chiark / gitweb /
8 years agodocs: install README files into /var/log and 7etc/rc.d/init.d
Lennart Poettering [Wed, 17 Oct 2012 18:03:21 +0000 (20:03 +0200)]
docs: install README files into /var/log and 7etc/rc.d/init.d

On systemd systems seasoned admins might be surprised to see that the
init scripts and log files are gone. To ease the transition let's place
some README files there, that hopefully help clearing up the situation.

8 years agologinctl: show pager also for status command
Lennart Poettering [Wed, 17 Oct 2012 01:15:16 +0000 (03:15 +0200)]
loginctl: show pager also for status command

8 years agotimedatectl: introduce new command line client for timedated
Lennart Poettering [Wed, 17 Oct 2012 00:50:09 +0000 (02:50 +0200)]
timedatectl: introduce new command line client for timedated

Much like logind has a client in loginctl, and journald in journalctl
introduce timedatectl, to change the system time (incl. RTC), timezones
and related settings.

8 years agoProperly handle device aliases used as dependencies
Eelco Dolstra [Wed, 10 Oct 2012 20:43:10 +0000 (16:43 -0400)]
Properly handle device aliases used as dependencies

If a device unit has aliases defined in udev rules, and there are
other units that depend on that alias, as in


then systemd will fail the start the alias, and any dependent units
will time out.  See

This is because unit_add_name() in device_add_escaped_name() will
return EEXIST.

The solution taken here is to call device_update_unit() on the alias
name.  Thus if a unit with the alias name already exists, we reuse it;
otherwise a new unit is created.  Creating multiple units for a single
device is perhaps suboptimal, but it's consistent with the treatment
of udev symlinks in device_process_new_device().

8 years agojournalctl: properly track rotated journals in follow mode
Lennart Poettering [Tue, 16 Oct 2012 21:00:04 +0000 (23:00 +0200)]
journalctl: properly track rotated journals in follow mode

8 years agosd-journal: fix bad memory access
Lennart Poettering [Tue, 16 Oct 2012 20:59:28 +0000 (22:59 +0200)]
sd-journal: fix bad memory access

8 years agoupdate TODO
Lennart Poettering [Tue, 16 Oct 2012 20:59:14 +0000 (22:59 +0200)]
update TODO

8 years agojournal: implement time-based rotation/vacuuming
Lennart Poettering [Tue, 16 Oct 2012 20:58:07 +0000 (22:58 +0200)]
journal: implement time-based rotation/vacuuming

This also enables time-based rotation (but not vacuuming) after 1month,
so that not more one month of journal is lost at a time per vacuuming.

8 years agojournal: sort data items of entries by offset
Lennart Poettering [Tue, 16 Oct 2012 19:40:48 +0000 (21:40 +0200)]
journal: sort data items of entries by offset

This should slightly optimize disk access patterns on rotating disks for
simple readers.

8 years agomount: don't try to initialize extra deps for mount units before initializing their...
Lennart Poettering [Tue, 16 Oct 2012 17:50:26 +0000 (19:50 +0200)]
mount: don't try to initialize extra deps for mount units before initializing their basic fields

Under some circumstances this could lead to a segfault since we we
half-initialized a mount unit, then tried to hook it into the network of
things and while doing that recursively ended up looking at our
half-initialized mount unit again assuming it was fully initialized.

8 years agologind: only release logind session from the PAM module if the same module instance...
Lennart Poettering [Tue, 16 Oct 2012 17:21:21 +0000 (19:21 +0200)]
logind: only release logind session from the PAM module if the same module instance actually created it

8 years agoid128: introduce new SD_ID128_CONST_STR() macro
Lennart Poettering [Tue, 16 Oct 2012 15:02:51 +0000 (17:02 +0200)]
id128: introduce new SD_ID128_CONST_STR() macro

8 years agojournal: don't export MESSAGE_ID() macro
Lennart Poettering [Tue, 16 Oct 2012 14:26:41 +0000 (16:26 +0200)]
journal: don't export MESSAGE_ID() macro

If it is exported it would need to be prefixed, but since we need it
exclusively internally so far, simply move it to an internal header.

8 years agoservice: Heuristically determine whether SysV scripts support reload
Michael Stapelberg [Tue, 16 Oct 2012 08:14:03 +0000 (10:14 +0200)]
service: Heuristically determine whether SysV scripts support reload

This commit checks for a usage line which contains [{|]reload[|}"] (to
not errnously match force-reload).

Heuristics like this suck, but it solves a real problem and there
appears to be no better way...

8 years agojournal: Set the last_unused pointer correctly when attaching an unused window
Colin Guthrie [Tue, 16 Oct 2012 11:03:37 +0000 (12:03 +0100)]
journal: Set the last_unused pointer correctly when attaching an unused window

It seems the previous code was copy/pasted from context_detach_window()
but not updated.

8 years agojournal: Properly track the number of allocated windows.
Colin Guthrie [Tue, 16 Oct 2012 11:03:36 +0000 (12:03 +0100)]
journal: Properly track the number of allocated windows.

Checks were already in place to make sure that the number of
windows was limited to 64, but the count was never incremented
or decremented.

8 years agosystemctl: append .service when unit does not have valid suffix
Lukas Nykryn [Tue, 16 Oct 2012 09:01:29 +0000 (11:01 +0200)]
systemctl: append .service when unit does not have valid suffix

systemctl status a and systemctl status a.service lead to same output but
systemctl status a.b and systemctl status a.b.service do not.

8 years agojournalctl: add --unit=/-u to match by unit name
Lennart Poettering [Tue, 16 Oct 2012 00:59:27 +0000 (02:59 +0200)]
journalctl: add --unit=/-u to match by unit name

This applies unit_name_mangle() to the specified unit names and hence
can handle weird characters nicely and will add unit suffixes as

8 years Add a --without-python option.
Wulf C. Krueger [Sat, 6 Oct 2012 09:37:44 +0000 (11:37 +0200)] Add a --without-python option.

Being able to be explicit about Python support (in addition to the
default of auto-detecting it) and acting upon the result, specifying
it as an option gains us more control about both dependencies and
the resulting build.

Furthermore, relying purely on auto-detection can lead to problems for
source-based distros. E. g. systemd being built before *both* 32-bit &
64-bit ABIs are installed will lead to build failures as systemd's
build system will pick up either 32-/64-bit Python, conclude both are
available and fail if that's not the case.

8 years agocore/swap.c: Do not add to swap units.
Thomas Bächler [Fri, 5 Oct 2012 23:11:53 +0000 (01:11 +0200)]
core/swap.c: Do not add to swap units.

The fstab generator adds by default, and when creating
a custom .swap unit, you can also add to the unit.

However, it is impossible to not have this ordering dependency right now.
Virtually all existing setups likely use the fstab generator, so this
change is unlikely to break anything.

8 years agocore: Refuse to run a user instance when the system hasn't been booted with systemd.
Thomas Bächler [Fri, 5 Oct 2012 23:11:52 +0000 (01:11 +0200)]
core: Refuse to run a user instance when the system hasn't been booted with systemd.

Running as a user instance won't work at all if systemd isn't running as system
manager, so refuse to start in that case.

8 years agoselinux: selabel_lookup_raw can return ENOENT and be a non failure mode.
Daniel J Walsh [Thu, 11 Oct 2012 21:08:47 +0000 (17:08 -0400)]
selinux: selabel_lookup_raw can return ENOENT and be a non failure mode.

8 years agoSELinux patch still broken, in that we are not checking the correct source context.
Daniel J Walsh [Thu, 11 Oct 2012 21:06:46 +0000 (17:06 -0400)]
SELinux patch still broken, in that we are not checking the correct source context.

This patch does the dbus calls correctly.

8 years agocore: allow Type=oneshot services to have ExecReload
Mantas Mikulėnas [Wed, 10 Oct 2012 17:18:42 +0000 (20:18 +0300)]
core: allow Type=oneshot services to have ExecReload

Use cases:

 * iptables.service – atomically reload rules without having to flush
   them beforehand (which may leave the system insecure if reload fails)

 * rpc-nfsd.service – reexport filesystems after /etc/exports update
   without completely stopping and restarting nfsd

(In both cases, the actual service is provided by a kernel module and
does not have any associated user-space processes, thus Type=oneshot.)

8 years agobuild-sys: check for name_to_handle_at declaration instead of its definition
Michael Olbrich [Tue, 9 Oct 2012 12:58:51 +0000 (14:58 +0200)]
build-sys: check for name_to_handle_at declaration instead of its definition

AC_CHECK_FUNCS may be successful, even though name_to_handle_at and
'struct file_handle'  are not available.

8 years agoman: Fix typo in udevadm documentation
Ferenc Wágner [Mon, 15 Oct 2012 17:38:21 +0000 (19:38 +0200)]
man: Fix typo in udevadm documentation

8 years agojournalctl: harmonise in-stream comments
Lennart Poettering [Mon, 15 Oct 2012 23:09:09 +0000 (01:09 +0200)]
journalctl: harmonise in-stream comments

From now on, always use ANSI-SQL-style comments in log streams, i.e.
prefix with --. We also suffix things with this, just to be nice...

8 years agoutil: properly handle -1 timespec/timeval
Lennart Poettering [Mon, 15 Oct 2012 23:02:36 +0000 (01:02 +0200)]
util: properly handle -1 timespec/timeval

8 years agoupdate TODO
Lennart Poettering [Mon, 15 Oct 2012 23:01:55 +0000 (01:01 +0200)]
update TODO

8 years agoReword sentences that contain psuedo-English "resp."
Andrew Eikum [Mon, 15 Oct 2012 18:59:12 +0000 (13:59 -0500)]
Reword sentences that contain psuedo-English "resp."

As you likely know, Arch Linux is in the process of moving to systemd.
So I was reading through the various systemd docs and quickly became
baffled by this new abbreviation "resp.", which I've never seen before
in my English-mother-tongue life.

Some quick Googling turned up a reference:

I guess it's a literal translation of the German "Beziehungsweise", but
English doesn't work the same way. The word "respectively" is used
exclusively to provide an ordering connection between two lists. E.g.
"the prefixes k, M, and G refer to kilo-, mega-, and giga-,
respectively." It is also never abbreviated to "resp." So the sentence
"Sets the default output resp. error output for all services and
sockets" makes no sense to a natural English speaker.

This patch removes all instances of "resp." in the man pages and
replaces them with sentences which are much more clear and, hopefully,
grammatically valid. In almost all instances, it was simply replacing
"resp." with "or," which the original author (Lennart?) could probably
just do in the future.

The only other instances of "resp." are in the src/ subtree, which I
don't feel privileged to correct.

Signed-off-by: Andrew Eikum <>
8 years agojournalctl: skip informational messages in export/json modes
Zbigniew Jędrzejewski-Szmek [Mon, 15 Oct 2012 16:14:09 +0000 (16:14 +0000)]
journalctl: skip informational messages in export/json modes

8 years agosystemctl: don't mangle name when it is a path
Lukas Nykryn [Mon, 15 Oct 2012 12:51:27 +0000 (14:51 +0200)]
systemctl: don't mangle name when it is a path

systemctl enable, disable, ... can also accept full path and in this case
we don't need to alter it.

8 years agoupdate TODO
Lennart Poettering [Mon, 15 Oct 2012 14:28:04 +0000 (16:28 +0200)]
update TODO

8 years agojournalctl: make --follow really work
Zbigniew Jędrzejewski-Szmek [Sun, 14 Oct 2012 15:29:38 +0000 (15:29 +0000)]
journalctl: make --follow really work

8 years agojournalctl: make --follow work again
Zbigniew Jędrzejewski-Szmek [Sat, 13 Oct 2012 19:21:07 +0000 (19:21 +0000)]
journalctl: make --follow work again

Stopped working after cfbc22ab 'journalctl: implement --since= and
--until for filtering by time'.

8 years agojournal-verify: get rid of an unused variable
Zbigniew Jędrzejewski-Szmek [Sat, 13 Oct 2012 17:50:53 +0000 (17:50 +0000)]
journal-verify: get rid of an unused variable

When compiling without gcrypt, gcc emits an annoying warning.

8 years agojournald: add missing include
Zbigniew Jędrzejewski-Szmek [Sat, 13 Oct 2012 16:09:09 +0000 (16:09 +0000)]
journald: add missing include

./src/journal/journald.h:123:114: warning: ‘struct ucred’ declared inside parameter list [enabled by default]

8 years agobuild-sys: always distribute file
Zbigniew Jędrzejewski-Szmek [Sat, 13 Oct 2012 11:13:50 +0000 (13:13 +0200)]
build-sys: always distribute file

8 years agojournal-gatewayd: rename variables to avoid -Wshadow warning
Zbigniew Jędrzejewski-Szmek [Sat, 13 Oct 2012 11:08:17 +0000 (13:08 +0200)]
journal-gatewayd: rename variables to avoid -Wshadow warning

8 years agojournald: properly update message size after stripping the identifier
Zbigniew Jędrzejewski-Szmek [Sat, 13 Oct 2012 09:31:54 +0000 (11:31 +0200)]
journald: properly update message size after stripping the identifier

Valgrind says:
==29176== Conditional jump or move depends on uninitialised value(s)
==29176==    at 0x412A85: cunescape_length_with_prefix (util.c:1565)
==29176==    by 0x40B351: dev_kmsg_record (journald-kmsg.c:301)
==29176==    by 0x40B653: server_read_dev_kmsg (journald-kmsg.c:347)
==29176==    by 0x40B701: server_flush_dev_kmsg (journald-kmsg.c:365)
==29176==    by 0x409DE7: main (journald.c:1535)

8 years agojournal: fix build in VALGRIND compatibility mode
Zbigniew Jędrzejewski-Szmek [Fri, 12 Oct 2012 12:56:19 +0000 (12:56 +0000)]
journal: fix build in VALGRIND compatibility mode

8 years agolog: introduce a macro to format message id
Zbigniew Jędrzejewski-Szmek [Mon, 8 Oct 2012 17:02:30 +0000 (17:02 +0000)]
log: introduce a macro to format message id

The MESSAGE_ID=... stanza will appear in countless number of places.
It is just too long to write it out in full each time.

Incidentally, this also fixes a typo of MESSSAGE is three places.

8 years agobuild-sys: check - fix typo
Kay Sievers [Thu, 11 Oct 2012 20:31:23 +0000 (22:31 +0200)]
build-sys: check - fix typo

8 years agoudev: path_id - add scm support
Sebastian Ott [Tue, 9 Oct 2012 13:57:54 +0000 (15:57 +0200)]
udev: path_id - add scm support

Add support for scm block devices. Introduced here:;a=commitdiff;h=f30664e2

Signed-off-by: Sebastian Ott <>
8 years agojournalctl: implement --since= and --until for filtering by time
Lennart Poettering [Thu, 11 Oct 2012 14:42:46 +0000 (16:42 +0200)]
journalctl: implement --since= and --until for filtering by time

8 years agorules: Fix persistent input symlinks for interface 00
Martin Pitt [Thu, 11 Oct 2012 06:21:17 +0000 (08:21 +0200)]
rules: Fix persistent input symlinks for interface 00

Commits 5e9eb156c and 32567f8 introduced persistent symlinks for input devices
with more than one interface. However, this does not ensure stability for the
"default" interface, i. e. with interface number 00 or a nonexisting one. If a
device with a higher interface number appears first, it'll claim the symlink
name without an interface number, and the "interface 00" device won't get any.

Fix this by creating the default symlink only for interface 00 or a nonexisting
one, so that we properly partition the two cases over the two rules.

8 years agojournal: support epxorting the journal in a format suitable for text/event-stream
Lennart Poettering [Thu, 11 Oct 2012 00:37:10 +0000 (02:37 +0200)]
journal: support epxorting the journal in a format suitable for text/event-stream

8 years agojournal: take scroll events in the entire white box in browse.html
Lennart Poettering [Wed, 10 Oct 2012 23:57:36 +0000 (01:57 +0200)]
journal: take scroll events in the entire white box in browse.html

8 years agojournal: make buttons bigger in browse.html
Lennart Poettering [Wed, 10 Oct 2012 23:52:35 +0000 (01:52 +0200)]
journal: make buttons bigger in browse.html

8 years agojournal: use localstorage instead of cookies in browse.html and store where the curre...
Lennart Poettering [Wed, 10 Oct 2012 22:38:20 +0000 (00:38 +0200)]
journal: use localstorage instead of cookies in browse.html and store where the current position

8 years agojournal: move buttons to the center in browse.html
Lennart Poettering [Wed, 10 Oct 2012 22:04:54 +0000 (00:04 +0200)]
journal: move buttons to the center in browse.html

8 years agojournal: add mousewheel scrolling to browse.html
Kay Sievers [Wed, 10 Oct 2012 21:59:29 +0000 (23:59 +0200)]
journal: add mousewheel scrolling to browse.html

8 years agojournal: add keyboard navigation to browse.html
Lennart Poettering [Wed, 10 Oct 2012 21:49:21 +0000 (23:49 +0200)]
journal: add keyboard navigation to browse.html

8 years agojournal: properly HTML escape more output in browse.html
Lennart Poettering [Wed, 10 Oct 2012 21:14:32 +0000 (23:14 +0200)]
journal: properly HTML escape more output in browse.html

8 years agojournal: properly escape HTML entities in browse.html
Mantas Mikulėnas [Wed, 10 Oct 2012 21:00:25 +0000 (23:00 +0200)]
journal: properly escape HTML entities in browse.html

8 years agojournal: when browsing the journal via browse.html allow clicking on entries to show...
Lennart Poettering [Wed, 10 Oct 2012 20:39:45 +0000 (22:39 +0200)]
journal: when browsing the journal via browse.html allow clicking on entries to show their details

8 years agojournal: drop path name from cursor strings
Lennart Poettering [Wed, 10 Oct 2012 20:24:28 +0000 (22:24 +0200)]
journal: drop path name from cursor strings

we had this mostly for debugging purposes and it was ignored when
parsing anyway, so let's get rid of it

8 years agoupdate TODO
Lennart Poettering [Wed, 10 Oct 2012 20:23:06 +0000 (22:23 +0200)]
update TODO

8 years agokeymap: Fix scan codes > 0x7FFFFFFF
Martin Pitt [Wed, 10 Oct 2012 16:56:01 +0000 (18:56 +0200)]
keymap: Fix scan codes > 0x7FFFFFFF

According to Linux drivers/input/evdev.c, scan codes are unsigned, not int nor

Thanks to Dejan Tosovic <> for reporting this!

8 years agogatewayd: beef up browse.html a bit
Lennart Poettering [Wed, 10 Oct 2012 00:41:27 +0000 (02:41 +0200)]
gatewayd: beef up browse.html a bit

8 years agojournal: fix seeking backwards in gateway daemon
Lennart Poettering [Tue, 9 Oct 2012 23:57:06 +0000 (01:57 +0200)]
journal: fix seeking backwards in gateway daemon

8 years agojournal: update comments a bit
Lennart Poettering [Tue, 9 Oct 2012 23:37:38 +0000 (01:37 +0200)]
journal: update comments a bit

8 years agojournalctl: properly parse --cursor= long getopt in addition to short -c
Lennart Poettering [Tue, 9 Oct 2012 23:35:24 +0000 (01:35 +0200)]
journalctl: properly parse --cursor= long getopt in addition to short -c

8 years agojournal: make sure sd_journal_seek_cursor() seeks to the specified entry if it exists...
Lennart Poettering [Tue, 9 Oct 2012 23:34:27 +0000 (01:34 +0200)]
journal: make sure sd_journal_seek_cursor() seeks to the specified entry if it exists, not one after it

8 years agodetect-virt: install with fs caps by default to allow unprivileged access
Lennart Poettering [Tue, 9 Oct 2012 20:25:32 +0000 (22:25 +0200)]
detect-virt: install with fs caps by default to allow unprivileged access

8 years agojournal: implement follow mode for HTTP GET in gatewayd
Lennart Poettering [Mon, 8 Oct 2012 23:31:27 +0000 (01:31 +0200)]
journal: implement follow mode for HTTP GET in gatewayd

8 years agojournal: add matching support to gatewayd
Lennart Poettering [Mon, 8 Oct 2012 23:17:29 +0000 (01:17 +0200)]
journal: add matching support to gatewayd

8 years agologind: expose missing signals in Session bus objects
Lennart Poettering [Mon, 8 Oct 2012 22:43:18 +0000 (00:43 +0200)]
logind: expose missing signals in Session bus objects

8 years agoudev: fix
Kay Sievers [Mon, 8 Oct 2012 22:49:15 +0000 (00:49 +0200)]
udev: fix

8 years agoudev: support multiple entries for ENV{SYSTEMD_ALIAS} and ENV{SYSTEM_WANTS}
Kay Sievers [Mon, 8 Oct 2012 22:16:50 +0000 (00:16 +0200)]
udev: support multiple entries for ENV{SYSTEMD_ALIAS} and ENV{SYSTEM_WANTS}

8 years agoREADME: update
Kay Sievers [Mon, 8 Oct 2012 20:02:01 +0000 (22:02 +0200)]
README: update

8 years agoudev: remove SYMLINK "unique" option
Kay Sievers [Mon, 8 Oct 2012 19:54:39 +0000 (21:54 +0200)]
udev: remove SYMLINK "unique" option

8 years agoudev: fix test
Kay Sievers [Mon, 8 Oct 2012 18:50:20 +0000 (20:50 +0200)]
udev: fix test

8 years agoupdate TODO
Lennart Poettering [Mon, 8 Oct 2012 18:07:50 +0000 (20:07 +0200)]
update TODO

8 years agoshared: fix broken string_is_safe
Martin Mikkelsen [Thu, 4 Oct 2012 12:27:29 +0000 (14:27 +0200)]
shared: fix broken string_is_safe

8 years agodbus-manager: check return of unit_dbus_path
Lukas Nykryn [Thu, 4 Oct 2012 07:49:04 +0000 (09:49 +0200)]
dbus-manager: check return of unit_dbus_path

Reply of dbus_message_new_method_return was check twice and
path from unit_dbus_path was not.

8 years agoShrink TODO
Zbigniew Jędrzejewski-Szmek [Sun, 7 Oct 2012 07:25:23 +0000 (09:25 +0200)]
Shrink TODO

8 years agoudev: allow firmware requests to bypass the dependency tracking
Kay Sievers [Sun, 7 Oct 2012 16:21:38 +0000 (18:21 +0200)]
udev: allow firmware requests to bypass the dependency tracking

The removal of the TIMEOUT= handling in udevd put firmware requests into the
devpath parent/child dependency tracking. Drivers which block in module_init()
asking userspace for firmware ran into a 30 sec device timeout.

The whole firmware loading willl hopefully move into the kernel and
the fragile-since-day-one fake async driver-core device dance involving
udev can be retired:;a=commit;h=abb139e75c2cdbb955e840d6331cb5863e409d0e

8 years agosysctl: avoiding exiting with error on -EEXIST
Dave Reisner [Sat, 6 Oct 2012 20:32:17 +0000 (16:32 -0400)]
sysctl: avoiding exiting with error on -EEXIST

If the final key in any sysctl.d file is a duplicate, systemd-sysctl
will exit with an error (and no explaination why). Ignore this, as
duplicate keys are to be expected when overriding settings in the
directory hierarchy.

8 years agoman/systemd-halt.service: fix typo
Dave Reisner [Thu, 4 Oct 2012 19:09:15 +0000 (15:09 -0400)]
man/systemd-halt.service: fix typo

Fix reference to "system-shutdown" as being a binary.

8 years agoupdate TODO
Lennart Poettering [Thu, 4 Oct 2012 18:10:02 +0000 (14:10 -0400)]
update TODO

8 years agodbus: expose Virtualization ID on PID1's bus interface
Lennart Poettering [Thu, 4 Oct 2012 18:08:17 +0000 (14:08 -0400)]
dbus: expose Virtualization ID on PID1's bus interface

This allows unprivileged clients to check for the used virtualization
even when lacking the privileges that some of the virtualization tests

8 years agoupdate TODO
Lennart Poettering [Thu, 4 Oct 2012 12:19:42 +0000 (08:19 -0400)]
update TODO

8 years agosystemctl: colorize number of units listed
Zbigniew Jędrzejewski-Szmek [Thu, 4 Oct 2012 11:12:23 +0000 (11:12 +0000)]
systemctl: colorize number of units listed

8 years agosystemctl: do not print header if no units listed
Zbigniew Jędrzejewski-Szmek [Thu, 4 Oct 2012 11:01:10 +0000 (11:01 +0000)]
systemctl: do not print header if no units listed

8 years agoman: handle-sleep-key is invalid, replace it with handle-suspend-key and handle-hiber...
Сковорода Никита Андреевич [Wed, 3 Oct 2012 01:48:19 +0000 (05:48 +0400)]
man: handle-sleep-key is invalid, replace it with handle-suspend-key and handle-hibernate-key in systemd-inhibit help and man.

8 years agobuild-sys: prepare 194 v194
Lennart Poettering [Wed, 3 Oct 2012 18:32:29 +0000 (14:32 -0400)]
build-sys: prepare 194

8 years agoman: document behaviour of ListenStream= with only a port number in regards to IPv4...
Lennart Poettering [Wed, 3 Oct 2012 18:18:55 +0000 (14:18 -0400)]
man: document behaviour of ListenStream= with only a port number in regards to IPv4/IPv6

8 years agodbus: add some more safety checks before accepting data from bus clients
Lennart Poettering [Wed, 3 Oct 2012 17:29:20 +0000 (13:29 -0400)]
dbus: add some more safety checks before accepting data from bus clients

8 years agojournal-send: simplification
Lennart Poettering [Wed, 3 Oct 2012 15:50:45 +0000 (11:50 -0400)]
journal-send: simplification

8 years agojournald: only accept fds from certain directories
Lennart Poettering [Wed, 3 Oct 2012 15:37:44 +0000 (11:37 -0400)]
journald: only accept fds from certain directories

8 years agoupdate TODO
Lennart Poettering [Wed, 3 Oct 2012 15:37:06 +0000 (11:37 -0400)]
update TODO

8 years agoselinux: properly free dbus error
Lennart Poettering [Wed, 3 Oct 2012 15:36:57 +0000 (11:36 -0400)]
selinux: properly free dbus error

8 years agobuild-sys: fix !HAVE_SELINUX case
Colin Walters [Tue, 2 Oct 2012 22:15:14 +0000 (18:15 -0400)]
build-sys: fix !HAVE_SELINUX case

8 years ago vconsole: default to the kernel compiled-in keymap
Tom Gundersen [Wed, 3 Oct 2012 13:57:40 +0000 (15:57 +0200)]
vconsole: default to the kernel compiled-in keymap

    No longer override the default kernel keymap if nothing is specified in

    The default should be to do nothing (i.e., use what is already in the
    kernel) unless the distro/admin has explicitly requested it.

8 years agovconsole: default to the kernel compiled-in font
Tom Gundersen [Tue, 25 Sep 2012 20:22:05 +0000 (22:22 +0200)]
vconsole: default to the kernel compiled-in font

No longer override the default kernel font if nothing is specified in

The default kernel font[0] provides ISO-8859-1 and box characters. Users
of Arabic, Cyrilic or Hebrew must set a different font manually as these
character sets were provided by the old default font [1], but are not
any longer.


 * it is counter-intuitive that an empty vconsole.conf file is different
   from adding FONT="";
 * the version of the default font shipped with Arch (which is the
   upstream one) behaves very badly during early boot[2] (which should
   admittedly be fixed in the font itself);
 * the kernel already supplies a default font, it seems reasonable to
   use that unless anything else is specified;
 * This also avoids a needless slow call to setfont; and
 * We don't want to work around problems in the kernel (in case the
   compiled-in font is not acceptable for whatever reason).

[0]: <>
[1]: <>
[2]: <>

8 years agoselinux: remove anything PID1-specific from selinux-access.[ch] so that we can reuse...
Lennart Poettering [Tue, 2 Oct 2012 21:56:54 +0000 (17:56 -0400)]
selinux: remove anything PID1-specific from selinux-access.[ch] so that we can reuse it in logind

8 years agoaudit: turn the audit fd into a static variable
Lennart Poettering [Tue, 2 Oct 2012 21:40:09 +0000 (17:40 -0400)]
audit: turn the audit fd into a static variable

As audit is pretty much just a special kind of logging we should treat
it similar, and manage the audit fd in a static variable.

This simplifies the audit fd sharing with the SELinux access checking
code quite a bit.

8 years agoselinux: rework selinux access check logic
Lennart Poettering [Tue, 2 Oct 2012 21:07:00 +0000 (17:07 -0400)]
selinux: rework selinux access check logic

a) Instead of parsing the bus messages inside of selinux-access.c
simply pass everything pre-parsed in the functions

b) implement the access checking with a macro that resolves to nothing
on non-selinux builds

c) split out the selinux checks into their own sources

d) this unifies the job creation code behind the D-Bus calls
Manager.StartUnit() and Unit.Start().

8 years agoTODO
Lennart Poettering [Tue, 2 Oct 2012 17:11:26 +0000 (13:11 -0400)]