GitLab

This is for kernels which have "block" built in instead of being a module.

Closes: #713877

We choose to not explicitly raise the required kernel version because
functional tests appear to be enough to validate real world kernels.

Use the upstream rules with a patch for the remaining Debian specific default
device permissions. Many thanks to Marco d'Itri for researching which
Debian-specific rules are obsolete!

Amongst other things, this now also reads the hwdb info for USB devices and
gets rid of some syntax errors.

Closes: #717405
Closes: #706221

Use the upstream rules with a patch for the sg module (see #657948). These now
stop calling modprobe and use the kmod builtin, giving some nice boot speed
improvement.

Closes: #717404

Move to using the upstream rules. They are compatible and do a superset of the
functionality.

In our 60-persistent-storage.rules:
- hd* rules are obsolete (only using PATA drivers these days with SCSI)
- id_dasd is obsolete (not in any package on any arch in any release)
- id_edd in Debian seems to not exist. Dell's edd package ships these rules by
  itself
- ID_SCSI_COMPAT isn't set by id_scsi any more, thus rules are obsolete

LP: #1274348

This avoids needing pt_chown (not available on all architectures).
Thanks Adam Conrad.

To avoid forgetting files when we move to new versions.

we don't want that in Debian and thus it shouldn't appear in dh_install
--list-missing output.

debian/extra/rule_generator.functions creates them in /run/udev/.

so that we don't drop the module from the PAM config when uninstalling
a foreign-arch package.  Related to Ubuntu bug #1295521.

Closes: #743158

Instead of tracking git upstream directly, the upstream branch now
contains the contents of the orig tarball available at
http://www.freedesktop.org/software/systemd/

This is necessary to properly deal with files like units/syslog.target,
which would not be deleted when merging otherwise.

This is a followup commit for e89ec25a

Instead of installing and removing tmp.mount later via debian/rules,
don't install it in the first place. This is a followup commit for
d151046c

closes #634472

This bug was introduced in aa57b6c5

We need to initialize the DBusError struct before it is used/freed.

from Ubuntu’s patches/pam-check-runtime-dir-user.patch:

From: Martin Pitt <martinpitt@gnome.org>
Date: Wed, 13 Nov 2013 13:02:28 +0100
Subject: [PATCH] pam: Check $XDG_RUNTIME_DIR owner

http://standards.freedesktop.org/basedir-spec/basedir-spec-latest.html requires
that $XDG_RUNTIME_DIR "MUST be owned by the user, and he MUST be the only one
having read and write access to it.".

Don't set an existing $XDG_RUNTIME_DIR in the PAM module if it isn't owned by
the session user. Otherwise su sessions get a runtime dir from a different user
which leads to either permission errors or scribbling over the other user's
files.

Bug-Fedora: https://bugzilla.redhat.com/show_bug.cgi?id=753882
Bug-Ubuntu: https://launchpad.net/bugs/1197395

closes: #739679

(Following Ben Hutching’s recommendation.)

closes #725422

closes #739113

from /etc/default/rcS and /etc/default/tmpfs, i.e. enable tmp.mount

Closes: #738687

fix systemctl enable/disable/… error message “Failed to issue method call: No such file or directory”

For real. This commit was accidentally reverted when merging
upstream-fixes into the debian branch *shakes fist at gitpkg*.

Closes: #738843

see #738799