chiark / gitweb /
8 years agoman: add man page for systemd-backlight@.service
Lennart Poettering [Wed, 14 Aug 2013 01:11:30 +0000 (03:11 +0200)]
man: add man page for systemd-backlight@.service

8 years agobacklight: instead of syspath use sysname for identifying backlight devices
Lennart Poettering [Wed, 14 Aug 2013 00:55:57 +0000 (02:55 +0200)]
backlight: instead of syspath use sysname for identifying backlight devices

This makes the description string of the backlight service a bit nicer.

8 years agobacklight: add minimal tool to save/restore screen brightness across reboots
Lennart Poettering [Tue, 13 Aug 2013 23:57:02 +0000 (01:57 +0200)]
backlight: add minimal tool to save/restore screen brightness across reboots

As many laptops don't save/restore screen brightness across reboots,
let's do this in systemd with a minimal tool, that restores the
brightness as early as possible, and saves it as late as possible. This
will cover consoles and graphical logins, but graphical desktops should
do their own per-user stuff probably.

This only touches firmware brightness controls for now.

8 years agologind: restore logic to kill user processes when session ends
Lennart Poettering [Tue, 13 Aug 2013 15:59:28 +0000 (17:59 +0200)]
logind: restore logic to kill user processes when session ends

8 years agogpt-auto-generator: fix swap unit generation
Lennart Poettering [Tue, 13 Aug 2013 15:48:42 +0000 (17:48 +0200)]
gpt-auto-generator: fix swap unit generation

8 years agoupdate TODO
Lennart Poettering [Tue, 13 Aug 2013 08:32:24 +0000 (10:32 +0200)]
update TODO

8 years agobuild-sys: add two makefile symlinks
Lennart Poettering [Tue, 13 Aug 2013 08:22:02 +0000 (10:22 +0200)]
build-sys: add two makefile symlinks

8 years agogpt-auto-generator: Skip /home mounting if /home is not empty
Lennart Poettering [Tue, 13 Aug 2013 08:21:16 +0000 (10:21 +0200)]
gpt-auto-generator: Skip /home mounting if /home is not empty

8 years agogpt-auto-generator: add basic auto-discovery of GPT partitions
Lennart Poettering [Tue, 13 Aug 2013 08:13:45 +0000 (10:13 +0200)]
gpt-auto-generator: add basic auto-discovery of GPT partitions

This adds a simple generator that is capable of automatically
discovering certain GPT partitions by their type UUID and mount/enable
them. This currently covers swap partitions and /home partitions, but is
expected to grow more features soon.

This currently doesn't handle LUKS encrypted /home.

This enables all swap partitions of type
0657fd6da4ab43c484e50933c84b4f4f, if found.

This mounts the first partition of type 933ac7e12eb44f13b8440e14e2aef915
as /home, if it is found.

8 years agomissing: use btrfs.h instead of defining our own btrfs structures
Lennart Poettering [Tue, 13 Aug 2013 08:12:35 +0000 (10:12 +0200)]
missing: use btrfs.h instead of defining our own btrfs structures

8 years agoTODO: update
Kay Sievers [Mon, 12 Aug 2013 22:00:32 +0000 (00:00 +0200)]
TODO: update

8 years agoshell-completion: tell bash about --state
Zbigniew Jędrzejewski-Szmek [Mon, 12 Aug 2013 16:01:55 +0000 (12:01 -0400)]
shell-completion: tell bash about --state

8 years agologs-show: limit to 3 lines and use dots if not showing full message
Zbigniew Jędrzejewski-Szmek [Sun, 11 Aug 2013 14:56:09 +0000 (10:56 -0400)]
logs-show: limit to 3 lines and use dots if not showing full message

So far, we would show up to 128 bytes from a message, simply
cutting of the rest. With multiline messages, it is quite common
for a message to be longer than that, and this model doesn't really
work anymore.

A new limit is added: up to 3 lines will be shown, unless --full is
used (c.f. first line below).  The limit for bytes is extended to 300
bytes. An ellipsis will always be used, if some form of truncation
occurs. If the tail of the message is cut off, either because of
length or line limit, dots will be shown at the end of the last
line. If this last line is short, the dots will be simply appended. If
the last line is too long for that, it will be ellipsized with dots at
the very end.

Note that the limits are in bytes, not characters, and we suck at
outputting unicode strings (c.f. last three lines below).

Aug 11 10:46:21 fedora python[67]: test message
Aug 11 10:50:47 fedora python[76]: test message word word word word word word word word word word word wor...
Aug 11 10:55:11 fedora python[83]: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx...
Aug 11 11:03:21 fedora python[90]: ąąąąąąąąąąąąąąąąąąąąąąąąąąąąąą...
Aug 11 11:03:53 fedora python[97]: aąąąąąąąąąąąąąąąąąąąąąąąąąąąąąą...
Aug 11 11:25:45 fedora python[121]: aąąąąąąąąąąąąąąąąąąąąąąąąąąąąąąąąąą�...

8 years agologs-show: fix indentation for 2nd and later lines, show lines in full
Zbigniew Jędrzejewski-Szmek [Thu, 8 Aug 2013 12:32:43 +0000 (08:32 -0400)]
logs-show: fix indentation for 2nd and later lines, show lines in full

Now --full will show long fields in full, like it already did
with --all.

8 years agosystemctl: show hint about --full when lines don't fit
Zbigniew Jędrzejewski-Szmek [Sat, 3 Aug 2013 23:38:13 +0000 (19:38 -0400)]
systemctl: show hint about --full when lines don't fit

8 years agosystemctl: add missing newline to --help output
Lennart Poettering [Thu, 8 Aug 2013 19:12:03 +0000 (21:12 +0200)]
systemctl: add missing newline to --help output

8 years agohwdb: map logitech R400 presenter keys from powerpoint hotkeys to sensible KEY_ defin...
Lennart Poettering [Thu, 8 Aug 2013 19:10:52 +0000 (21:10 +0200)]
hwdb: map logitech R400 presenter keys from powerpoint hotkeys to sensible KEY_ definitions

8 years agoupdate TODO
Lennart Poettering [Wed, 7 Aug 2013 10:36:59 +0000 (12:36 +0200)]
update TODO

8 years agofstab-generator: read rd.fstab=on/off switch correctly
WANG Chao [Fri, 9 Aug 2013 09:01:50 +0000 (17:01 +0800)]
fstab-generator: read rd.fstab=on/off switch correctly

8 years agoservice: prohibit Restart= set when Type=oneshot
Maciej Wereski [Fri, 9 Aug 2013 10:54:10 +0000 (12:54 +0200)]
service: prohibit Restart= set when Type=oneshot

8 years agoservice: always unwatch PIDs before forgetting old ones
Lennart Poettering [Fri, 9 Aug 2013 14:40:57 +0000 (16:40 +0200)]
service: always unwatch PIDs before forgetting old ones

8 years agofstab-generator: respect noauto/nofail when adding sysroot mount
WANG Chao [Thu, 8 Aug 2013 07:18:11 +0000 (15:18 +0800)]
fstab-generator: respect noauto/nofail when adding sysroot mount

Currently we don't respect noauto/nofail root mount options (from
rootflags kernel cmdline). We should map these two flags to the
corresponding boolean variable noauto and nofail when calling

8 years agoudevd: respect the log-level set in /etc/udev/udev.conf
Tom Gundersen [Thu, 8 Aug 2013 19:44:02 +0000 (21:44 +0200)]
udevd: respect the log-level set in /etc/udev/udev.conf

A regression introduced when we moved to systemd's logging is that the only
way to adjust the log-level of the udev daemon is via the env var, kernel
commandline or the commandline.

This reintroduces support for specifying this in the configuration file.

8 years agodbus: use _cleanup_free_ instead of freeing ourself
Ronny Chevalier [Thu, 25 Jul 2013 15:36:01 +0000 (17:36 +0200)]
dbus: use _cleanup_free_ instead of freeing ourself

8 years agohwdb: update
Kay Sievers [Wed, 7 Aug 2013 14:49:18 +0000 (16:49 +0200)]
hwdb: update

8 years agoTODO: add hwdb timestamp check
Kay Sievers [Wed, 7 Aug 2013 14:46:53 +0000 (16:46 +0200)]
TODO: add hwdb timestamp check

8 years agoudev: hwdb - try reading modalias for usb before falling back to the composed one
Kay Sievers [Wed, 7 Aug 2013 11:10:01 +0000 (13:10 +0200)]
udev: hwdb - try reading modalias for usb before falling back to the composed one

8 years agohwdb: add GIGABYTE U2442 mute key to the force-release list
Zbigniew Jędrzejewski-Szmek [Mon, 5 Aug 2013 02:46:38 +0000 (22:46 -0400)]
hwdb: add GIGABYTE U2442 mute key to the force-release list

8 years agokeymap: add HP EliteBook 2570p's wlan switch to hwdb
Christoph Junghans [Tue, 6 Aug 2013 01:54:00 +0000 (19:54 -0600)]
keymap: add HP EliteBook 2570p's wlan switch to hwdb

8 years agosystemd: fix segv in snapshot creation
Zbigniew Jędrzejewski-Szmek [Wed, 7 Aug 2013 01:30:34 +0000 (21:30 -0400)]
systemd: fix segv in snapshot creation

8 years agojournald: remove unused variable
Zbigniew Jędrzejewski-Szmek [Wed, 7 Aug 2013 01:02:35 +0000 (21:02 -0400)]
journald: remove unused variable

8 years agoTODO: add weird instance unit enable behavior
Kay Sievers [Tue, 6 Aug 2013 13:03:53 +0000 (15:03 +0200)]
TODO: add weird instance unit enable behavior

8 years agojournal: handle multiline syslog messages
Harald Hoyer [Thu, 1 Aug 2013 10:14:02 +0000 (12:14 +0200)]
journal: handle multiline syslog messages

Since the journal can handle multiple lines just well natively,
and rsyslog can be configured to handle them as well, there is no need
to truncate messages from syslog() after the first newline.


1. Add following four lines to /etc/rsyslog.conf

   $EscapeControlCharactersOnReceive off
   $ActionFileDefaultTemplate RSYSLOG_SysklogdFileFormat
   $SpaceLFOnReceive on
   $DropTrailingLFOnReception off

3. Restart rsyslog
  # service rsyslog restart

4. Compile and run the following program

   #include <stdio.h>
   #include <syslog.h>

   int main()
    syslog(LOG_INFO, "aaa%caaa", '\n');
    return 0;

Actual results:
Below message appears in /var/log/messages.

   Sep  7 19:19:39 localhost test2: aaa

Expected results:
Below message, which worked prior to systemd-journald
appears in /var/log/messages.

   Sep  7 19:19:39 localhost test2: aaa aaa

8 years agoman: fix typo in documentation of systemd-machined
Michael Scherer [Sun, 4 Aug 2013 20:36:42 +0000 (22:36 +0200)]
man: fix typo in documentation of systemd-machined

8 years agoman: wording and grammar updates
Jan Engelhardt [Sun, 4 Aug 2013 23:13:28 +0000 (01:13 +0200)]
man: wording and grammar updates

This includes regularly-submitted corrections to comma setting and
orthographical mishaps that appeared in man/ in recent commits.

8 years agosystemd-delta: Only print colors when on a tty
Daniel Schaal [Fri, 2 Aug 2013 05:59:02 +0000 (07:59 +0200)]
systemd-delta: Only print colors when on a tty

This make systemd-delta follow the behaviour of systemctl
and journalctl.

[zj: unify color query methods between those three programs.]

8 years agogitignore: add test-journal-init
Zbigniew Jędrzejewski-Szmek [Sun, 4 Aug 2013 12:07:55 +0000 (08:07 -0400)]
gitignore: add test-journal-init

8 years agobuild-sys: link with librt if linking with libsd-daemon-int
Zbigniew Jędrzejewski-Szmek [Sat, 3 Aug 2013 14:18:21 +0000 (10:18 -0400)]
build-sys: link with librt if linking with libsd-daemon-int

In fba1ea0 'build: do not link everything with -lrt (and
therefore -pthread)' librt was removed from the list of
libraries. But libsd-daemon-internal also uses symbols from
librt and librt must thus be added everywhere where
libsd-daemon-interal is used, or otherwise linking might

/usr/bin/ld: ./.libs/libudev-core.a(sd-daemon.o): undefined reference to symbol 'mq_getattr@@GLIBC_2.3.4'
/usr/bin/ld: note: 'mq_getattr@@GLIBC_2.3.4' is defined in DSO /lib64/ so try adding it to the linker command line

8 years agozsh completion: add _kernel-install
Daniel Wallace [Sat, 3 Aug 2013 02:38:14 +0000 (21:38 -0500)]
zsh completion: add _kernel-install

8 years agopkg-config: export systemd{system,user}generatordir and catalogdir
Zbigniew Jędrzejewski-Szmek [Fri, 2 Aug 2013 02:58:03 +0000 (22:58 -0400)]
pkg-config: export systemd{system,user}generatordir and catalogdir

We export the location of a bunch of directories this way,
so it makes sense to add those three. Especially catalogdir
is something that we want people to add things to.

Note on the naming: the first two are tied closely to systemd
itself, so I prefixed them with "systemd". The third one is
rather more generic, so no prefix.

8 years agojournal: fix hashmap leak in mmap-cache
George McCollister [Thu, 1 Aug 2013 17:40:01 +0000 (12:40 -0500)]
journal: fix hashmap leak in mmap-cache

hashmap_free() wasn't being called on m->contexts and m->fds resulting
in a leak.

To reproduce do:
while(1) {
sd_journal_open(&j, SD_JOURNAL_LOCAL_ONLY);

Memory usage will increase until OOM.

8 years agotests: add a program for repetitive opening and closing of the journal
Zbigniew Jędrzejewski-Szmek [Fri, 2 Aug 2013 15:01:10 +0000 (11:01 -0400)]
tests: add a program for repetitive opening and closing of the journal

Basically wraps an example provided by George McCollister.
Should help with leaks in the future.

8 years agozsh_completion: add "gdb" verb to coredumpctl
Zbigniew Jędrzejewski-Szmek [Fri, 2 Aug 2013 11:58:56 +0000 (07:58 -0400)]
zsh_completion: add "gdb" verb to coredumpctl

8 years agocoredumpctl: add more debug output
Zbigniew Jędrzejewski-Szmek [Fri, 2 Aug 2013 11:58:26 +0000 (07:58 -0400)]
coredumpctl: add more debug output

It can be quite useful when somebody confuses _PID with COREDUMP_PID :).

8 years agozsh_completion: Speed up noncached perf of _journalctl
William Giokas [Fri, 2 Aug 2013 02:35:16 +0000 (21:35 -0500)]
zsh_completion: Speed up noncached perf of _journalctl

Splitting things unnecessarily at newlines causes tab completion to take
an extremely long time. Also add a note saying that caching is not good
for journalctl's completion.

8 years agozsh_completion: Split out zsh _systemd-tmpfiles
William Giokas [Wed, 31 Jul 2013 00:05:31 +0000 (19:05 -0500)]
zsh_completion: Split out zsh _systemd-tmpfiles

You can choose to have systemd-tmpfiles at configuration time, so only
install the completion for this if configured to do so.

8 years agozsh_completion: Fix journalctl's --boot
William Giokas [Mon, 29 Jul 2013 19:37:35 +0000 (14:37 -0500)]
zsh_completion: Fix journalctl's --boot

Actually displays a list of boot ID's and offsets to the user

8 years agozsh_completion: Remove unused functions
William Giokas [Mon, 29 Jul 2013 19:13:57 +0000 (14:13 -0500)]
zsh_completion: Remove unused functions

8 years agozsh_completion: Split out zsh _machinectl
William Giokas [Mon, 29 Jul 2013 19:13:29 +0000 (14:13 -0500)]
zsh_completion: Split out zsh _machinectl

8 years agozsh_completion: Split out zsh _systemd-analyze
William Giokas [Mon, 29 Jul 2013 19:12:55 +0000 (14:12 -0500)]
zsh_completion: Split out zsh _systemd-analyze

8 years agozsh_completion: Split out zsh _systemd-inhibit
William Giokas [Mon, 29 Jul 2013 19:12:15 +0000 (14:12 -0500)]
zsh_completion: Split out zsh _systemd-inhibit

8 years agozsh_completion: Split out zsh _systemd-nspawn
William Giokas [Wed, 31 Jul 2013 22:11:17 +0000 (17:11 -0500)]
zsh_completion: Split out zsh _systemd-nspawn

Also fix the random lack of completion

8 years agozsh_completion: Split out zsh _udevadm
William Giokas [Mon, 29 Jul 2013 19:10:39 +0000 (14:10 -0500)]
zsh_completion: Split out zsh _udevadm

8 years agozsh_completion: Split out zsh _timedatectl
William Giokas [Mon, 29 Jul 2013 19:09:48 +0000 (14:09 -0500)]
zsh_completion: Split out zsh _timedatectl

8 years agozsh_completion: Split out zsh _coredumpctl
William Giokas [Mon, 29 Jul 2013 19:09:10 +0000 (14:09 -0500)]
zsh_completion: Split out zsh _coredumpctl

8 years agozsh_completion: Split out zsh _localectl
William Giokas [Mon, 29 Jul 2013 19:08:07 +0000 (14:08 -0500)]
zsh_completion: Split out zsh _localectl

8 years agozsh_completion: Split out zsh _journalctl
William Giokas [Mon, 29 Jul 2013 19:07:21 +0000 (14:07 -0500)]
zsh_completion: Split out zsh _journalctl

Re-ordered some of the options and added a few that were missing
previously as well.

8 years agozsh_completion: Split out zsh _hostnamectl
William Giokas [Mon, 29 Jul 2013 19:05:48 +0000 (14:05 -0500)]
zsh_completion: Split out zsh _hostnamectl

8 years agozsh_completion: Split out zsh _loginctl
William Giokas [Mon, 29 Jul 2013 19:03:43 +0000 (14:03 -0500)]
zsh_completion: Split out zsh _loginctl

8 years agozsh_completion: Split out zsh _systemctl
William Giokas [Mon, 29 Jul 2013 19:02:01 +0000 (14:02 -0500)]
zsh_completion: Split out zsh _systemctl

8 years agozsh_completion: fix zsh completion installation
William Giokas [Thu, 25 Jul 2013 05:57:05 +0000 (00:57 -0500)]
zsh_completion: fix zsh completion installation

Moved zsh shell completion to shell-completion/zsh/_systemd for
automake's sake. Also allow users to specify where the files should go

  ./configure --with-zshcompletiondir=/path/to/some/where

and by default going to `$datadir/zsh/site-functions`

8 years ago80-net-name-slot.rules: only rename network interfaces on ACTION=="add"
Harald Hoyer [Thu, 1 Aug 2013 10:31:38 +0000 (12:31 +0200)]
80-net-name-slot.rules: only rename network interfaces on ACTION=="add"

Otherwise systemd-udevd will rename on "change" and "move" events,
resulting in weird renames in combination with biosdevname

systemd-udevd[355]: renamed network interface eth0 to em1
systemd-udevd[355]: renamed network interface eth1 to p3p2
systemd-udevd[357]: renamed network interface eth0 to p3p1
systemd-udevd[429]: renamed network interface p3p2 to ens3f1
systemd-udevd[428]: renamed network interface p3p1 to ens3f0
systemd-udevd[426]: renamed network interface em1 to enp63s0


systemd-udevd[356]: renamed network interface eth0 to em1
systemd-udevd[356]: renamed network interface eth0 to p3p1
systemd-udevd[420]: renamed network interface p3p1 to ens3f0
systemd-udevd[418]: renamed network interface em1 to enp63s0
systemd-udevd[421]: renamed network interface eth1 to p3p1

8 years agoAdd /usr/share/keymaps to localectl supported locations.
Michał Górny [Wed, 31 Jul 2013 21:12:17 +0000 (23:12 +0200)]
Add /usr/share/keymaps to localectl supported locations.

This is the standard upstream location where kbd installs keymaps.

8 years agobash-completion: simplify udevadm completion
Dave Reisner [Tue, 30 Jul 2013 17:00:00 +0000 (13:00 -0400)]
bash-completion: simplify udevadm completion

The AA is unnecessary and only adds needless complexity. Replace it
with a case statement instead of repeatedly calling __contains_word to
overglorify string equalities.

8 years agobash-completion: use a better definition of __contains_word
Dave Reisner [Tue, 30 Jul 2013 16:46:23 +0000 (12:46 -0400)]
bash-completion: use a better definition of __contains_word

- scope the iterator var
- use the correct, quoted, non-expansion prone positional parameter
- prevent expansion on RHS of comparison
- remove unneeded explicit returns.

This really should be defined only once...

8 years agotest-unit-file: return error without dumping core on permission error
Christian Hesse [Tue, 30 Jul 2013 13:07:20 +0000 (09:07 -0400)]
test-unit-file: return error without dumping core on permission error

8 years agotest-fileio: use random name for written file
Zbigniew Jędrzejewski-Szmek [Mon, 29 Jul 2013 23:49:55 +0000 (19:49 -0400)]
test-fileio: use random name for written file

If two instances of test-fileio were run in parallel,
they could fail when trying to write the same file.
This predictable name in /tmp/ wasn't actually a security
issue, because write_env_file would not follow symlinks,
so this could be an issue only when running tests in

8 years agosimplify bash completion for kernel-install
Harald Hoyer [Mon, 29 Jul 2013 15:48:29 +0000 (17:48 +0200)]
simplify bash completion for kernel-install

8 years agoudev-rules: report rule parsing errors from get_key
Dave Reisner [Mon, 29 Jul 2013 22:18:43 +0000 (18:18 -0400)]
udev-rules: report rule parsing errors from get_key

8 years agobuild-sys: support old glibc versions without clock_gettime()
Kay Sievers [Tue, 30 Jul 2013 01:38:55 +0000 (03:38 +0200)]
build-sys: support old glibc versions without clock_gettime()

8 years agologind: make sure login sessions are terminated with SIGHUP
Lennart Poettering [Tue, 30 Jul 2013 00:51:12 +0000 (02:51 +0200)]
logind: make sure login sessions are terminated with SIGHUP

bash ignores SIGTERM, and can only be terminated cleanly via SIGHUP.
Hence make sure that we the scope unit for the session is created with
SendSIGHUP enabled.

8 years agoupdate TODO
Lennart Poettering [Tue, 30 Jul 2013 00:51:07 +0000 (02:51 +0200)]
update TODO

8 years agocore: make sure scope attributes survive a reload
Lennart Poettering [Tue, 30 Jul 2013 00:50:44 +0000 (02:50 +0200)]
core: make sure scope attributes survive a reload

8 years agocore: open up SendSIGHUP property for transient units
Lennart Poettering [Tue, 30 Jul 2013 00:28:22 +0000 (02:28 +0200)]
core: open up SendSIGHUP property for transient units

8 years agobuild: do not link everything with -lrt (and therefore -pthread)
Shawn Landden [Mon, 22 Jul 2013 03:57:35 +0000 (20:57 -0700)]
build: do not link everything with -lrt (and therefore -pthread)

8 years agodo not pass-along the environment from the kernel or initrd
Kay Sievers [Tue, 30 Jul 2013 00:02:45 +0000 (02:02 +0200)]
do not pass-along the environment from the kernel or initrd

8 years agocore: optionally send SIGHUP in addition to the configured kill signal
Lennart Poettering [Mon, 29 Jul 2013 23:54:59 +0000 (01:54 +0200)]
core: optionally send SIGHUP in addition to the configured kill signal

This is useful to fake session ends for processes like shells.

8 years agoupdate TODO
Lennart Poettering [Mon, 29 Jul 2013 21:08:31 +0000 (23:08 +0200)]
update TODO

8 years agoupdate TODO
Lennart Poettering [Mon, 29 Jul 2013 16:43:27 +0000 (18:43 +0200)]
update TODO

8 years agoupdate TODO
Lennart Poettering [Mon, 29 Jul 2013 14:39:22 +0000 (16:39 +0200)]
update TODO

8 years agoman: reowrk list of documented unit search paths
Lennart Poettering [Wed, 24 Jul 2013 15:31:44 +0000 (17:31 +0200)]
man: reowrk list of documented unit search paths

The generator paths are internal implementation details, they should not
be documented explicitly.

We should document where private user units are found however.

8 years agoman: there is no session mode, only user mode
Lennart Poettering [Wed, 24 Jul 2013 15:31:37 +0000 (17:31 +0200)]
man: there is no session mode, only user mode

8 years agoman: link up scope+slice units from systemd.unit(5)
Lennart Poettering [Wed, 24 Jul 2013 15:31:17 +0000 (17:31 +0200)]
man: link up scope+slice units from systemd.unit(5)

8 years agoman: systemd.unit: fix volatile path
Brandon Philips [Sat, 27 Jul 2013 16:49:58 +0000 (09:49 -0700)]
man: systemd.unit: fix volatile path

The volatile path was '/run/systemd/systemd' when it should be
'/run/systemd/system'. Fix.

8 years agoTODO: remove kernel env var importing fix
Kay Sievers [Sat, 27 Jul 2013 17:54:29 +0000 (19:54 +0200)]
TODO: remove kernel env var importing fix

8 years agosystemctl.8: fix typo in SEE ALSO
Dave Reisner [Sat, 27 Jul 2013 00:01:33 +0000 (20:01 -0400)]
systemctl.8: fix typo in SEE ALSO

8 years agologind: update the session state file before we send out the CreateSession() reply
Lennart Poettering [Fri, 26 Jul 2013 16:59:46 +0000 (18:59 +0200)]
logind: update the session state file before we send out the CreateSession() reply

8 years agorework systemd's own process environment handling/passing
Kay Sievers [Fri, 26 Jul 2013 03:22:22 +0000 (05:22 +0200)]
rework systemd's own process environment handling/passing

Stop importing non-sensical kernel-exported variables. All
parameters in the kernel command line are exported to the
initial environment of PID1, but suppressed if they are
recognized by kernel built-in code. The EFI booted kernel
will add further kernel-internal things which do not belong
into userspace.

The passed original environ data of the process is not touched
and preserved across re-execution, to allow external reading of
/proc/self/environ for process properties like container*=.

8 years agojournalctl: use _COMM= match for scripts
Zbigniew Jędrzejewski-Szmek [Fri, 19 Jul 2013 08:02:50 +0000 (04:02 -0400)]
journalctl: use _COMM= match for scripts

In case of scripts, _EXE is set to the interpreter name, and
_COMM is set based on the file name. Add a match for _COMM,
and _EXE if the interpreter is not a link (e.g. for yum,
the interpreter is /usr/bin/python, but it is a link to
/usr/bin/python2, which in turn is a link to /usr/bin/python2.7,
at least on Fedora, so we end up with _EXE=/usr/bin/python2.7).
I don't think that such link chasing makes sense, because
the final _EXE name is more likely to change.

8 years agologind: update state file after generating the session fifo, not before
Lennart Poettering [Fri, 26 Jul 2013 15:32:19 +0000 (17:32 +0200)]
logind: update state file after generating the session fifo, not before

8 years agosystemd-python: fix gcc warning
Zbigniew Jędrzejewski-Szmek [Fri, 26 Jul 2013 15:02:27 +0000 (11:02 -0400)]
systemd-python: fix gcc warning

src/python-systemd/_reader.c: In function Reader_get_catalog:
src/python-systemd/_reader.c:912:53: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
             assert(mid_len > l);

8 years agobuild-sys: use pkg-config for python compilation flags
Zbigniew Jędrzejewski-Szmek [Fri, 26 Jul 2013 14:08:15 +0000 (10:08 -0400)]
build-sys: use pkg-config for python compilation flags

Python 2.7, and 3.2 and higher support querying compilation
flags through pkg-config. This makes python support follow
rules similar to various other optional compilation-time
libraries. New flags are called PYTHON_DEVEL_CFLAGS and
PYTHON_DEVEL_LIBS, because PYTHON (without _DEVEL), is
already used for the python binary name, and things would
be confusing if the same prefix was used for two things.
configure has --disable-python-devel to disable python modules.

One advantage is that CFLAGS for modules gets smaller:
- -I/usr/include/python3.3m -I/usr/include/python3.3m -Wno-unused-result -DDYNAMIC_ANNOTATIONS_ENABLED=1 -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -grecord-gcc-switches -m64 -mtune=generic -D_GNU_SOURCE -fPIC -fwrapv
+ -I/usr/include/python3.3m
as does LIBS:
- -lpthread -ldl -lutil -lm -lpython3.3m
+ -lpython3.3m

Support for Python 2.6 is removed, but can be easily
restored by using
PYTHON_DEVEL_CFLAGS="$(python2.6-config --cflags)",
etc., as ./configure parameters.

8 years agosystemd-python: use modern C
Zbigniew Jędrzejewski-Szmek [Thu, 25 Jul 2013 22:55:08 +0000 (18:55 -0400)]
systemd-python: use modern C

8 years agosystemctl: rearrange --help output a bit
Lennart Poettering [Fri, 26 Jul 2013 15:11:41 +0000 (17:11 +0200)]
systemctl: rearrange --help output a bit

8 years agosystemctl: remove "load" command
Lennart Poettering [Fri, 26 Jul 2013 15:05:43 +0000 (17:05 +0200)]
systemctl: remove "load" command

"systemctl load" has always been racy since the GC could hit any time,
before making use of the loaded unit. Very recent systemd will run GC
immeidately after all unit state changes which has the effect that the
the effect of "systemctl load" is completely gone now, so let's remove
the support for it in "systemctl" for good.

8 years agosystemctl: move set-log-level to systemd-analyze
Lennart Poettering [Fri, 26 Jul 2013 14:59:55 +0000 (16:59 +0200)]
systemctl: move set-log-level to systemd-analyze

"systemctl set-log-level" is a command for analysis and tracing hence
"systemd-analyze" should be the better home for it, thus allowing us to
make the overly large "systemctl" a bit smaller.

8 years agosystemctl: move "dump" command from systemctl to systemd-analyze
Lennart Poettering [Fri, 26 Jul 2013 14:34:52 +0000 (16:34 +0200)]
systemctl: move "dump" command from systemctl to systemd-analyze

It's an analysis command and its format is explicitly not covered by any
stability guarantees, hence move away from systemctl and into
systemd-analyze, minimizing the already large interface of systemctl a

This patch also adds auto-paging to the various systemd-analyze commands
where that makes sense

8 years agocore: allow setting RemainAfterExit= for transient services
Lennart Poettering [Fri, 26 Jul 2013 14:09:25 +0000 (16:09 +0200)]
core: allow setting RemainAfterExit= for transient services

8 years agoshared: split mkdir_*() and mkdir_*_label() from each other
Kay Sievers [Fri, 26 Jul 2013 01:34:18 +0000 (03:34 +0200)]
shared: split mkdir_*() and mkdir_*_label() from each other

Avoid pulling-in selinux for tools which just create directories
but not need to fix the selinux label.

8 years agoshell-completion: add kernel-install
Kay Sievers [Fri, 26 Jul 2013 00:33:51 +0000 (02:33 +0200)]
shell-completion: add kernel-install