chiark / gitweb /
elogind.git
8 years agogudev: docs - work around the broken gtk-doc mess
Kay Sievers [Wed, 8 Aug 2012 16:16:50 +0000 (18:16 +0200)]
gudev: docs - work around the broken gtk-doc mess

gtk-doc add CFLAGS/LDFLAGS multiple times to the gcc command line,
which breaks options that must be listed only once.

For now, clear CFLAGS/LDFLAGS for the intermediate documentation
binary.

8 years agosystemd: introduced new timeout types
Michal Sekletar [Tue, 7 Aug 2012 12:41:48 +0000 (14:41 +0200)]
systemd: introduced new timeout types

Makes possible to specify separate timeout for start and stop of
the service.

[ Improved the manpage. Coding style fix. -- michich ]

8 years agoudev: re-initialize builtins in the daemon process, not in the worker
Kay Sievers [Wed, 8 Aug 2012 14:27:11 +0000 (16:27 +0200)]
udev: re-initialize builtins in the daemon process, not in the worker

8 years agoudev: initialize rules dir timestamps when reading rules
Kay Sievers [Wed, 8 Aug 2012 12:45:16 +0000 (14:45 +0200)]
udev: initialize rules dir timestamps when reading rules

On Wed, Aug 8, 2012 at 11:48 AM, Michael Schroeder <mls@suse.de> wrote:
> if rules are installed in the first 3 seconds after the udev start,
> the stamps will all be zero, so the [first] call to check_rules_timestamp()
> will just copy the current mtime [and not cause a rules re-load].

8 years agoautogen.sh: disable _FORTIFY_SOURCE, we want -O0, which is incompatible
Kay Sievers [Wed, 8 Aug 2012 11:16:57 +0000 (13:16 +0200)]
autogen.sh: disable _FORTIFY_SOURCE, we want -O0, which is incompatible

8 years agobuild-sys: link internal selinux lib to systemd-remount-fs
Kay Sievers [Wed, 8 Aug 2012 10:28:45 +0000 (12:28 +0200)]
build-sys: link internal selinux lib to systemd-remount-fs

8 years agobuild-sys: pass param to stack protector
Lennart Poettering [Wed, 8 Aug 2012 10:26:53 +0000 (12:26 +0200)]
build-sys: pass param to stack protector

8 years agobuild-sys: drop obsolete gcc switch
Lennart Poettering [Wed, 8 Aug 2012 10:26:41 +0000 (12:26 +0200)]
build-sys: drop obsolete gcc switch

8 years agobuild-sys: typo fix
Lennart Poettering [Wed, 8 Aug 2012 10:26:27 +0000 (12:26 +0200)]
build-sys: typo fix

8 years agoTODO: misleading socket warning
Kay Sievers [Wed, 8 Aug 2012 10:09:33 +0000 (12:09 +0200)]
TODO: misleading socket warning

8 years agobuild-sys: enable a couple of security features
Lennart Poettering [Wed, 8 Aug 2012 10:03:34 +0000 (12:03 +0200)]
build-sys: enable a couple of security features

Most distributions enable these downstream anyway, but it probably makes
sense to enable them unconditionally upstream too.

8 years agounits: remove prefdm
Lennart Poettering [Tue, 7 Aug 2012 15:07:35 +0000 (17:07 +0200)]
units: remove prefdm

It's time to get rid of prefdm. Distributions which still want to use
this should maintain this downstream, but it's probably better to just
provide proper units for the various display managers, like Fedora is
doing this, for example:

https://fedoraproject.org/wiki/Features/DisplayManagerRework

8 years agoupdate TODO
Lennart Poettering [Mon, 6 Aug 2012 23:01:33 +0000 (01:01 +0200)]
update TODO

8 years agojournalctl: include corrupted files in output
Lennart Poettering [Mon, 6 Aug 2012 22:51:17 +0000 (00:51 +0200)]
journalctl: include corrupted files in output

If a journal file was rotated away because it was corrupted or dirty we
should still show its contents via "journalctl".

8 years agomount-setup: change system mount propagation to shared by default
Lennart Poettering [Mon, 6 Aug 2012 16:28:42 +0000 (18:28 +0200)]
mount-setup: change system mount propagation to shared by default

In order to make containers work nicely out of the box it is highly
desirable to have the mount propagation mode for the root fs is set as
"shared" by default so that containers receive system mounts by default.
(See mount --make-shared for more information).

As it is unlikely that the kernel will change the default to "shared"
for this, do this early at boot-up from PID 1. Setups which prefer the
default of "private" should undo this change via invoking "mount
--make-private /" or a similar command after boot.

In the long run /etc/fstab should take the propagation mode as a mount
option like any other, so that this may be used to change the default
mode. However, if fstab is not around or doesn't list / we still should
default to shared as propagation mode, hence this change now.

8 years agosysctl: apply configuration at once
Michal Sekletar [Fri, 3 Aug 2012 14:20:31 +0000 (16:20 +0200)]
sysctl: apply configuration at once

https://bugzilla.redhat.com/show_bug.cgi?id=767795

[ Simplified by iterating the config files in the backwards order -
  no need for hashmap_update(). Other minor cleanups. -- michich ]

8 years agomain: fix typo
Lennart Poettering [Mon, 6 Aug 2012 14:41:29 +0000 (16:41 +0200)]
main: fix typo

8 years agocontinue work with error messages, log_oom()
Shawn Landden [Sat, 4 Aug 2012 00:22:09 +0000 (17:22 -0700)]
continue work with error messages, log_oom()

Adds messages for formally silent errors: new "Failed on cmdline argument %s: %s".

Removes some specific error messages for -ENOMEM in mount-setup.c. A few specific
ones have been left in other binaries.

8 years agoupdate TODO
Lennart Poettering [Mon, 6 Aug 2012 14:35:44 +0000 (16:35 +0200)]
update TODO

8 years agoshutdown: allow to specify broadcast message when cancelling shutdown
Michal Sekletar [Mon, 30 Jul 2012 15:25:39 +0000 (17:25 +0200)]
shutdown: allow to specify broadcast message when cancelling shutdown

makes shutdown behaviour more compatible

8 years agoTODO: /etc/timezone
Kay Sievers [Mon, 6 Aug 2012 13:09:46 +0000 (15:09 +0200)]
TODO: /etc/timezone

8 years agoupdate TODO
Lennart Poettering [Fri, 3 Aug 2012 19:49:06 +0000 (21:49 +0200)]
update TODO

8 years agocryptsetup: add keyfile-size= support
Tom Gundersen [Fri, 3 Aug 2012 10:47:24 +0000 (12:47 +0200)]
cryptsetup: add keyfile-size= support

This is useful e.g. if the keyfile is a raw device, where only parts of it
should be read. It is typically used whenever the keyfile-offset= option is
specified.

Tested-by: Erik Westrup <erik.westrup@gmail.com>
8 years agobuild-sys: do not link systemctl against journal-internal and id128-internal
Simon Peeters [Fri, 3 Aug 2012 15:09:21 +0000 (17:09 +0200)]
build-sys: do not link systemctl against journal-internal and id128-internal

systemctl does not need the internal journal and id128 libraries to
function, so don't link against them.

8 years agologs-show: fix off-by-one error
Shawn Landen [Fri, 3 Aug 2012 02:28:28 +0000 (02:28 +0000)]
logs-show: fix off-by-one error

Ellipsize lines that are one character too long.

8 years agoman: document three new cgtop options
Zbigniew Jędrzejewski-Szmek [Fri, 3 Aug 2012 15:42:04 +0000 (17:42 +0200)]
man: document three new cgtop options

8 years agotest: allow deletion of temporary files from normal fs
Zbigniew Jędrzejewski-Szmek [Wed, 1 Aug 2012 15:48:29 +0000 (17:48 +0200)]
test: allow deletion of temporary files from normal fs

Not everybody has /tmp on tmpfs, and this was breaking 'make check'.

8 years agokeyboard-force-release: Fix previous commit
Martin Pitt [Fri, 3 Aug 2012 11:52:48 +0000 (13:52 +0200)]
keyboard-force-release: Fix previous commit

Was missing a * for the globbing.

8 years agokeyboard-force-release: Apply to all Fujitso AMILO models
Martin Pitt [Fri, 3 Aug 2012 11:50:46 +0000 (13:50 +0200)]
keyboard-force-release: Apply to all Fujitso AMILO models

https://launchpad.net/bugs/1011323 reports more AMILO models which need this
quirk; enough to assume that all of them need it, and applying it on working
models does not really hurt.

8 years agoudev: path_id - add comment about the unrteliable rebase logic
Kay Sievers [Fri, 3 Aug 2012 08:52:24 +0000 (10:52 +0200)]
udev: path_id - add comment about the unrteliable rebase logic

8 years agoupdate TODO
Lennart Poettering [Wed, 1 Aug 2012 17:53:01 +0000 (19:53 +0200)]
update TODO

8 years agojournal: add sd_journal_perror() to API
Lennart Poettering [Tue, 31 Jul 2012 14:09:01 +0000 (16:09 +0200)]
journal: add sd_journal_perror() to API

8 years agoupdate TODO
Lennart Poettering [Tue, 31 Jul 2012 12:35:42 +0000 (14:35 +0200)]
update TODO

8 years agojournald: fixed memory leak
Artur Zaprzala [Tue, 31 Jul 2012 12:14:51 +0000 (14:14 +0200)]
journald: fixed memory leak

8 years agolibudev-queue.c: simplified open_queue_file
Arvydas Sidorenko [Tue, 31 Jul 2012 09:38:08 +0000 (11:38 +0200)]
libudev-queue.c: simplified open_queue_file

There is no apparent justification for using util_strscpyl
on the filename since it's a plain hardcoded path.
Older versions used:
util_strscpyl(filename, sizeof(filename), SOME_DIR, "/queue.bin", NULL);
and when changed nobody bothered to simplify it.

8 years agodoc: clarification in system.unit man page
Zbigniew Jędrzejewski-Szmek [Wed, 13 Jul 2011 12:13:31 +0000 (14:13 +0200)]
doc: clarification in system.unit man page

8 years agoautomount: print mount point in debug message
Zbigniew Jędrzejewski-Szmek [Wed, 13 Jul 2011 12:06:42 +0000 (14:06 +0200)]
automount: print mount point in debug message

Old: systemd[1]: Got direct mount request for ffff88003bb10c00, triggered by 14476 (fuser)
New: systemd[1]: Got direct mount request on /dev/mqueue, triggered by 2177 (ls)

8 years agoAdd a few more dependencies to README
Zbigniew Jędrzejewski-Szmek [Tue, 12 Jul 2011 11:57:48 +0000 (13:57 +0200)]
Add a few more dependencies to README

8 years agobuild-sys: fix path to src/gudev/gudevenumtypes.h
Zbigniew Jędrzejewski-Szmek [Mon, 30 Jul 2012 22:06:12 +0000 (00:06 +0200)]
build-sys: fix path to src/gudev/gudevenumtypes.h

This file is generated, so it should be referred to as
$(top_builddir)/src/gudev/gudevenumtypes.h. It could only appear in
$(top_srcdir) as a result of previous build in $(top_srcdir). Better
to just let automake add the prefix for us, so there's no need to
spell it out.

Remove the prefix from other source files too, $(top_srcdir) is the
default anyway.

8 years agobuild-sys: always create the output directory first
Zbigniew Jędrzejewski-Szmek [Mon, 30 Jul 2012 21:57:50 +0000 (23:57 +0200)]
build-sys: always create the output directory first

$(MKDIR_P) is added where missing, and rules are standardized on one
form of $(MKDIR_P), to make it easier to spot when it is missing.
Single line $(MKDIR)&&command form is broken into two line form.

https://bugs.freedesktop.org/show_bug.cgi?id=49459

For compilation in a separate build directory to work, when a file is
generated, the rule must include an explicit mkdir first, unless the
file is created at the top level. Even when building in a separate
build-dir, automake would normally create all directories as a side
result of creating the dependencies files. Therefore the bug was only
visible with -C (turning off dependency generation).

8 years agoudev: set log.c's log level from udev's kernel command line option
Kay Sievers [Mon, 30 Jul 2012 20:34:07 +0000 (22:34 +0200)]
udev: set log.c's log level from udev's kernel command line option

8 years ago99-systemd.rules.in: ignore nbd in the "add" uevent
Harald Hoyer [Mon, 30 Jul 2012 19:21:37 +0000 (21:21 +0200)]
99-systemd.rules.in: ignore nbd in the "add" uevent

8 years agorules/99-systemd.rules.in: ENV{SYSTEMD_READY}="0" for incomplete md
Harald Hoyer [Mon, 30 Jul 2012 18:27:52 +0000 (20:27 +0200)]
rules/99-systemd.rules.in: ENV{SYSTEMD_READY}="0" for incomplete md

8 years agoupdate TODO
Kay Sievers [Mon, 30 Jul 2012 18:03:00 +0000 (20:03 +0200)]
update TODO

8 years agoudev: add missing newline when writing to /dev/kmsg
Kay Sievers [Mon, 30 Jul 2012 15:07:49 +0000 (17:07 +0200)]
udev: add missing newline when writing to /dev/kmsg

8 years agologind: add new loginctl lock-sessions command
Lennart Poettering [Sun, 29 Jul 2012 15:08:47 +0000 (17:08 +0200)]
logind: add new loginctl lock-sessions command

8 years agologind: add LockSessions() call to lock all local sessions at once
Lennart Poettering [Sun, 29 Jul 2012 14:17:06 +0000 (16:17 +0200)]
logind: add LockSessions() call to lock all local sessions at once

8 years agoRevert "man: ship systemd-udevd as the real manpage"
Lennart Poettering [Sun, 29 Jul 2012 09:58:41 +0000 (11:58 +0200)]
Revert "man: ship systemd-udevd as the real manpage"

We want to keep things uniform, and hence treat udevd's man page like
any other in the repo. What matters is how users primarily interface
with a service, and that is not the binary path in /usr/lib/systemd but
the service name.

This reverts commit 6c1f3ba54ac1f42598c543b86daaf294b29294b1.

8 years agoman: grammar fixes for systemd-udev(8)
Zbigniew Jędrzejewski-Szmek [Sat, 28 Jul 2012 21:10:15 +0000 (23:10 +0200)]
man: grammar fixes for systemd-udev(8)

8 years agoselinux: avoid inclusion of null_log without selinux
Dave Reisner [Sat, 28 Jul 2012 00:02:00 +0000 (20:02 -0400)]
selinux: avoid inclusion of null_log without selinux

Avoids a compile time warning:

  warning: 'null_log' defined but not used [-Wunused-function]

8 years agoman: ship systemd-udevd as the real manpage
Dave Reisner [Sat, 28 Jul 2012 00:46:40 +0000 (20:46 -0400)]
man: ship systemd-udevd as the real manpage

Instead of making systemd-udevd a so-link to systemd-udevd.service,
ship the real page as systemd-udevd to integrate better with distros
where udevd might be run standalone.

8 years agobuild-sys: always populate EXTRA_DIST
Bryan Kadzban [Sat, 28 Jul 2012 02:43:38 +0000 (19:43 -0700)]
build-sys: always populate EXTRA_DIST

"make dist" can build a different tarball depending on the flags passed
to ./configure and the (optional) dependencies found on the system.
Move all append-to-EXTRA_DIST operations out of automake conditionals to
fix this.

Introduce a polkitpolicy_files so that the policy files built still
correctly depend on the automake conditionals, but the .in files that
get distributed do not.

8 years agosystemctl: append .service to unit names lacking suffix
Lennart Poettering [Sat, 28 Jul 2012 11:20:35 +0000 (13:20 +0200)]
systemctl: append .service to unit names lacking suffix

https://bugs.freedesktop.org/show_bug.cgi?id=39386

8 years agoupdate TODO
Lennart Poettering [Fri, 27 Jul 2012 10:21:39 +0000 (12:21 +0200)]
update TODO

8 years agojournalctl: add --priority= switch for filtering by priority
Lennart Poettering [Fri, 27 Jul 2012 08:31:33 +0000 (10:31 +0200)]
journalctl: add --priority= switch for filtering by priority

8 years agocgtop: add --version option
Zbigniew Jędrzejewski-Szmek [Thu, 26 Jul 2012 21:09:02 +0000 (23:09 +0200)]
cgtop: add --version option

8 years agocgtop: use full terminal width
Zbigniew Jędrzejewski-Szmek [Thu, 26 Jul 2012 18:23:28 +0000 (20:23 +0200)]
cgtop: use full terminal width

8 years agoAdd a 'b' option to cgtop, equivalent to the same option in top
David Strauss [Wed, 25 Jul 2012 23:33:07 +0000 (16:33 -0700)]
Add a 'b' option to cgtop, equivalent to the same option in top

[zj: use static]

8 years agoAdd an 'n' option to cgtop (equivalent to top)
David Strauss [Wed, 25 Jul 2012 23:33:06 +0000 (16:33 -0700)]
Add an 'n' option to cgtop (equivalent to top)

8 years agojournald: log driver messages at LOG_INFO
Lennart Poettering [Thu, 26 Jul 2012 14:51:58 +0000 (16:51 +0200)]
journald: log driver messages at LOG_INFO

8 years agojournalctl: hightlight log lines by priority
Lennart Poettering [Thu, 26 Jul 2012 14:50:35 +0000 (16:50 +0200)]
journalctl: hightlight log lines by priority

warn/notice = bright white
< error = red

8 years agologs-show: fix OOM path
Lennart Poettering [Thu, 26 Jul 2012 14:05:26 +0000 (16:05 +0200)]
logs-show: fix OOM path

8 years agobuild-sys: use correct cpp
Peeters Simon [Thu, 26 Jul 2012 13:41:02 +0000 (15:41 +0200)]
build-sys: use correct cpp

this solves issues where $(CPP) has differend include paths than 'cpp'

8 years agoupdate README
Lennart Poettering [Thu, 26 Jul 2012 13:56:43 +0000 (15:56 +0200)]
update README

8 years agologin: lose policy a bit: allow user suspend unauthenticated when multiple sessions...
Lennart Poettering [Thu, 26 Jul 2012 13:56:36 +0000 (15:56 +0200)]
login: lose policy a bit: allow user suspend unauthenticated when multiple sessions exist

8 years agolog: out-of-line __log_oom()
Michal Schmidt [Thu, 26 Jul 2012 12:23:49 +0000 (14:23 +0200)]
log: out-of-line __log_oom()

The callers' code gets smaller.

8 years agolog: log_oom() must be a macro
Michal Schmidt [Thu, 26 Jul 2012 12:07:27 +0000 (14:07 +0200)]
log: log_oom() must be a macro

8 years agoudev: cdrom_id, accelerometer - enable debug output for --debug
Kay Sievers [Thu, 26 Jul 2012 10:03:25 +0000 (12:03 +0200)]
udev: cdrom_id, accelerometer - enable debug output for --debug

8 years agolog.h: new log_oom() -> int -ENOMEM, use it
Shawn Landden [Wed, 25 Jul 2012 21:55:59 +0000 (14:55 -0700)]
log.h: new log_oom() -> int -ENOMEM, use it

also a number of minor fixups and bug fixes: spelling, oom errors
that didn't print errors, not properly forwarding error codes,
few more consistency issues, et cetera

8 years agoupdate TODO
Kay Sievers [Thu, 26 Jul 2012 08:20:47 +0000 (10:20 +0200)]
update TODO

8 years agounit-name: style fix in unit_name_is_template()
Michal Schmidt [Thu, 26 Jul 2012 08:13:26 +0000 (10:13 +0200)]
unit-name: style fix in unit_name_is_template()

to make it look like the newly added unit_name_is_instance()

8 years agosystemd: enable/disable instances of template
Michal Sekletar [Fri, 13 Jul 2012 13:59:26 +0000 (15:59 +0200)]
systemd: enable/disable instances of template

https://bugzilla.redhat.com/show_bug.cgi?id=752774

8 years agounit: add missing deps in unit_dependency_table
Michal Schmidt [Thu, 26 Jul 2012 07:46:40 +0000 (09:46 +0200)]
unit: add missing deps in unit_dependency_table

also reorder the table according to the enum order.

8 years agounit: make the table of inverse deps symmetric
Michal Schmidt [Thu, 26 Jul 2012 07:42:26 +0000 (09:42 +0200)]
unit: make the table of inverse deps symmetric

8 years agodbus-unit: expose PartOf/ConsistsOf properties
Michal Schmidt [Thu, 26 Jul 2012 07:38:39 +0000 (09:38 +0200)]
dbus-unit: expose PartOf/ConsistsOf properties

8 years agoman: rewrite the description of PartOf
Michal Schmidt [Thu, 26 Jul 2012 07:34:27 +0000 (09:34 +0200)]
man: rewrite the description of PartOf

8 years agosystemd: added new dependency PartOf
Michal Sekletar [Fri, 20 Jul 2012 13:55:01 +0000 (15:55 +0200)]
systemd: added new dependency PartOf

This should address TODO item "new dependency type to "group" services
in a target". Semantic of new dependency is as follows. Once configured
it creates dependency which will cause that all dependent units get
stopped if unit they all depend on is stopped or restarted.  Usual use
case would be configuring PartOf=some.target in template unit file
and WantedBy=some.target in [Install] section and enabling desired
number of instances. In this case starting one instance won't pull in
target but stopping or starting target(in case of WantedBy is properly
configured) will cause stop/start of all instances.

8 years agobuild-sys: create ntp-units.d dropin directories on install
Michal Schmidt [Thu, 26 Jul 2012 05:47:12 +0000 (07:47 +0200)]
build-sys: create ntp-units.d dropin directories on install

8 years agobuild-sys: really make gtk-doc optional & some portability fixes
Mantas Mikulėnas [Wed, 25 Jul 2012 19:39:17 +0000 (22:39 +0300)]
build-sys: really make gtk-doc optional & some portability fixes

The failing check caused autogen.sh to die mysteriously due to `set -e`
and the unquoted `test -z $VAR`.

Also, some syntax fixes to make it work with /bin/sh, which most other
autogen.sh's use.

[zj: implemented simplification suggested by Peters Simon
 and fixed the case where gtkdocize was available but is not anymore.]

8 years agoupdate TODO
Lennart Poettering [Wed, 25 Jul 2012 16:13:32 +0000 (18:13 +0200)]
update TODO

8 years agoupdate TODO
Kay Sievers [Wed, 25 Jul 2012 09:40:14 +0000 (11:40 +0200)]
update TODO

8 years agomake-man-index.py: compatibility with Python 3
Mantas Mikulėnas [Tue, 24 Jul 2012 23:59:28 +0000 (02:59 +0300)]
make-man-index.py: compatibility with Python 3

Despite its name, ElementTree.tostring() returns already-encoded
bytes() instead of a string in Python 3; sys.stdout doesn't like that.

8 years agouse "Out of memory." consistantly (or with "\n")
Shawn Landden [Wed, 25 Jul 2012 04:12:43 +0000 (21:12 -0700)]
use "Out of memory." consistantly (or with "\n")

glibc/glib both use "out of memory" consistantly so maybe we should
consider that instead of this.

Eliminates one string out of a number of binaries. Also fixes extra newline
in udev/scsi_id

8 years agounits: add the modules-load cmdline parameters to the unit conditions
Michal Schmidt [Wed, 25 Jul 2012 06:48:50 +0000 (08:48 +0200)]
units: add the modules-load cmdline parameters to the unit conditions

For 'modules-load=' and 'rd.modules-load=' to be effective,
systemd-modules-load.service must be started. It is currently
conditional on the existence of config files. Add the presence of the
cmdline parameters to the triggering conditions.

8 years agomodules-load: fix kernel cmdline parsing
Michal Schmidt [Tue, 24 Jul 2012 23:45:16 +0000 (01:45 +0200)]
modules-load: fix kernel cmdline parsing

8 years agomain: set PR_SET_CHILD_REAPER for MANAGER_USER
Auke Kok [Tue, 24 Jul 2012 22:13:11 +0000 (00:13 +0200)]
main: set PR_SET_CHILD_REAPER for MANAGER_USER

Become the reaper for all children part of the user session. Tested
with several forking services.

8 years agoconf-files: continue searching if one dir fails
Michal Schmidt [Tue, 24 Jul 2012 21:18:25 +0000 (23:18 +0200)]
conf-files: continue searching if one dir fails

A problem with systemd-tmpfiles has been observed where the service
failed just because one of the configuration directories could not be
read due to SELinux policy.

Complain about the failure, but try to go on.

https://bugzilla.redhat.com/show_bug.cgi?id=839736

8 years agouaccess: add ID_SECURITY_TOKEN device class for USB authentication keys
Kay Sievers [Tue, 24 Jul 2012 18:13:46 +0000 (20:13 +0200)]
uaccess: add ID_SECURITY_TOKEN device class for USB authentication keys

8 years agorules: avoid mounting raid devices too early
Michal Schmidt [Tue, 24 Jul 2012 07:47:33 +0000 (09:47 +0200)]
rules: avoid mounting raid devices too early

/dev/md0 appears as soon as the first component of the raid array is
added by incremental assembly rules. This is too early for systemd to
attempt to mount the device. The device should be considered plugged
after the raid array becomes active.

https://bugzilla.redhat.com/show_bug.cgi?id=767561

8 years agosystemctl: use color specification understood by dot
Zbigniew Jędrzejewski-Szmek [Mon, 23 Jul 2012 16:26:31 +0000 (18:26 +0200)]
systemctl: use color specification understood by dot

grey66 is aproximately equal to dark grey. Not understanding dark grey is really
a bug in dot, but trivial to work around.

Closes https://bugs.freedesktop.org/show_bug.cgi?id=45706

8 years agojournalctl: fix ellipsization with PAGER=cat
Zbigniew Jedrzejewski-Szmek [Fri, 20 Jul 2012 07:06:26 +0000 (09:06 +0200)]
journalctl: fix ellipsization with PAGER=cat

There are other reasons for not opening the pager then the --no-pager
or --follow options (described below). If the pager is not used,
messages must be ellipsized.

On Fri, Jul 20, 2012 at 05:42:44AM +0000, Shawn Landen wrote:
> "Pager to use when --no-pager is not given; overrides $PAGER.
> Setting this to an empty string or the value cat is equivalent to passing --no-pager."

8 years agokeymap: Broaden ThinkPad X2.. tablet match to also apply to X230
Terence Honles [Mon, 23 Jul 2012 04:32:46 +0000 (06:32 +0200)]
keymap: Broaden ThinkPad X2.. tablet match to also apply to X230

https://bugs.freedesktop.org/show_bug.cgi?id=52371

8 years agobuild-sys: man - add missing dependency
Kay Sievers [Sun, 22 Jul 2012 23:58:24 +0000 (01:58 +0200)]
build-sys: man - add missing dependency

8 years agosd-daemon: remove unnecessary memset calls
Jim Meyering [Thu, 16 Jun 2011 06:51:14 +0000 (08:51 +0200)]
sd-daemon: remove unnecessary memset calls

There is no point in clearing the bits of a "struct stat" when the very
next statement just calls stat or fstat to fill in that same memory.

[zj: two more places]

8 years agologind: fix operation precedence mix-up
Zbigniew Jędrzejewski-Szmek [Sun, 22 Jul 2012 13:24:25 +0000 (15:24 +0200)]
logind: fix operation precedence mix-up

Since + has higher precedence than ?:, and u+b is unlikely to be exactly zero,
the timestamp would usually be set to IDLE_THRESHOLD_USEC. Fix it by
returning either 'last activity', or 'last activity+IDLE_THRESHOLD_USEC'.

8 years agotest: correct enum misassignment
Zbigniew Jędrzejewski-Szmek [Sun, 22 Jul 2012 12:48:52 +0000 (14:48 +0200)]
test: correct enum misassignment

CC     src/test/test_job_type-test-job-type.o
../src/test/test-job-type.c:40:25: warning: implicit conversion from enumeration type
      'enum UnitType' to different enumeration type 'ServiceType' (aka 'enum ServiceType')
      [-Wconversion]
                .type = UNIT_SERVICE
                        ^~~~~~~~~~~~

8 years agocollect, keymap, systemctl: use _noreturn_
Zbigniew Jędrzejewski-Szmek [Sun, 22 Jul 2012 12:42:09 +0000 (14:42 +0200)]
collect, keymap, systemctl: use _noreturn_

8 years agoman: typo
Shawn Landden [Fri, 22 Jun 2012 11:45:29 +0000 (04:45 -0700)]
man: typo

8 years agobuild-sys: distribute a few more man pages
Zbigniew Jędrzejewski-Szmek [Sun, 22 Jul 2012 21:32:21 +0000 (23:32 +0200)]
build-sys: distribute a few more man pages

8 years agobuild-sys: hide error message from grep
Zbigniew Jędrzejewski-Szmek [Sun, 22 Jul 2012 12:10:51 +0000 (14:10 +0200)]
build-sys: hide error message from grep