chiark / gitweb /
elogind.git
10 years agodbus-timer: fix bus_timer_vtable to have the correct times
Djalal Harouni [Mon, 24 Feb 2014 00:42:23 +0000 (01:42 +0100)]
dbus-timer: fix bus_timer_vtable to have the correct times

next_elapse_monotonic() should map to the "NextElapseUSecMonotonic"
property and next_elapse_realtime() to "NextElapseUSecRealtime" one.

This makes "systemctl list-timers" compute and show the correct times.

https://bugs.freedesktop.org/show_bug.cgi?id=75272

10 years agoupdate TODO
Lennart Poettering [Mon, 24 Feb 2014 03:04:23 +0000 (04:04 +0100)]
update TODO

10 years agogetty-generator: verify ttys before we make use of them
Lennart Poettering [Mon, 24 Feb 2014 02:59:50 +0000 (03:59 +0100)]
getty-generator: verify ttys before we make use of them

The ttyS[0-3] devices are weird. They may be enumerated, but when one
actually tries to open and use them they return EIO, because they don't
actually exist. Because they may be enumerated they may be specified on
the kernel command line as console=. And some people do that as default.
As response to that we'll spawn a getty on the tty that will quickly
fail, and we retry a couple of time before giving up. That is quite
noisy.

With this new change we will validate all serial terminals configured
with console= on the kernel cmdline before adding gettys on them, and
remove the invalid ones. THis should remove the noise later on.

This should make Eric Paris happy!

10 years agocgroup: certain cgroup attributes are not available in the root cgroup, hence don...
Lennart Poettering [Mon, 24 Feb 2014 02:38:24 +0000 (03:38 +0100)]
cgroup: certain cgroup attributes are not available in the root cgroup, hence don't bother

10 years agocore: gc half-created stub units
Lennart Poettering [Mon, 24 Feb 2014 02:20:54 +0000 (03:20 +0100)]
core: gc half-created stub units

10 years agocore: expose root control group on the bus
Lennart Poettering [Mon, 24 Feb 2014 02:20:24 +0000 (03:20 +0100)]
core: expose root control group on the bus

We expose the control group of the units on the bus, so let's also
expose the root control group.

10 years agoupdate TODO
Lennart Poettering [Mon, 24 Feb 2014 02:03:05 +0000 (03:03 +0100)]
update TODO

10 years agonspawn: make use of the devices cgroup controller by default
Lennart Poettering [Mon, 24 Feb 2014 02:00:38 +0000 (03:00 +0100)]
nspawn: make use of the devices cgroup controller by default

10 years agocgroup: parse array cgroup properties correctly when they aren't at the end of the...
Lennart Poettering [Mon, 24 Feb 2014 01:59:57 +0000 (02:59 +0100)]
cgroup: parse array cgroup properties correctly when they aren't at the end of the message

10 years agosd-dhcp-client: respect TP_STATUS_CSUMNOTREADY
Tom Gundersen [Mon, 24 Feb 2014 00:09:21 +0000 (01:09 +0100)]
sd-dhcp-client: respect TP_STATUS_CSUMNOTREADY

If an UDP packet has not passed through a hardware device, its checksum may not
have been computed. This is exposed through the TP_STATUS_CSUMNOTREADY sockopt.

When using raw sockets, skip checksum validation when TP_STATUS_CSUMNOTREADY
is set.

This is necessary for dhcp to work directly over a veth tunnel, e.g. as done
in systemd-nspawn.

10 years agoupdate TODO
Lennart Poettering [Mon, 24 Feb 2014 01:31:01 +0000 (02:31 +0100)]
update TODO

10 years agoNEWS: mention 99-default.link and NamePolicy
Zbigniew Jędrzejewski-Szmek [Sun, 23 Feb 2014 23:11:48 +0000 (18:11 -0500)]
NEWS: mention 99-default.link and NamePolicy

10 years agosd-dhcp-client: fix max size of REQUEST message
Tom Gundersen [Sun, 23 Feb 2014 21:09:23 +0000 (22:09 +0100)]
sd-dhcp-client: fix max size of REQUEST message

10 years agosd-dhcp: message_init - only set secs in the client
Tom Gundersen [Sun, 23 Feb 2014 21:07:07 +0000 (22:07 +0100)]
sd-dhcp: message_init - only set secs in the client

10 years agosd-dhcp: don't reject packets with the 'wrong' source port
Tom Gundersen [Sun, 23 Feb 2014 18:21:50 +0000 (19:21 +0100)]
sd-dhcp: don't reject packets with the 'wrong' source port

The RFC does not specify that the packets from the DHCP server must come from
the DHCP server port, only that that's where they should be sent.

This fixes a problem when running networkd in VirtualBox.

Thanks to Sébastien Luttringer for reporting the bug and very patiently testing
various fixes.

10 years agosd-dhcp: use FIONREAD to get correct size of incoming DHCP packet
Tom Gundersen [Sun, 23 Feb 2014 16:30:13 +0000 (17:30 +0100)]
sd-dhcp: use FIONREAD to get correct size of incoming DHCP packet

This avoids the problem of broken DHCP servers sending us too big packets that don't fit in our buffer.

10 years agojournal: downgrade vaccuum message to debug level
Lennart Poettering [Sun, 23 Feb 2014 16:05:37 +0000 (17:05 +0100)]
journal: downgrade vaccuum message to debug level

https://bugzilla.redhat.com/show_bug.cgi?id=1047148

10 years agoman: document that per-interface sysctl variables are applied as network interfaces...
Lennart Poettering [Sun, 23 Feb 2014 15:43:19 +0000 (16:43 +0100)]
man: document that per-interface sysctl variables are applied as network interfaces show up

https://bugzilla.redhat.com/show_bug.cgi?id=1062955

10 years agosd-dhcp: equally verify udp and raw dhcp messages
Tom Gundersen [Sun, 23 Feb 2014 13:15:05 +0000 (14:15 +0100)]
sd-dhcp: equally verify udp and raw dhcp messages

Also be more explicit about why packages are ignored.

10 years agoupdate TODO
Lennart Poettering [Sun, 23 Feb 2014 02:18:46 +0000 (03:18 +0100)]
update TODO

10 years agocore: clean up some confusing regarding SI decimal and IEC binary suffixes for sizes
Lennart Poettering [Sun, 23 Feb 2014 02:13:54 +0000 (03:13 +0100)]
core: clean up some confusing regarding SI decimal and IEC binary suffixes for sizes

According to Wikipedia it is customary to specify hardware metrics and
transfer speeds to the basis 1000 (SI decimal), while software metrics
and physical volatile memory (RAM) sizes to the basis 1024 (IEC binary).
So far we specified everything in IEC, let's fix that and be more
true to what's otherwise customary. Since we don't want to parse "Mi"
instead of "M" we document each time what the context used is.

10 years agodocs: add new gudev functions to documentation
Zbigniew Jędrzejewski-Szmek [Sun, 23 Feb 2014 01:52:18 +0000 (20:52 -0500)]
docs: add new gudev functions to documentation

10 years agosrc/shared/ cannot reference symbols from libraries
Kay Sievers [Sat, 22 Feb 2014 23:46:26 +0000 (00:46 +0100)]
src/shared/ cannot reference symbols from libraries

../src/shared/unit-name.c:462: error: undefined reference to 'sd_bus_label_escape'
../src/shared/unit-name.c:477: error: undefined reference to 'sd_bus_label_unescape'
collect2: error: ld returned 1 exit status

10 years agosd-dhcp: be more detailed about invalid headers
Tom Gundersen [Sun, 23 Feb 2014 00:34:05 +0000 (01:34 +0100)]
sd-dhcp: be more detailed about invalid headers

This may be a common problem, so let's make it simpler to debug,
at least for now.

10 years agobuild-sys: remove libsystemd.sym duplicate
David Herrmann [Sat, 22 Feb 2014 23:26:29 +0000 (00:26 +0100)]
build-sys: remove libsystemd.sym duplicate

sd_memfd_new is available twice. Remove the second one.

10 years agobuild-sys: fix libsystemd --enable-kdbus symbol export
Kay Sievers [Sat, 22 Feb 2014 22:27:03 +0000 (23:27 +0100)]
build-sys: fix libsystemd --enable-kdbus symbol export

10 years agobuild-sys: build compat libs from sources, avoiding the not generally available ifunc...
Kay Sievers [Sat, 22 Feb 2014 18:39:12 +0000 (19:39 +0100)]
build-sys: build compat libs from sources, avoiding the not generally available ifunc wrapping

10 years agonetworkd: handle SIGINT and SIGTERM
Tom Gundersen [Sat, 22 Feb 2014 19:19:49 +0000 (20:19 +0100)]
networkd: handle SIGINT and SIGTERM

10 years agotest: always use assert_se in tests
Tom Gundersen [Sat, 22 Feb 2014 19:22:41 +0000 (20:22 +0100)]
test: always use assert_se in tests

10 years agosd-dhcp: add very basic debug logging
Tom Gundersen [Sat, 22 Feb 2014 18:53:45 +0000 (19:53 +0100)]
sd-dhcp: add very basic debug logging

10 years agobuild-sys: add seccomp includes where needed
Kay Sievers [Sat, 22 Feb 2014 17:47:28 +0000 (18:47 +0100)]
build-sys: add seccomp includes where needed

10 years agogudev: add device::get_sysfs_attr_keys and device::has_sysfs_attr
Andreas Fuchs [Sat, 22 Feb 2014 11:02:35 +0000 (12:02 +0100)]
gudev: add device::get_sysfs_attr_keys and device::has_sysfs_attr

For gudev -> gudevdevice:
- Add support for get_sysfs_attr_keys()
- Add support for has_sysfs_attr()

10 years agoudev: net - allow MTU and Speed to be specified with units
Tom Gundersen [Sat, 22 Feb 2014 17:23:19 +0000 (18:23 +0100)]
udev: net - allow MTU and Speed to be specified with units

This also changes the names to MTUBytes and BitsPerSecond, respectively. Notice
that the speed was mistakenly documented to be in bytes before this change.

10 years agonetworkd/nspawn: enable dhcp client on veth device created in guest
Tom Gundersen [Fri, 21 Feb 2014 21:44:42 +0000 (22:44 +0100)]
networkd/nspawn: enable dhcp client on veth device created in guest

When starting systemd-nspawn with --network-veth, we create a veth device called
host0 in the guest. Pick up on this and start a dhcp client on it. We will also
pick up host0 netdevs created by other containers should they chose to use the
same name.

10 years agobuild-sys: create /var/log/journal/ in journald section
Kay Sievers [Sat, 22 Feb 2014 16:58:05 +0000 (17:58 +0100)]
build-sys: create /var/log/journal/ in journald section

10 years agobuild-sys: Also move libsystemd-journal to rootlibdir
Armin K [Sat, 22 Feb 2014 14:22:00 +0000 (15:22 +0100)]
build-sys: Also move libsystemd-journal to rootlibdir

10 years agohwdb: Update database of Bluetooth company identifiers
Marcel Holtmann [Sat, 22 Feb 2014 05:41:01 +0000 (21:41 -0800)]
hwdb: Update database of Bluetooth company identifiers

10 years agocgroup: Extend DeviceAllow= syntax to whitelist groups of devices, not just particula...
Lennart Poettering [Sat, 22 Feb 2014 01:47:29 +0000 (02:47 +0100)]
cgroup: Extend DeviceAllow= syntax to whitelist groups of devices, not just particular devices nodes

10 years agoupdate gitignore
Thomas Hindoe Paaboel Andersen [Fri, 21 Feb 2014 22:21:26 +0000 (23:21 +0100)]
update gitignore

10 years agonet-util: match on the driver as exposed by ethtool if DRIVER not set
Tom Gundersen [Fri, 21 Feb 2014 21:29:25 +0000 (22:29 +0100)]
net-util: match on the driver as exposed by ethtool if DRIVER not set

Also fix a copy-paste error that broke matching on interface name.

10 years agoudev: net_setup - import ID_NET_DRIVER
Tom Gundersen [Fri, 21 Feb 2014 20:45:47 +0000 (21:45 +0100)]
udev: net_setup - import ID_NET_DRIVER

This will do until all net properties are imported.

10 years agoMakefile: fix build on clang
Thomas Hindoe Paaboel Andersen [Fri, 21 Feb 2014 21:21:50 +0000 (22:21 +0100)]
Makefile: fix build on clang

needed after libsystemd-daemon was merged into libsystemd in
0ebee8818404adb95a0b8a01416aad3a16f64ae1

10 years agoupdate TODO
Tom Gundersen [Fri, 21 Feb 2014 20:41:48 +0000 (21:41 +0100)]
update TODO

10 years agonet: initialize structs when declaring
Tom Gundersen [Fri, 21 Feb 2014 20:32:38 +0000 (21:32 +0100)]
net: initialize structs when declaring

10 years agonet: initialize structs when declaring them/do not clobber caller variables
Lennart Poettering [Fri, 21 Feb 2014 20:23:40 +0000 (21:23 +0100)]
net: initialize structs when declaring them/do not clobber caller variables

Just two minor style fixes...

10 years agologind: when we wake up from suspend and the lid is still closed, go to sleep immedia...
Lennart Poettering [Fri, 21 Feb 2014 20:10:00 +0000 (21:10 +0100)]
logind: when we wake up from suspend and the lid is still closed, go to sleep immediately again

This is quite useful on laptops such as the Lenovo Yoga, where the power
button is placed on the front side of the laptop and can be pressed by
accident even if the lid is closed.

This reworks a bit of the logind logic to repeatedly try to suspend the
system as long as a lid is closed. We use the new "post" event source
for this, so that we don't keep things busy.

This also adds some code to check the lid status on boot, so that a
powered-off machine that is accidentaly powered on goes into suspend
immediately.

Yay! From now on I can put my Yoga safely in my backpack without fearing
that it might turn itself on and drain the battery.

10 years agoevent: add new kind of event source called "post"
Lennart Poettering [Fri, 21 Feb 2014 20:06:09 +0000 (21:06 +0100)]
event: add new kind of event source called "post"

This new event source is triggered by the dispatching of any non-post
event source. It can thus be used to do clean-up or recheck work,
triggered by any other event source having been executed.

This is different from "defer" event source which are unconditionally
triggered as long as they are enabled. A "defer" event source that does
nothing will result in the event loop busy looping unless it is turned
off eventually. This is different for "post" event sources that will be
only triggered when some other non-post ran, and will thus not keep the
event loop busy on its own.

10 years agoman: networkd typo fixes
Thomas Hindoe Paaboel Andersen [Fri, 21 Feb 2014 19:54:30 +0000 (20:54 +0100)]
man: networkd typo fixes

10 years agoudev - link-setup - expose ID_NET_DRIVER
Tom Gundersen [Fri, 21 Feb 2014 15:54:00 +0000 (16:54 +0100)]
udev - link-setup - expose ID_NET_DRIVER

This is the same as shown by 'ethtool -i <ifname>', and is sometimes
set even though DRIVER is not.

10 years agologind-session: Fix invalid free in the error case
Jasper St. Pierre [Fri, 21 Feb 2014 16:22:42 +0000 (11:22 -0500)]
logind-session: Fix invalid free in the error case

utf8 needs to be initialized to NULL for the free for the early return,
otherwise we try to free invalid data.

10 years agoinstall: networkd.service depend on !--disable
Jason A. Donenfeld [Fri, 21 Feb 2014 15:36:59 +0000 (16:36 +0100)]
install: networkd.service depend on !--disable

It doesn't make any sense to symlink this unit file into /etc when the
unit file itself isn't even installed, with --disable-networkd. This
moves the GENERAL_ALIASES logic into the right "if" block.

10 years agoupdate TODO
Lennart Poettering [Fri, 21 Feb 2014 18:22:00 +0000 (19:22 +0100)]
update TODO

10 years agoman: suffix networkd config file options with "="
Lennart Poettering [Fri, 21 Feb 2014 18:19:31 +0000 (19:19 +0100)]
man: suffix networkd config file options with "="

That's what we do for all options in the other man pages. It helps
clarifying that these are options that values need to be assigned to.

10 years agoman: split settings documentaiton in systemd.unit(5) into two sections
Lennart Poettering [Fri, 21 Feb 2014 17:57:47 +0000 (18:57 +0100)]
man: split settings documentaiton in systemd.unit(5) into two sections

10 years agoman: don't document ".include" in configuration files anymore as first step to deprec...
Lennart Poettering [Fri, 21 Feb 2014 17:57:15 +0000 (18:57 +0100)]
man: don't document ".include" in configuration files anymore as first step to deprecate them one day

10 years agoupdate TODO
Lennart Poettering [Fri, 21 Feb 2014 17:54:59 +0000 (18:54 +0100)]
update TODO

10 years agojournald: ignore failure to watch hostname_fd on older kernels
Dave Reisner [Fri, 21 Feb 2014 15:31:26 +0000 (10:31 -0500)]
journald: ignore failure to watch hostname_fd on older kernels

Prior to 3.2, /proc/sys/kernel/hostname isn't a pollable file and
sd_event_add_io will return EPERM. Ignore this failure, since it isn't
critical to journald operation.

Reported and tested by user sraue on IRC.

10 years agoptyfwd: reset nonblocking mode
Lennart Poettering [Fri, 21 Feb 2014 17:04:29 +0000 (18:04 +0100)]
ptyfwd: reset nonblocking mode

Apparently bash doesn't turn off non-blocking mode on stdin/stdout when
reading from it, so be nice to bash. Ideally bash would do this on its
own for robustness reasons, though.

https://bugs.freedesktop.org/show_bug.cgi?id=70622

10 years agoNEWS: add name_to_handle_at() requirement
Kay Sievers [Fri, 21 Feb 2014 17:11:47 +0000 (18:11 +0100)]
NEWS: add name_to_handle_at() requirement

10 years agobuild-sys: make network files respect --rootprefix
Samuli Suominen [Fri, 21 Feb 2014 14:14:51 +0000 (16:14 +0200)]
build-sys: make network files respect --rootprefix

With --rootprefix= systemd-udevd gets installed to /lib/systemd, and since
the network configuration is also required during early boot, it should be
available there with it. Using --prefix= is not an option since it would
put everything, including pkg-config files, man pages, documentation, to /
which is not wanted. This commit puts 99-default.link to
/lib/systemd/network/ when required.

10 years ago.network/.netdev/.link: allow to match on architecture
Tom Gundersen [Fri, 21 Feb 2014 13:51:19 +0000 (14:51 +0100)]
.network/.netdev/.link: allow to match on architecture

10 years agoconfigure: show if networkd is enabled in status
Jason A. Donenfeld [Fri, 21 Feb 2014 14:21:21 +0000 (15:21 +0100)]
configure: show if networkd is enabled in status

10 years agoman: there is no ExecStopPre= for service units
Lennart Poettering [Fri, 21 Feb 2014 14:06:44 +0000 (15:06 +0100)]
man: there is no ExecStopPre= for service units

https://bugs.freedesktop.org/show_bug.cgi?id=73177

10 years agoinstall: enable networkd by default
Tom Gundersen [Fri, 21 Feb 2014 13:20:14 +0000 (14:20 +0100)]
install: enable networkd by default

This treats it similarly to getty@tty1.service and remote-fs.target, which can be
enabled/disabled, but defaults to enabled.

10 years agoinstall: do not statically enable systemd-networkd
Jason A. Donenfeld [Fri, 21 Feb 2014 03:38:17 +0000 (04:38 +0100)]
install: do not statically enable systemd-networkd

[tomegun: pruned the commit message as not to contradict the follow-up commit]

10 years agosd-event: Fix systemd crash when using timer units.
Thomas Bächler [Fri, 21 Feb 2014 10:20:24 +0000 (11:20 +0100)]
sd-event: Fix systemd crash when using timer units.

There was a copy-paste error introduced in commit c2ba3ad6604ef2e189d7e0a36d6911116e84d3ab
which causes the following error when using timer units:

Assertion '(x->type == SOURCE_MONOTONIC && y->type == SOURCE_MONOTONIC) || (x->type == SOURCE_REALTIME && y->type == SOURCE_REALTIME)'
failed at src/libsystemd/sd-event/sd-event.c:264, function latest_time_prioq_compare(). Aborting.

10 years agosystemctl: fix compiler warning in list_timers()
Daniel Mack [Fri, 21 Feb 2014 11:05:38 +0000 (12:05 +0100)]
systemctl: fix compiler warning in list_timers()

get_next_elapse() will always fill 'next' with values when it
returns >= 0. Hence, the compiler is wrong about this warning.
Initialize 'next' nevertheless.

src/systemctl/systemctl.c: In function ‘list_timers’:
src/systemctl/systemctl.c:953:43: warning: ‘next.monotonic’ may be used
uninitialized in this function [-Wmaybe-uninitialized]
                                 converted = nw.realtime - (nw.monotonic - next.monotonic);
                                           ^
In file included from ./src/shared/log.h:30:0,
                 from src/systemctl/systemctl.c:46:
./src/shared/macro.h:137:38: warning: ‘next.realtime’ may be used
uninitialized in this function [-Wmaybe-uninitialized]
                         _a < _b ? _a : _b;      \
                                      ^
src/systemctl/systemctl.c:933:32: note: ‘next.realtime’ was declared here
                 dual_timestamp next;
                                ^

10 years agoshared/architecture.h: fix preprocessor directives
Daniel Mack [Fri, 21 Feb 2014 10:39:38 +0000 (11:39 +0100)]
shared/architecture.h: fix preprocessor directives

Use 'if defined()', not 'ifdef defined()'. Fixes the following warning.

  CC       src/shared/architecture.lo
In file included from src/shared/architecture.c:24:0:
src/shared/architecture.h:89:17: warning: extra tokens at end of #ifdef
directive [enabled by default]
 #  ifdef defined(WORDS_BIGENDIAN)
                 ^

10 years agoman: xinclude pkg-config note
Zbigniew Jędrzejewski-Szmek [Fri, 21 Feb 2014 03:39:26 +0000 (22:39 -0500)]
man: xinclude pkg-config note

10 years agoman: xinclude --host/--machine
Zbigniew Jędrzejewski-Szmek [Fri, 21 Feb 2014 00:47:03 +0000 (19:47 -0500)]
man: xinclude --host/--machine

As usual, those common options are pushed to the end.

10 years agoman: xinclude --user/--system
Zbigniew Jędrzejewski-Szmek [Thu, 13 Feb 2014 19:40:40 +0000 (14:40 -0500)]
man: xinclude --user/--system

10 years agoman: xinclude --help/--version/--no-pager
Zbigniew Jędrzejewski-Szmek [Thu, 13 Feb 2014 19:33:51 +0000 (14:33 -0500)]
man: xinclude --help/--version/--no-pager

10 years agobuild-sys: add conditionals and regenerate manpage list
Zbigniew Jędrzejewski-Szmek [Wed, 19 Feb 2014 21:04:30 +0000 (16:04 -0500)]
build-sys: add conditionals and regenerate manpage list

The list of man pages is auto generated, based on conditonal='...'
attributes in the man page itself.

10 years agoman: ignore non-manpage files when generating makefile
Zbigniew Jędrzejewski-Szmek [Wed, 19 Feb 2014 21:06:10 +0000 (16:06 -0500)]
man: ignore non-manpage files when generating makefile

10 years agomanager: be more precise in message about restart
Zbigniew Jędrzejewski-Szmek [Thu, 13 Feb 2014 22:10:12 +0000 (17:10 -0500)]
manager: be more precise in message about restart

10 years agoupdate TODO
Lennart Poettering [Fri, 21 Feb 2014 03:04:47 +0000 (04:04 +0100)]
update TODO

10 years agonspawn: when adding a veth interface to a bridge, use the "vb-" rather than "ve-...
Lennart Poettering [Fri, 21 Feb 2014 03:02:12 +0000 (04:02 +0100)]
nspawn: when adding a veth interface to a bridge, use the "vb-" rather than "ve-" interface name prefix

This way we can recognize the interfaces later on to apply different
host-side configuration to them.

10 years agoreplace tabs with spaces in some files
Jason St. John [Thu, 20 Feb 2014 23:09:27 +0000 (18:09 -0500)]
replace tabs with spaces in some files

Files:
* hwdb/60-keyboard.hwdb
* shell-completion/zsh/_systemd-coredumpctl
* src/test/test-helper.h

10 years agocore: Add AppArmor profile switching
Michael Scherer [Thu, 20 Feb 2014 15:19:44 +0000 (16:19 +0100)]
core: Add AppArmor profile switching

This permit to switch to a specific apparmor profile when starting a daemon. This
will result in a non operation if apparmor is disabled.
It also add a new build requirement on libapparmor for using this feature.

10 years agoselinux: Don't attempt to load policy in initramfs if it doesn't exist
Colin Walters [Fri, 21 Feb 2014 02:29:00 +0000 (03:29 +0100)]
selinux: Don't attempt to load policy in initramfs if it doesn't exist

Currently on at least Fedora, SELinux policy does not come in the
initramfs.  systemd will attempt to load *both* in the initramfs and
in the real root.

Now, the selinux_init_load_policy() API has a regular error return
value, as well as an "enforcing" boolean.  To determine enforcing
state, it looks for /etc/selinux/config as well as the presence of
"enforcing=" on the kernel command line.

Ordinarily, neither of those exist in the initramfs, so it will return
"unknown" for enforcing, and systemd will simply ignore the failure to
load policy.

10 years agobuild-sys: don't use -Wcast-align anymore
Lennart Poettering [Fri, 21 Feb 2014 02:13:57 +0000 (03:13 +0100)]
build-sys: don't use -Wcast-align anymore

We do a lot of these casts, and they are all OK, so let's get rid of the
warning, to make things quieter.

http://lists.freedesktop.org/archives/systemd-devel/2014-February/017174.html

10 years agobuild-sys: Fix compilation of nspawn when seccomp is not enabled
Michael Scherer [Thu, 20 Feb 2014 15:07:08 +0000 (16:07 +0100)]
build-sys: Fix compilation of nspawn when seccomp is not enabled

10 years agomissing: simplifications
Lennart Poettering [Fri, 21 Feb 2014 02:07:42 +0000 (03:07 +0100)]
missing: simplifications

10 years agobuild-sys: Add setns() functions if not in the C library.
Holger Schurig [Thu, 20 Feb 2014 13:39:13 +0000 (14:39 +0100)]
build-sys: Add setns() functions if not in the C library.

Debian Stable is still using glibc 2.13, which doesn't provide the setns().
So we detect this and provide a tiny wrapper that issues the setns syscall
towards the kernel.

10 years agohostnamectl: should the sanitized arch, not the native uname() one
Lennart Poettering [Fri, 21 Feb 2014 01:28:54 +0000 (02:28 +0100)]
hostnamectl: should the sanitized arch, not the native uname() one

10 years agocore: add new ConditionArchitecture() that checks the architecture returned by uname...
Lennart Poettering [Fri, 21 Feb 2014 01:06:04 +0000 (02:06 +0100)]
core: add new ConditionArchitecture() that checks the architecture returned by uname()'s machine field.

10 years agoman: networkd/udev - add to [Match] documentation
Tom Gundersen [Thu, 20 Feb 2014 22:26:07 +0000 (23:26 +0100)]
man: networkd/udev - add to [Match] documentation

10 years agonet-util: match - allow globbing
Tom Gundersen [Thu, 20 Feb 2014 21:55:42 +0000 (22:55 +0100)]
net-util: match - allow globbing

For now support globbing for interface name and path.

10 years agobootctl: add bash completion
Thomas Hindoe Paaboel Andersen [Thu, 20 Feb 2014 22:12:02 +0000 (23:12 +0100)]
bootctl: add bash completion

10 years agosd-dhcp-client: prevent timer related memory leaks
Umut Tezduyar Lindskog [Thu, 20 Feb 2014 20:04:03 +0000 (21:04 +0100)]
sd-dhcp-client: prevent timer related memory leaks

10 years agonetworkd: netdev - allow filtering on kernel cmdline, host and virt
Tom Gundersen [Thu, 20 Feb 2014 18:56:50 +0000 (19:56 +0100)]
networkd: netdev - allow filtering on kernel cmdline, host and virt

10 years agonetwork/link: Match - filter on kernel cmdline, host and virt
Tom Gundersen [Thu, 20 Feb 2014 18:39:49 +0000 (19:39 +0100)]
network/link: Match - filter on kernel cmdline, host and virt

10 years agocondition: split out most of condition handling into libsystemd-shard
Tom Gundersen [Thu, 20 Feb 2014 18:26:54 +0000 (19:26 +0100)]
condition: split out most of condition handling into libsystemd-shard

The parts that require linknig to libcap, libselinux and friends stays in libsystemd-core.

10 years agocondition: refactor a bit
Tom Gundersen [Thu, 20 Feb 2014 18:09:49 +0000 (19:09 +0100)]
condition: refactor a bit

No functional change, just move code around a bit to make it simpler to split out
some functions.

10 years agopython: reindent _reader.c
Lennart Poettering [Thu, 20 Feb 2014 18:14:52 +0000 (19:14 +0100)]
python: reindent _reader.c

All files should follow our coding style, and that means 8ch indenting.
Let's correct that.

10 years agomacro: add nice macro for disabling -Wnonnull temporarily
Lennart Poettering [Thu, 20 Feb 2014 17:44:51 +0000 (18:44 +0100)]
macro: add nice macro for disabling -Wnonnull temporarily

10 years agomacro: introduce nice macro for disabling -Wmissing-prototypes warnigs
Lennart Poettering [Thu, 20 Feb 2014 17:35:03 +0000 (18:35 +0100)]
macro: introduce nice macro for disabling -Wmissing-prototypes warnigs

10 years agomacro: introduce a nice macro for disabling -Wformat-nonliteral temporarily
Lennart Poettering [Thu, 20 Feb 2014 17:18:32 +0000 (18:18 +0100)]
macro: introduce a nice macro for disabling -Wformat-nonliteral temporarily

10 years agoupdate TODO
Lennart Poettering [Thu, 20 Feb 2014 17:06:35 +0000 (18:06 +0100)]
update TODO

10 years agoutil: get rid of warnings around assert_cc() macro
Lennart Poettering [Thu, 20 Feb 2014 17:05:56 +0000 (18:05 +0100)]
util: get rid of warnings around assert_cc() macro

Suggested by Holger Schurig.