (as on desktop).

Changed in systemd (Ubuntu):
status: In Progress → Fix Committed
Revision history for this message
Martin Pitt (pitti) wrote :

The "empty /etc/writable/" dir should be fixed with https://launchpad.net/ubuntu/+source/initramfs-tools-ubuntu-touch/0.52

Changed in initramfs-tools-ubuntu-touch (Ubuntu):
status: In Progress → Fix Released
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package systemd - 204-0ubuntu15

---------------
systemd (204-0ubuntu15) saucy; urgency=low

  * 0029-Support-phablet-etc-writable.patch: Update patch to also work for
    relative symlinks to /etc/writable, as livecd-rootfs 2.191 creates them
    relative now. (LP: #1227520)
 -- Martin Pitt <email address hidden> Tue, 01 Oct 2013 17:53:13 +0200

Changed in systemd (Ubuntu):
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.